HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Mattei, B. Articles by Schmied, A. Search for Related Content PubMed PubMed Citation Articles by Mattei, B. Articles by Schmied, A.

Journal of Physiology (2002), 538.3, pp. 849-865
© Copyright 2002 The Physiological Society
DOI: 10.1113/jphysiol.2001.012701

Delayed and prolonged effects of a near threshold EPSP on the firing time of human -motoneurones

Benjamin Mattei and Annie Schmied

	ABSTRACT

Top Abstract Introduction Methods Results Discussion References

In order to investigate the effects of near-threshold excitatory inputs on the precise timing of the action potentials during the tonic discharge of human motoneurones, the activity of single motor units was recorded in the extensor carpi radialis muscles while tendon taps (indentation, 0.1 mm; duration, 1 ms) were being delivered irregularly at a mean rate of 0.8 s^-1. New methods of analysis, such as the phase response function, were used to study the relative changes in the interspike interval (ISI₁) during which the stimulus was being delivered and in the three subsequent intervals (ISI₂, ISI₃, ISI₄) as a percentage of the pre-stimulus interspike interval (ISI₀). The consistency of the effects of the actual stimulus as regards the spontaneous variability was assessed by comparing the data with those obtained with virtual stimulation. When the stimulus occurred at the end of ISI₁, and triggered a spike, ISI₁ and ISI₃ were generally shortened, whereas ISI₂ was lengthened, probably due to the negative correlation induced by the summation of the after-hyperpolarisations (AHPs). When the stimulus occurred in the middle of ISI₁ without triggering a spike, ISI₁, ISI₂ and more rarely ISI₃ were shortened. Lastly, when the stimulus occurred during the AHP scoop in ISI₁, ISI₂ was shortened although ISI₁ remained unchanged. ISI₄ was not consistently affected in any of these cases. The present results show that the tendon tap-induced inputs (probably from muscle spindle primary endings) mediated delayed and prolonged shortening effects of the ISIs on most of the -motoneurones tested (n = 16). These effects undetected in classic peri-stimulus histogram analysis may involve long-lasting conductance changes although the contribution of polysynaptic pathways cannot be excluded. The changes in ISI were quite moderate (< 15 % of ISI) but highly consistent. Their functional involvement in the synchronisation or desynchronisation processes and/or the mechanisms of optimisation of muscle contraction still remains to be explored.

(Received 8 May 2001; accepted after revision 17 October 2001)
Corresponding author B. Mattei: CNRS-DPM, 31 chemin Joseph Aiguier, 13402 Marseille CEDEX 20, France. Email: jacopasto7{at}yahoo.com

	INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

The temporal structure of a neuronal discharge depends on the electrophysiological properties of that neurone, which are mainly determined by the interplay between the various membrane conductances and the synaptic inputs. The present study focuses on the effects of transient near-threshold excitatory synaptic inputs on the timing of the firing of human -motoneurones (MNs).

A tonically discharging MN can be compared to an oscillator. In fact, between two action potentials, the membrane potential takes approximately a saw-tooth trajectory (Schwindt & Calvin, 1972). After being hyperpolarised just after the occurrence of the action potential, it increases gradually until it crosses the firing threshold. Moreover, when a neurone is subjected to a steady synaptic input, it will discharge at a fairly steady firing rate as shown clearly by the frequency-intensity curves (Granit et al. 1963) within a range of variability resulting mainly from the synaptic noise (Calvin & Stevens, 1968). The important point here is that in the case of a neurone receiving a fairly constant input, which thus discharges tonically at a fairly constant frequency, one can predict how long an interspike interval (ISI) is likely to last. This expected ISI can be used as a reference to calculate the effects of a transient input.

It has been assumed until recently that an excitatory post-synaptic potential (EPSP) has one of two effects on a motoneurone spike train; an EPSP will either trigger a spike or leave the spike train unchanged (Fetz & Gustafsson, 1983). According to the experimental results on which the ramp model was based, the spike-triggering probability of an EPSP increases as the membrane potential rises towards the firing threshold, i.e. at the end of the expected ISI. Therefore, during the first part of the ISI, when the membrane potential is highly hyperpolarised, the EPSP is generally thought to have no effect.

These assumptions have been tested on cat neocortical slices (Reyes & Fetz, 1993a,b). The neurones, from which the normal synaptic drive is lacking in this preparation, were made to discharge steadily by injecting current. Pseudo-EPSPs produced by injecting transient excitatory current and EPSPs induced by stimulating the white matter have been tested with similar results. Two kinds of effects were observed on the timing of the spike in response to the transient input. When the 'EPSP' produced by the transient input occurred in the late part of the expected ISI, the spike triggered was time-locked to the stimulus, and the ISI was shortened accordingly. When the 'EPSP' occurred in the earlier part of the expected ISI, when no effect was expected to occur, the ISI was also found to be shortened, but the spike was not triggered directly, and it was therefore not time-locked to the stimulus. This effect is therefore subliminal in relation to the firing threshold. The fact that the ISI was shortened without any spike being triggered directly suggested the crossing of a lower threshold possibly associated with the activation of a conductance accelerating the spike-generating processes (Reyes & Fetz, 1993a,b). This was the first evidence that an EPSP might advance the next firing either with a precise and near-synchronous coupling or with a delayed and variable coupling. The shortening effects produced either by direct spike triggering or by the advanced spiking unlocked to the stimulus are illustrated in Fig. 1. The theoretical membrane potential trajectory is shown with an EPSP crossing the firing threshold (Fig. 1A) or occurring too early to cross the threshold but able to induce a change in the membrane properties which advances the next firing (Fig. 1B). The resulting changes in the duration of the ISI depending on the delay between the EPSP and the previous spike are plotted in the case of the spike-triggering effect alone (Fig. 1B) and in the case of the combined spike triggering and advanced spiking effects (Fig. 1D).

		View larger version [in this window] [in a new window]
Figure 1 . Possible effects of an EPSP on the firing time A and C, the theoretical membrane potential trajectory is shown with an EPSP crossing the firing threshold (A) or occurring too early to cross the threshold but able to induce a change in the membrane properties which advances the next firing (C). B and D, the resulting changes in the ISI duration depending on the delay between the EPSP and the previous spike are plotted in the cases of the spike-triggering (B) and both spike-triggering and advanced spiking effects (D).

A similar duality in the immediate or delayed effects of an EPSP on the firing time has been recently reported in the case of quiescent pyramidal neurones in hippocampal slices (Fricker & Miles, 2000).

In the present study, we investigated the possibility that similar effects might occur when human MNs are discharging steadily under voluntary control. Here the transient excitatory input was provided by applying a very light tendon tap known to activate the primary endings of the muscle spindles (Ia fibres) quite selectively, and thus to excite the corresponding MNs monosynaptically. We questioned whether these composite EPSPs might shorten the ISIs without triggering directly any spikes. In addition to the immediate or delayed effect produced by a transient input on the subsequent spike, the possibility that this effect might be prolonged, involving the subsequent ISIs, was examined, in keeping with the possibility that persistent changes may occur in the electrophysiological properties of the MN membrane (Schwindt & Crill, 1980; Hounsgaard et al. 1984; Bennett et al. 1998). Part of this study has been previously published in abstract form (Mattei et al. 2000).

	METHODS

Top Abstract Introduction Methods Results Discussion References

Experiments were performed on two healthy right-handed human subjects, one 28-year-old male and one 34-year-old female, with the approval of the Ethics Committee of the local Medical University (CCPPRB-Marseilles I, approval no. 92/74). The subjects gave their written informed consent to the experimental procedure as required by the Helsinki Declaration (1964).

Experimental device

The subjects were seated on an adjustable armchair with their right forearm placed in a cushioned groove. The distal end of the forearm was immobilised in a U-shaped device leaving the wrist joint free and maintaining the hand in a semi-prone position, flexed at an angle of 10 deg, with the back of the hand in permanent contact with an isometric force transducer. The subjects were asked to selectively contract their wrist extensor muscles by pushing on the force transducer with the back of their hand while keeping their finger muscles relaxed. They had to adjust their muscle contractions so that the motor units kept firing tonically as steadily as possible, using auditory and visual feedback information about the motor unit activity. The recording sequences lasted 300-800 s. Recording sequences were separated by pauses lasting at least 5 min.

Single unit recording

The net force produced by the wrist extension calibrated in Newtons was recorded in the form of direct (DC) and filtered signals (AC, 0.1 Hz to 1 kHz). The overall electromyographic (EMG) activity of the extensor carpi radialis longus and brevis muscles (ECR) was recorded by means of a pair of non-polarisable single-use surface electrodes (16 mm², Ag-AgCl) placed 2 cm apart.

The action potentials generated by single motor units (MUs) were recorded using a single-use metal microelectrode (impedance 12 M tested at 1000 Hz, Frederick Haer and Co, Bowdoinham, ME, USA) previously sterilised in formaldehyde vapour. The microelectrode was inserted transcutaneously into the muscle and then moved in minute steps until a stable recording of the activity of a clearly identifiable single motor unit was obtained. The subject was connected to the ground via an electrode (6 cm²) placed on the upper arm close to the elbow. The motor unit spike trains were amplified, filtered (300-3000 Hz) and discriminated with dual-window discriminators (BAK Electronics, Germantown, MD, USA). The impulses triggered by the spikes of each motor unit were fed into a loudspeaker providing auditory feedback and were concurrently used for preliminary on-line computer analyses using the software Spike2 (CED, Cambridge, UK).

The surface electrodes and the microelectrodes were connected to amplifiers via probes with an isolated ground to protect the subjects (current leakage was less than 3 µA).

Stimulation

While the motor unit recorded from was discharging steadily, gentle taps were applied to the distal tendons of the extensor carpi radialis muscles by means of an electromagnetic hammer (LDS 201 vibration generator, LDS, Royston, UK) driven by a pulse generator (S88 stimulator, Grass Instruments, Quincy, MA, USA), which delivered constant voltage pulses lasting 1 ms, via a power amplifier. The hammer was positioned over the distal part of the tendons, between the proximal end of the second and third metacarpal bones and the distal condyle of the radius. When the V-shaped extremity of the hammer (width, 10 mm; contact surface, 30 mm²), was not positioned exactly on the tendons of the ECR, the stimulation was ineffective, therefore the stimulation applied was highly selective.

The taps had an average amplitude of 0.1 mm (range, 0.07- 0.14 mm) and were delivered randomly; the time lag between two stimuli obeyed a uniform probability distribution within 0.8-1.6 s. Virtual stimulation pulses were generated on-line by the computer, 0.5 s after each actual stimulation pulse with the same probability distribution. This allowed us to compare the changes in ISI duration caused by the actual stimulation with any spontaneous changes occurring concurrently with the virtual stimulation.

An average of 451 stimuli of each type (actual and virtual) was delivered per recording (range, 346-645). The DC and AC force signals, the surface EMG signal, the microelectrode recording signal (sampling rates of 1 kHz, 3 kHz and 20 kHz, respectively) and the train of actual and virtual stimuli (transistor transistor logic (TTL) pulses) were recorded and filed, using the software Spike2 as an analog-to-digital converter, for off-line analysis.

Data analysis

The recorded muscle action potentials were discriminated off-line using the waveform recognition tool that is part of the Spike2 program. This operation was carried out with the greatest care, since the smallest mistake would have invalidated the subsequent analyses. In order to double-check that no muscular action potentials from other motor units were recorded, we systematically constructed the autocorrelograms of the motor units involved. The spike discrimination was completed on the basis of the instantaneous frequency of the motor unit discharge throughout the recording period. This made it possible to detect and include a posteriori any spikes which might have been missed in the waveform recognition process, therefore generating ISIs which were twice as long as the mean ISI value.

The first step of the analyses was to construct a standard post-stimulus time histogram (PSTH) (i.e. including all spikes before and after the stimulus) and the corresponding cumulative sum (CUSUM) analysis (Ellaway, 1978), which cumulates and thus enhances the changes in the PSTH spike counts with respect to a pre-stimulus baseline (Fig. 2A). This allowed detection of the sharp increase in the MN firing probability occurring with a monosynaptic-like latency above the baseline mean and measurement of its duration and amplitude (peak area).

		View larger version [in this window] [in a new window]
Figure 2. Post-stimulus time histograms (PSTHs) A, standard PSTH in the case of actual stimulation only, with cumulative sum analysis (CUSUM) superimposed in MU141. B and C, reduced PSTH in the cases of actual (B) and virtual (C) stimulation in the same motor unit (MU141). The time distribution of the first spike following the stimulus was plotted (in 0.25 ms bins). In comparison with B, the peak present in A showed the increase in the motoneurone spike counts consistent with a monosynaptic excitation. The latency of the peak was used to estimate the peripheral conduction time .

The discriminated spike trains yielded the times at which the muscular action potentials were generated, and the trains of stimulus pulses (TTL) yielded those at which the tendon taps were delivered. In order to evaluate at the spinal level the effects of the EPSPs generated by applying the tendon taps on the timing of the action potentials generated by the motoneurone, it was necessary to take the peripheral delays into account. The time between the onset of the tendon tap and the first detected change in motoneurone firing probability includes peripheral delays composed of the time taken for the afferent volley to reach the spinal cord (₁) and for the motoneurone action potential to reach the muscle (₂). To estimate the total peripheral delay = ₁ + ₂, we constructed the distribution of the first spike following the stimulus in the form of a 'reduced' PSTH for both types of stimulation (actual, Fig. 2B; virtual, Fig. 2C). This reduced PSTH corresponds to what has been called a cross-interval histogram in its more general form by Moore et al. (1966). Comparisons between the two spike distributions of the reduced PSTHs (Fig. 2B and C) with a ² test at a significance threshold of 0.01 made it possible to further assess the peak latency, from which the delay could be estimated. The stimulus trains were then shifted by adding this delay.

The two previous and the four subsequent spikes with respect to each stimulus (actual or virtual) were selected, as shown in Fig. 3. We then measured, the duration of the ISI prior to the stimulus (ISI₀), the interval during which the stimulus occurred (ISI₁) and the three subsequent intervals (ISI₂, ISI₃, ISI₄). The shortening of the ISIs was computed as a percentage of the reference interval, ISI₀. This approach was adopted because the MN discharges tonically with a quasi-constant frequency, and the shortening of the ISIs can be measured and compared with the expected ISI (i.e. that which would have been given by the mean of the ISI₀ values). However, some slow modulation of the discharge frequency generally occurs due to variation of the effective synaptic current. It is therefore advisable with each type of stimulus to take the previous ISI₀ as an estimate of the expected ISI.

		View larger version [in this window] [in a new window]
Figure 3. Notations and definitions Two spikes before and four spikes after each stimulus were selected. The corresponding ISIs and the changes occurring in response to the stimulus were defined as shown in the equations. The stimulus class was defined as a normalised index to the time elapsing between the stimulus and the previous spike ( 1).

We defined 10 classes of stimulus, depending on the moment at which they occurred during ISI₁. A stimulus was taken to belong to class i if it occurred between 10 (i - 1) % of ISI₀ and 10 i % of ISI₀ during ISI₁. Note that we did not include stimuli occurring after a duration longer than ISI₀ in ISI₁ (class 10 upwards). However, this selection criterion did not lead to rejecting more than 10 % of all the stimuli. Most of the analyses performed here were based on this definition.

A spike was said to be 'triggered' by the stimulus if it occurred with a latency that was within the limits of the monosynaptic response peak in the PSTH, and to be 'untriggered' otherwise. Based on this definition, we therefore used the term 'triggered' for such spikes even in the case of virtual stimulation, since it was useful for making comparisons between the effects of actual and virtual stimulation.

To study the triggering effect of the EPSPs, we computed the spike-triggering frequency trajectory, which is the frequency of the spikes triggered per class of stimulus. A frequency of 1 for stimulus class i, means that the EPSPs occurring in class i always triggered a spike.

To obtain further insights into the overall effects of an EPSP on ISI₁ and ISI₂, we calculated the experimental phase response functions for each motor unit tested following the procedure used by Reyes & Fetz (1993a). This approach was first introduced by Perkel et al. (1964) in terms of 'delay-function' in order to study the effect of synaptic input on the firing time of pacemaker neurones. In the present study, the phase response function gives the shortening of the ISI_i tested as a percentage of the ISI₀ per stimulus class, defined as follows: 100(ISI₀ - ISI_i)/ISI₀. Although this kind of analysis can bring subtle effects to light, it showed a bias when the stimulus occurred in the latest classes (9 and 10) with respect to ISI₀. In these latest classes, by definition, if the stimulus had no effect, ISI₁ could only be equal to or longer than ISI₀. This possibly explains the slight lengthening consistently observed in classes 9 and 10 of the response function computed in the virtual stimulation conditions. However in the case of actual stimulation, ISI₁ was generally shortened and thus compensated for the bias. This bias was probably enhanced because each stimulus occurred 0.8-1.6 s after the previous one with a uniform probability. In the case of an ISI which comes within this time period, the longer the ISI is, the higher the probability that it will include the stimulus (i.e. ISI = ISI₁). One might therefore expect ISI₁ to be often longer than the other ISIs. We calculated experimental phase response functions, taking on the one hand all the stimuli, and on the other hand only those which did not have a triggering effect.

To estimate the latencies of the non-triggering effects of the EPSP on ISI₁, we also plotted graphs of the ISI₁ shortening as defined previously, against the time of occurrence of the post-stimulus spike (Fig. 6). The net mean changes in ISI₁ to ISI₄ were calculated by subtracting the mean shortening calculated for virtual stimuli from that obtained in response to the actual stimuli. These analyses were performed on all the trials combined, as well as separately on the trials in which the stimulus triggered a spike and those in which it did not. Moreover, to investigate the overall effects of the stimulation, we calculated the distribution of the net dephasing effect occurring throughout the trials, where the net dephasing effect was calculated as follows: (ISI₁ + ISI₂ + ISI₃ + ISI₄) - 4 ISI₀.

Selection of the ISIs

Since the instantaneous frequency of the MN discharge has a certain variability due to the synaptic noise and the effects of the stimulation, it was necessary to select the ISIs where no interference occurred between the virtual and actual stimulation. In fact, if we define ISI_a and ISI_v as being ISIs associated with actual and virtual stimulus, respectively, it sometimes happens that ISI_a,i = ISI_v,j, where i and j are indices ranging between 0 and 5, denoting the five ISIs tested with each type of stimulus. For instance, if ISI_a,3 = ISI_v,0, we cannot keep ISI_v,0 as the reference ISI relative to the virtual stimulation in our computations, because it may have been shortened by the actual stimulation. We therefore selected the ISI_vs having durations which could not be affected by the actual stimulation. We assumed that ISI_a,i (i > 4) could not be lengthened or shortened by the actual stimulation. Depending on the type of analysis performed, we then selected the stimuli with which the set of ISI_v did not overlap with the set of ISI_a,i (i [1, 4]). For the analyses focusing on ISI₁ and ISI₂, the set of ISI_v was ISI_v,i (i [0, 2]), and for the analyses focusing on the long-lasting effects, the set of ISI_v was ISI_v,i (i [0, 4]).

Motor unit characteristics

At the beginning of each recording session, the subjects performed a stereotyped ramp contraction (0.25 N s^-1), where the force level at which the motor unit started to fire was measured and taken to be its recruitment threshold. The macro-potential (macro-MUP), which is the electrical activity of all the muscle fibres of the motor unit under study, was educed by applying the spike-triggered averaging procedure to the surface EMG activity. The change in force (twitch) selectively associated with the activity of each motor unit was derived by applying the spike-triggered averaging procedure (Stein et al. 1972) to the net filtered extension force, and the rise time (ms) of the twitch was measured. Whenever possible, spike-triggered averaging was performed during two or three different periods in the recording session, always including at least 100 action potentials. The contraction time of the twitch of each motor unit was calculated by averaging the contraction times measured on the different periods.

Statistical methods

To assess the significance of the differences in the reduced PSTH bins and in the triggering frequency in the various stimulus classes between the actual and virtual stimulation conditions, we performed subdividing chi-square analyses with correction for continuity (Zar, 1996), taking the significance threshold to be 0.01.

Because of the non-Gaussian distribution of the shortening of the ISI, we used an improved normal approximation of the non-parametric Mann Whitney U test (Zar, 1996), with a significance threshold of 0.01, to assess the significance of the differences in the 10 classes of experimental response function between the actual and virtual stimulation conditions. The same test was used to compare the mean ISI_i changes and the distributions of the net dephasing effects between the two conditions of stimulation.

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Properties of the motor units

Out of the 24 motor units we tested, only 16 were selected on the basis of the reduced PSTH analysis, which showed a pure single peak without any evidence of a polysynaptic response. Table 1 summarises the response properties (latency, duration, amplitude), the discharge characteristics (ISI, coefficient of variation) and the biomechanical properties (twitch rise time, twitch amplitude, recruitment threshold) of the motor units.

Immediate effects of the EPSPs on ISI₁

(1) The triggering effect of the EPSPs. In order to characterise the spike-triggering effect of the EPSP generated by the actual stimulation, reduced PSTHs obtained under the actual and virtual stimulation conditions are presented in the case of the motor unit MU141 (Fig. 2B-C). The reduced PSTHs were based on a total of 423 stimuli. No peak occurred under virtual stimulation conditions. It can be seen that under actual stimulation conditions, there was a significant peak lasting for 3.25 ms, which started 22.75 ms after the stimulus was delivered. In view of its latency, the occurrence of this peak can be attributed to a triggering effect of the monosynaptic EPSPs generated in the wrist extensor muscles by the tendon taps (Schmied et al. 1997). Note that we thus obtained an estimate of the peripheral delay () of 22.75 ms in the case of the motor unit illustrated in Fig. 2B and C. For the sake of comparison, the standard PSTH and its CUSUM computed for the same unit are shown in Fig. 2A. Besides the steep rise corresponding to the primary peak, a secondary rise is observed in the CUSUM 110 ms after the stimulus occurred, as expected due to autocorrelation features (absent in the reduced PSTH).

With each motor unit, the spike-triggering frequency trajectory (i.e. the probability that a spike would occur after the actual or virtual stimulus within the boundaries of the monosynaptic-like response in the successive classes) was calculated. This is illustrated in the case of two units, MU141 and MU148, in Fig. 4A and B. As was to be expected, the later the stimulus was delivered in ISI₁, the more the triggering frequency increased. Here, in the case of both MU141 and MU148, the triggering frequency started to differ significantly between the actual and virtual types of stimulation in class 7. The maximum value of the spike-triggering frequency was almost 1 in class 10, which means that an EPSP occurring at the end of the expected ISI₁ triggered a spike almost every time. The first class during which the actual stimulus consistently triggered a spike at a monosynaptic-like latency varied from 5 to 7 among the 16 motor units tested; the most commonly obtained value was 6 (i.e. the tendon taps had a triggering effect during the last 40 % of the ISI).

		View larger version [in this window] [in a new window]
Figure 4. Comparison of the spike-triggering effect and the ISI1-shortening effect between two motor units, MU141(A, C and E) and MU148 (B, D and F) The S.E.M. values are shown upwards in the case of actual stimulation and downwards for virtual stimulation. A and B, triggering frequency trajectory versus stimulus classes. In both cases, the first class in which the triggered spike frequency differed significantly between actual and virtual stimulation was class 7. Note that the triggered-like spikes in the case of virtual stimulation were due to frequency variability. C-F, experimental phase response functions. These give the ISI1 shortening as a percentage of the reference interval ISI0 depending on the class in which the stimulus occurred. C and D, in the case of MU141 (C), the first class in which a significant shortening of ISI1 was induced by the actual stimulation in comparison with the virtual stimulation was class 6. In the case of MU148 (D), it was class 7. E and F, only the trials in which the stimulation did not trigger a spike are given here. In the case of MU148 (F), there was no significant shortening, whereas in the case of MU141 (E), a significant shortening of ISI1 occurred in classes 6 and 7, where the stimulation induced an advanced spiking effect. Note that in E, there were no class 10 trials, meaning that the actual stimulation always triggered a spike. The response function with virtual stimulation (F) was associated with an out of range lengthening of ISI1 in class 10 (not shown).

(2) The experimental phase response function. Once the triggering effect had been characterised, we constructed the experimental phase response function for ISI₁. This is shown in Fig. 4C-D in the case of the two units illustrated in Fig. 4A-B. The changes in ISI₁ observed under actual simulation conditions were generally positive (i.e. they took the form of a shortening). They increased up to about class 7, and then decreased along a straight line, reaching a value of almost 0 at class 10. For the last three classes, the standard errors in the experimental phase response functions of MU141, where the spike triggering effect of the EPSP was particularly strong, were very low. Under the virtual stimulation conditions, the ISI₁ shortenings were almost 0 during classes 1-7, and then decreased progressively, as was to be expected because of the bias described in the Methods. To quantify the net ISI₁ shortening, we subtracted the phase response function obtained with actual stimulation from that obtained with virtual stimulation. The maximum value observed in the case of MU141 and MU148 (Fig. 4C-D) was 23 % and 11 % of ISI₀, respectively. The results obtained with the 16 motor units tested are summarised in Table 2.

In the case of MU148, class 7 was the first class in which both the spike triggering and the ISI₁ shortening became significant. In this case, the ISI₁ shortening seems to have been entirely due to the spike-triggering effect. In the case of MU141, on the contrary, the response function started to increase in class 4 and reached significant values in class 6, (i.e. one class before the spike triggering effect occurred). We obtained similar results on 11 out of the 16 motor units tested. This therefore shows that the shortening of ISI₁ was not due to the spike triggering. The spike occurred earlier without being directly triggered.

(3) The advanced spiking effect. In order to investigate the advanced spiking effect of the excitatory input generated by the tendon taps on the MN discharge timing more thoroughly, we calculated the ISI₁ experimental phase response functions without including the trials in which the stimulus triggered a spike. This is shown for the same units in Fig. 4E and F.

In 11 out of the 16 motor units tested, the shortening of the ISI₁ induced by the actual stimulus was still significant. It generally started at class 5 and disappeared at class 7, and its maximum amplitude was just over 10 % of ISI₀. We called this effect 'advanced spiking'. The five motor units that did not show significant advanced spiking effects tended to have a higher recruitment threshold, shorter twitch rise time, and larger twitch amplitude than the 11 others.

By definition, the spikes involved in this effect occurred outside the peak in the reduced PSTH. In order to further investigate their distribution after the stimulus, reduced PSTHs have been computed for each class of stimulation as shown in Fig. 5. From class 3-5, the spike distribution in the reduced PSTHs computed with actual stimulation (continuous line) is shifted leftwards compared to that with virtual stimulation (dotted line).

		View larger version [in this window] [in a new window]
Figure 5. Post-stimulus spike latency in different stimulation classes. Reduced PSTHs have been computed for each of the 10 classes of actual (continuous lines) and virtual (dotted lines) stimulus, in the case of MU141 Note that they have been corrected for the peripheral delay, as explained in the Methods. The stimulus occurrences (filled triangles at the origin of each PSTH) are roughly located on the ramp of the membrane potential trajectory of ISI1 (dashed line). The spikes directly triggered by the stimulus are grouped into narrow peaks at the very beginning of the reduced PSTH from class 6 to class 10. Besides the peaks, the spike distribution with actual stimulus is shifted to the left from class 3 to class 7, corresponding to the advanced spiking effect.

In order to estimate the latency of this effect, we plotted the ISI₁ shortenings as defined previously against the time of occurrence of the post-stimulus spike under the actual and virtual stimulation conditions, as shown in Fig. 6. The spikes triggered in the monosynaptic peak can be clearly seen to have clustered together during the first 5 ms after the stimulus was applied. However 20-50 ms after the stimulus, the great majority of the values recorded were still positive under the actual stimulation conditions, contrary to what was observed with virtual stimulation. This distribution was therefore that of the spikes where the ISI₁ shortening was not directly due to the spike-triggering effect. This pattern was observed in 11 out of the 16 motor units tested.

		View larger version [in this window] [in a new window]
Figure 6. Distribution of the changes in ISI1 with respect to the post-stimulus time This figure shows the changes in the inter-spike interval ISI1 during which the stimulus occurred as a percentage of the reference interval ISI0 with respect to the time elapsing between the stimulus and the subsequent spike. In comparison with the virtual stimulation values (open circles), most of the actual stimulation values (filled circles) were positive in the two regions shown by the arrows, which indicates that a shortening of ISI1 had occurred, that was either time-locked (within less than 5 ms) to the stimulus (spike-triggering effect) or occurred later on (20-50 ms later) and was broadly dispersed (advanced spiking effect).

Long-lasting effects

(1) Effects on ISI₂. In motoneurones as well as in cortical neurones, it has been reported that if an ISI is shortened in comparison with the mean ISI values, the following ISI will be lengthened due to summation of the AHPs (Baldissera et al. 1978; Reyes & Fetz, 1993a,b). This hypothesis was tested here by computing experimental phase response functions expressing the changes in ISI₂ in relation to the timing of the stimulus occurrence within ISI₁, under the conditions of both actual and virtual stimulation (Fig. 7). In view of its shape, the function could be divided into two parts. During the first part (class 1-5), ISI₂ was shortened. In the second part (class 5-10), ISI₂ was lengthened. These two parts suggested the existence of a dual effect that did not always reach significance level. In fact, four motor units showed a significant shortening of ISI₂, four motor units showed a lengthening of ISI₂, and one showed both, but the same tendencies were observed consistently with all 16 motor units tested. The two motor units shown here are typical examples. In the case of MU1410, the response function reached significantly negative values in class 8, whereas in the case of MU161, the response function reached significantly positive values in classes 2 and 3 under the conditions of actual stimulation compared to virtual stimulation.

		View larger version [in this window] [in a new window]
Figure 7. Experimental response functions computed with all trials in the case of ISI1 and ISI2 for two motor units, MU1410 (A and C) and MU161 (B and D) The S.E.M. values are shown upwards in the case of actual stimulation and downwards in the case of virtual stimulation. A and B, ISI1 phase response functions. As observed in Fig. 4, the shortening of ISI1 that occurred in response to the actual stimulation involved only the last classes (6-10). C and D, ISI2 phase response functions. In the case of MU1410 (C), a lengthening of ISI2 was observed in class 8, reflecting the existence of a negative correlation between ISI1 and ISI2. In the case of MU161 (D), a shortening effect was observed in classes 2 and 3, while the stimulation had been delivered during the AHP peak period. It is worth noting that a similar but non-significant trend was also observed with MU1410 (C).

(2) Net effects on ISI₁, ISI₂, ISI₃ and ISI₄. The effects of stimulation could be investigated up to ISI₄ in the case of 14 out of the 16 motor units tested (without overlapping between successive actual and virtual stimuli, see Methods). In Fig. 8, the net values (actual minus virtual) of the mean changes observed with each of the four successive ISIs associated with the occurrence of the stimulus are linked by a straight line in the case of each motor unit. The curves obtained with 14 units are superimposed. The procedure used to calculate these values is described in the Methods. The circles indicate the values at which the actual and virtual differences were significant. In Fig. 8A, all trials are included, showing that the mean tendency was a shortening of ISI₁ and ISI₃. When the stimulus triggered a spike (Fig. 8B), ISI₁ was shortened as expected, ISI₂ was lengthened, and ISI₃ was shortened again. No clear tendency was observed for ISI₄. This shows either that the stimulation had very long-lasting effects (over 300 ms) and/or that a strong correlation between subsequent ISIs existed. If we look at the trials in which the stimulus did not trigger any spikes (Fig. 8C), ISI₁ and ISI₂ were shortened; a similar tendency was seen in ISI₃, although the effect was not significant. This again showed that the stimulation had prolonged effects, even when it did not trigger a monosynaptic-like response.

		View larger version [in this window] [in a new window]
Figure 8. Stimulation-induced changes in ISI A-C, the various curves give the results obtained with 14 different motor units. ISI0 is the reference inter-spike interval, ISI1 the interval during which the stimulation occurred, ISI2, ISI3 and ISI4 are the subsequent inter-spike intervals. With each MU, the relative changes in ISI1 to ISI4 were computed by subtracting the mean change in ISI expressed as a percentage of the ISI0 values obtained with virtual stimulation from those obtained with actual stimulation. The circles indicate the relative changes observed in the cases where the mean changes in ISI (%ISI0) differed significantly between actual and virtual stimulation. A, all trials are included. A long-lasting effect can be seen to have occurred from the fact that the stimulation occurring during ISI1 sometimes affected ISI2, ISI3, and ISI4. ISI1, ISI2 and ISI3 were shortened whereas ISI4 was lengthened. B, only the trials in which the stimulation triggered a spike have been taken into consideration. ISI1 and ISI3 were always shortened whereas ISI2 and more rarely ISI4 tended to be lengthened. There was a clear-cut negative correlation between successive intervals, probably partly due to AHP summation. C, only the trials in which the stimulation did not trigger a spike were taken into consideration. ISI1 and ISI2 were generally shortened, whereas ISI4 tended to be lengthened and ISI3 was not visibly affected. This shows that a long-lasting shortening effect occurred, which was possibly masked by the negative correlation shown in B.

The distributions of the net dephasings calculated as (ISI₁ + ISI₂ + ISI₃ + ISI₄) - 4 ISI₀ are shown in the case of the unit 149b (Fig. 9A), and for the 14 units together (Fig. 9B). Although the envelopes of the two distributions were quite similar, the average dephasing value was centred around 0 under the virtual stimulation conditions, whereas it was centred around -10 ms under the actual stimulation conditions. In the case of eight motor units, the average dephasing value observed under the actual stimulation conditions differed significantly from that obtained under the virtual stimulation conditions. In the other motor units, although a similar tendency was observed, it did not reach significance level.

		View larger version [in this window] [in a new window]
Figure 9. Comparisons between the net dephasing effects obtained with actual and virtual stimulation A and B, the net dephasing effects are estimated as follows: (ISI1 + ISI2 + ISI3 + ISI4) - 4 ISI0 (s). A 0.02 s bin was used to build the distributions. A, data obtained on motor unit MU149b. The two distributions differed significantly. The net dephasing effect shifted to the left with actual stimulations, meaning that there was an overall shortening of the four intervals ISI1, ISI2, ISI3 and ISI4. B, data obtained on eight motor units in which the net dephasing distributions between the actual and virtual stimulation were significantly different. The relative stimulation counts were normalised depending on the total number of stimuli (actual or virtual) applied. The curves obtained with the various motor units have been combined to form the filled and open areas, corresponding to the actual and virtual conditions of stimulation, respectively, in order to show up the overall pattern of the dephasing effects. The net dephasing generated by the actual stimuli was negative on the whole, as shown by the shift of the filled area towards negative values.

It was unexpected that the dephasing distributions were slightly skewed leftwards even in the virtual stimulation condition. The explanation for this could be that in most of our recordings the ISI distributions were significantly skewed towards the right. Therefore, when comparing (ISI₁ + ISI₂ + ISI₃ + ISI₄) - 4 ISI₀, the probability of picking four consecutive intervals in the skewed part of the distribution is low, whereas if ISI₀ is taken in this part, the impact of its long duration will be enhanced by multiplying by 4. This explains why the dephasing distribution is generally skewed towards negative values. Note that this 'bias' can be neglected since the net dephasing effects were compared between actual and virtual stimulation conditions.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

The experimental phase response functions computed in this study show that transient excitatory inputs, such as those which could be expected to be generated by the very tiny tendon taps used, can advance the subsequent spikes of a steadily discharging MN when delivered during the first half of the ISIs in a region where the membrane potential is too hyperpolarised for a direct threshold-crossing process to occur. These effects were prolonged during two successive interspike intervals. The inputs generated by the tendon taps thus effectively altered the subsequent firing times of the motoneurone over a broader part of the ISI than expected. It is worth noting here that experimental phase response function analysis can reveal effects that are not apparent in the PSTH.

With the appropriate precautions due to the limited size of the sample of motor units studied here, it was observed that the strongest spike-triggering and advanced spiking effects generated by the tendon taps occurred with slowly contracting motor units recruited at the lowest force threshold, which is in keeping with the 'size principle' (Henneman, 1977).

We will discuss in turn the nature and characteristics of the synaptic input generated by the stimulation, its effects on the subsequent firing time (ISI₁), the following ones (ISI₂, ISI₃ and ISI₄), and lastly, the possible origins of these delayed, long-lasting effects.

Nature of the synaptic input

The transient muscle stretching induced by applying tendon taps has classically been used as a tool to stimulate the muscle spindles in order to elicit a reflex activation of the homonymous MNs. The possibility that cutaneous afferents may have been involved can be ruled out in view of the fact that whenever the hammer was not positioned exactly on the tendon, it was completely ineffective, although cutaneous receptors were probably still activated. Although the possibility that contributions from other proprioceptive inputs may have been involved, such as those mediated by spindle secondary afferents and tendon organ afferents, cannot be completely ruled out (Burke et al. 1983), data obtained in animal experiments (Stuart et al. 1970; Matthews, 1972; Fetz et al. 1979) as well as from human microneurographic recordings (Vallbo, 1973; Burke et al. 1978, 1983; Vallbo et al. 1979; Roll & Vedel, 1982) have indicated that the muscle spindle primary afferents are those most readily activated by this type of stimulation. The Ia afferent volleys can induce both monosynaptic and polysynaptic excitatory responses (Watt et al. 1976; Malmgren & Pierrot-Deseilligny, 1988; Romaiguère et al. 1991), but in the present study we selected the motor units showing no evidence at all of polysynaptic responses in the PSTH.

In an early study, Lundberg & Winsbury (1960) reported that in motoneurones from the soleus muscle of anaesthetised cats, the EPSPs generated by applying electrical stimulation to Ia afferents or by inducing phasic muscle stretching (100 µm) could be strikingly similar, with a duration of about 15 ms. This provides an indication to the time course of the EPSPs which could be expected to be generated by the very tiny tendon taps we used (70-140 µm), with all due reservations about comparing data obtained on anaesthetised animals and awake humans. It is worth noting here that taps such as those applied transversely in our study to the tendon are liable to have induced much smaller muscle stretches than those used in Lundberg & Winsbury's study (1960). Another clue about the time course of the EPSP generated by Ia afferents during voluntary contraction in human muscle was provided by Ashby & Zilm (1982), who delivered double stimuli within an ISI, and thereby showed that the duration of the falling phase of homonymous group I EPSPs in tibialis anterior motoneurones lasted less than 20 ms.

Detailed analysis of the effects of the stimulation on ISI₁

In the first classes of the ISI₁ phase response function, the stimulus had neither shortening nor lengthening effects on ISI₁. The lack of shortening effects can be easily understood. In fact, after a spike, the membrane potential is hyperpolarised by the activation of potassium conductances. Moreover, the AHP involves a massive increase in the calcium-dependent potassium conductances (Rekling et al. 2000), i.e. a decrease in the membrane resistance, and the EPSPs generated consequently have lower amplitudes. Therefore, one can expect the MN to be less sensitive to depolarisations. In some cortical neurones (Reyes & Fetz, 1993a,b), as in the model of Hodgkin & Huxley (1952), a lenthening of ISI₁ has been observed in the phase response in the first classes of the phase response function. The reason for this lengthening is still unknown, although some hypotheses have been proposed, such as the activation of a low threshold transient potassium conductance (Spain et al. 1991), or the deactivation of an anomalous rectifying conductance (Spain et al. 1987; Schwindt et al. 1988).

In the last classes, the phase response functions decreased along a straight line and the standard error was conspicuously lower than in the previous classes. This was the logical consequence of the spike-triggering effect of the EPSPs. In fact, let us consider the behaviour of a hypothetical neurone in which all the EPSPs trigger a spike. In this case, the phase response function would be a straight line having a y-axis intercept which is 100 % of ISI₀ and an x-axis intercept which is 0 % of ISI₀. This defines the upper boundary of any phase response function. In the last classes, we observed that the frequency of triggered spikes increased up to 1 (or nearly 1). The phase response function then tended towards its upper boundary. The fact that the standard error decreased resulted directly from this behaviour. In fact, most of the spikes were triggered and were therefore generated in response to the stimulation with nearly identical latencies, within the very narrow limits of the PSTH peak. The shortening was therefore quite constant, as long as the duration of ISI₀ did not vary too much.

In the examples presented here, as well as in most of the other motor units, the triggering effect was significant in class 6, which corresponded to an average effective period of 40 % of ISI. In other words, only 40 % of the ISI₁ was significantly affected by the stimulation spike-triggering effect. This period is shorter on average than reported values obtained in tibialis, gastrocnemius, soleus and flexor carpi radialis (FCR) muscles (Table 3). A longer effective interval could be explained in terms of either shorter AHPs or larger EPSPs. The 'shorter AHP' hypothesis can be ruled out for two reasons. (1) Among the leg muscles tested, the motoneurone AHP of the soleus muscle is known to have a particularly long duration (Eccles et al. 1958). (2) Although the properties of the ECR and FCR motoneurones have never been directly compared, both groups of muscles have been described as 'fast' muscles containing a fair proportion of fast contracting muscle fibres which are probably associated with motoneurones having rather short AHPs.

The EPSPs generated by the types of stimulation we used here therefore seem to have been much smaller than those in the latter studies. This point was confirmed by the narrowness of the PSTH peaks (3.74 ± 1 ms) and their small amplitudes (28.30 ± 4.50 %; i.e. only about 1 out of 3 tendon taps induced direct threshold crossing).

Although the advanced spiking effect was observed in 11 out of the 16 motor units, its effectiveness in terms of ISI₁ shortening differed depending on the motor units, as can be seen from Table 2. Based on the examples of MU169 and MU141, we noted that the stronger the spike-triggering effect, the stronger the advanced spiking effect turned out to be. This suggests that the advanced spiking effect might depend on the EPSP amplitude as does the direct threshold crossing effect.

Possible explanations of the advanced spiking effect

The first possible explanation of the advanced spiking effect put forward recently by Türker & Powers (1999) on the basis of the data obtained by Gustafsson & McCrea (1984) was that the synaptic noise may summate and trigger a spike in the late part of a small EPSP. This explanation is based on the assumption that the EPSP is of the same order of amplitude as the synaptic noise. The first argument against this idea is that we observed a stronger advanced spiking effect in motor units showing a stronger triggering effect, and in which we can therefore assume that the EPSPs were much larger than the synaptic noise. The second argument against this hypothesis is that the advanced spiking effect occurred 20-50 ms after the stimulation, which, on the basis of the animal experiments discussed above (Lundberg & Winsbury, 1960), is much longer than the expected duration of the EPSPs likely to be generated by the tendon taps. Moreover, if a triggering effect occurred during the late part of the EPSP, we would expect the effects to show some continuity, matching the shape of the EPSP in Fig. 6, as proposed by Türker & Powers (1999), which was not the case. Lastly, we would not expect the stimulation to have any shortening effects on ISI₂ and ISI₃. This hypothesis can therefore be ruled out.

Another possible explanation involves the contribution of polysynaptic pathways, as suggested in an early study (Kanda, 1972). In fact, the earliest latencies of the advanced spiking effect were in the range of the latencies of the transcortical response observed in wrist muscles (Palmer & Ashby, 1992; Inglis et al. 1997). In this case, our analyses would have allowed the detection of a low and widely dispersed polysynaptic response that remained undetected in the PSTH and the CUSUM analyses. It is worth noting that in the eight motor units not included in the present study, which produced a second peak about 10 ms later in the PSTH, the advanced spiking effect could still be observed in one region of the phase response function, while no spikes were triggered in the boundaries of either peaks.

Finally, the contribution of some intrinsic properties of the motoneurone might be proposed to account for the advanced spiking effect on ISI₁. This could involve the acceleration of the spike-generating mechanisms as suggested by Gutkin & Ermentrout (1999). In a theoretical study based on phase response function analyses, they suggested that the advanced spiking effect on ISI₁ may simply reflect the fact that the sodium conductance involved in spike generation has been only partially activated. This might occur when the EPSP brings the membrane potential close enough to the firing threshold to partially activate the sodium conductance. The evolution of the membrane potential on its trajectory would therefore be accelerated and the spike advanced. Alternatively, it could involve the activation of an inward current below the firing threshold as proposed by Reyes & Fetz (1993a) in the case of cortical neurones. In mammalian motoneurones, the actual contribution of such a low threshold conductance to motoneurone discharge behaviour is not completely clear (Powers & Binder, 2001). The advanced spiking effect might be an expression of it in physiological conditions.

In fact, we observed that the strengths of the advanced spiking and spike-triggering effects did vary in the same way, which suggested that both effects were dependent on the amplitude of the EPSP. This could be taken to indicate the existence of two separate thresholds as proposed by Reyes & Fetz (1993a), the lower one being that of a conductance or any other process able to advance after some delay the activation of the spike-generating conductances, the higher one being the firing threshold itself.

Detailed analysis of the effects of the stimulation on the subsequent ISIs

One particularly interesting point worth mentioning was the dual effect of the stimulation on the ISI₂ following the ISI during which the stimulus occurred. In fact, two parts could be distinguished in the ISI₂ experimental phase response functions; the first was a positive one, which means that ISI₂ was shortened, and the second one was negative, which means that ISI₂ was lengthened.

First we will focus on the lengthening effect, since a similar process has been observed by Reyes & Fetz (1993a,b) in cat neocortical neurones. This effect was generally significant in the classes where the frequency of the spikes triggered was high. A simple explanation may be provided by the summation of AHPs, as described in motoneurones (Ito & Oshima, 1962; Calvin & Stevens, 1968). During the AHP phase, the persistent calcium-dependent potassium conductance (I_K,Ca,SK) is activated and cannot be deactivated by a spike. Its gradual deactivation is responsible for the ramp trajectory of the membrane potential between two spikes. When the firing is advanced, the AHP phase is not complete when the spike occurs, and many I_K,Ca,SK channels still remain to be deactivated. The intracellular calcium concentration increases because high voltage-activated calcium conductances are activated during the firing process (Rekling et al. 2000), and further I_K,Ca,SK channels are activated. The AHP will therefore be longer than usual in the case of ISI₂ when ISI₁ is shortened.

However, as shown in Fig. 8A, among the two opposite effects of the stimulation on the duration of ISI₂, the shortening effects seem to prevail. These effects have never been previously observed as far as we know, which makes it all the more interesting. Since we noted that it was stronger when the stimulus had not triggered a spike, it is tempting to suggest that it might be correlated with the advanced spiking effect on ISI₁. However, the shortening of ISI₂ was also present when the EPSP occurred during the AHP scoop, that is when it had no effect on the duration of ISI₁.

In addition, we observed that in our recordings that the negative correlation between the changes in ISI₁ and ISI₂ was weak in comparison with that observed by Reyes & Fetz (1993a,b) in cat neocortical neurones. This may have been due either to the low firing frequency of the MN present under our conditions or to the competition between the two effects. In fact, it is possible that the two effects (lengthening and shortening) may have been superimposed on ISI₂ when the stimulus occurred in the intermediate and last classes of ISI₁. The lengthening effect might then be preponderant during the last classes of ISI₁.

Unlike ISI₂, which tended to be shortened in the case of the advanced spiking effect on ISI₁ and lengthened in case of the triggering effect on ISI₁, ISI₃ always tended to be shortened. Here again, the idea of a persistent shortening affecting successively ISI₁ (advanced spiking effect), ISI₂ (an effect partly masked by AHP summation), and ISI₃ is quite tempting. Note that the possibly negative correlation between ISI₂ and ISI₃ does not explain the changes in the duration of ISI₃, since the latter was affected in the same way in both cases (advanced spiking and triggering effects on ISI₁), whereas ISI₂ was not.

Possible explanations for the delayed and prolonged effects

When the EPSPs generated by the tendon taps had no spike-triggering effect, ISI₂ and ISI₃ were shortened, which means that an effect was prolonged for more than 100 ms after the occurrence of the stimulus in the case of ISI₂ and more than 200 ms in the case of ISI₃. Such widely dispersed and prolonged effects might of course involve polysynaptic pathways, assumming the contribution of many reverberating loops through spinal or supraspinal pathways. However, other explanations involving the intrinsic properties of the motoneurone must be kept in mind.

The shortening effect on ISI₂ was observed even when the stimulus occurred in the very first classes of the phase response function with no effect at all on ISI₁. This might suggest the contribution a non-inactivating and persistent conductance change with an activation time too long to affect the next firing time (i.e. ISI₁). Although no definite relationship can be established at this stage, this hypothesis is reminiscent of the existence of non-inactivating persistent conductance changes involved in the plateau potential phenomenon that can be triggered in spinal motoneurones by volleys of Ia afferent stimulation or phasic muscle stretches in anaesthetised cats (Bennett et al. 1998).

A final hypothesis to explain the delayed and prolonged effects involves an interaction between polysynaptic pathways and specific conductances. As recently established in lamprey reticulospinal neurones (Di Prisco et al. 2000), sensory stimulation can polysynaptically activate local AMPA receptors, thus contributing to the activation of NMDA receptors. The inward calcium current generated may cause intracellular calcium to be released and the increase in the intracellular calcium concentration may activate the persistent non-voltage-dependent conductance (I_CAN; calcium-activated non-selective cation current) that may be responsible for plateau potentials. Here, this conductance might not be completely activated, and would therefore not generate a real plateau potential but a prolonged depolarisation facilitating the crossing of the firing threshold, and thus advancing the time of occurrence of the spikes. Given the long latency of the activation of these conductances, which is due to the complexity of the mechanisms involved, they might explain the prolonged effects observed on ISI₂ and ISI₃.

To conclude, it has been established that a tendon tap can modify the human -motoneurone discharge pattern for a few hundred milliseconds during voluntary contraction of the corresponding muscle. Besides the precise timing effect due to threshold crossings time-locked to the EPSP, the changes in firing times consist firstly of a shortening of the interspike interval during which the stimulus occurred, without any direct spike triggering, and secondly, of a prolonged shortening of the two subsequent interspike intervals. Although no definite explanations are yet available for these findings, the time course of this process is taken to suggest that non-inactivating persistent conductance changes might be involved, although the contribution of polysynaptic pathways cannot be excluded at this stage.

These findings might have functional implications as regards the synchronising or desynchronising processes affecting the activity of the motor units and/or the optimisation of muscle contraction efficiency. Considering an assembly of tonically discharging motoneurones innervated by common Ia afferents, it can be predicted that the changes in firing times induced by Ia EPSPs should actually be quite dispersed throughout two to three interspike intervals except in the very instances when precisely timed synchronous spikes are generated by the EPSPs occurring concurrently in the spike-triggering zone of the motoneurone phase response functions.

Further studies are now required to explore all these possibilities. The present data were obtained using specific methods of analysis, including the phase response function procedure which is commonly used in the field of theoretical physics (Gutkin & Ermentrout, 1999; Stoop et al. 2000). The fact that these methods have been applied to the field of motoneurone behaviour for the first time here might explain why these effects have never been observed so far, and the results therefore show that it is necessary to analyse spike trains in new ways to be able to extract all the information they contain. The use of the phase response function as introduced by Reyes & Fetz (1993a,b) proved to be a very powerful although indirect procedure, which can be used to characterise both the electrophysiological properties of a neurone and the activity of its various afferents. It would therefore be possible to perform analyses of this kind to study pathological conditions affecting human motoneurones and/or their afferents in order to obtain further insights into the impairments induced by these diseases.

	REFERENCES

Top Abstract Introduction Methods Results Discussion References

ASHBY, P. & ZILM, D. (1982). Characteristics of postsynaptic potentials produced in single human motoneurons by homonymous group 1 volleys. Experimental Brain Research 47, 41-48	[Medline]
BALDISSERA, F., GUSTAFSSON, B. & PARMIGGIANI, F. (1978). Saturating summation of the afterhyperpolarization conductance in spinal motoneurones: a mechanism for 'secondary range' repetitive firing. Brain Research 146, 69-82	[Medline]
BENNETT, D. J., HULTBORN, H., FEDIRCHUK, B. & GORASSINI, M. (1998). Synaptic activation of plateaus in hindlimb motoneurons of decerebrate cats. Journal of Neurophysiology 80, 2023-2037	[Abstract/Full Text]
BURKE, D., GANDEVIA, S. C. & MCKEON, B. (1983). The afferent volleys responsible for spinal proprioceptive reflexes in man. Journal of Physiology 339, 535-552	[Medline]
BURKE, D., HAGBARTH, K. E. & LOSFTEDT, L. (1978). Muscle spindle activity in man during shortening and lengthening contractions. Journal of Physiology 277, 131-142	[Abstract]
CALVIN, W. H. & STEVENS, C. F. (1968). Synaptic noise and other sources of randomness in motoneuron ISI. Journal of Neurophysiology 31, 574-587	[Medline]
DI PRISCO, G. V., PEARLSTEIN, E., LE RAY, D., ROBITAILLE, R. & DUBUC, R. (2000). A cellular mechanism for the transformation of a sensory input into a motor command. Journal of Neuroscience 20, 8169-8176	[Abstract/Full Text]
ECCLES, J. C., ECCLES, R. M. & LUNDBERG, A. (1958). The action potentials of the alpha motoneurones supplying fast and slow muscles. Journal of Physiology 142, 275-291
ELLAWAY, P. H. (1978). Cumulative sum technique and its application to the analysis of peristimulus time histograms. Electroencephalography and Clinical Neurophysiology 45, 302-304	[Medline]
FETZ, E. E. & GUSTAFSSON, B. (1983). Relation between shapes of post-synaptic potentials and changes in firing probability of cat motoneurons. Journal of Physiology 341, 387-410	[Medline]
FETZ, E. E., JANKOWSKA, E., JOHANNISSON, T. & LIPSKI, J. (1979). Autogenetic inhibition of motoneurones by impulses in group Ia spindle afferents. Journal of Physiology 293, 173-195	[Abstract]
FRICKER, D. & MILES, R. (2000). EPSP amplification and the precision of spike timing in hippocampal neurons. Neuron 28, 559-569	[Medline]
GRANIT, R., KERNELL, D. & SHORTESS, G. K. (1963). Quantitative aspects of repetitive firing of mammalian motoneurones, caused by injected currents. Journal of Physiology 168, 911-931
GUSTAFSSON, B. & McCrea, D. (1984). Influence of stretch-evoked synaptic potentials on firing probability of cat spinal motoneurons. Journal of Physiology 347, 431-451	[Abstract]
GUTKIN, B. S. & ERMENTROUT, G. B. (1999). Dynamics of action potential generation predict the relationship between the timing of excitatory inputs and the timing of spikes in cortical pyramidal neurons. Society For Neuroscience Abstracts 21, 691.1
HENNEMAN, E. (1977). Functional organization of motoneuron pools: the size principle. Proceedings for the International Union of Physiological Sciences 12, 50
HODGKIN, A. L. & HUXLEY, A. F. (1952). A quantitative description of membrane current and its application to conduction and excitation in nerve. Journal of Physiology 117, 500-544
HOUNSGAARD, J., HULTBORN, H. & KIEHN, O. (1984). Intrinsic membrane properties causing a bistable behaviour of -motoneurones. Experimental Brain Research 55, 391-394	[Medline]
INGLIS, J. T., MEUNIER, S., LEEPER, J. B., BURKE, D. & GANDEVIA, S. C. (1997). Weak short-latency spinal projections to the long flexor of the human thumb. Experimental Brain Research 115, 165-168	[Medline]
ITO, M. & OSHIMA, T. (1962). Temporal summation of AHP following a motoneurone spike. Nature 195, 910-911
JONES, K. E. & BAWA, P. (1995). Responses of human motoneurons to Ia inputs: effects of background firing rate. Journal of Physiology and Pharmacology 73, 1224-1234
KANDA, K. (1972). Contribution of polysynaptic pathways to the tonic vibration reflex. Japanese Journal of Physiology 22, 367-377	[Medline]
KUDINA, L. P. (1988). Excitability of firing motoneurones tested by Ia afferent volleys in human triceps surae. Electroencephalography and Clinical Neurophysiology 69, 576-580	[Medline]
LUNDBERG, A. & WINSBURY, G. (1960). Selective adequate activation of large afferents from muscle spindles and golgi tendon organs. Acta Physiologica Scandinavica 49, 155-164
MALMGREN, K. & PIERROT-DESEILLIGNY, E. (1988). Evidence for non-monosynaptic Ia excitation of human wrist flexor motoneurones, possibly via propriospinal neurones. Journal of Physiology 405, 747-764	[Abstract]
MATTEI, B., SCHMIED, A., VEDEL, J. P. & SALIN, P. (2000). Different modes of inter-spike interval shortening induced by juxtathreshold Ia inputs on steadily discharging human motoneurones. Society For Neuroscience Abstracts 26, 258.15
MATTHEWS, P. B. C. (1972). Mammalian muscle receptors and their central actions. Edward Arnold, London
MOORE, G. P., PERKEL, D. H. & SEGUNDO, J. P. (1966). Statistical analysis and functional interpretation of neuronal spike data. Annual Review of Physiology 28, 493-522	[Medline]
PALMER, E. & ASHBY, P. (1992). Evidence that a long latency stretch reflex in humans is transcortical. Journal of Physiology 449, 429-440	[Abstract]
PERKEL, D. H., SCHULMAN, J. H., BULLOCK, T. H., MOORE, G. P. & SEGUNDO, J. P. (1964). Pacemaker neurons: effects of regularly spaced synaptic input. Science 145, 61-63
POWERS, R. K. & BINDER, M. D. (2001). Input-output functions of mammalian motoneurons. Reviews of Physiology, Biochemistry and Pharmacology 143, 137-263	[Medline]
REKLING, J. C., FUNK, G. D., BAYLISS, D. A., DONG, X.-W. & FELDMAN, J. L. (2000). Synaptic control of motoneuronal excitability. Physiological Reviews 80, 767-852	[Abstract/Full Text]
REYES, A. D. & FETZ, E. E. (1993a). Two modes of interspike interval shortening by brief transient depolarization in cat neocortical neurons. Journal of Neurophysiology 69, 1661-1672	[Medline]
REYES, A. D. & FETZ, E. E. (1993b). Effects of transient depolarizing potentials on the firing rate of cat neocortical neurons. Journal of Neurophysiology 69, 1673-1682	[Medline]
ROLL, J. P. & VEDEL, J. P. (1982). Kinesthetic role of muscle afferents in man studied by tendon vibration and microneurography. Experimental Brain Research 41, 177-190
ROMAIGUÈRE, P., VEDEL, J. P., AZULAY, J. P. & PAGNI, S. (1991). Differential activation of motor units in the wrist extensor muscles during the tonic vibration reflex in man. Journal of Physiology 444, 645-669	[Abstract]
SCHMIED, A., MORIN, D., VEDEL, J. P. & PAGNI, S. (1997). The "size principle" and synaptic effectiveness of muscle afferent projections to human extensor carpi radialis motoneurones during wrist extension. Experimental Brain Research 113, 214-229	[Medline]
SCHWINDT, P. C. & CALVIN, W. H. (1972). Membrane-potential trajectory between spikes underlying motoneuron firing rates. Journal of Neurophysiology 35, 311-325	[Medline]
SCHWINDT, P. C. & CRILL, W. E. (1980). Properties of a persistent inward current in normal and TEA-injected motoneurons. Journal of Neurophysiology 43, 1700-1724	[Medline]
SCHWINDT, P. C., SPAIN, W. J. & CRILL, W. E. (1988). Influence of anomalous rectifier activation on afterhyperpolarization of neurons on cat sensorimotor cortex in vitro. Journal of Neurophysiology 59, 468-481	[Medline]
SPAIN, W. J., SCHWINDT, P. C. & CRILL, W. E. (1987). Anomalous rectification in neurons from cat sensorimotor cortex in vitro. Journal of Neurophysiology 57, 1555-1576	[Medline]
SPAIN, W. J., SCHWINDT, P. C. & CRILL, W. E. (1991). Two transient potassium currents in layer V pyramidal neurones from cat sensorimotor cortex in vitro. Journal of Physiology 434, 591-607	[Abstract]
STEIN, R. B., FRENCH, A. S., MANNARD, A. & YEMM, R. (1972). New methods for analysing motor function in man and in animals. Brain Research 40, 187-192	[Medline]
STOOP, R., SCHINDLER, K. & BUNIMOVICH, L. A. (2000). When pyramidal neurons lock, when they respond chaotically, and when they like to synchronize. Neuroscience Research 36, 81-91	[Medline]
STUART, D. G., MOSCHER, C. G., GERLACH, R. L. & REINKIND, R. M. (1970). Selective activation of Ia afferents by transient muscle stretch. Experimental Brain Research 10, 477-487	[Medline]
TÜRKER, K. S. & POWERS, R. K. (1999). Effects of large excitatory and inhibitory inputs on motoneuron discharge rate and probability. Journal of Neurophysiology 82, 829-840	[Abstract/Full Text]
VALLBO, A. B. (1973). Muscle spindle afferent discharge from resting and contracting muscles in normal subjects. In New Developments in EMG and Clinical Neurophysiology, vol. 3, ed. DESMEDT, J. E., pp. 251-262. Karger, Basel
VALLBO, A. B., HAGBARTH, K. E., TOREBJORK, H. E. & WALLIN, B. G. (1979). Somatosensory, proprioceptive, and sympathetic activity in human peripheral nerves. Physiological Reviews 59, 919-957	[Medline]
WATT, D. G. D., STAUFFER, E. K., TAYLOR, A., REINKIND, R. M. & STUART, D. G. (1976). Analysis of muscle receptor connections by spike triggered averaging: spindle primary and tendon organ afferents. Journal of Neurophysiology 38, 1375-1392
ZAR, J. H. (1996). Biostatistical Analysis, 3rd edn. Simon & Schuster, London

Acknowledgements

We are grateful to Dr J. Blanc for correcting the English and to J. P. Vedel for helpful discussions. This research was supported by Grants from the Association Française contre les Myopathies (A.F.M.), the Fondation pour la Recherche Médicale (F.R.M.), and STTC-D.G.A.

This article has been cited by other articles:

				J. G. Mancilla, T. J. Lewis, D. J. Pinto, J. Rinzel, and B. W. Connors Synchronization of Electrically Coupled Pairs of Inhibitory Interneurons in Neocortex J. Neurosci., February 21, 2007; 27(8): 2058 - 2073. [Abstract] [Full Text] [PDF]

				B. S. Gutkin, G. B. Ermentrout, and A. D. Reyes Phase-Response Curves Give the Responses of Neurons to Transient Inputs J Neurophysiol, August 1, 2005; 94(2): 1623 - 1635. [Abstract] [Full Text] [PDF]

Y. Prut and S. I. Perlmutter Firing Properties of Spinal Interneurons during Voluntary Movement. I. State-Dependent Regularity of Firing J. Neurosci., October 22, 200