During and after voluntary exercise there are changes in the behaviour of the motoneurones and motor cortex. During sustained maximal voluntary isometric contractions, the discharge rate of motoneurones declines from about 30 Hz to below 20 Hz within a minute (e.g. Bigland-Ritchie et al. 1986; Gandevia et al. 1990), and voluntary activation of the muscle also declines as can be demonstrated with motor nerve or motor cortical stimulation (e.g. Gandevia et al. 1996). Thus, the stimuli delivered during the continuing ‘maximal’ effort generate larger force increments from the muscle at the end than at the start of the contraction, that is, central fatigue develops (for reviews see Gandevia et al. 1995; Gandevia, 1998). However, during the contraction the muscle's excitatory response to motor cortical stimulation increases, even when corrected for the increase in size of the maximal compound muscle action potential (Taylor et al. 1996, 1999). After the contraction when the muscle is relaxed, the responses to transcranial magnetic stimulation of the motor cortex are depressed (e.g. Brasil-Neto et al. 1993; Zanette et al. 1995; McKay et al. 1995; Liepert et al. 1996). This post-contraction depression is believed to reflect a purely cortical phenomenon because estimates of the excitability of spinal motoneurones using transcranial electrical stimulation of the cortex were said to be unaltered by the contraction (e.g. Brasil-Neto et al. 1993; cf. McKay et al. 1995).
The present studies were designed to assess the possibility that voluntary contractions induce neural changes at subcortical levels within the motor pathways from cortex to motoneurone. The selected test input was a descending volley excited by electrical stimulation to the cervicomedullary junction via electrodes placed on the mastoids. Based on collision experiments with transcranial electrical stimulation, the transmastoid stimulus is likely to generate a single excitatory volley in corticospinal axons (Ugawa et al. 1991; see also Rothwell et al. 1994). This descending input was also chosen (rather than a reflex input) because corticospinal synapses appear to lack presynaptic inhibition (Nielsen & Petersen, 1994), and thus the response to transmastoid stimulation is likely to reflect only changes in the corticospinal path and motoneurones. To ensure that any changes in the evoked EMG were not due to changes at the muscle fibre membrane or neuromuscular junction (see Cupido et al. 1996), we also recorded responses to supramaximal nerve stimulation. Stimuli were delivered before a sustained isometric maximal contraction and immediately after it. The corticospinal stimuli were deliberately given during muscle relaxation because changes in motoneurone discharge rates with fatigue make it impossible to match the intracellular firing trajectories of motoneurones during voluntary contraction.
The responses to transmastoid stimulation are initially reduced and later facilitated after a voluntary contraction. This initial reduction requires voluntary activity in descending motor paths and is not produced simply by motoneuronal activation. Preliminary findings have been reported (Petersen et al. 1998; Gandevia et al. 1999).
- Top of page
The maximal voluntary torques for elbow flexion in the male and female subjects were within predicted normal ranges (e.g. Allen et al. 1994). During the sustained maximal voluntary contractions (MVCs) lasting 2 min, maximal torque declined by 65 ± 1 %, whereas it declined minimally for MVCs of 5 s duration. Sets of stimuli to descending motor paths (electrical stimulation between transmastoid electrodes), the motor cortex and the brachial plexus (at Erb's point) were delivered to the relaxed muscles before and after a test maximal contraction. The sets of stimuli were arranged in such a way that the responses to transmastoid stimulation could be corrected for any changes in size of the maximal M-wave (see Methods).
Transmastoid stimulation evoked compound muscle action potentials in the elbow flexor muscles with onset latencies of 7.8 ± 0.2 ms for biceps brachii and 10.2 ± 0.2 ms for brachioradialis. The amplitude of the responses increased during a weak voluntary contraction of the elbow flexors. Such behaviour is consistent with activation of rapidly conducting paths to motoneurones and inconsistent with direct activation of motor axons. The latter possibility was excluded in each study because the response latencies to transmastoid stimulation decreased by about 2 ms if stimulus intensities were increased from the usual stimulus level of ∼500 V to 750 V (see Methods). At the intensities used for the main studies, responses were larger (relative to the maximal M-wave) in biceps than brachioradialis.
Changes in the responses to transmastoid stimulation following muscle contraction
Figure 2 shows typical responses to transmastoid stimulation (i.e. cervicomedullary motor evoked potentials, CMEPs), motor cortical stimulation (i.e. motor cortical evoked potentials, MEPs), and the maximal M-wave. Voluntary contraction of the elbow flexors on one side markedly depressed their responses to transmastoid stimulation. This post-contraction depression of the response to transmastoid stimulation was evident when tested as early as 2–5 s after the contraction and occurred in all subjects. The depression did not occur for the homologous muscles on the non-contracting side (Fig. 2). In contrast, voluntary contraction consistently increased the size of the maximal M-wave in muscles involved in the MVC (Cupido et al. 1996; see also Taylor et al. 1999).
Changes in absolute area of the responses to transmastoid stimulation and those to supramaximal stimulation of the motor nerve are pooled for the group of subjects in Fig. 3. The growth of the M-wave reached its peak about 1–2 min after the contraction and recovered towards the control value after about 7 min. By contrast, the responses to transmastoid stimulation were maximally depressed immediately after the 2 min MVC and recovered to control levels about 2 min after the contraction. The pattern of changes was similar for the biceps and brachioradialis muscles. Because changes in the magnitude of the M-wave must influence the apparent efficacy of responses elicited by transmastoid and motor cortical stimulation, in subsequent analyses we expressed the size of responses relative to the maximal M-wave recorded in the same set of stimuli (see Methods).
Figure 3. Responses to supramaximal stimulation of the motor nerve (M-wave) and to transmastoid stimulation before and after a 2 min MVC
Absolute size of the responses elicited in brachioradialis by supramaximal stimulation of the motor nerve (‘M-wave’, •) and by transmastoid stimulation (i.e. cervicomedullary motor evoked response, ‘CMEP’, ○). The MVC lasted 2 min and is shown by the shaded area. Time ‘zero’ corresponds to the end of the MVC. Means ±s.e.m. for 7 subjects. Following the contraction the M-wave increased above control levels but the response to transmastoid stimulation was depressed. A similar pattern of change was observed for recordings from biceps brachii (not shown).
Download figure to PowerPoint
To determine whether the size of the depression of the responses to transmastoid stimulation was directly related to the duration of the contraction (and hence the metabolic and cardiorespiratory demands), we measured responses when the duration of the contraction was reduced from 2 min to 5 s (Fig. 4, right panels). In these studies, the first assessment of the responses to transmastoid stimulation occurred 2 s after the contraction ended. Again, the responses were markedly depressed in the motoneurones of biceps and brachioradialis and they recovered to pre-contraction levels with a similar time course to those after MVCs lasting 2 min (Fig. 4, left panels). Additional studies showed that some depression of responses occurred even with contractions lasting 1 s.
Figure 4. Grouped data for the responses to transmastoid stimulation before and after MVCs
Upper panels show the normalized responses (CMEPs) from brachioradialis and the lower panels show the responses from biceps brachii. Data for the 2 min MVC on the left and for the 5 s MVC on the right. Because the compound muscle action potential can change as a result of the voluntary contraction, responses to transmastoid stimulation have been normalized to the area of the corresponding maximal M-wave (Mmax) recorded in the same set of stimuli. The shaded region indicates the period of the MVC. Means ±s.e.m. (n= 7). Time ‘zero’ corresponds to the end of the MVC.
Download figure to PowerPoint
One potentially confounding factor was that the absolute size of the initial response to transmastoid stimulation was usually greater than that of the MEP (see below) so that the same motoneurones were not sampled with both descending inputs. However, the post-contraction depression of the CMEP still occurred when the two stimuli activated about the same fraction of the biceps motoneurone pool. An example is shown in Fig. 5.
Figure 5. Matched responses in biceps with transmastoid and motor cortical stimulation
Responses from one subject in whom the intensity of the transmastoid stimulus was adjusted so that the size of the initial CMEP approximated the size of the MEP produced by motor cortical stimulation (≈5–10 % of the maximal M-wave). Responses are shown (superimposed) for sets of stimuli before and at various times after a sustained MVC of 2 min duration. Immediately after the MVC, the responses to transmastoid stimulation were reduced while those to motor cortical stimulation were initially enhanced and then declined. Note that the largest MEPs occurred immediately after the MVC, when the CMEPs were almost abolished.
Download figure to PowerPoint
To examine whether the post-contraction depression of responses to transmastoid stimulation simply depended on activation of the motoneurones rather than descending voluntary inputs to them, we repeated the study but activated the motoneurones using supramaximal tetanic stimulation of the peripheral nerve. In one experiment three subjects made MVCs of 10 s duration, and, in the other, the motor nerve to biceps brachii was stimulated for 10 s at 30 Hz in the same subjects while they made no voluntary effort to contract. Results for both experiments are compared in Fig. 6. As expected from the depression after 5 s MVCs, responses to transmastoid stimulation decreased after the 10 s MVC and did not return to control levels until 2 min after the contraction. However, the responses to transmastoid stimulation showed no depression when the muscle and its motoneurones were activated as a result of tetanic peripheral stimulation at 30 Hz. The simplest explanation is that post-contraction depression of responses to transmastoid stimulation reflects changes in the descending motor pathway at a pre-motoneuronal level (see Discussion).
Figure 6. Comparison of responses in biceps to transmastoid stimulation after a voluntary contraction and after a tetanic contraction produced by stimulation of its motor nerve
Responses to transmastoid stimulation for 3 subjects after a 10 s MVC (○) and after a 10 s tetanus at 30 Hz (•). Means ±s.e.m. The initial depression of responses to transmastoid stimulation was absent for the tetanic but not the voluntary contraction. This suggests that activation of the motoneurones through peripheral nerve stimulation is not sufficient to depress their responsiveness to a descending volley produced by transmastoid stimulation.
Download figure to PowerPoint
Apart from the obvious post-contraction depression of the responses to transmastoid stimulation, we noted a small long-lasting facilitation after the 2 min MVC (e.g. Fig. 4). This was also evident after supramaximal stimulation of the motor nerve lasting 10 s (Fig. 6). In the main studies some facilitation also occurred on the non-contracting side (Fig. 2, left panels).
Changes in the responses to transcranial magnetic stimulation of the motor cortex
Following a maximal voluntary effort, the response to stimulation of the motor cortex decreased. However, the reduction had a different time course from that with transmastoid stimulation. After a 2 min MVC, responses were not initially reduced but declined over about 2 min to reach a stable depressed level of 30–40 % of the initial control values (Fig. 7A). This depression was still present 12 min after the contraction. With a 5 s MVC the changes were more variable but there was significant depression over the first 2 min after the contraction (Fig. 7B). Because responses to motor cortical stimulation may be affected by the state of spinal excitability revealed by the responses to transmastoid stimulation, the data have been expressed as a ratio in Fig. 7C. When the changes at a spinal level are taken into account, the responses to motor cortical stimulation are initially markedly increased above control levels but then depressed for many minutes, especially following the longer maximal contractions.
Figure 7. Responses in biceps to transcranial magnetic stimulation of the cortex after MVCs
Responses are shown before and after MVCs of the elbow flexor muscles. Means ±s.e.m. Shaded area denotes the duration of the MVC in each panel. A, subjects (n= 7) performed a 2 min MVC. B, subjects (n= 4) performed a 5 s MVC. C, data for the 2 min MVC (•) and 5 s MVC (○), with the responses to transcranial magnetic stimulation of the cortex expressed as a percentage of the corresponding responses to transmastoid (cervicomedullary) stimulation and normalized to the control values before the contraction. Data for pairs of stimuli have been averaged.
Download figure to PowerPoint
Combined transmastoid and motor cortical stimulation
Transcranial magnetic stimulation activates the corticospinal output particularly to distal muscles of the upper limb (e.g. Palmer & Ashby, 1992). To determine whether the responses to transmastoid stimulation were mediated by the same axons, we delivered cortical stimuli, transmastoid stimuli, and both stimuli at different interstimulus intervals (Fig. 8). Subjects were relaxed. When the cortical stimuli were delivered 0–3 ms before the transmastoid stimuli, the combined response was no larger than that to the transmastoid stimulus alone and its latency was the same as that of the transmastoid response. This interaction was not present when the cortical stimuli were delivered at earlier intervals. The apparent occlusion (at 0–3 ms) is consistent with collision of a descending excitatory volley by an antidromic volley set up by the transmastoid stimulation. If this analysis is correct, the two forms of stimulation are likely to activate a common set of corticospinal axons to elbow flexor motoneurones. Of course, both forms of stimulation may also activate other neural paths.
Figure 8. Apparent occlusion of the responses in biceps with combined magnetic cortical and transmastoid stimuli
Responses from a typical recording in one subject for studies in which transcranial magnetic stimuli (Cortex), transmastoid (Transmastoid), and combined stimuli (Both) were applied. Responses were 20–30 % of the maximal M-wave. Each trace is the average of 3 responses. The lowest traces represent the subtraction of the response to transmastoid stimulation from the response to combined stimulation (Both - transmastoid). Vertical dashed lines indicate the timing of the cortical stimulus and the filled arrows indicate the timing of the transmastoid stimulus. When the cortical stimulus preceded the transmastoid stimulus by 4 ms the combined response was slightly facilitated. However, when the interstimulus interval was reduced to 3 ms, there was occlusion, with the response to combined stimulation being similar to the response to transmastoid stimulation alone. Vertical calibration: 2 mV.
Download figure to PowerPoint
- Top of page
The main novel finding is that immediately after a sustained voluntary contraction activation of descending motor tracts at the cervicomedullary level by an electrical stimulus delivered between the mastoids produced greatly reduced EMG responses in the muscles which had performed the contraction. This effect recovered within 2 min and was followed by a longer-lasting facilitation of the response. The post-contraction depression was demonstrated in responses that ranged from 5 to 50 % of the maximal M-wave (Figs 2, 4 and 5). The time course of the post-contraction depression was quite different from that produced by transcranial magnetic stimulation of the motor cortex. Before consideration of possible mechanisms for this phenomenon, several technical and measurement issues need to be resolved.
First, what is being activated by the transmastoid stimulation? The study by Ugawa and colleagues (1991) showed that the evoked EMG response to transmastoid stimulation was likely to arise from a single descending volley in corticospinal axons. The strongest evidence for this was the apparent occlusion of the responses to transmastoid stimuli with appropriately timed transcranial electrical stimulation. Our study confirmed this result for the corticospinal output to elbow flexor muscles. Overall, these results suggest that the depression of responses to transmastoid stimuli after a contraction is likely to involve corticospinal axons and their actions on motoneurones.
Second, the way in which EMG responses are quantified in studies of muscle fatigue needs scrutiny. There are variable changes in the maximal M-wave during repetitive stimulation and fatiguing exercise and these may take some time to recover (e.g. Cupido et al. 1996; Taylor et al. 1999). A sustained increase in the size of the maximal M-wave produced by supramaximal stimulation of the brachial plexus at Erb's point occurred after the 2 min MVC and took about 7 min to recover fully. This change did not occur for non-contracting muscles (Taylor et al. 1999; confirmed here). Thus, we usually used a normalization procedure so that the responses to transmastoid stimulation could be assessed in the light of any changes in the maximal M-wave. The depression in the response to transmastoid stimulation after a contraction was evident in raw traces (measured as either peak-to-peak amplitudes or areas; Figs 2, 3 and 5). However, to measure this effect, responses were normalized to the area of the maximal M-wave produced no more than 10 s before or after the transmastoid stimulation. This normalization is likely to be inaccurate in the few seconds immediately after the MVC when the responses to transmastoid stimulation were changing rapidly.
Because the post-contraction depression of responses to transmastoid stimulation was not seen after non-voluntary activation of motoneurones by electrical stimulation of the motor nerve, the depression probably has a premotoneuronal origin. It is likely that the tetanic stimulation will activate early-recruited motoneurones (of low voluntary threshold) trans-synaptically because their conduction velocity will be less than that of the excitatory Ia volley. Motoneurones of higher voluntary threshold will also be activated antidromically. It was not possible to mimic the normal activation of all motoneurones by an artificial form of stimulation. However, the frequency of electrical stimulation matched the mean maximal firing rates of motor units for the biceps during MVCs (Bellemare et al. 1983). Further evidence in favour of a premotoneuronal origin for the depression of responses to transmastoid stimulation comes from the observation that the depression was not influenced by maintenance of the muscle in an ischaemic state (for 2 min) after the contraction, a procedure which will produce sustained activation of its group III and IV afferents (Petersen et al. 1998). If the effect had been mediated by postsynaptic inhibition at a motoneuronal level produced by group III and IV afferents, then the post-contraction depression should not have recovered when the muscle was briefly held ischaemic.
There are at least two possible sites for the depression of responses to transmastoid stimulation. First, the changes may be occurring at corticospinal synapses on motoneurones. The properties of this synapse have been studied in detail in non-human primates (for review see Phillips & Porter, 1968; Porter & Lemon, 1993; see also Lawrence et al. 1985). While we are not aware that single responses have been assessed after a period of repetitive stimulation of the corticospinal system in animals, the EMG responses to single transcranial electrical stimuli are reduced immediately after repetitive magnetic stimulation of the human motor cortex (1 s at 10 Hz) (Modugno et al. 1998). By contrast, responses to transcranial magnetic stimulation were enhanced. Given that the response to weak transcranial electrical stimulation is likely to be dominated by the response to direct excitation of corticospinal axons (D-waves) in awake subjects (Rothwell et al. 1991), the pattern of changes after repetitive cortical activation is similar to that following voluntary contractions. A complex homosynaptic depression with a recovery time course of about 10 s has been described for the Ia afferent motoneurone synapse so that synaptic efficiency of inputs to motoneurones is not necessarily fixed (Hultborn et al. 1996; see also Curtis & Eccles, 1960; Crone & Nielsen, 1989; Lev-Tov & Pinco, 1992). However, a remarkable feature of the post-contraction depression reported here is its duration, which is much longer than that reported for effects at the spindle afferent synapse on human motoneurones.
Second, the absolute size of the evoked descending volley may decline because the excitability of corticospinal axons is reduced due to activity-dependent changes produced by voluntary action. This possibility is hard to assess directly, as it is not possible to record the relevant descending volley. However, an activity-dependent hyperpolarization of peripheral motor axons occurs as a result of voluntary contractions (e.g. Bergmans, 1970; Vagg et al. 1998). Variable changes including increases and decreases in the excitability of single corticospinal axons were reported in monkeys doing voluntary tasks (Schmied & Fetz, 1987). However, there is some evidence that a simple activity-dependent change in axonal behaviour is not responsible for the post-contraction depression, but it is not conclusive. The post-contraction depression of responses to corticospinal stimulation reported here was not directly proportional to the duration of the contraction. Indeed, the depression was surprisingly similar for contractions lasting between 5 and 120 s. In addition, the depression was absent if tested during brief voluntary contractions after the sustained conditioning MVC, rather than at rest as in the present study (Butler et al. 1999).
As expected, we found that responses to transcranial magnetic stimulation of the motor cortex were not depressed immediately after contractions but became depressed and remained depressed for many minutes following sustained fatiguing MVCs (see Introduction and Fig. 7A). The different behaviour of the responses to transmastoid and motor cortical stimulation after contractions not only occurred under the standard testing conditions (e.g. Fig. 2) but also when the initial response to transmastoid stimulation was decreased to match that to motor cortical stimulation. If recruitment of motoneurones by the two forms of stimulation begins with low-threshold motoneurones, then the post-contraction depression of the response to transmastoid stimulation is unlikely to result (for example) from selective inhibition of high-threshold motoneurones. Hence, given our evidence that there may be substantial changes occurring ‘downstream’ of the cell bodies in the motor cortex, the present results highlight the difficulty in interpretation of changes in responses to motor cortical stimulation. The possibility that the corticospinal connection with motoneurones shows large activity-dependent changes following voluntary activity requires further investigation.