Evidence that a transcortical pathway contributes to stretch reflexes in the tibialis anterior muscle in man

Authors


Corresponding author J. Nielsen: Physiologisches Institut, Christian-Albrechts-Universität zu Kiel, Olshausenstrasse 40–60, 24098 Kiel, Germany. Email: j.b.nielsen@physiologie.uni-kiel.de

Abstract

  • 1In human subjects, stretch applied to ankle dorsiflexors elicited three bursts of reflex activity in the tibialis anterior (TA) muscle (labelled M1, M2 and M3) at mean onset latencies of 44, 69 and 95 ms, respectively. The possibility that the later of these reflex bursts is mediated by a transcortical pathway was investigated.
  • 2The stretch evoked a cerebral potential recorded from the somatosensory cortex at a mean onset latency of 47 ms in nine subjects. In the same subjects a compound motor-evoked potential (MEP) in the TA muscle, evoked by magnetic stimulation of the motor cortex, had a mean onset latency of 32 ms. The M1 and the M2 reflexes thus had too short a latency to be caused by a transcortical pathway (minimum latency, 79 ms (47 + 32)), whereas the later part of the M2 and all of the M3 reflex had a sufficiently long latency.
  • 3When the transcranial magnetic stimulation was timed so that the MEP arrived in the TA muscle at the same time as the M1 or M2 reflexes, no extra increase in the potential was observed. However, when the MEP arrived at the same time as the M3 reflex a significant (P < 0.01) extra-facilitation was observed in all twelve subjects investigated.
  • 4Peaks evoked by transcranial magnetic stimulation in the post-stimulus time histogram of the discharge probability of single TA motor units (n= 28) were strongly facilitated when they occurred at the same time as the M3 response. This was not the case for the first peaks evoked by electrical transcranial stimulation in any of nine units investigated.
  • 5We suggest that these findings are explained by an increased cortical excitability following TA stretch and that this supports the hypothesis that the M3 response in the TA muscle is - at least partly - mediated by a transcortical reflex.

In human subjects, stretch of an active muscle evokes a series of electromyographic reflex bursts (Lee & Tatton, 1975; Sinkjær et al. 1988; Toft et al. 1991). The earliest of the reflex peaks (termed M1 in this study) is known to be mediated dominantly by the monosynaptic I a pathway to the spinal motoneurones. For a long time there has been considerable controversy over the mechanism for the later occurring reflex peaks, but at least for muscles in the distal upper limb it now seems well established that the M2 reflex (or long-latency reflex) is mediated by a transcortical reflex pathway (Matthews et al. 1990; Day et al. 1991; Capaday et al. 1991; Palmer & Ashby, 1992). For more proximal muscles and lower limb muscles the controversy, however, still persists. Thilmann et al. (1991) have argued that M2 responses only disappear in hand muscles in patients with lesions of supraspinal pathways, whereas M2 responses in proximal arm and lower limb muscles persist. In contrast to findings in distal arm muscles, M2 responses are also not seen in contralateral proximal arm muscles in patients with mirror movement (Fellows et al. 1996). Finally, convincing evidence is now accumulating that the M2 response in the ankle muscles is mediated by group II afferents and that it is probably spinal in origin (Corna et al. 1995; Schieppati & Nardone, 1997).

Does this mean that the transcortical reflex loops play no role in the control of lower limb and proximal arm muscles? This would be in agreement with the long held view that the function of the hand depends largely on direct cortical control, whereas lower limb muscles are assumed to be controlled to a larger extent by brainstem and spinal mechanisms. However, in ankle dorsiflexors a third reflex burst, termed M3, may be observed. This reflex burst is mainly seen during active voluntary contraction (Toft et al. 1989) and has a latency which seems to be compatible with a ‘long-loop’ transcortical reflex pathway. The present study examined this possibility by measuring the effect of muscle stretch on muscular responses in the tibialis anterior (TA) muscle to magnetic and electrical stimulation of the motor cortex. This method has been used to suggest that the M2 response in distal muscles in the upper limb is mediated by a transcortical pathway (Day et al. 1991; Palmer & Ashby, 1992). A preliminary account of the work has been published in abstract form (Petersen et al. 1996).

METHODS

General experimental set-up

The experiments were performed on seventeen healthy subjects (aged 25–37 years). All the subjects gave informed consent to the experimental procedure, which was approved by the local ethics committee.

The subjects were seated in an armchair with the right leg semiflexed at the hip (120 deg), the knee flexed to 160 deg and the ankle in 110 deg plantar flexion. The left foot was attached to a footplate which could be rotated by a motor (Sinkjær et al. 1988). At the start of each experiment the subjects' maximal voluntary dorsiflexion effort was measured as the maximal dorsiflexion torque that they could exert on the footplate. The torque measured during the recordings was expressed as a percentage of this maximal voluntary dorsiflexion torque.

Stretch reflexes

Stretch reflexes evoked by a quick stretch of the ankle dorsiflexors were measured in the form of rectified electromyograms (EMGs) from the TA muscle using bipolar non-polarizable surface Ag-AgCl disc electrodes (1 cm2, 1 cm distance between poles). The axis of rotation of the ankle joint was aligned to the axis of rotation of the footplate. The force from the flexor and extensor muscles around the ankle joint was measured as the torque on the footplate using strain gauges. The position of the footplate was measured by a potentiometer. In most experiments the imposed stretch had an amplitude of 4 deg, a rise time of 40 ms and a maximal velocity of 170 deg s−1 (Toft et al. 1989). The duration of the stretch was 500 ms. The size of the stretch reflexes was measured as the area under the full-wave rectified signal. The beginning and end of the different reflex components were determined visually with the use of markers. In experiments on single motor units a stretch with an amplitude of 1 deg and a rise time of 50–60 ms was used, since selective recording from a single unit was not possible with larger and faster stretches.

Magnetic stimulation

Magnetic stimulation (Magstim 200; The Magstim Company Ltd, UK) was applied over the motor cortex. At the beginning of each experiment the position of the coil (a figure-of-eight prototype) was systematically changed to find the optimum location for activation of the TA muscle. In general a position 1–2 cm lateral to the vertex was chosen. The intensity of the stimulation was initially adjusted to be just above the threshold for evoking a compound motor potential (MEP) in the TA muscle during a steady voluntary dorsiflexion (approximately 10 % of maximal voluntary effort). In experiments on single motor units the intensity of the stimulation was adjusted to be just below the threshold for the compound MEP. The intensity of the stimulation was expressed as a percentage of the maximal stimulator output.

In nine experiments the brain was stimulated electrically. The cathode was placed 5 cm in front of the vertex and the anode was placed 2 cm lateral to the vertex. A Digitimer D180A stimulator (Digitimer Co. Ltd, Welwyn Garden City, UK) was used. The intensity of the stimulation was expressed as a percentage of the maximal stimulator output and adjusted in the same way as for the magnetic stimulation.

In some experiments MEPs were elicited to coincide with different components of the stretch reflexes. In these experiments stretch reflexes and MEPs were elicited either separately or in combination. The interval between the stretch and the transcranial magnetic stimulation was varied in each experiment between 10 and 100 ms at 5–10 ms steps. At least twenty responses of each alternative (stretch alone, transcranial magnetic stimulation alone, combined stretch and transcranial magnetic stimulation) were randomly alternated for each inter-stimulus interval. The averaged signals were stored on a computer (5 kHz sampling rate) and the sizes of the responses were measured as the area under the full-wave rectified signal. Student's t test was used to test whether combined stretch and transcranial magnetic stimulation evoked a larger response than the algebraic sum of the responses to separate stretch and transcranial magnetic stimulation.

Post-stimulus time histogram (PSTH)

This technique has been described in detail elsewhere (Fournier et al. 1986) and will only be briefly summarized here.

The firing probability of single TA motor units was measured following stimulation of either the motor cortex (magnetic or electrical) or stretch of the ankle dorsiflexors. The activity of the motor units was recorded by monopolar needle electrodes (type 13K88; Dantec, Denmark), amplified (1000–5000 times) and bandpass filtered (5 Hz to 1 kHz). The subjects were provided with visual and auditory feedback of the discharge of the unit. They were instructed to maintain a steady discharge of the unit, but they were not given any instructions as to the specific discharge rate. All recorded units were recruited at a low contraction level (around 1–5 % of maximal voluntary dorsiflexion effort) and the discharge rate during the recording was relatively low (7–10 Hz).

To reduce the number of stimuli necessary to evoke significant peaks or troughs in the PSTH, the stimuli were delivered with a variable delay following the previous firing of the investigated motor unit in each experiment. The motor action potentials were converted into standard pulses by a spike processor (Digitimer D130; Digitimer Co. Ltd), which were then used to trigger a computer. At variable delays after the triggering the computer activated the different stimulators.

The probability of the discharge in the PSTH does not only reflect the effect of the stimulation, but also the membrane trajectory in the motoneurone during the interspike interval. PSTHs were therefore constructed for both the occurrence of spikes following stimulation and for the background firing probability, i.e. without stimulation. The histograms obtained in the control situation without stimulation were subtracted from the histograms obtained following stimulation. The PSTHs were constructed for the period 20–80 ms (1 ms bin width) following a stimulation. Measurements without stimulation or with different combinations of stimuli (stretch alone, transcranial magnetic stimulation alone, combined stretch and transcranial magnetic stimulation) were alternated randomly. This was also the case in experiments in which the effect of prior stretch on peaks in the PSTHs evoked by electrical and magnetic cortical stimulation was investigated. At least 100 triggers were sampled for each alternative. Stimulations were delivered every 2–3 s. The statistical significance of changes in the firing probability of the motor unit was evaluated by a χ2 test.

Somatosensory-evoked potentials (SEPs)

The cortical potentials evoked by the stretch of the ankle dorsiflexors were recorded by needle electrodes placed on the scalp of the subjects. The active electrode was placed 2 cm behind the vertex, whereas the indifferent electrode was placed 5 cm in front. The signals were amplified (50 000 times) and bandpass filtered (1 Hz to 1 kHz). At least 200 traces were averaged for a period from 50 ms before, until 150 ms after, the onset of the stretch.

Results are given as means ±s.d.

RESULTS

Reflex effects evoked by stretch of the ankle dorsiflexors

In all of the seventeen subjects investigated stretch of the ankle dorsiflexors (4 deg, 40 ms rise time) evoked three bursts of reflex activity during weak to moderate tonic dorsiflexion (approximately 10–30 % of maximal voluntary dorsiflexion effort) at mean onset latencies of 44 ± 2, 69 ± 6 and 95 ± 9 ms, respectively. An example from one subject is illustrated in Fig. 1C. These reflex bursts were named M1, M2 and M3 in accordance with Toft et al. (1989). In the subject used for the illustration in Fig. 1, the cerebral potentials (SEPs) evoked by the stretch had an onset latency of 46 ms (Fig. 1A), whereas the MEP evoked by transcranial magnetic stimulation in the TA muscle had a latency of 30 ms (Fig. 1B). The total conduction time of a possible transcortically mediated stretch reflex would thus be at least 76 ms, when the SEP latency (Fig. 1A) is used as a measure of afferent conduction time and the MEP as a measure of efferent conduction time (Fig. 1B). Since some central processing must be necessary, an extra delay of around 10 ms probably has to be added (Deuschl et al. 1989). It follows that a transcortical reflex would require at least 87 ms, which exceeds both the latency of M1 and M2 (latencies of 42 and 64 ms, respectively), but is less than the latency of M3 (latency, 94 ms; Fig. 1C). This calculation was made in a total of nine subjects. In these subjects, the SEPs and MEPs had mean onset latencies of 47 and 32 ms, respectively, giving an estimated minimum conduction time for a possible transcortical pathway of 79 ms (47 + 32 ms). In the same subjects, the M2 and M3 reflex responses had mean latencies of 69 and 94 ms, respectively. From these calculations a transcortical pathway can be rejected for M1 and at least the first part of M2, but it remains an open possibility for M3.

Figure 1.

Calculation of the minimum conduction time necessary for a transcortical stretch reflex in the tibialis anterior muscle

A, somatosensory cortical potential (amplified 50 000 times; SEP) evoked by stretching of the ankle dorsiflexors (4 deg, 40 ms rise time) during tonic dorsiflexion (20 % of maximal voluntary effort). The latency of the SEP (46 ms) is indicated by the double-arrow above the trace; 400 traces were averaged. B, motor-evoked potential (MEP) in the tibialis anterior (TA) EMG evoked by magnetic stimulation of the motor cortex. The latency of the MEP (30 ms) is indicated below the trace; 25 traces were averaged. C, reflex responses in the TA EMG after the same stretch of the dorsiflexors as in A. Three clearly distinct peaks are seen in the averaged EMG at latencies of 42 ms (labelled M1), 64 ms (labelled M2) and at 94 ms (labelled M3). The vertical dotted line marks the onset of M3. 400 traces were averaged. The ankle joint position is shown below the EMG trace.

Extra-facilitation of MEP when evoked simultaneously with M3

Transcranial magnetic stimulation was applied at different delays in relation to the stretch of the ankle dorsiflexors to evoke a MEP in the TA muscle at the same latency as M1 (Fig. 2D), at the same latency as M2 (Fig. 2E) or at the same latency as M3 (Fig. 2F). The intensity of the magnetic stimulation was adjusted so that it evoked only a very small MEP when applied alone (arrows in Fig. 2D–F). The MEP was therefore mainly witnessed from the subsequent silent period in the EMG. When the MEP was evoked 20 ms after the stretch (so that it occurred at the latency of M1) there was no change in the size of the MEP compared with when it was evoked without a previous stretch (compare Fig. 2G with Fig. 2A and D). This was also the case when the MEP was evoked at the latency of M2 (compare Fig. 2H with Fig. 2B and E; delay, 40 ms). However, when the MEP was evoked with a delay of 75 ms, so that it appeared at the same latency as M3, a pronounced increase in its size was observed (compare Fig. 2I with Fig. 2C and F). The size of the MEP when preceded by the stretch was divided by the algebraic sum of the reflex response and the MEP following separate stimulation for each of the different delays (Fig. 2J). As can be seen, the MEP when conditioned by the stretch had the same size as the algebraic sum of the potentials following separate stretch and magnetic stimulation until the magnetic stimulation was delayed by around 65 ms. This corresponds to the onset of M3 (latency, 95 ms) when 30 ms, which was the latency of the MEP in this subject, was added. Similar experiments were performed in twelve subjects. In all of them a significant (P < 0.01) extra-facilitation of the MEP was observed when it was evoked at the latency of M3. On average, the combination of M3 and the MEP was 346 ± 93 % larger than the algebraic sum of the two potentials when evoked separately. The corresponding values for M1 and M2 were 122 ± 9 and 125 ± 9 %, respectively. The latency of the extra-facilitation was, on average, 53 ± 8 ms in the twelve subjects, which is only a little longer than the mean latency of the SEP (47 ms, see above). The extra-facilitation was thus observed shortly after the arrival of the afferent volley at the cortical level, which is what should be expected if it is caused by increased cortical excitability.

Figure 2.

Evidence of extra-facilitation of magnetically induced TA MEP following stretch of ankle dorsiflexors

All data are from a single subject. The subject performed a tonic dorsiflexion corresponding to 20 % of the maximal voluntary dorsiflexion effort. A-C, averaged TA EMGs (n= 25) following stretch of the ankle dorsiflexors (4 deg, 40 ms rise time) that were used as conditioning stimulation to MEP. The three reflex responses M1, M2 and M3 are indicated by arrows. The stretch was given 20 ms (A), 40 ms (B), and 75 ms (C) before the magnetic stimulation. D–F, TA test MEPs (n= 25) evoked by magnetic stimulation of the motor cortex (45 % of maximal stimulator output). The intensity of the stimulation was adjusted so that it was just at the threshold for the MEP (indicated by arrows). Note that the traces in D and F are identical, because the effect of the stretch on the MEP at conditioning-test intervals of 20 and 75 ms was recorded in the same experimental run. G-I, effect of combined dorsiflexion stretch and magnetic cortical stimulation. The magnetic stimulation was delayed in relation to the stretch by 20 ms (G), 40 ms (H) and 75 ms (I). In this way, the MEPs were evoked at the time interval corresponding to the M1, M2 and M3 intervals, respectively. J, time course of the size of the TA MEP conditioned by previous stretch of the ankle dorsiflexors at different time intervals. The conditioning-test intervals were varied between 10 and 120 ms. The size of the conditioned MEP is expressed as a percentage of the algebraic sum of the stretch-induced reflex activity and the control MEP at corresponding time intervals. The filled circle indicates the measurement at a conditioning-test interval of 20 ms (corresponding to the traces in A, D and G); the open square the measurement at a conditioning-test interval of 40 ms (corresponding to the traces in B, E and H) and the open pentagon the measurement at a conditioning-test interval of 75 ms (corresponding to the traces in C, F and I). The asterisks indicate that the conditioned MEP was significantly different (P < 0.05) from the sum of the stretch-induced reflex activity and the control MEP.

Evidence of increased cortical excitability corresponding to M3 from single motor units

In order to investigate the effect of stretch of the ankle dorsiflexors on single TA motor units, it was necessary to decrease the amplitude of the stretch to just under the threshold for eliciting compound reflex potentials. In the different subjects this was obtained with stretches of 0.2–1 deg and rise times of 50–60 ms. Figure 3 demonstrates data from one subject. As was the case when studying the MEP, the stretch (1 deg, 50 ms rise time) was advanced in relation to the magnetic stimulus by 55 ms. The increased firing probability observed in the histogram in Fig. 3A at an interval from approximately 45 to 65 ms thus corresponds to M3 (latency after the stretch, 45 + 55 = 100 ms; in this subject the M3 reflex in the surface EMG had a latency of approximately 98 ms). The intensity of the transcranial magnetic stimulation was decreased well below MEP threshold so that it induced only a small increase in the discharge probability of the motor unit (Fig. 3B). In this unit the magnetic stimulation induced what could be two peaks of increased firing probability at latencies of 45 and 51 ms, respectively, each lasting 3–4 ms. These peaks may be explained by multiple volleys in the corticospinal fibres evoked by the magnetic stimulation (Day et al. 1989; Edgley et al. 1990, 1997; Burke et al. 1993; Nielsen et al. 1995). As can be seen in Fig. 3C, combination of the cortical stimulus and the muscle stretch resulted in a pronounced increase in the size of the first of these peaks. When the sum of the firing probability of the motor unit following separate stretch (Fig. 3A) and magnetic stimulation (Fig. 3B) was subtracted from the firing probability following combined stretch and magnetic stimulation (Fig. 3C), it could be seen that the combination of stretch and magnetic stimulation resulted in an increased firing probability which was much larger than that expected from their effects when applied separately (Fig. 3D; C - (A+B)). Notice also that the peak was somewhat broader than the peak induced by the magnetic stimulation alone. When the number of counts in each bin contained in the peak in the histogram in Fig. 3D (marked by vertical lines) was added together it was found that the combination of stretch and magnetic stimulation increased the number of counts by 38 counts per 100 triggers over that which should be expected from their separate effects.

Figure 3.

Evidence of extra-facilitation of magnetically induced PSTH peaks following stretch of ankle dorsiflexors

Post-stimulus time histograms (PSTHs) showing the changes in discharge probability of a single TA motor unit following separate or combined stretch of the ankle dorsiflexors and transcranial magnetic stimulation. The histograms were obtained by subtracting the discharge probability of the motor unit in a control situation without stimulation from the discharge probability following the stimulation. The stretch had a rise time of 50 ms and an amplitude of 1 deg. The cortical stimulation was applied using a magnetic stimulator at an intensity of 45 % of the maximal stimulator output. The magnetic stimulation was delayed by 55 ms in relation to the stretch. The histograms only show the time after the magnetic stimulation. In A, only stretch was applied; in B, only magnetic stimulation was applied; and in C, the two stimuli were combined. The histogram in D was obtained by adding the counts in A and B bin per bin (A+B) and subtracting the result of this calculation from the histogram in C (C - (A+B)). The two vertical dashed lines indicate the time interval during which the combined stretch and magnetic stimulation induced a larger increase of firing probability than the algebraic sum of their separate effects. When the number of counts in this time interval was calculated, it was found that the combined stretch and magnetic stimulation resulted in 38 counts per 100 triggers more than the algebraic sum of their separate effects. All data are from a single subject. The total number of triggers for each alternative was 100.

Similar calculations were done for twenty-eight units from seven subjects. On average, 8.6 more counts per 100 triggers (range: −2 to +38 counts per 100 triggers) were observed when stretch and magnetic stimulation were combined (with an interval between the stretch and magnetic stimulation of 55–70 ms) compared with when the two stimuli were applied separately. In nineteen of the twenty-eight motor units the peak evoked by combined stretch and magnetic stimulation was significantly (P < 0.05) larger than the algebraic sum of the effects induced by separate stretch and magnetic stimulation.

In contrast to this a significant increase of the magnetically evoked peaks was never observed in any of ten units from four subjects, when the delay of the magnetic stimulus was adjusted so that the peak occurred at the latency of M1 or M2.

Comparison of magnetic and electrical stimulation

In the motor unit used for the illustration in Fig. 4, magnetic stimulation evoked a peak at a latency of 38 ms and with a duration of 3–4 ms (Fig. 4B, left panel). When the magnetic stimulation was preceded by a weak stretch of the ankle dorsiflexors at a latency of 70 ms, this peak was very strongly facilitated (Fig. 4C, left panel). As was shown also in Fig. 3, this facilitation was much stronger than that expected from the effects of the stretch and magnetic stimulation when applied separately (Fig. 4D, left panel). When the corticospinal tract was instead stimulated electrically, a peak was observed at a latency of 35 ms and with a duration of 4 ms (Fig. 4B, right panel). No increase in this peak could be observed when the same stretch as in Fig. 4A–D (left panels) was applied (Fig. 4D, right panel). To take the different latency of the peaks evoked by magnetic and electrical stimulation into account the stretch was advanced by an extra 3 ms in Fig. 4A-D (right panels).

Figure 4.

Comparison of the effect of a stretch of the ankle dorsiflexors on magnetically (left panels) and electrically (right panels) induced PSTH peaks

Histograms are organized as in Fig. 3. All data are from a single motor unit The histograms show the changes in firing probability after the cortical stimuli. A, changes in the firing probability of the motor unit following stretch of the ankle dorsiflexors (1 deg, 50 ms rise time). B, changes in the firing probability of the motor unit following cortical stimulation. C, changes in firing probability following combined stretch and cortical stimulation. In B and C, left panels show magnetic stimulation (45 % of maximal stimulator output) and right panels electrical stimulation (30 % of maximal stimulator output). D, histogram obtained by adding the counts in A and B bin per bin and subtracting the result of this calculation from the histogram in C. The stretch was advanced by 70 ms in relation to the magnetic cortical stimulation and by 73 ms in relation to the electrical cortical stimulation. The total number of triggers was 100 for each alternative in all cases.

In all of nine motor units from four subjects in which the effects of magnetic and electrical brain stimulation were compared, an increase in the peaks evoked by magnetic stimulation was observed following stretch at the latency of M3. In contrast, the earlier occurring peaks evoked by electrical transcranial stimulation were never seen to increase. On average, the magnetically evoked peaks were (significantly; P < 0.01) facilitated by 9.8 counts per 100 triggers, whereas the electrically evoked peaks were (not significantly; P > 0.1) depressed by −1.2 counts per 100 triggers.

DISCUSSION

This study has demonstrated that TA MEPs and peaks of increased firing probability in the PSTH of single TA motor units evoked by magnetic stimulation of the motor cortex are facilitated corresponding to the M3 reflex response evoked by stretch of the TA muscle. Several arguments suggest that this observation is best explained by an increased cortical excitability induced by the muscle stretch.

It may, first, be taken as suggestive evidence that the onset of the M3 reflex corresponded to the latency expected for a transcortical reflex. When the latency of the cortical potential evoked by the stretch was added to the latency of the MEP in the TA muscle evoked by the magnetic stimulus, the minimum conduction time required for a transcortical reflex (mean onset latency, 79 ms) was found to be substantially longer than the M1 and M2 reflex responses, but reasonably shorter than the M3 response (mean onset latency, 94 ms). The 15 ms interval between the sum of the latencies of the SEP and MEP and the latency of the M3 response may be explained by the delay required for central processing of the afferent input, as suggested also for musculo-cutaneous, presumed transcortical, reflexes in the leg and distal arm (Marsden et al. 1976; Deuschl et al. 1989; Nielsen et al. 1997).

It was, second, observed that a significant extra-facilitation of the MEP was only present corresponding to the M3 reflex response, but not corresponding to the earlier occurring M1 and M2 reflex responses. If the facilitation had been caused simply by a non-linear summation of the stretch and descending inputs to the spinal motoneurones, a facilitation corresponding to all three reflex responses would have been expected. It may be argued against this reasoning that it rests on the assumption that the same motoneurones are recruited in the three reflex responses, which is not necessarily the case. However, and this is our third argument, we observed that magnetically evoked peaks in the PSTH of single motor units were also facilitated corresponding to M3, whereas there was no such facilitation at earlier intervals. This indicates that spinal motoneuronal mechanisms are not likely to explain the extra-facilitation of the responses following combined stretch and magnetic stimulation, but that the explanation is more likely to be found at a premotoneuronal, possibly cortical, site.

The peaks evoked in the PSTH by transcranial magnetic stimulation are usually assumed to be caused mainly by the direct corticomotoneuronal pathway (Rothwell et al. 1991), but a contribution from non-monosynaptic (indirect) pathways should not be disregarded (Nielsen et al. 1993; Burke et al. 1994; Gracies et al. 1994). However, we did find that the very first bin in the PSTH peaks (see Fig. 3) was already facilitated. It is unlikely that non-monosynaptic pathways should make a major contribution to the PSTH peak so shortly after its onset, and the most likely explanation of the increase in the PSTH peaks when stretch and magnetic stimulation were combined is therefore that the stretch induced an increased excitability at a cortical level.

This was strengthened very considerably by the observation that a similar facilitation was not seen for peaks in the PSTH evoked by electrical stimulation of the motor cortex. Electrical and magnetic stimulation of the corticospinal tract have been shown to activate the same corticospinal fibres, but at different sites (Edgley et al. 1990, 1997). Whereas the electrical stimulation easily penetrates deep into the brain and, even at weak stimulus intensities, activates the corticospinal axons in the white matter, the magnetic stimulation activates the cells indirectly or at a site close to or at the cell soma (Day et al. 1989; Edgley et al. 1990, 1997; Burke et al. 1993; Nielsen et al. 1995). The initial peaks in the PSTH induced by electrical cortical stimulation are therefore not sensitive to changes in the excitability of the cortical cells, whereas this is the case for the peaks induced by magnetic stimulation (Nielsen et al. 1993, 1997). The different effect of the muscle stretch on the peaks evoked by the two types of cortical stimulation is thus most easily explained by an increased cortical excitability following the stretch at the latency of the M3 reflex response.

This raises the possibility that a transcortical reflex pathway may at least make a contribution to the M3 reflex response. Around 20 years ago, Melvill Jones and co-workers similarly argued that a supraspinal pathway made a significant contribution to stretch reflex responses in the leg also during more functional tasks than the one studied here (Melvill Jones & Watt, 1971a, b; Chan, 1983). They used the term ‘the functional stretch reflex’ to designate the long-latency stretch reflex responses (latency around 120 ms) in the ankle plantar flexors, since they found these to be of greater functional significance in human subjects than the earlier spinal (monosynaptic) reflexes. Based on latency measurements, their ability to be modified by the voluntary intent of the subjects and their absence in patients with lesions of the corticospinal tract, it was suggested that the responses were caused by a supraspinal reflex pathway (Chan, 1983). There is thus a possibility that a transcortical reflex pathway similar to the one we have provided evidence for here in the case of ankle dorsiflexors is also involved in the generation of ‘the functional stretch reflex’ in ankle plantar flexors. It is, however, not clear to what extent our findings also apply for the ankle plantar flexors and during other tasks than the one we investigated and we therefore do not wish to put too much emphasis on a comparison between our findings and those of Melvill Jones and co-workers.

It should also be pointed out that we do not wish to suggest that M3 is exclusively mediated by a transcortical reflex pathway and that other (subcortical) pathways make no contribution. Indeed, the relative contribution of transcortical and other mechanisms may vary according to the situation, such as the voluntary intent of the subject, and it may change dramatically after lesions of the central nervous system. The demonstration of modulation of the M3 reflex response in one task compared with another (i.e. postural task as compared with voluntary movement), therefore, does not necessarily provide any information of the activity in the transcortical pathway.

Comparison with the upper limb

In the upper limb it has been demonstrated that the M2 response is probably transcortical (Day et al. 1991; Palmer & Ashby, 1992), whereas the data in the present study, using similar methodology, suggest that M3, but not the M2 response, is - at least partly - transcortical. This emphasizes that responses in different muscles are not comparable, as also suggested by Thilmann et al. (1991), but contrary to those authors we believe that this is not necessarily due to a difference in the organization of the control of the muscles, but rather to an unfortunate naming of the responses. What is called an M2 response in the upper limb thus seems to correspond, at least in terms of the underlying mechanism, to what is called an M3 response in the lower limb. Such misleading labelling of the responses is naturally bound to cause confusion and may also lead to wrong conclusions. Using the alternative labelling, short-latency, medium-latency and long-latency responses, also does not help much in this respect. From our point of view the most reasonable would therefore be to maintain the present labelling, while realizing that responses with the same name are not necessarily equivalent - at least when comparing the lower and upper limbs.

Implications for motor control

It is difficult to draw conclusions regarding the functional significance of the transcortical pathway to TA. One important reason for this is that it is not known whether the pathway we have described is closed, or perhaps even more facilitated, during tasks that are more ‘natural’ than the tonic contraction in sitting subjects investigated here. However, since M3 was generally of the same size as M1 and M2 in the precontracted muscle (see Fig. 1; see also Toft et al. 1989), it seems reasonable to assume that the reflex mediated by the pathway does make a rather substantial contribution to the regulation of the muscle and ankle joint stiffness at least in this particular task. We speculate that one advantage of a transcortical reflex pathway is that it allows more flexibility than a purely segmental pathway and that the automatic response to stretch may be set in advance according to motivational and environmental influences (see also Chan, 1983). In this context it is of interest that Marple-Horvat et al. (1993) found that corticospinal cells in the cat reacted strongly to unexpected perturbations at the onset of stance. In those experiments, the cats walked on a horizontal ladder, which requires a significant visumotor co-ordination, and it was suggested that the reaction of the corticospinal cells constitutes an automatic transcortical reaction when the support unexpectedly gives way. It may be speculated that the transcortical reflex that we have provided evidence for here serves a similar function as the reactions described by Marple-Horvat et al. (1993).

Acknowledgements

This work was supported by grants from the Danish Medical Research Council, the Danish Sports Research Council, the Danish Research Foundation and the Danish Society of Multiple Sclerosis.

Ancillary