Group II muscle afferents probably contribute to the medium latency soleus stretch reflex during walking in humans

Authors


Corresponding author M. J. Grey: Center for Sensory-Motor Interaction, Aalborg University, Fredrik Bajersvej 7-D3, DK-9220, Aalborg, Denmark. Email: mg@smi.auc.dk

Abstract

  • 1The objective of this study was to determine which afferents contribute to the medium latency response of the soleus stretch reflex resulting from an unexpected perturbation during human walking.
  • 2Fourteen healthy subjects walked on a treadmill at approximately 3.5 km h−1 with the left ankle attached to a portable stretching device. The soleus stretch reflex was elicited by applying small amplitude (∼8 deg) dorsiflexion perturbations 200 ms after heel contact.
  • 3Short and medium latency responses were observed with latencies of 55 ± 5 and 78 ± 6 ms, respectively. The short latency response was velocity sensitive (P < 0.001), while the medium latency response was not (P= 0.725).
  • 4Nerve cooling increased the delay of the medium latency component to a greater extent than that of the short latency component (P < 0.005).
  • 5Ischaemia strongly decreased the short latency component (P= 0.004), whereas the medium latency component was unchanged (P= 0.437).
  • 6Two hours after the ingestion of tizanidine, an α2-adrenergic receptor agonist known to selectively depress the transmission in the group II afferent pathway, the medium latency reflex was strongly depressed (P= 0.007), whereas the short latency component was unchanged (P= 0.653).
  • 7An ankle block with lidocaine hydrochloride was performed to suppress the cutaneous afferents of the foot and ankle. Neither the short (P= 0.453) nor medium (P= 0.310) latency reflexes were changed.
  • 8Our results support the hypothesis that, during walking the medium latency component of the stretch reflex resulting from an unexpected perturbation is contributed to by group II muscle afferents.

The central nervous system (CNS) takes advantage of a network of complex neural pathways and mechanisms in the control of normal human gait. One such mechanism is the use of afferent feedback from muscle, cutaneous and joint receptors. Our knowledge of the contribution of afferent information in human gait is still limited, although this has been an area of active research for many years (e.g. Dietz et al. 1985; Yang et al. 1991; Sinkjær et al. 1996). Yang et al. (1991) and Sinkjær et al. (1996) have shown that afferent-mediated feedback is used by the CNS in the control of gait when an unexpected stretch of the ankle extensors is imposed. More recently, Sinkjær et al. (2000) provided evidence that during walking, up to 50 % of the background EMG from the soleus muscle can be attributed to afferent feedback. However, the relative importance of the separate afferent pathways may differ for the background locomotor EMG and the EMG that results from an imposed stretch.

When the human soleus muscle is stretched in a seated subject, two distinct bursts, with average peak latencies of 59 and 86 ms are evident in the EMG (Toft et al. 1989). These bursts are often referred to as the short (SLR) and medium (MLR) reflex responses, respectively, and have also been labelled the M1 and M2 stretch reflex responses, respectively. The short latency response has an onset latency of approximately 40 ms and is attributed to monosynaptic excitation of spinal motoneurones from the large diameter group Ia afferent fibres (Taylor et al. 1985; Matthews, 1991). The medium latency response may be mediated by group II afferents through an oligosynaptic spinal pathway (Berger et al. 1984; Dietz et al. 1985, 1992; Nardone et al. 1996), and possibly via group Ib afferents (Dietz et al. 1998). In addition, evidence suggesting that group II afferents contribute to the medium latency response during standing has also been reported (Dietz et al. 1985; Corna et al. 1995; Nardone et al. 1996). Dietz et al. (1985) originally proposed that group II afferents mediate the medium latency response during walking, although the evidence at that time in support of this suggestion was indirect. Evidence for the involvement of group Ib force-sensitive afferents has also been shown in the cat (Duysens & Pearson, 1980; Pearson & Collins, 1993), and may contribute to the regulation of human stance (Dietz & Colombo, 1996; Dietz, 1998) and gait (Stephens & Yang, 1999).

The purpose of this study was to determine which afferents mediate the medium latency reflex response in the soleus electromyogram when the ankle is perturbed during walking. This was investigated using five independent techniques. The first technique exploited the velocity sensitivity of the afferents. Given that group Ia fibres are known to be velocity sensitive, whereas group II fibres are more amplitude sensitive (Houk & Rymer, 1981), it can be expected that the short latency stretch reflex during walking will increase when a faster perturbation is imposed onto the muscle, whereas the medium latency response will depend less on stretch velocity. The second technique involved nerve cooling similar to that used in standing subjects (Schieppati & Nardone, 1997); this technique was employed to investigate whether or not the short and medium latency components of the stretch reflex are mediated by the same afferents. The technique exploits the different diameters of the fibres. When the fibres are cooled their conduction velocity is slowed to an extent that is inversely proportional to their diameter (Paintal, 1965); consequently, the small diameter group II fibres are slowed more readily than are the large diameter group Ia fibres. Thirdly, an ischaemic block of the large diameter afferent fibres was performed. This method also exploits the different fibre diameters; the larger diameter Ia fibres are blocked before the smaller fibres. This experiment was conducted in order to exclude the possibility that group Ia afferents make an important contribution to the medium latency component. Fourthly, the α2-adrenergic receptor agonist tizanidine was used to selectively depress transmission in the polysynaptic spinal pathway from group II afferents. The effect of tizanidine on afferent pathways has been investigated by recording spinal focal field potentials resulting from electrical stimulation of the quadriceps nerve in the cat (Bras et al. 1989, 1990; Skoog, 1996). While tizanidine was shown to strongly depress the potentials from group II afferents, no effect was shown on group I afferents. In the fifth technique a regional ankle block was generated using a local anaesthetic to depress the transmission of cutaneous afferents in the foot and ankle. This experiment was conducted in order to exclude the possibility that cutaneous afferents from the foot and ankle make an important contribution to the medium latency component.

METHODS

Fourteen healthy subjects (9 males and 5 females, age 22-44 years), with no history of neuromuscular disorder, participated in this study. Subjects gave their written informed consent prior to their participation and the local ethics committee approved the experiments. All experiments were conducted in accordance with the Declaration of Helsinki.

Apparatus and instrumentation

Subjects were instrumented with bipolar surface EMG electrodes over the soleus and tibialis anterior muscles of their left leg. EMG signals were amplified and bandpass filtered from 20 Hz to 1 kHz. Perturbations were imposed using a portable stretching device capable of rotating the ankle joint while the subjects walked on a treadmill. Complete details regarding the mechanics of the portable stretching device are reported elsewhere (Andersen & Sinkjær, 1995). Briefly, the device consisted of a functional joint attached to the subject's left ankle joint with a polypropylene plaster cast and connected to a powerful AC servomotor via flexible Bowden cables. The angle of the ankle was measured with an optical encoder incorporated within the portable stretcher. All data were sampled at 2 kHz and the ankle velocity was determined off-line by the differentiation of the ankle angle record.

Stretch reflex as a function of stretch velocity

Each subject walked with a natural cadence between 3.5 and 4 km h−1 for an adaptation period of approximately 5 min. During this period they chose a comfortable walking speed which was maintained for the remainder of the experiment. Dorsiflexion perturbations of 8 deg were generated by the portable stretching device during stance 200 ms after heel contact. Stretches were applied pseudo-randomly every three to five steps. EMG activity was recorded from the soleus and tibialis anterior muscles for 1000 ms, starting 200 ms prior to the stretch. A control step was recorded immediately prior to each perturbed step. Perturbations were applied for a particular set until 15-20 records were acquired. The perturbation velocity was then changed and the protocol was repeated. Each subject was presented with 8 to 12 perturbation velocities in a range between 85 and 300 deg s−1 relative to the ongoing ankle angular velocity.

Nerve cooling

In eight subjects, nerve cooling was effected by wrapping large diameter plastic tubing around the thigh from the groin to the knee, whilst circulating an ice-cold mixture of water and antifreeze through the tubing (maintained at 0 ± 5 °C). Skin temperature and coolant temperature were monitored throughout the experiment. Baseline stretch reflexes were measured prior to the cooling by generating constant amplitude and velocity perturbations at (8 deg, 300 deg s−1) at 200 ms following heel contact. Subjects remained seated while the cooling was applied. On average, the skin temperature was maintained at 12 ± 2 °C. To avoid warming by the surrounding air, the tubing was insulated by wrapping the thigh in thick foam. After 1.5-2 h of cooling, the foam insulation was removed and the subjects walked at the same velocity as before and were exposed to the same perturbations. When the subjects stood up and walked, the increased circulation of blood from the trunk to the leg progressively warmed the nerve. To decrease this warming effect, the coolant circulation was maintained as the subject was walking. In addition, data collection was discontinued within 5 min of the subject standing up,

Ischaemic block

In four subjects, an ischaemic nerve block was generated by an occlusion of the blood supply with a tourniquet. The block was induced by applying a pneumatic cuff approximately 10 cm above the knee and inflating it to a pressure of 240 mmHg. The subject sat with the knee held at 90 deg while the cuff was applied and inflated. The effect of the ischaemia was monitored by eliciting an M-wave, H-reflex and a stretch reflex in the soleus muscle at regular intervals. The subject remained seated until the H-reflex and stretch reflex amplitudes had decreased to less than 10 % of the baseline amplitude while the M-wave was unchanged. Typically a period of 15-25 min of ischaemia was required to block the Ia afferents to this level. When the stretch reflex amplitude decreased to the required level the subject walked on the treadmill with the same cadence as during normal walking. Constant amplitude and velocity perturbations (8 deg, 300 deg s−1) were presented 200 ms after heel contact. A selective ischaemic block could be applied for only a short time, consequently fewer records (8-10) were recorded compared with normal walking. The experiment was terminated when the subject could no longer walk at the required speed or when the amplitude of the M-wave decreased.

Tizanidine depression of group II afferents

In three subjects a depression of the pathway mediating the group II afferents was achieved with the use of tizanidine. Each subject was given an oral dose of fast acting tizanidine (Sirdalud, 150 μg kg−1). The ankle was perturbed whilst walking in a similar manner to that described above. Twenty constant amplitude and velocity perturbations (8 deg, 300 deg s−1) were presented 200 ms after heel contact as the subjects walked on the treadmill. Data records were recorded before the ingestion of tizanidine and then again 2 h later, when the action of the drug was fully effective.

Anaesthetic depression of cutaneous afferents

In three subjects, the transmission of cutaneous afferents from the foot and ankle was blocked by an injection of lidocaine (lignocaine) hydrochloride solution (250-500 mg) around the nerves supplying the skin of the foot and ankle. The efficacy of the block was evaluated by the subject's sensation from the skin (tested with the hub and point of a needle) as well as somatosensory evoked potentials (SEPs). Surface electrical stimulation was applied to three areas of the foot (anterior, mediodorsal and laterodorsal aspects) covering the three main sensory areas of the foot. A series of 500 stimuli were delivered at 2 Hz to each of these areas at an intensity corresponding to two times the paraesthetic threshold while EEG was recorded (31 channels: bandpass, 0.05-500 Hz; sampling rate, 2 kHz, bilateral-ears reference). A combination of global field power and visual inspection of the topographies were used for isolating the peak maxima and electrode sites. SEPs were recorded for each of the three areas and compared to the pre-block measures.

When the block was considered efficient (i.e. when the subject was no longer able to feel anything from any part of the skin on the foot and ankle and the SEPs were depressed), the subject was instructed to walk and the ankle was perturbed as before. Twenty constant amplitude and velocity perturbations (8 deg, 300 deg s−1) were presented 200 ms after heel contact. Data records were recorded before and after the ankle block.

Data analysis

Signal processing and analysis were carried out off-line. The EMG recordings were rectified and filtered with a 20 Hz first-order low pass filter to extract an amplitude envelope. For the cooling experiments, a 40 Hz low pass filter was used for better resolution of the peaks in the reflex responses allowing more accurate latency measurements. The individual records for a particular trial were ensemble averaged producing a single record for each perturbation velocity.

The peak and onset latencies of the soleus stretch reflex were determined by visual inspection using a cursor on the display. The onset latency was defined as the first major deflection in the EMG record following the perturbation. Unlike the short latency response, the onset of the medium latency response is not always easily defined. Because the EMG does not always drop to the level of background EMG, the beginning of this response is often difficult to determine. Consequently we did not attempt to measure the onset latency of the medium latency response.

The area under the EMG curve was calculated over a 50 ms window starting at the reflex onset. The area under the curve for the corresponding averaged control step was calculated using the same window. The increased area attributable to the stretch-mediated response was determined by subtracting the perturbed area from the control area. This increase was expressed as a percentage change with respect to the control area.

The 50 ms window provided a measure of the overall reflex response. To examine the short and medium latency components of the stretch reflex response, two 20 ms windows were defined as shown in Fig. 1. The first window started at the onset of the short latency stretch reflex and the second window started 30 ms later. The 10 ms division between the two windows ensured a clear separation. The short and medium EMG responses were analysed in the same manner as for the 50 ms window.

Figure 1.

Determination of windows defining the short and medium latency muscle responses to a stretch in the soleus EMG

The beginning of the short latency response (SLR) window was placed coincidental with the onset of the stretch reflex response. Each window was 20 ms wide. The beginning of the medium latency response (MLR) window was placed 10 ms after the end of the short latency window. The reflex responses were defined by the area between the perturbed record (thick line) and control record (thin line). The time base in this figure is presented such that zero corresponds with the stretch onset.

Statistical analysis

A linear regression analysis was carried out to determine the relationship between the area increments and relative perturbation velocities for all subjects. A relationship was determined to be significant if the regression produced a slope that was different from zero. The slopes were then pooled and Student's paired t test was performed to determine if the short latency response was greater than the medium latency response. For the cooling experiment, a one-way repeated measures analysis of variance test (ANOVA) was performed on the differences between the latencies of the two responses, i.e. (MLR-SLR)before and (MLR-SLR)after. Two-way repeated measures ANOVA tests were used to determine the effects of ischaemia, tizanidine, and anaesthesia on the areas of the short and medium latency reflex responses. All statistical tests were conducted with a significance level of 0.05. Results are shown as means (±s.d.).

RESULTS

A typical set of averaged data for one subject is shown in Fig. 2. In this case the subject was walking at 3.5 km h−1 and a stretch reflex was applied 200 ms following heel contact (Fig. 2A). The perturbed step (thick line) is shown superimposed over the control step (thin line). The stretch onset occurs at time zero and is indicated with a vertical line through each record. The ankle was stretched for 33 ms and then held for 200 ms. It returned to the position of the control step approximately 150 ms after it was released. The perturbation velocity was 302 deg s−1 and the control step velocity at that time was 39 deg s−1; giving a relative perturbation velocity of 263 deg s−1.

Figure 2.

Example of an averaged data record for one subject with the perturbed trials and control trials superimposed

A full step cycle for the left leg is shown. The perturbed (thick line) and control (thin line) records are shown superimposed. The time base is shifted so that zero corresponds with the stretch onset. A, ankle angular position with an offset such that 0 deg corresponds to the position at the time of stretch onset. B, soleus (SOL) EMG. C, tibialis anterior (TA) EMG.

Prior to the stretch reflex, the soleus EMG for the control and perturbed steps were similar (Fig. 2B). The onset of the stretch reflex at 40 ms is evident and a second burst of activity with an onset latency of 79 ms is also noticeable in this trace. The soleus activity resulting from the perturbation was greater than that of the control step for approximately 250 ms. The perturbation did not produce any change in the tibialis anterior activity preceding or during the time period for which the present analyses were carried out. However, the perturbation did result in an earlier onset of the tibialis anterior EMG, although this occurred approximately 400 ms after the stretch and had no consequence on the present analysis.

Averaged across all subjects, the onset latency for the short latency response was 39 ± 2 ms. As previously discussed, the onset of the medium latency response is not always clearly defined and therefore was not determined. In contrast, the peak latencies for both responses are clearly evident and easily defined. The mean peak latencies for the short and medium latency bursts were 55 ± 6 ms and 86 ± 11 ms, respectively. Across all subjects and conditions, the peak latencies of the short latency burst observed in this study did not change in response to the relative perturbation velocity (P < 0.001).

EMG responses at different stretch velocities

Figure 3 shows the area increment in response to the relative perturbation velocity over the 50 ms window (Fig. 3A) and over the two 20 ms windows (Fig. 3B). These are representative data from one subject with perturbations presented 200 ms after heel contact. In this example the reflex response over the 50 ms window increased with stretch velocity, having a significant positive slope (0.22 ± 0.06 % s deg−1; r2= 0.75; P = 0.005). Over the 20 ms window the short latency response also showed an increase with perturbation velocity (0.41 ± 0.08 % s deg−1; r2= 0.86; P < 0.001), whereas the slope of the medium response was not significantly different from zero (0.08 ± 0.09 % s deg−1; r2= 0.22; P = 0.48).

Figure 3.

Area increments from the muscle responses resulting from ankle stretches at various relative perturbation velocities for a single subject

A, area increment for a reflex response over a 50 ms window (▪). In this case the slope was 0.26 ± 0.08 and significantly different from zero (P = 0.008). B, comparison of the short (•) and medium (○) latency muscle responses. The slopes were 0.54 ± 0.06 (P < 0.001) and -0.04 ± 0.12 (P = 0.725), respectively.

Across all subjects the average slope for the short latency response was 0.33 ± 0.04 % s deg−1 and that for the medium response was 0.11 ± 0.07 % s deg−1 (Fig. 4). The slope of the short latency response was significantly different from zero (P < 0.001), whereas the slope of the medium latency response was not (P = 0.725). Student's paired t test indicated that the slope for the short latency response was significantly greater than that for the medium response (P = 0.016).

Figure 4.

Comparison of the slope of the stretch velocity-stretch reflex relationship for the short and medium latency responses

The slopes are shown for the short and medium latency reflex responses. Although both responses show positive velocity sensitivity, only the short latency response is statistically different from zero (P < 0.001).

The velocity at which the short latency response becomes greater than the medium latency response can be estimated by the point of intersection of the two regression lines. When this is calculated for each subject, the average intersection point is 132 ± 70 deg s−1 above the velocity of the ongoing movement.

To further investigate the velocity sensitivity, the data from all subjects and conditions were pooled and separated into two groups based on the relative perturbation velocity. Velocities less than 132 deg s−1 above that of the ongoing movement were placed in a ‘low’ group and velocities greater than 132 deg s−1 above the ongoing movement were placed in a ‘high’ group. The low group represents velocities in the normal operational range of ankle velocities during walking (e.g. Winter, 1990), whereas the high group represents ankle velocities that may be expected during an unexpected perturbation. For the low velocities the area increment was greater over the medium latency window than over the short latency window (P < 0.001). For the high velocities, the opposite effect was found (P = 0.002).

Nerve cooling

Nerve cooling was applied to eight subjects to investigate whether or not the short and medium latency reflex responses are mediated by different afferents. The effect of nerve cooling on the peak latencies of the two responses is shown in Fig. 5A. On the left, the soleus EMG record from a single subject is presented together with the corresponding ankle position records. For clarity, only the records for the perturbed steps are shown. On the right, the mean data across all subjects is shown. Prior to cooling the average peak latency of the short and medium latency responses were 55 ± 5 and 78 ± 6 ms, respectively. These average latencies increased to 58 ± 5 and 86 ± 6 ms, respectively, after 1.5-2 h of cooling. Although both the short and medium latency reflexes were delayed by cooling (P < 0.001), the medium latency reflex was delayed to a greater extent as indicated by the significant interaction effect (P < 0.005). For each subject the MLR-SLR delay was calculated before and after cooling. Cooling increased this difference by 5 ± 2 ms (P < 0.005).

Figure 5.

The effect of nerve cooling, ischaemia, tizanidine and an ankle block on the short and medium latency muscle responses

The left side of each panel shows a soleus (SOL) EMG and ankle angular position recording from a single subject during perturbed steps. Control steps have been omitted for clarity. Averaged data across all subjects are shown on the right side of each panel. A, left side, data records before (thin line) and after (thick line) nerve cooling. The arrows shown above the soleus EMG highlight the latency differences before and after cooling. Right side, short (•) and medium (○) latency responses averaged across all subjects (n = 8). Both responses were delayed although the medium response is delayed to a greater extent than the short latency response (P < 0.005). B, during ischaemia the short latency response was reduced to the level of the background EMG determined just prior to the stretch (P < 0.001). The medium response decreased but the decrease was not significant (P = 0.437). Left side, data records before (thin line) and after (thick line) the ischaemic block. Right side, average across all subjects (n = 4) before (▪) and after (□) ischaemia. C, 2 h after the ingestion of tizanidine the medium latency response was significantly depressed (P = 0.007). The short latency response decreased, although the change was not statistically significant (P = 0.653). Left side, data records before (thin line) and after (thick line) tizanidine. Right side, average across all subjects (n = 3) before (▪) and after (□) the ingestion of tizanidine. D, after an ankle block with subcutaneously administered lidocaine, there were no significant changes in either the short (P = 0.453) or medium (P = 0.310) latency components of the stretch reflex. Left side, data records before (thin line) and after (thick line) the ankle block. Right side, average across all subjects (n = 3) before (▪) and after (□) the ankle block. The filled and open rectangles shown immediately below the soleus EMG records in panels B, C and D represent the 20 ms windows for the short and medium latency responses, respectively.

Ischaemic block

An ischaemic block was applied on four subjects to investigate if the large diameter afferents contribute to the medium latency response. Approximately 15-25 min after the inflation of the tourniquet, the short latency stretch reflex was reduced to less than 10 % of its normal value, providing evidence that the Ia afferents were indeed blocked. In Fig. 5B a soleus EMG record from a single subject is presented together with the corresponding ankle position records. From these data a clear depression of the short latency reflex response in the ischaemic condition is evident, with no change in the medium latency component. Figure 5B also shows the effect of the ischaemic block averaged across the four subjects tested. During ischaemia, the short latency reflex response was decreased to the level of the background EMG (P = 0.004), whereas the small decrease of the medium latency response was not statistically significant (P = 0.437).

Tizanidine depression of group II afferents

Tizanidine was used in three subjects to depress transmission in the group II afferent pathway. The effect of tizanidine on the short and medium latency reflex responses is shown in Fig. 5C. A depression of the medium latency reflex is visible in the soleus EMG record from a single subject. The mean results across the three subjects tested are presented on the right of this panel. After the ingestion of tizanidine the stretch-evoked increase in EMG activity corresponding with the medium latency response was reduced with respect to the EMG activity measured in the control step. Prior to the ingestion of tizanidine a 94 % increase was observed and this was reduced to 55 % 2 h later (P = 0.007). The small decrease of the short latency response observable in this figure was not statistically significant (P = 0.653).

Anaesthetic depression of cutaneous afferents

Lidocaine was used in three subjects to depress the transmission of information from the cutaneous afferents of the foot and ankle. The effect of the anaesthesia on the short and medium latency reflex responses is shown in Fig. 5D. Across all subjects anaesthesia resulted in a decrease in the somatosensory evoked potentials of 74 % (P = 0.016), indicating that the cutaneous afferents were strongly depressed. Despite the effectiveness of the block, no change in the responses of either the short or medium latency stretch reflex are evident in the example data presented on the left panel of Fig. 5D. Average data across all subjects are presented on the right. Neither the short (P = 0.453) nor medium (P = 0.310) latency components of the stretch reflex were changed by blocking the cutaneous afferents.

DISCUSSION

The results presented in this paper demonstrate that during walking, the stretch-mediated response in the soleus electromyogram can be decoupled into two responses based on velocity sensitivity, nerve cooling, ischaemic block and the effects of tizanidine and lidocaine. Furthermore, we have provided evidence that the medium latency component of the stretch reflex is contributed to by group II muscle spindle afferents.

Which afferents are involved?

The five independent techniques used in this study indicate that different afferent pathways contribute to the stretch reflex when the ankle is perturbed during walking. The onset and peak latencies for the short and medium latency responses are consistent with values previously reported in the literature (Dietz et al. 1984; Sinkjær et al. 1999). The short and medium latency responses investigated in this study correspond with the range of conduction velocities from a monosynaptic pathway mediated by group Ia fibres and an oligosynaptic pathway mediated by group II fibres (Schieppati & Nardone, 1997). Other afferents with conduction velocities within this range include group Ib afferents and fast conducting cutaneous afferents (Aα and Aβ fibres). The slower conduction velocities of groups III and IV afferents exclude them as potential contributors to the short and medium latency reflex responses.

The velocity sensitivity of the short latency response observed in the present study is consistent with the observation that it is mediated by group Ia afferent fibres (Houk & Rymer, 1981). The depression of the short latency stretch reflex during ischaemia indicates that the Ia afferents were effectively blocked, whereas the lack of depression of the medium latency response shows that it is not mediated by the same Ia afferents. It is possible that small diameter group Ia fibres with slower conduction velocities and/or polysynaptic interneuronal connections may contribute to the medium latency response. However, given the fact that the velocity sensitivity for the medium response was small, the contribution from slower Ia fibres is likely to be minimal.

The lack of velocity sensitivity of the medium latency reflex response is consistent with results in the cat which show that group II muscle afferents are not as velocity sensitive as group Ia afferents (Nichols & Houk, 1976). The strong depression of the medium latency stretch reflex with tizanidine suggests that group II afferents contribute to the medium latency reflex response and the lack of depression of the short latency response suggests that they do not contribute to the short latency response. These results are in contrast to the results of Eriksson et al. (1996) for the stretch reflex in the quadriceps muscle. The discrepancy between their findings and ours is probably due to the shorter distance between the quadriceps muscle and the spinal cord. With the shorter conduction distance for the quadriceps reflex loop, the effects of group I and group II fibres are probably less clearly differentiated than the soleus muscle.

It is important to note that although tizanidine has been suggested to selectively depresses transmission in the polysynaptic spinal pathway of group II afferents in the cat (Bras et al. 1989, 1990; Skoog, 1996), their investigations were restricted to groups I and II muscle spindle afferents. The effect of tizanidine on either the group Ib tendon afferents or cutaneous afferents were not investigated. In addition, tizanidine has been shown to depress responses of dorsal horn neurons to noxious, but not to innocuous, skin stimuli (Davies et al. 1984), indicating that in some cases α2-adrenergic agonists may have actions other than group II inhibition. Because the depression was only shown for a noxious pathway, these results are not directly relevant to the conclusions of the present study.

The effect of cutaneous reflexes during human walking has recently received considerable attention (for review see Zehr & Stein, 1999). It has been suggested that these reflexes are important in the regulation of gait (van Wezel et al. 2000) and in stumbling reactions (Zehr et al. 1997). In contrast to the present study, these studies have focused on the transitions of the ‘to’ and ‘from’ swing phase. The conduction velocity of the large diameter Aα and medium diameter Aβ cutaneous afferents is such that they may also contribute to the reflex responses observed in the present study. However, van Wezel et al. (2000) showed that it is the Aβ fibres that are probably more responsible for mediating cutaneous reflexes during walking.

In the present study, transmission in cutaneous afferents from the foot and ankle was strongly depressed by local anaesthesia. The lack of change in either the short or medium latency components of the stretch reflex after the foot and ankle were anaesthetised suggests that, at least during the stance phase of walking, cutaneous afferents do not contribute strongly to the stretch reflex.

Group Ib musculotendon afferent fibres have been shown to inhibit homonymous motoneurones in decerebrate cats (Nichols & Houk, 1976). However, recent evidence in cats has shown that the effect of these afferents is reversed during locomotion, such that they produce an excitatory response (Pearson & Collins, 1993; Gossard et al. 1994). In humans, it has been suggested that group Ib afferents may also have a facilitative effect (Dietz, 1998; Stephens & Yang, 1999). It is not possible to assess the contribution of the force-sensitive group Ib afferents to the stretch reflex responses based on the protocol used in the present study. Although the conduction velocity of group Ib fibres in human subjects is unknown, Ib afferents are slightly smaller than the Ia afferents and have both disynaptic and oligosynaptic connections with homonymous α-motoneurones (Pierrot-Deseilligny et al. 1981). Consequently, any effects in the electromyogram from group Ib fibres should be expected to be delayed with respect to group Ia fibres and earlier with respect to the contributions from group II fibres. Because their conduction velocity is unknown, it is unclear whether the Ib afferents should contribute more to the short latency response or to the medium latency response.

From the results of the present study, it cannot be stated conclusively that group II afferents are solely responsible for mediating the medium latency component of the stretch reflex. This is because it is unknown whether or not tizanidine has a depressive action on group Ib afferents as well as the known effect in the group II afferent pathway. Further study, particularly with respect to load receptors, is required to determine and quantify the contribution of these afferents to the corrective action of the stretch reflex response in the walking human. Moreover, additional investigation is needed to determine if these afferents contribute in the same way to the background locomotor EMG as they do to the corrective response observed in the present study.

Acknowledgements

The authors thank Dr Ken Yoshida for his comments on the experimental design and the manuscript, Mr David Niddam for his work with the EEG recordings, Dr Prem Bajaj and Dr Jens Haase for performing the ankle blocks, and Mr Knud Larsen for his technical assistance. The Danish National Research Foundation is acknowledged for financial support for the studies reported in this paper.

Ancillary