Spinal inhibition of descending command to soleus motoneurons is removed prior to dorsiflexion

Non-technical summary

The coordination of antagonistic muscle activity starts well in advance of the onset of voluntary movement. We recently demonstrated that antagonist muscle responses evoked by stimulation of the brain were increased prior to voluntary contraction at the ankle. Although our data indicated that this was explained by activation of a subcortical motor program, the neural pathways involved are unknown. Here we probe the transmission in the underlying neuronal networks by peripheral nerve stimulation in order to investigate the neural pathways responsible for this facilitation of antagonist muscle responses. We demonstrate that this stimulation produces a spinal inhibition of the antagonist muscle responses, which is removed prior to voluntary contraction. We propose that the removal of this inhibition might explain the increased antagonist muscle responses prior to voluntary contraction at the ankle. This mechanism might enable the direction of movement to be changed quickly during functional motor tasks such as dribbling.

Abstract

Abstract

It has recently been demonstrated that soleus motor-evoked potentials (MEPs) are facilitated prior to the onset of dorsiflexion. The purpose of this study was to examine if this could be explained by removal of spinal inhibition of the descending command to soleus motoneurons. To test this, we investigated how afferent inputs from the tibialis anterior muscle modulate the corticospinal activation of soleus spinal motoneurons at rest, during static contraction and prior to movement. MEPs activated by transcranial magnetic stimulation (TMS) and Hoffmann reflexes (H-reflexes), activated by electrical stimulation of the posterior tibial nerve (PTN), were conditioned by prior stimulation of the common peroneal nerve (CPN) at a variety of conditioning–test (CT) intervals. MEPs in the precontracted soleus muscle were inhibited when the TMS pulse was preceded by CPN stimulation with a CT interval of 35 ms, and they were facilitated for CT intervals of 50–55 ms. A similar inhibition of the soleus H-reflex was not observed. To investigate which descending pathways might be responsible for the afferent-evoked inhibition and facilitation, we examined the effect of CPN stimulation on short-latency facilitation (SLF) and long-latency facilitation (LLF) of the soleus H-reflex induced by a subthreshold TMS pulse at different CT intervals. SLF is known to reflect the excitability of the fastest conducting, corticomotoneuronal cells whereas LLF is believed to be caused by more indirect descending pathways. At CT intervals of 40–45 ms, the LLF was significantly more inhibited compared to the SLF when taking the effect on the H-reflex into account. Finally, we investigated how the CPN-induced inhibition and facilitation of the soleus MEP were modulated prior to dorsiflexion. Whereas the late facilitation (CT interval: 55 ms) was similar prior to dorsiflexion and at rest, no inhibition could be evoked at the earlier latency (CT interval: 35 ms) prior to onset of dorsiflexion. The observation that the CPN-induced inhibition of soleus MEPs disappears prior to onset of dorsiflexion may explain why soleus MEPs are facilitated prior to onset of dorsiflexion contraction. A possible mechanism involves the removal of inhibition of the descending command to the motoneurons at a spinal interneuronal level because the inhibition was seen in LLF and not in SLF, and the MEP inhibition was not observed in the H-reflex. The data illustrate that spinal interneuronal pathways modify descending commands to human spinal motoneurons and influence the size of MEPs elicited by TMS.

Introduction

The coordination of antagonistic muscle activity starts well in advance of the onset of voluntary movement, as reflected in excitability changes in several pathways projecting to the agonist and antagonist. In primates, motor cortex cells with direct monosynaptic projections to the motoneurons (corticomotoneuronal cells) begin to fire about 100 ms prior to the onset of contraction (Cheney & Fetz, 1980). Similarly, in humans the excitability of corticomotoneuronal cells projecting to soleus motoneurons is increased 70–100 ms prior to the onset of plantar flexion (Nielsen & Petersen, 1992), while soleus H-reflexes are increased 40–50 ms prior to plantar flexion (Nielsen & Kagamihara, 1993). With respect to dorsiflexion, soleus H-reflexes are strongly depressed from about 50 ms prior to the onset of contraction (Crone & Nielsen, 1989), whereas soleus MEPs are facilitated (Geertsen et al. 2010). The inhibition of soleus H-reflexes at this time is partly due to supraspinal facilitation of the disynaptic reciprocal inhibition pathway (Kots & Zhukov, 1971; Crone & Nielsen, 1989). The MEP facilitation is suggested to be caused by subcortical mechanisms because responses evoked at the level of the brainstem were similarly facilitated and since the direct corticomotoneuronal connections were not facilitated (Geertsen et al. 2010), although the pathways involved are unknown. The facilitation could result from either reduced excitability of an inhibitory pathway or increased excitability of a facilitatory pathway. One way to investigate the mechanism is to probe transmission in the underlying neuronal networks by peripheral nerve stimulation. Such stimulation typically evokes several responses in the muscles of the stimulated limb. Early responses have a definite spinal origin whereas some of the later responses have a sufficiently long latency to be caused by supraspinal structures like the brainstem, the cerebellum or the cortex.

Stimulation of peripheral sensory afferents in the hand is known to activate both excitatory and inhibitory intracortical circuits to potentially facilitate and inhibit descending motor commands. When afferent inputs from electrical stimulation first arrive at the motor cortex, MEPs from TMS are depressed in what is called short-latency afferent inhibition (SAI) (Tokimura et al. 2000). However, a recent study investigating how afferent inputs modulate the excitability of cortical circuits in the leg area of M1 showed that spinal rather than cortical circuits mediate SAI of leg MEPs and that only later facilitatory effects had a cortical origin (Roy & Gorassini, 2008). Therefore, there appears to be some differences between the cortical control of leg and arm muscles.

In forearm muscles, activation of the median nerve afferents can suppress the excitability of cortical areas controlling the antagonist forearm extensor muscles (Bertolasi et al. 1998; Hortobagyi et al. 2006), but so far no studies have reported the effect of activation of leg afferent inputs on corticospinal outputs to antagonist muscles in humans.

In this study, we used CPN stimulation, which has both inhibitory and facilitatory effects on soleus MEPs, to investigate the origin of soleus MEP facilitation prior to the onset of dorsiflexion. We hypothesized that either the inhibition would be removed or that the facilitation would be increased prior to the onset of dorsiflexion. We then used TMS conditioning of the soleus H-reflex during tonic plantar flexion to further investigate the mechanism responsible for these effects.

Methods

Participants

Eighteen healthy human subjects (7 female) aged 18–42 years (25 ± 6 years, mean ± SD) participated in this study. Some subjects participated in all of the three experiments that comprise this study. All subjects gave their written, informed consent to the experimental procedures, which were approved by the ethics committee for the Capital Region of Denmark. The study was performed in accordance with theDeclaration of Helsinki.

General organization of the studies

In all experiments, subjects were comfortably seated in an armchair and the left leg was positioned with the hip semi-flexed (120 deg), the knee flexed (110 deg) and the ankle at a slightly plantar flexed position (140 deg) for the duration of the experiment. The left foot was firmly attached to a force pedal using adjustable straps. The torque exerted on the foot plate was measured with a strain gauge.

At the beginning of each experiment, subjects were instructed to perform a maximal dorsiflexion (1–2 s) to measure their maximal voluntary contraction strength (MVC). Subjects were verbally encouraged to produce maximal torque. The torque was displayed as a moving vertical bar on the monitor. At least three trials were performed, separated by 30 s rest periods, and the peak torque was used as the dorsiflexion MVC. Subsequently, the same procedure was used to measure the plantar flexor MVC.

EMG measurements

EMG activity was recorded from the anterior tibial (TA) and soleus muscles by non-polarisable bipolar electrodes (diameter 0.5 cm; Blue Sensor, Ambu, Ølstykke, Denmark) placed over the belly of the muscles with an inter-electrode distance of 2 cm. The EMG signals were amplified (500–5000×) using custom-built EMG amplifiers, filtered (band-pass, 5 Hz to 1 kHz), sampled at 2 kHz, and stored on a PC for offline analysis (CED 1401+ with Signal 4.05 software, Cambridge Electronic Design, Cambridge, UK).

Soleus H-reflexes

Soleus H-reflexes were induced by stimulation (1 ms rectangular pulses; model DS7A Digitimer, UK) of the posterior tibial nerve (PTN) using a ball-shaped mono-polar electrode (Simon electrode) placed in the popliteal fossa. The anode was placed proximal to the patella. The unconditioned test reflex size was adjusted to 15% of the maximal M-response (M_max) because reflexes of this size have previously been shown to be most sensitive to facilitatory and inhibitory inputs (Crone et al. 1990).

Common peroneal nerve stimulation

The common peroneal nerve (CPN) was stimulated (1 ms rectangular pulses; model DS7A Digitimer) through bipolar surface electrodes (diameter 0.5 cm; Blue Sensor) placed 1–3 cm distal to the neck of the fibula. Care was taken to ensure that the conditioning stimulus was applied at a position where the threshold for an M-response (motor threshold, MT) in TA was lower than the MT in the peroneal muscle. A stimulation strength of 1.1× MT was used in all trials.

Maximal M-response

All MEP and H-reflex data were normalized to M_max in the respective muscles. In TA and soleus, M_max was evoked by stimulation of CPN and PTN, respectively, as described above. In these measurements, the intensity of stimulation was increased from a subliminal level until there was no further increase in the peak-to-peak amplitude of the M-response with increasing stimulation intensity.

Transcranial magnetic stimulation

Magnetic stimuli were delivered by a Magstim 200 stimulator connected to a custom-made figure-of-eight-shaped bat-wing coil with an individual wing diameter of 90 mm (The Magstim Company, Whitland, Dyfed, UK). The site where stimuli of slightly suprathreshold intensity consistently produced the largest MEPs in the pre-contracted (10% of plantar flexion MVC) soleus muscle (referred to as ‘motor hot spot’) was marked using frameless stereotaxy (BrainSight 1.7, Rogue Research software, Canada). The coil was then secured in place and the active motor threshold was found by reducing the stimulus intensity to a level that elicited an MEP in 3 out of 5 trials in soleus (58 ± 4% of maximal stimulator output, mean ± SD). In study 3, the same procedure was used to find the resting motor threshold (70 ± 6% of maximal stimulator output, mean ± SD). Note that the resting threshold is usually lower in TA than in soleus, which means that an MEP was also evoked in TA in trials involving suprathreshold stimulations (Geertsen et al. 2010). One could fear that the soleus MEPs were simply a mirror-image of the larger TA MEPs, but this is not very likely, since the shape of TA and soleus MEPs to the same stimuli was clearly different, as also discussed in Geertsen et al. (2010). Furthermore, it has been shown by cross-correlation analysis that there is hardly any cross-talk between the TA and soleus recordings with similar electrode arrangements and amplification as in the present study (Hansen et al. 2005).

Study 1: Time-course of afferent CPN conditioning of soleus MEPs

In six subjects, we investigated how the corticospinal projections to the soleus muscle (soleus MEPs) were affected by a preceding activation of afferents from the antagonistic TA muscle (CPN stimulation).

Single stimuli (1.2× active MT) were applied to the hotspot of the soleus muscle while the subject held an isometric plantar flexion contraction of 10% of MVC. A conditioning stimulation of the CPN was applied at different intervals prior to the TMS pulse. The inter-stimulus interval between TMS pulses was 5 s. The conditioning–test (CT) intervals used were: 0, 10, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65 and 70 ms. At least 10 MEPs were measured at each CT interval.

Study 2: Afferent CPN conditioning of short-latency facilitation and long-latency facilitation

To investigate how CPN afferent input modulates descending transmission in specific pathways, we conditioned H-reflexes with subthreshold TMS at different CT intervals. This elicits a short-latency facilitation of the H-reflex (SLF), which reflects the excitability of the direct corticomotoneuronal connections to soleus spinal motoneurons and a long-latency facilitation (LLF), which probably reflects the excitability of indirect pathways to soleus spinal motoneurons (Nielsen et al. 1993; Nielsen & Petersen, 1995a,b;).

TMS conditioning of the H-reflex

A time-course of the TMS conditioning effect on the test H-reflex was obtained for each of the 10 subjects who participated in this study (Nielsen et al. 1993). Soleus H-reflexes were evoked as described above. The unconditioned test reflex size was kept at 15% of M_max while the reflex was conditioned by TMS at different CT intervals. The intensity of the conditioning TMS stimulation was adjusted to 2–3% below the active MT. The conditioning pulse was initially given at 1 ms intervals after the test H-reflex (–5 to 0 ms) and 10 ms before the test H-reflex during tonic plantar flexion at 10% MVC. A more precise time-course (0.5 ms steps) was then used to find the exact onset of the facilitation. The earliest CT interval where the TMS pulse had a significant facilitatory effect on the H-reflex was determined as the SLF (–1.5 ms in Fig. 3). For the LLF, a CT interval of 10 ms was always used (Nielsen et al. 1993; Nielsen & Petersen, 1995a,b;) except in one subject where a CT interval of 6 ms was used instead as no facilitation could be obtained with a CT interval of 10 ms. SLF and LLF were expressed as the size of TMS-conditioned H-reflexes as a percentage of the unconditioned test H-reflex size (Fig. 3).

Figure 3. Time-course of the effect of TMS conditioning of the soleus H-reflex (example from one subject).

The soleus H-reflex was conditioned by a subthreshold TMS pulse applied to the M1 soleus hotspot at different time-points to determine the short-latency facilitation (SLF) and the long-latency facilitation (LLF). In each subject, the earliest conditioning–test (CT) interval that produced a significant increase in the soleus H-reflex was used as the SLF (–1.5 ms in example above). A CT interval of 10 ms was used for the LLF. The SLF and LLF of each subject were used for the following experiments. *, **, ***Significant difference between the conditioned H-reflex and the test H-reflex size. Each vertical bar is one SEM.

Effect of a conditioning CPN stimulation on SLF and LLF

A time-course of the effect of a conditioning CPN stimulation on the test H-reflex (cH-reflex), SLF (cSLF) and LLF (cLLF) was then obtained in each subject at CT intervals from 0 to 70 ms (see Fig. 4). Each CT interval was measured in different sessions that included measures of the unconditioned test H-reflex, SLF and LLF (6 alternatives in total):

Figure 4. Time-course of the effect of conditioning CPN stimulation on SLF and LLF.

Example from a single subject of the modulation of short-latency facilitation (SLF) and long-latency facilitation (LLF) when preceded by a conditioning CPN stimulus.A, top panel (SLF): soleus EMG showing the unconditioned test H-reflex response (average of 10 reflexes) following stimulation of the posterior tibial nerve (PTN) at time 0 ms (black trace). Note that this trace is the same in all traces inAandC, as all measurements for each CT interval were performed in the same session. The overlaid blue trace shows the H-reflex when conditioned by a subthreshold TMS pulse at –1.5 ms (black arrow) producing SLF. In the second panel (cH), the test H-reflex is instead conditioned by a CPN stimulation 45 ms prior to the PTN stimulation (grey trace). Bottom panel (cSLF): the H-reflex is conditioned by both subthreshold TMS at –1.5 ms and CPN stimulation at 45 ms (red trace).B, time-course of SLF (blue line), cSLF (red line) and cH (dashed line) in percentage of the test H-reflex size.C, top panel (LLF): like inA, but here the H-reflex was conditioned by a subthreshold TMS pulse at 10 ms (black arrow) producing LLF (green trace). Second panel (cH): same trace as inA. Bottom panel (cLLF): the H-reflex is conditioned by both subthreshold TMS at 10 ms and CPN stimulation at 45 ms (magenta trace). Note the reduced amplitude of the conditioned reflex compared to LLF in the top panel.D, time-course of LLF (green line), cLLF (magenta line) and cH (dashed line) in percentage of the test H-reflex size. *, **, ***Significant difference between SLF and cSLF or between LLF and cLLF.

1. Unconditioned test H-reflex

2. SLF: H-reflex conditioned by TMS at short interval

3. LLF: H-reflex conditioned by TMS at long interval

4. cH-reflex: H-reflex conditioned by CPN at given interval (0–70 ms)

5. cSLF: H-reflex conditioned by TMS at short interval (SLF) and CPN

6. cLLF: H-reflex conditioned by TMS at long interval (LLF) and CPN

At least 10 responses of each alternative were measured at each CT interval that lasted about 5 min. At the beginning of each session, the intensity of the PTN stimulation was adjusted (if necessary) to elicit an unconditioned H-reflex size of about 15% of M_max.

In order to obtain an estimate of the effect of the CPN stimulation on SLF and LLF (i.e. the corticospinal transmission in the direct monosynaptic pathway and longer-latency indirect pathway, respectively) independent of the effect of the CPN stimulation on the H-reflex (i.e. the spinal motoneurons), we subtracted the effect of CPN stimulation on the H-reflex (alternative 4, above) from the combined effect of TMS and CPN stimulation (alternatives 5 and 6) and compared the result of these subtractions to the effect of TMS on the H-reflex (alternatives 2 and 3). This makes it possible to make a direct comparison in Fig. 5 between the size of SLF/LLF with and without CPN stimulation.

Figure 5. CPN stimulation inhibits LLF at spinal latency.

In order to obtain an estimate of the effect of the CPN stimulation on SLF and LLF independent of the effect of the CPN stimulation on the H-reflex, we subtracted the effect of CPN stimulation on the H-reflex (cH) from the combined effect of TMS and CPN stimulation (cSLF and cLLF) and compared the result of these subtractions to the effect of TMS on the H-reflex (SLF and LLF). This makes it possible to make a direct comparison between the size of SLF/LLF with and without CPN stimulation.A, time-course of the difference between conditioned reflexes in percentage of the test H-reflex (group data, n = 10). Filled circles, difference between the size of cSLF subtracted by cH and SLF at different conditioning–test (CT) intervals. Open squares, difference between the size of cLLF subtracted by cH and LLF at different CT intervals. **Significant difference between the conditioning CPN effect on SLF and LLF at this CT interval. Each vertical bar is one SEM.B, control experiments in four subjects where the CPN conditioning stimulation was applied to match the latency of both SLF and LLF in the same trial for each subject. Given the 45 ms CT interval was the only one showing a significant difference between the CPN effect on SLF and LLF in Fig. 5A, we focused on CT intervals 40, 45 and 50 ms in this control experiment. For the subject example in Fig. 4 with a SLF of –1.5 ms and LLF of 10 ms, it means that at the CT 45 ms interval the CPN stimulation was now applied 33.5 ms prior to the test H-reflex (instead of 45 ms) for the cSLF whereas the cLLF interval was still 45 ms (thereby taking into account the 11.5 ms difference in the application of the subthreshold TMS pulse). An extra cH alternative where the CPN was applied 33.5 ms before the test H-reflex was also included in order to make a direct comparison between the size of SLF/LLF with and without CPN stimulation as in Fig. 5A. For the 45 ms CT interval, the ‘SLF’ was thus: (cSLF_33.5– cH_33.5) – SLF, and the ‘LLF’ was: (cLLF₄₅– cH₄₅) – LLF.

One has to be cautious when interpreting the time-course of CPN effects on SLF and LLF (Fig. 4 and 5A), as the latency between the conditioning CPN stimulation and the conditioning TMS pulse is 11–14 ms shorter for LLF compared to the SLF at each interval. Therefore, we performed control experiments in four subjects where we matched the latency between the stimuli (Fig. 5B).

Study 3: CPN conditioning of soleus MEPs prior to dorsiflexion

In order to investigate if previous observations of facilitation of soleus MEPs prior to dorsiflexion (Geertsen et al. 2010) may be explained by a change in corticospinal transmission through premotoneurons contacted by CPN afferents, we investigated the modulation of the inhibition and facilitation of soleus MEPs elicited by CPN stimulation prior to dorsiflexion.

Time-course of afferent CPN conditioning of soleus MEPs at rest

We first made a time-course of the effect of afferent input from CPN on soleus MEPs (1.2× resting MT) at rest to find the CT intervals that produced an early inhibition and a later facilitation of the MEP (n = 9). As in study 1, the CT intervals used were: 0, 10, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65 and 70 ms. At least 10 MEPs were measured at each CT interval. The CT intervals at which the largest early inhibition (36 ± 3 ms; mean ± SD) and late facilitation (around 55 ± 4 ms; mean ± SD) were seen, were chosen for the subsequent experiment.

Modulation of CPN conditioned soleus MEPs prior to dorsiflexion

Subjects then performed a reaction time task where an auditory warning cue (C_warning) prepared the subjects to be ready. They were instructed to dorsiflex their ankle to 30% MVC as fast as possible (the dynamic phase lasted around 100–200 ms) when a second auditory ‘go’ cue (C_go) was delivered 2.8–3.2 s after C_warning. Prior to the experiment, subjects practiced the reaction time task in order to reduce the variability of the reaction time and the effect of learning. After about 5 min practice, subjects could complete the task with little variation in their reaction time.

In random manner, a TMS pulse (1.2× resting soleus MT) was applied about 25–50 ms prior to onset of dorsiflexion (–25 ms) or 100 ms prior to C_go (<C_go) either alone (unconditioned) or preceded by the conditioning CPN stimulation that produced inhibition or facilitation in that subject at rest.

Control trials in which the cues were presented but no TMS applied were interspersed 33% of the time. Thereby it was possible to measure reaction times, the time between C_go and the onset of contraction (onset of agonist EMG), during the course of the experiment to ensure the validity of the estimates. The average reaction time was calculated offline for each subject and the onset of EMG activity in the agonist muscle was set as time 0. Peak-to-peak amplitudes of the conditioned and unconditioned MEPs were then averaged, and the size of the conditioned MEPs were expressed as a percentage of the unconditioned MEP size at <C_go and at –25 ms (Fig. 6).

Figure 6. CPN inhibition of soleus MEPs disappear prior to onset of dorsiflexion.

A, soleus MEPs were conditioned with a CPN stimulation at an interval that produced inhibition at rest (35 ms CT interval in most subjects). Subjects dorsiflexed their ankle to 30% of their MVC immediately after an auditory ‘go’ cue (C_go). The size of the conditioned soleus MEP (in percentage of the unconditioned soleus MEP size) is shown for each subject at (<C_go) and at –25 ms (n = 9). The thick grey dotted line is the group average.B, same as inA, but here the effect of the facilitatory CT interval is plotted (n = 8). In one subject, there was only strong inhibitory effects at all CT intervals, so this subject is not included in this figure. *Significant difference between the size of the conditioned MEP <C_go and at –25 ms.

Trials in which the subject made a ‘false start’ or ‘missed’ C_go were omitted from the analyses. See Geertsen et al. (2010) for a more detailed description of this set-up.

Statistics

Individual MEP and H-reflex amplitudes computed within a 30–60 ms window after stimulation were averaged for each condition in every subject. The average value of the peak-to-peak amplitudes for each condition was then divided (normalized) by the value of the average control amplitude and this ratio was multiplied by 100.

Statistical analysis was conducted with SigmaStat 2.03 (SPSS Inc.). Before statistical comparison, all data sets were tested for normal distribution by a Kolmogorov–Smirnov test. A one-way repeated measures ANOVA was used to investigate the effect of CPN stimulation on soleus MEPs. A two-way repeated measures ANOVA with factors CT interval and stimulation type (TMS or PTN stimulation) was used to compare the effect of CPN stimulation on MEP and H-reflex responses (study 1). A two-way ANOVA with factors CT interval (0–70 ms) and MEP type (SLF or cSLF subtracted by cH-reflex) was used to investigate the effect of CPN stimulation on SLF in each subject. All pair-wise comparisons between the group mean responses were performed with Bonferroni correction. The same analysis was conducted for LLF (study 2). It was not possible to perform this test on group data from all subjects because data were not obtained for all CT intervals in some subjects. In cases where the sample size did not allow a Gaussian distribution of data or when the variance between the control and test experimental condition differed, data were statistically analysed by the non-parametric two-sample Mann–Whitney U test. The modulation of the conditioned MEPs prior to dorsiflexion (study 3) was investigated by comparing the conditioned MEP size (as a percentage of test MEP size) at <C_go with conditioned MEPs (as a percentage of test MEP size) evoked at –25 ms using a paired Student'sttest.

Statistical significance was given forPvalues smaller than 0.05 (*), 0.01 (**) and 0.001 (***). Data are presented as means ± standard error of the mean (SEM) unless reported otherwise.

Results

Study 1: Time-course of afferent CPN conditioning of soleus MEPs and H-reflexes

Stimulation of the CPN had both inhibitory and facilitatory effects on the size of the soleus MEP depending on the CT interval (see example in Fig. 1A). Figure 1C shows the averaged group data of the effect of CPN stimulation on the test MEP responses (n = 6). Stimulating the CPN 35 ms before applying a TMS pulse to the motor cortex depressed the test MEP (77 ± 5% of the test MEP size;P = 0.016), whereas a conditioning stimulation at CT intervals of 50 ms (130 ± 9% of the test MEP size;P< 0.001) and 55 ms (129 ± 6% of the test MEP size;P = 0.004) facilitated the test MEP (Fig. 1C).

Figure 1. Time-course of the effect of conditioning CPN stimulation on soleus MEP size (n = 6).

A, raw sweeps showing average soleus MEPs (black arrow, time of test TMS pulse) in one subject when preceded by a conditioning CPN stimulus. The CT interval influenced the size of the MEP by producing inhibition at 35 ms and facilitation at 50 ms.B, individual subject data at CT intervals from 0 to 70 ms in percentage of M_max.C, group data fromB(in percentage of the unconditioned test MEP size) showing a significant inhibition at 35 ms and a significant facilitation at 50 and 55 ms. Each vertical bar is one SEM.

In four of the subjects, a time-course of the effect of CPN stimulation was also investigated for the soleus H-reflex. Figure 2 shows the average group data of the effect of CPN stimulation on the soleus H-reflex and soleus MEPs at CT intervals from 0 to 70 ms. The inhibition of soleus MEPs observed around 30–35 ms is not seen in the soleus H-reflex (significant difference between MEP and H-reflex at 35 ms;P = 0.041). In contrast, the soleus H-reflex size is increased from CT intervals of 35 ms. From CT intervals of 65 ms, the soleus H-reflex size starts to decrease (significant difference between MEP and H-reflex at 70 ms;P = 0.021), which might be due to late presynaptic (D2 inhibition) effects (Mizuno et al. 1971).

Figure 2. Time-course of the effect of conditioning CPN stimulation on the soleus MEP and the soleus H-reflex size (n = 4).

In four of the subjects from Fig. 1, a time-course of CPN effects on the soleus H-reflex was also made. Filled circles, time-course of conditioned soleus MEPs (in percentage of the unconditioned test MEP size). Open squares, time-course of conditioned soleus H-reflexes (in percentage of the unconditioned H-reflex size). *Significant difference between the conditioned MEP and the conditioned H-reflex. Each vertical bar is one SEM.

The amount of D1 inhibition (another measure of presynaptic inhibition), which is often seen at CT intervals from approximately 7–30 ms, was rather small (92% of the test H-reflex size at CT intervals of 10 and 20 ms) and variable between subjects (see Fig. 4B, ‘cH’ trace) compared to reported values in some studies (Achache et al. 2010), but similar to what has been reported in other studies (Perez et al. 2003). However, low intensity and only one stimulus were used in the present study as compared to three to five stimuli and stronger stimulation used in the previous studies reporting larger D1 inhibition (El-Tohamy & Sedgwick, 1983; Faist et al. 1996; Achache et al. 2010).

Study 2: Afferent CPN conditioning of short-latency facilitation and long-latency facilitation

The MEP is a compound response, which is influenced by transmission in direct and indirect facilitatory and inhibitory pathways to the spinal motoneurons. To investigate how CPN afferent input modulates descending transmission in specific pathways, we conditioned H-reflexes with subthreshold TMS at different CT intervals. In the beginning of each experiment, a time-course of the TMS conditioning effect on the test H-reflex was made in order to determine the onset of the earliest (presumed monosynaptic) facilitation (Nielsen et al. 1993; Nielsen & Petersen, 1995a,b;).

Figure 3 shows a time-course of the effect of TMS on the H-reflex in a single subject. The SLF had an onset around –1.5 ms (217 ± 34% of the test H-reflex size;P = 0.008). A CT interval of 10 ms (245 ± 34% of the test H-reflex size;P< 0.001) was used to examine the LLF, which is believed to be caused by different descending pathways than the SLF (Nielsen & Petersen, 1995b). In one subject, in whom a CT interval of 10 ms did not produce facilitation, a CT interval of 6 ms was used for the LLF. The CT intervals producing SLF and LLF were then used in the following experiments.

Figure 4 shows in a single subject the changes in SLF and LLF when preceded by a CPN stimulus at different CT intervals. Each CT interval was measured in different sessions of 5 min. In Fig. 4A, the conditioning effects of subthreshold TMS (SLF), CPN stimulation (cH) and both subthreshold TMS and CPN stimulation (cSLF) on the test H-reflex are illustrated from one subject at a CT interval of 45 ms. The size of SLF (blue trace) at this interval was 142 ± 9% and the size of cH (grey trace) was 132 ± 21% of the unconditioned test H-reflex size (black trace). The combination of both conditioning stimuli produced a cSLF (red trace) of 151 ± 15%, but this was not significantly different from SLF alone (P = 0.57). Figure 4B shows the time-course of the conditioning effects at CT intervals from 0 to 70 ms for this subject. The only CT interval in which a significant difference between SLF and cSLF was observed was at 60 ms (125 ± 8%vs.158 ± 17%;P = 0.044).

The panels in Fig. 4C are from the same 45 ms CT interval session as Fig. 4A, but here the LLF and cLLF are illustrated instead. The size of cLLF (magenta trace) was significantly smaller than LLF (green trace) at this interval (142 ± 14%vs.191 ± 24%;P = 0.049). This was also the case for CT intervals 10, 25, 30, 35 and 40 ms (Fig. 4D). However, at CT intervals 60 and 70 ms cLLF was significantly larger than LLF (Fig. 4D).

It thus appears that the conditioning CPN stimulation has a different effect on SLF and LLF. Nevertheless, it is not evident from Fig. 4 if these differences were simply due to the effect of the CPN stimulation on the H-reflex (dashed line in Fig. 4BandD). To obtain an estimate of the effect of the CPN stimulation on SLF and LLF (i.e. the corticospinal transmission in the direct monosynaptic pathway and indirect longer-latency pathway, respectively) independent of the effect of the CPN stimulation on the H-reflex (i.e. the spinal motoneurons), we subtracted the effect of CPN stimulation on the H-reflex (cH) from the combined effect of TMS and CPN stimulation (cSLF and cLLF) and compared the result of these subtractions to the effect of TMS on the H-reflex (SLF and LLF). This makes it possible to make a direct comparison between the size of SLF/LLF with and without CPN stimulation (Fig. 5).

At a group level, this comparison revealed a significant difference between the CPN effect on SLF (0 ± 9%) and LLF (–47 ± 16%) at a CT interval of 45 ms (P = 0.004; Fig. 5A). There was no difference at any other CT interval. As the latency between the conditioning CPN stimulation and the conditioning TMS pulse was 11–14 ms shorter for LLF than SLF, we performed a control experiment in four subjects at each interval in which the CPN stimuli were matched with the TMS latencies producing SLF and LLF in the same trial for each subject (Fig. 5B). Given that the 45 ms CT interval was the only interval with a significant difference between the CPN effect on SLF and LLF in Fig. 5A, we focused on CT intervals 40, 45 and 50 ms in this control experiment. For the subject example in Fig. 4 with a SLF of –1.5 ms and LLF of 10 ms, it means that at the CT 45 ms interval the CPN stimulation was now applied 33.5 ms prior to the test H-reflex (instead of 45 ms) for the cSLF whereas the cLLF interval was still 45 ms (thereby taking into account the 11.5 ms difference in the application of the subthreshold TMS pulse). An extra cH alternative where the CPN was applied 33.5 ms before the test H-reflex was also included in order to make a direct comparison between the size of SLF/LLF with and without CPN stimulation as in Fig. 5A. In all four subjects, the CPN stimulation reduced the LLF compared to the SLF at the 40 ms CT interval and in three of the subjects at the 45 ms CT interval. For the 50 ms CT interval there was no clear effect (Fig. 5B). The tendency was thus the same for the control experiments, but the effect was more pronounced for the 40 ms CT interval, which is probably due to the matching of latencies used here or differences from subject to subject, as one subject had a stronger effect at the 45 ms CT interval (red triangle). Such inter-individual variability could be due to differences in nerve conduction time or height of the subjects.

It thus appears that CPN stimulation can activate spinal inhibitory mechanisms that influence the indirect corticospinal pathway to soleus.

Study 3: CPN-induced inhibition of soleus MEPs disappears prior to dorsiflexion contraction

We next investigated if this spinal inhibitory mechanism could explain our previous observation that soleus MEPs are increased prior to the onset of dorsiflexion (Geertsen et al. 2010).

The average reaction time was 135 ± 10 ms (mean ± SD). The CT intervals producing an inhibition (36 ± 3 ms; mean ± SD) and facilitation (55 ± 4 ms; mean ± SD) of the soleus MEP at rest was evoked prior to the onset of dorsiflexion (–25 ms) and in a control condition (<C_go). The CT interval that produced inhibition of the soleus MEPs at <C_go did not produce an inhibition at –25 ms (80 ± 8%vs.103 ± 8%;P = 0.01; Fig. 6A). There was no significant difference between the conditioned MEP size at –25 ms and <C_go at the CT interval producing facilitation (119 ± 10%vs.107 ± 6%;P = 0.12; Fig. 6B).

These findings suggest that there is a removal of spinal inhibition of soleus MEPs prior to the onset of dorsiflexion.

Discussion

Here, we demonstrate that CPN stimulation evokes an inhibition of the soleus MEP, which is in all likelihood mediated by a spinal inhibition of the descending corticospinal input to soleus motoneurons. This inhibition is either absent or greatly reduced prior to dorsiflexion. This observation provides one explanation for the facilitation of the soleus MEPs prior to dorsiflexion, which has been recently reported (Geertsen et al. 2010).

What is the origin of the afferent-induced facilitation of soleus MEPs?

In a recent study, Roy & Gorassini (2008) reported that TA MEPs were facilitated when preceded by tibial nerve stimulation (at the ankle) at CT intervals around 45–50 ms, while H-reflexes and responses evoked from corticospinal tract stimulation at the brainstem were depressed at these intervals. This result led them to suggest that the TA MEP facilitation was of cortical origin whereas the inhibition was in all likelihood explained by a spinal inhibitory mechanism (Roy & Gorassini, 2008). In our study, a facilitation of soleus MEPs was observed at CT intervals around 50–55 ms, which corresponds quite well to the latency reported by Roy & Gorassini (2008), as well as by other studies in which transcortical reflex pathways have been explored (Nielsen et al. 1997; Petersen et al. 1998). The increase of the TMS-induced short-latency facilitation (SLF) of the soleus H-reflex at a corresponding latency following the CPN stimulation is also in line with a cortical origin of the facilitation, since SLF has been shown to be caused by activation of direct monosynaptic corticospinal projections to the soleus motoneurons (Nielsen et al. 1993). However, it should be pointed out that we also observed a facilitation of the control soleus H-reflex as well as the TMS-induced long-latency facilitation (LLF) of the reflex. Therefore, we cannot exclude the possibility that the facilitation may be partially or fully explained by a subcortical mechanism.

What is the origin of the afferent-induced inhibition of soleus MEPs?

The earlier occurring inhibition of the MEP had such a short latency that it cannot be explained by a cortical mechanism, but more probably, as was also suggested by Roy & Gorassini (2008) in the case of their tibial-induced inhibition of the TA MEP, by a spinal inhibitory mechanism. This notion is also supported by the observed lack of SLF inhibition, because a spinal inhibition would not influence transmission in the direct corticomotoneuronal pathway. As the control H-reflex was not inhibited by the CPN stimulation at these intervals it also seems reasonable to assume that the inhibition does not affect the motoneuronal excitability directly, but rather that it affects transmission in the corticospinal pathways at a premotoneuronal site. In order for this to happen, part of the corticospinal input to the soleus motoneurons that is elicited by TMS must be mediated through subcortical, i.e. spinal, neurons. It is well known that the majority of the corticospinal tract terminates in the intermediate zone of the spinal cord and that the motor command is mediated through spinal interneurons to the motoneurons (Kuypers, 1981; Ralston & Ralston, 1985; Isa et al. 2007). Evidence that activation of such pathways may contribute to the size of the soleus MEP especially has been presented previously (Nielsen et al. 1993; Nielsen & Petersen, 1995a,b;). For example, Nielsen et al. (1995b) suggested that the LLF was mediated by an indirect corticospinal pathway. The observation that the LLF was reduced at a latency corresponding to the inhibition of the MEP suggests that it is inhibition of transmission through this indirect corticospinal pathway which is responsible for the MEP inhibition.

We cannot make firm conclusions regarding the more precise nature of the neurons responsible for the inhibition or of the indirect corticospinal pathway that is being inhibited. However, one likely candidate for the latter is the lumbar propriospinal pathway mediating non-monosynaptic group I and group II excitation (Cavallari et al. 1987; Edgley & Jankowska, 1987; Forget et al. 1989; Marchand-Pauvert et al. 1999; Marchand-Pauvert & Nielsen, 2002; see also Pierrot-Deseilligny & Burke, 2005). These non-monosynaptic group I/II interneurons mediate part of the corticospinal command to lumbar motoneurons and are also under efficient inhibitory control by local inhibitory interneurons (Edgley & Jankowska, 1987; Marchand-Pauvert et al. 1999; Simonetta-Moreau et al. 1999). We propose that our observations may be explained by the activation of inhibitory interneurons, which inhibit specifically corticospinal transmission through group I/II interneurons, as illustrated in Fig. 7.

Figure 7. Diagram sketching the proposed mechanism of the CPN-induced inhibition and its removal prior to dorsiflexion.

The corticospinal input to soleus motoneurons is mediated both by direct, corticomotoneuronal pathways and by indirect pathways through spinal neurons – probably lumbar propriospinal neurons (PNs). These PNs receive strong excitatory input from group I/II afferents, but are also under inhibitory control from group I afferents. The inhibition of the soleus MEP is probably due to inhibition of transmission in this indirect pathway, through spinal inhibitory interneurons (INs) activated by group I afferents from the conditioning CPN stimulation. Prior to onset of dorsiflexion, the spinal inhibition is removed, which is most probably due to the descending motor command. Several mechanisms can be responsible for this disinhibition: (1) Postsynaptic inhibition of the IN mediating the inhibition of the PN (mediated by INs, activated by the corticospinal tract). (2) Presynaptic inhibition of the group I afferents that are activating the IN mediating the inhibition of the PN (mediated by local INs activated by the corticospinal tract – dashed line). (3) A reduction in the feedforward inhibition of the PN from the corticospinal tract, by reducing the input to the IN mediating the inhibition. (4) In cats, the lumbar propriospinal neurons receive strong input from the reticulospinal tract. It is possible that the removal of inhibition prior to onset of dorsiflexion is mediated from this tract through one of the above-mentioned mechanisms. The classical disynaptic reciprocal inhibition pathway, mediated by the Ia inhibitory interneurons (Ia) that receive convergent input from the corticospinal tract and the Ia afferents from TA, is also illustrated.

Removal of inhibition prior to onset of dorsiflexion

The removal of the inhibition prior to the onset of dorsiflexion is probably due to the descending motor command for dorsiflexion. We can only speculate about the mechanism responsible for this disinhibition, and there are several possible candidate mechanisms as illustrated in Fig. 7. First, it is possible that the local inhibitory interneurons, mediating the inhibition of the group I/II neurons in the indirect corticospinal pathway, are themselves inhibited or at least disfacilitated. The group I/II neurons have been shown to receive feed-forward inhibition, through local inhibitory interneurons, from the corticospinal tract (Simonetta-Moreau et al. 1999).

In cats, the lumbar propriospinal neurons receive strong input from the reticulospinal tract (Davies & Edgley, 1994; Matsuyama et al. 1999). It is possible that the removal of inhibition prior to onset of dorsiflexion is mediated from this tract through one of the above-mentioned mechanisms. This could be part of a subcortical motor program activated prior to onset of dorsiflexion as recently suggested (Geertsen et al. 2010).

What is the functional significance of this removal of inhibition prior to onset of dorsiflexion?

Prior to the onset of dorsiflexion, it appears that central motor areas send supraliminal commands to the TA and subliminal collateral influences in the interneuronal pathways directed to the soleus muscle. It seems somewhat contradictory that part of the supraspinal control of agonist–antagonist pairs appears to involve a parallel activation pattern and another part of the classical reciprocal inhibitory activation pattern (Fig. 7). Why this apparent competition of facilitatory and inhibitory influences?

As discussed in our recent study (Geertsen et al. 2010), one possibility is that the facilitation of the antagonist may help to ensure that quick transitions from dorsiflexion into plantar flexion may be made more efficiently than if the plantar flexors would need first to be strongly depolarized from a hyperpolarized state before becoming active. Removal of spinal inhibition prior to onset of dorsiflexion makes sense in this perspective, as it brings the spinal interneurons closer to threshold, so that only small changes in the descending drive through this indirect pathway to soleus motoneurons would be necessary to switch from one movement direction to another.

Such quick switches between antagonistic muscle activities may be of special importance in relation to high-performance sports.

Conclusion

We propose that the removal of spinal inhibition prior to the onset of dorsiflexion might explain why soleus MEPs are increased prior to the onset of dorsiflexion. We suggest that it is the indirect corticospinal pathways, possibly through group I/II propriospinal interneurons, to soleus motoneurons that are affected by the conditioning stimulus, since the inhibition induced by the conditioning CPN stimulation is observed in the LLF pathway and not in the SLF pathway, and because MEPs but not H-reflexes were inhibited. The data illustrate that spinal interneuronal pathways modify descending commands to human spinal motoneurons and influence the size of MEPs elicited by TMS. Changes in the transmission and excitability of spinal interneuronal pathways should therefore be considered when evaluating changes in MEPs during different tasks.

Glossary

Abbreviations

C_go

auditory go cue

C_warning

auditory warning cue

cH-reflex

CPN conditioned Hoffmann reflex

cLLF

CPN conditioned long-latency facilitation

CPN

common peroneal nerve

cSLF

CPN conditioned short-latency facilitation

CT

conditioning–test

EMG

electromyography

H-reflex

Hoffmann reflex

ICF

intracortical facilitation

ICI

intracortical inhibition

LLF

long-latency facilitation

M_max

maximal M-response

MEP

motor-evoked potential

MT

motor threshold

MVC

maximal voluntary contraction

PTN

posterior tibial nerve

SAI

short-latency afferent inhibition

SLF

short-latency facilitation

TA

tibialis anterior

TMS

transcranial magnetic stimulation

Author contributions

The experimental work was performed in the Copenhagen Neural Control of Movement (CPH-NCM) laboratory at the Panum Institute, the University of Copenhagen, Copenhagen, Denmark. All authors contributed to the concept and design of the experiments as well as to the collection, analysis and interpretation of data. S.S.G. and J.B.N. drafted the manuscript and all authors critically revised the manuscript and approved the final version for publication.