Reciprocal Ia inhibition contributes to motoneuronal hyperpolarisation during the inactive phase of locomotion and scratching in the cat

Non-technical summary

During a movement, the contraction of a given muscle group is often coordinated with the simultaneous relaxation of its antagonist muscles. The neural basis of this antagonist relaxation has been investigated in both animal and human experiments for decades and it is believed that activation of the Ia inhibitory interneurones by central motor programmes plays a major role in this relaxation of antagonist muscles. The alternating movements during locomotion would seem to especially require reciprocal actions, but recent studies have raised significant questions about the role of this inhibition. We found that inhibition evoked by these inhibitory interneurones is largest when their target motoneurones are inactive – even in the absence of supraspinal influence. The results of this work provide new evidence for the role of the Ia inhibitory interneurones during rhythmic motor activity. This supports the classical view of reciprocal inhibition as a basis for antagonist relaxation.

Abstract

Despite decades of research, the classical idea that ‘reciprocal inhibition’ is involved in the hyperpolarisation of motoneurones in their inactive phase during rhythmic activity is still under debate. Here, we investigated the contribution of reciprocal Ia inhibition to the hyperpolarisation of motoneurones during fictive locomotion (evoked either by electrical stimulation of the brainstem or by l-DOPA administration following a spinal transection at the cervical level) and fictive scratching (evoked by stimulation of the pinna) in decerebrate cats. Simultaneous extracellular recordings of Ia inhibitory interneurones and intracellular recordings of lumbar motoneurones revealed the interneurones to be most active when their target motoneurones were hyperpolarised (i.e. in the inactive phase of the target motoneurones). To date, these results are the most direct evidence that Ia inhibitory interneurones contribute to the hyperpolarisation of motoneurones during rhythmic behaviours. We also estimated the amount of Ia inhibition as the amplitude of Ia IPSC in voltage-clamp mode. In both flexor and extensor motoneurones, Ia IPSCs were always larger in the inactive phase than in the active phase during locomotion (n= 14) and during scratch (n= 11). Results obtained from spinalised animals demonstrate that the spinal rhythm-generating network simultaneously drives the motoneurones of one muscle group and the Ia interneurones projecting to motoneurones of the antagonist muscles in parallel. Our results thus support the classical view of reciprocal inhibition as a basis for relaxation of antagonist muscles during flexion–extension movements.

Abbreviations 

CDP

potentials recorded from the dorsum of the spinal cord

ENG

electroneurogram

FDHL

flexor digitorum and hallucis longus

GS

lateral and medial gastrocnemius and soleus

IN

interneurone

LGS

lateral gastrocnemius plus soleus

MG

medial gastrocnemius

MLR

mesencephalic locomotor region

MN

motoneurone

PBSt

posterior biceps and semitendinosus

Pl

plantaris

Q

quadriceps

RF

rectus femoris

Sart

Sartorious

SmAB

semimembranosus and anterior biceps

TAEDL

tibialis anterior and extensor digitorum longus

T

threshold

Tib

posterior tibial

Vasti

vastus lateralis, medialis and intermedialis

Introduction

During a movement, the contraction of a given muscle group is often coordinated with the simultaneous relaxation of its antagonist muscles. The neural basis of this antagonist relaxation has been investigated in both animal and human experiments for decades (Orlovsky et al. 1999; Pierrot-Deseilligny & Burke, 2005). The majority of this research has focused on the potential role that the spinal interneurones (INs) mediating reciprocal inhibition from primary spindle endings (Ia afferents) play in this relaxation of antagonist muscles (Eccles et al. 1956; Hultborn et al. 1971; Jankowska & Roberts, 1972). These INs receive converging excitatory and inhibitory inputs from a large number of segmental and descending pathways (see Hultborn, 1972; Baldissera et al. 1981). Importantly, the functional pattern of excitatory inputs received by motoneurones (MNs) to one set of muscles appears to be replicated onto the Ia inhibitory INs projecting to the MNs of the antagonists (Hongo et al. 1969; Lundberg, 1970). This similarity in the convergence onto ‘corresponding’ MNs and Ia inhibitory INs led Lundberg to hypothesise that both MNs and Ia INs are activated in parallel during voluntary movements in order to secure a coordinated contraction of agonists and relaxation of antagonists (Hongo et al. 1969; Lundberg, 1970). This hypothesis entails that activation of the Ia INs would contribute to the hyperpolarisation (i.e. inactivation) of their target MNs.

In the cat, direct recording of the activity of Ia inhibitory INs is possible in relation to fictive locomotion (Feldman & Orlovsky, 1975; Pratt & Jordan, 1987) or scratching (Deliagina & Orlovsky, 1980). Consistent with Lundberg's hypothesis, Ia inhibitory INs have been found to be more active when their (assumed) target MNs were silent as monitored by electroneurograms (ENGs). Non-invasive human experiments have also provided indirect evidence of a similar functional regulation of disynaptic reciprocal Ia inhibition in relation to walking (Petersen et al. 1999) and bicycling (Pyndt et al. 2003).

Studies focusing on the transmission in the pathway mediating the disynaptic Ia inhibition (by recording Ia IPSPs in the target MNs) throughout the locomotor cycle in decerebrate cats, however, have produced inconclusive results (Pratt & Jordan, 1987). In most cases the Ia IPSPs had the largest amplitudes during the active phase. These findings are contrary to what would be expected given the results of the studies directly recording from the Ia inhibitory INs and also contrary to what would be expected if Ia inhibitory INs do indeed contribute to the hyperpolarisation of MNs during their inactive phase. Pratt & Jordan (1987) have suggested that these findings may not reflect the excitability of the pool of Ia inhibitory INs, but rather be explained by the cyclic depolarisation of the MN itself, with IPSPs being larger as the membrane potential is further away from the IPSP equilibrium potential (Coombs et al. 1955). Another confounding factor in studies measuring Ia IPSP amplitude is related to variations in the subliminal range/discharge zone in the pool of Ia inhibitory INs during the step cycle. If the excitatory drive from the locomotor network is very strong during a part of the locomotor cycle it could actually cause a decrease of the test Ia IPSP due to occlusion.

There is evidence from the spinal cord adult turtle preparation (in which connections from the brainstem to the spinal cord are removed) that the motoneurones receive strong inhibition during the phase of the scratch cycle in which they are inactive (Robertson & Stein, 1988; Alaburda et al. 2005; reviewed in Stein, 2010). However, recent evidence from the same preparation revealed that inhibitory input is also present in the active phase of MNs and in terms of inhibitory conductances, it is in fact larger during the active than the inactive phase of the MNs during scratching (Berg et al. 2007). This challenges the classical view of the role of reciprocal inhibition and the parallel drive of MNs and Ia inhibitory INs. There may be several explanations for this apparent inconsistency between the findings in the turtle (Berg et al. 2007) and the hypothesised role of Ia inhibition in the cat. For example, it might be a consequence of species differences or it could be related to the lack of descending input from supraspinal centres in the turtle preparation since all previous data relating to reciprocal Ia inhibition during fictive motor activity in the cat is from preparations with intact brainstem–spinal cord connectivity.

In the present study, we set out to resolve these issues and to provide conclusive data regarding the activity of Ia inhibitory INs during fictive rhythmic activity (locomotion and scratching) and their contribution to inhibition in the inactive (hyperpolarised) phase of their target MNs. By recording from Ia INs simultaneously with a target MN during fictive motor behaviour in both spinalised and decerebrate cats we could demonstrate that individual Ia inhibitory INs become (more) active during the inactive phase of their target MNs. By using a voltage-clamp protocol, we eliminated the confounding factor of the cyclic fluctuations in membrane potential and could demonstrate that the reciprocal inhibition (the Ia IPSCs) evoked by stimulation of antagonist nerves is always largest when the MNs are in their inactive phase. Taken together, these results provide conclusive evidence in support of the parallel activation of the MNs innervating a muscle group and the Ia INs projecting to the antagonist muscles during rhythmic motor activity.

Methods

Experiments were performed on 12 adult cats of either sex weighing 3.5–6.4 kg. All surgery and experimental protocols were conducted in accordance with EU regulations (Council Directive 86/609/EEC) and with National Institutes of Health guidelines for the care and use of laboratory animals (National Institutes of Health publication no. 86-23 revised 1985). All procedures were approved by the Danish Animal Experimentation Inspectorate. Our experimental procedures also comply with the polices set out by The Journal of Physiology (Drummond, 2009).

Preparation

After the induction of anaesthesia with halothane/isoflurane–nitrous oxide (2–3% halothane, 70% N₂O and 30% O₂), the animals were intubated, and cannulas were inserted into the right carotid artery and the right jugular vein (near the trachea) for the monitoring of blood pressure and the administration of fluid and drugs. Atropine (0.1 mg kg⁻¹, s.c.) and solumedrol (2.5 mg kg⁻¹, i.v.) were administered at the beginning of the experiment whilst a buffer solution (10% dextrose and 1.7% NaHCO₃) was infused continuously (∼4.5 ml h⁻¹) after cannulation. The cats were decerebrated, either by an anaemic decerebration (n= 10) or by a precollicular transection and removal of the rostral parts of the brain (n= 2). The experiments using precollicular transections were mainly used for data collection for other studies.

Anaemic decerebrations were performed under anaesthesia by ligating the basilar and both common carotid arteries, a procedure that has been shown to produce a decerebration that includes all cortical tissue above the pons and the anterior part of the cerebellum (Pollock & Davis, 1930; Crone et al. 1988). Following the transfer of the cat to the recording frame, the last carotid artery was ligated and the anaesthetic removed. The decerebration was verified to be clinically complete by the development of tonic extensor muscle tone, lack of spontaneous movements, and large non-reactive pupils (Crone et al. 1988). In 1/10 cats, stereotyped stepping movements developed after the removal of anaesthetics and in this case the anaesthesia was immediately reinstated and the brain was removed rostral to a section at the level of the superior colliculi. In this case, the brain was found to be necrotic, and it was concluded that the stepping movements must have originated from the caudal brainstem centres. Following these procedures and tests, pancuronium bromide (0.6 mg h⁻¹) was administered to block neuromuscular transmission, and artificial respiration was initiated.

A precollicular decerebration (2 cats) was made, during anaesthesia, at the end of the surgical procedure after the animal had been transferred to the recording frame with a stereotactic head-holder (Gossard et al. 1994). Following this decerebration, pancuronium bromide (0.6 mg h⁻¹) was administered to block neuromuscular transmission, and artificial respiration was initiated.

The following nerves from the left hindlimb were cut and dissected for stimulation and recording: quadriceps (Q), sartorius (Sart), semimembranosus and anterior biceps (SmAB), posterior biceps and semitendinosus (PBSt), medial gastrocnemius (MG), lateral gastrocnemius plus soleus (LGS), tibialis anterior and extensor digitorum longus (TAEDL), flexor digitorum and hallucis longus (FDHL) and plantaris (Pl). On the right side, typically the following nerves were dissected: Sart, rectus femoris (RF), vastus lateralis, medialis, and intermedialis (Vasti), lateral and medial gastrocnemius and soleus (GS), and posterior tibial (Tib). The Q, RF, Vasti and Sart nerves were placed in cuff electrodes, and the other nerves were mounted on bipolar silver–silver chloride hook electrodes. The nerve branches of the right tibial and common peroneal nerves were sectioned but were usually not mounted. Other hindlimb nerves including the obturator nerves were all denervated bilaterally. The tendons around the hip were cut bilaterally to avoid sensory information of limb position reaching the spinal cord as it may interfere with the expression of fictive locomotion (Kriellaars et al. 1994). To expose the lumbar spinal cord a laminectomy over the L4–L7 vertebral columns was performed. As the animals were transferred to the recording frame (including a stereotactic head-holder), both the spinal column and head were mechanically secured. The skin flaps around the exposed areas of the spinal cord and the hindlimb were sewn and retracted to form pools that were filled with warm paraffin oil. A pneumothorax was performed bilaterally prior to recording to reduce movement artefacts. A decrease of blood pressure to <80 mmHg was compensated for by i.v. administration of additional amounts of buffer solution. The expired CO₂ was maintained between 3 and 6% and the animals were kept at around 37°C core temperature by feedback-controlled heating lamps.

In 9/10 cats in which anaemic decerebration was performed, a complete spinalisation was performed at the first cervical (C1) level following a local injection of lidocaine (SAD, 20 mg ml⁻¹) into the spinal cord.

Experiments were terminated by an overdose of pentobarbital.

Fictive motor activity

Nialamide, a monoamine oxidase inhibitor (i.v. 50 mg kg⁻¹, Sigma), was administered at a slow rate during the hindlimb surgery in experiments aimed at investigating DOPA-locomotion (anaemic decerebration was used in these instances). l-DOPA (50 mg kg⁻¹) (Grillner & Zangger, 1979) was administered i.v. after the animals were mounted in the frame and spinalised at the C1 level. In some animals, a second dose of l-DOPA (given >6 h after the first dose) was also successful in evoking rhythmic activity lasting for another 4–6 h period.

In animals in which brainstem stimulation was used to evoke fictive locomotion, tungsten electrodes (0.1–0.3 MΩ, World Precision Instruments, Sarasota, FL, USA) were placed according to stereotaxic coordinates of the mesencephalic locomotor region (MLR) (Shik et al. 1969; Gossard et al. 1994) and the final electrode positioning was determined by monitoring rhythmic hindlimb ENG activity. Pulses of 0.1–0.2 ms duration were applied at frequencies of 10–20 Hz and at strengths of 50–150 μA.

In all animals, a combination of 4-AP (i.v. 100 μg kg⁻¹) and naloxone (i.v. 100–200 μg kg⁻¹) was used to enhance the fictive locomotor activity (see e.g. Pearson et al. 1992; Perreault et al. 1999).

The C1–C2 spinal segments were exposed in order to apply d-tubocurarine (0.1–0.3%) (Domer & Feldberg, 1960), and/or bicucculine (0.1–1%) (Baev et al. 1981) topically at the C1 and C2 dorsal root entry zones prior to mechanical stimulation of the skin covering the ears and the surrounding face in order to evoke fictive scratch activity.

Recording of interneurone activity

Extracellular recordings from identified Ia inhibitory INs were collected in three experiments. The criteria for identifying Ia inhibitory INs included the following: (1) action potentials could be evoked by peripheral nerve stimulation (low threshold group I fibres) at latencies <0.9 ms and could follow >200 Hz stimuli, (2) an inhibition of IN activity could be evoked by ventral root stimulation (i.e. recurrent inhibition; Hultborn et al. 1971; Hultborn, 1972). We preferentially aimed to record from INs receiving input from hip or ankle flexors and MNs innervating hip or ankle extensors. After identification, a bout of fictive motor activity was evoked and repeated typically 2–3 times. In a few cases, locomotion and scratch activity could be evoked consecutively while maintaining stable recording conditions for the same IN.

As one of the aims of this experiment was to clarify the relationship between the activity of Ia INs and MNs, we also recorded pairs of identified INs and MNs (Jankowska & Roberts, 1972) during bouts of fictive motor activity. This consisted of identifying a Ia IN (activated from Sart or Q) in the 5th lumbar segment at a depth of 2.9–3.8 mm, by the criteria stated above, and then intracellularly recording from an antidromically identified MN (SmAB or PBSt) in the 7th lumbar segment using a separate glass microelectrode.

Recording of disynaptic IPSPs/IPSCs during the locomotor cycle

Intracellular recordings were obtained (with a microelectrode amplifier; Axoclamp 2A, Molecular Devices, Sunnyvale, CA, USA) from antidromically identified MNs innervating the left hindlimb. Glass micropipettes (tip diameter, 1.2–2 μm; resistance, 2–4 MΩ) filled with the lidocaine derivative QX-314 (100–200 mm) in potassium acetate (2 m) were used. QX-314 blocks sodium spikes, allowing the recording of Ia IPSPs throughout the step/scratch cycle without interference of spike potentials. Only MNs with a stable membrane potential more hyperpolarised than −50 mV and spike potentials of more than 60 mV (before the action by QX-314) were accepted.

The recordings from MNs were used for recording (1) locomotor drive potentials, (2) unitary IPSPs as part of the most rigid identification of the Ia inhibitory INs (see preceding section), (3) Ia IPSPs to assess the transmission in the Ia inhibitory pathway, and (4) Ia IPSCs in voltage clamp mode.

The threshold (T) was defined as the intensity required for activating the largest diameter afferents in each individual nerve. The disynaptic reciprocal Ia IPSPs were evoked by stimulating group I afferents in the nerves from antagonist muscles. The electrical stimulus strength (single pulse) was adjusted to evoke IPSPs both at minimal amplitude (with stimulation about 1.1 ×T) and also as large as possible (with supramaximal stimulation for the Ia afferents; i.e. 2–5 ×T, see below).

One confounding factor when judging the convergence onto the pool of Ia inhibitory INs using changes in the amplitude of the Ia IPSPs recorded from the MNs is introduced when the membrane potential of the recorded MN is changing significantly. This is indeed the case during fictive motor behaviour. When the membrane potential is depolarised (i.e. during the active phase of the locomotor and scratch drive potentials) the difference between the current membrane potential and the equilibrium potential for IPSPs is of course increased resulting in larger amplitudes even with activation of the same number of inhibitory INs (Coombs et al. 1955). In order to resolve this complication we also recorded the Ia inhibition at a fixed membrane potential by recording the Ia IPSCs in single electrode voltage-clamp mode of the microelectrode amplifier (Stuart & Redman, 1990). The switching frequency was typically 3 kHz, and thus the efficiency of the clamp was not optimal for the beginning of the IPSP.

Another confounding factor is related to changes in the subliminal fringe (the number of neurones that are excited, but have not yet reached firing threshold) versus the discharge zone (the number of neurones that have reached firing threshold) in the pool of Ia inhibitory INs during the step/scratch cycle. To allow a facilitation of Ia IPSPs as the pool of Ia INs are excited by the locomotor circuits, it is important to keep input from the Ia afferents as weak as possible. Indeed, if the Ia volley is large enough to discharge most of the INs, an additional excitation in a particular phase of locomotion or scratch may actually cause a decreased Ia IPSP due to occlusion. In order to assess this factor, we often used two kinds of test Ia IPSPs, the first by weak stimulation of the antagonist nerve to evoke a Ia IPSP of minimal amplitude, and the second by a supramaximal stimulation for Ia afferents (2–5 ×T) to discharge most (or all) Ia inhibitory INs.

Data capture and analysis

ENG recordings were digitized at a rate of 10 kHz and filtered (5 Hz to 1 kHz). Potentials recorded from the dorsum of the spinal cord (CDP), IN recordings (AC coupled) and MN intracellular (i.c.) recordings (DC coupled) were digitized at a rate of 10–20 kHz. All recordings were collected and stored on a PC for offline analysis (CED 1401+ with Spike 2.5 software, Cambridge Electronic Design, Cambridge, UK). The ENG recordings from each nerve were rectified and smoothed in order to determine the onset of flexor and extensor bursts by visual inspection.

The extracellular recordings of the activity of Ia INs were low-pass filtered. Built-in spike sorting algorithms within the Spike 2.5 software were used to follow the activity of the neuron identified by the criteria listed above as a Ia inhibitory IN. Extracellular (e.c.) IN recordings as well as i.c. recordings from MNs were selected for further analysis from a bout of rhythmic activity with a minimum of five consecutive cycles with alternating flexion and extension. Cycle-based analysis of IN firing and IPSP amplitude was based on flexor bursts as determined by the onset of a flexor (either Sart or TA) ENG when passing a threshold, which was selected by visual inspection of the bout. Comparisons across different experiments were possible by normalizing the step cycles. Cycle duration was determined as the time between consecutive onsets of flexor activity; flexor phase duration was determined as the time period when the flexor ENG activity was larger than the determined thresholds. The time points when the minimum and maximum MN membrane potentials were measured were determined for each cycle and expressed as a percentage of the cycle (i.e. normalised). The time points at which IN firing started and stopped as well as the time point of maximal instantaneous firing were determined for each cycle and they were also normalised. Spike counts are reported as per 50 ms long bins during non-normalised cycles and as per cycle for normalised cycles and as mean firing frequency for flexion and extension averaged separately from several cycles. We verified that IPSPs were evoked disynaptically by confirming that the onset latency was between 1.2 and 1.5 ms from the arrival of the incoming volley at the cord dorsum. The membrane potential just prior to stimulation and the peak amplitude of the IPSPs were measured using averaged records during the active (i.e. depolarised) and the inactive (i.e. hyperpolarised) phases of each MN. Detailed analysis of IPSP modulation was performed in some cases by dividing the step or scratch cycles into 10 bins. The IPSPs evoked by near threshold stimulation were expressed both in absolute values (mV) and as percentages of the maximal IPSPs evoked by supramaximal stimulation. In voltage clamp mode, the peak amplitude of the IPSCs was measured during both the active and the inactive phases. Statistical comparisons between the amplitudes measured in the two phases were performed using Student's t test for paired data during each type of rhythmic motor activity. The IPSCs during rest (i.e. no ENG activity in any of the recorded nerves) and during tonic flexor activity preceding rhythmic scratch were also compared to the two phases during rhythmic scratch when these periods were sufficiently long that at least 10 single records could be averaged.

Results

Identified Ia inhibitory IN–MN pairs recorded during fictive locomotion and scratch

Twelve Ia inhibitory INs were investigated during fictive motor activity. They were all monosynaptically activated by low threshold muscle afferents, followed stimulation frequencies >200 Hz and received Renshaw inhibition following stimulation of the central part of sectioned ventral roots (Fig. 1A). In 5 of the 12 investigated INs, simultaneous recordings of one or more of their target MNs were obtained. We recorded the activity of these IN–MN pairs during fictive locomotion mostly in cats with intact spinal cords (spontaneous locomotion was enhanced by pharmacological agents). One of these pairs was recorded after spinalisation at the C1 level and administration of l-DOPA to evoke locomotion. Three of the IN-MN pairs were also examined during fictive scratch.

Figure 1. Simultaneous recordings from a Ia inhibitory interneurone and its target motoneurone during fictive locomotion and scratch A, schematic illustration of the methodology used in this study. The identification of Ia interneurones (INs) was performed by stimulation of peripheral nerves and of partially dissected ventral roots (VR). Intracellular (i.c.) recordings from motoneurones (MNs) were collected simultaneously while recording the extracellular (e.c.) activity of identified Ia INs and monitoring the rhythmic activity by electroneurogram (ENG) recordings. B, identification of Ia INs by stimulation of the Sart muscle nerve with a train of 22 pulses (filled arrows) at 250 Hz. The lower trace illustrates recordings from the dorsum of the spinal cord (CDP) at the 7th lumbar (L7) segmental level. C, stimulation of VRs partially dissected at the L7 level (open arrow) inhibited the activity of this IN. D, spike triggered average of the interneuronal spike and a SmAB MN membrane potential showing the unitary IPSP. The first and second dotted lines indicate the onset of the spike and the onset of the unitary IPSP, respectively, and the period in between them represents the onset latency of the unitary IPSP, 0.8 ms. The arrow shows the point where the peak amplitude (0.02 mV) was measured. E and F, interneuronal activity (Ia IN e.c.), the instantaneous firing frequency (Inst. f.f.), and intracellular recordings from a SmAB MN and raw ENGs from the flexor Sart and the extensor SmAB during a 5.5 s long bout of fictive locomotion developed spontaneously and enhanced by i.v. administration of 4-AP and naloxone in a cat with intact spinal cord (E) and during a 2 s long bout of fictive scratch evoked by stimulation of the pinna after curare application to the dorsal roots at the first and second cervical level (F). Note that the SmAB MNs illustrated in E and F are different. G and H, averaged MN i.c. recordings on top and rectified ENGs from the same nerves as in panels E and F from 5 cycles (mean duration of 2.70 ± 1.67 s) during locomotion (G) and from 10 cycles (mean duration of 0.30 ± 0.09 s) during scratch (H). Non-normalised averages are aligned by the onset of Sart activity with the vertical dotted line indicating the onset of flexion. The firing during each cycle is shown in the raster plot.

One of the INs examined during such spontaneously developing fictive locomotion in a spinal intact animal is illustrated in Fig. 1. This Ia IN received monosynaptic input from the hip flexor Sart muscle nerve (Fig. 1B) and its projections were to hip extensor (SmAB) MNs. It followed high frequency stimulation of the Sart nerve but stimulation of the ventral roots inhibited the evoked activity (Fig. 1C). The unitary IPSP evoked in this SmAB MN had a latency of 0.4 ms and peak amplitude of 0.025 mV (Fig. 1D). Three additional Ia IN and target MN pairs showed similar characteristics of unitary IPSPs with 0.3, 0.7 and 0.8 ms latency and 0.04, 0.06 and 0.1 mV peak amplitude. The composite (near maximal) Ia IPSP evoked by Sart muscle nerve stimulation (not illustrated) with a latency of 1.45 ms (measured from the incoming volley) had a peak amplitude of 1.9 mV. As can be seen from Fig. 1E, this Ia IN fired mostly during the efferent activity in the Sart nerve corresponding to the hyperpolarised phase of the SmAB MN recorded simultaneously. The raster plot, below the averaged ENGs in Fig. 1G, illustrates the pattern of IN firing during each of the cycles used for averaging.

The activity of the same Ia IN recorded simultaneously with another SmAB MN during fictive scratch is illustrated in Fig. 1F and H. Note that the rhythmic ENG activity during scratch consists of shorter duration flexion and extension bursts than during locomotion (Fig. 1F) resulting in an average cycle duration of 0.30 ± 0.09 s (Fig. 1H). The activity of the Ia IN was completely restricted to the flexor phase and it was silent during the extensor phase of fictive scratch as depicted by the raster plots in Fig. 1H. Thus the reduction of activity during the extensor phase of scratching was even stronger than during the extensor phase of fictive locomotion.

The activity profile of Ia inhibitory INs during fictive locomotion and scratch is similar with or without a spinal transection

The activity of individual INs in relation to ENG activity during locomotion and scratch is summarised in Table 1. Figure 2 illustrates the pooled data of the mean firing frequencies during locomotion (Fig. 2A) and scratch (Fig. 2B and C) after normalizing the cycles. Note the similar modulation of the firing frequency of flexor-coupled INs (Fig. 2A and B) in preparations with intact spinal cord (filled circles) and in those spinalised at the C1 level (squares). Three of the six INs recorded during fictive locomotion were also recorded during fictive scratch (see a, b and c in Table 1). During fictive locomotion, Ia INs with input from flexors (4 from Sart and 1 from TAEDL) were mainly active during the flexion phase and the firing frequency was reduced completely at some point during extension (e.g. Fig. 1). The activity of one Ia IN recorded after spinalisation at the C1 level showed a similar pattern of modulation to those recorded prior to the spinalisation (Fig. 2A).

Table 1.  Summary of Ia inhibitory IN activity during fictive DOPA-locomotion and scratching

Locomotion input	Cycle (s)	Max ff (Hz)	Mean FLX (Hz)	Mean EXT (Hz)	Cycles (n)
Columns: IN input; cycle duration; the maximal instantaneous firing frequency (max ff); the mean firing frequency during flexion (mean FLX) and extension (mean EXT); the number of cycles used for the analysis. The asterisks indicate INs recorded in cats after spinalisation at C1 level. The letters (a, b and c) indicate those INs which were recorded during both fictive locomotion and scratch.
Sart^a	2.70 ± 1.67	576 ± 153	216 ± 40	79 ± 28	5
Sart^b	1.73 ± 0.68	79 ± 20	34 ± 4	2 ± 5	5
Sart	2.61 ± 1.39	96 ± 58	45 ± 8	8 ± 11	13
TAEDL	1.82 ± 0.60	347 ± 99	187 ± 37	33 ± 15	11
Sart*	2.93 ± 0.62	305 ± 123	81 ± 22	30 ± 19	12
Q^c	3.11 ± 2.12	172 ± 126	57 ± 11	58 ± 18	10
Scratch input	Cycle (s)	Max ff (Hz)	Mean FLX (Hz)	Mean EXT (Hz)	Cycles (n)
Sart^a	0.30 ± 0.09	341 ± 375	195 ± 52	54 ± 62	5
Sart^b	0.32 ± 0.06	111 ± 29	53 ± 4	0	7
Sart*	0.16 ± 0.00	399 ± 12	281 ± 13	154 ± 77	14
Sart*	0.17 ± 0.01	143 ± 44	77 ± 12	0	12
Sart*	0.18 ± 0.01	601 ± 262	245 ± 31	144 ± 26	13
Q*	0.18 ± 0.01	443 ± 62	257 ± 13	154 ± 23	15
Q*	0.20 ± 0.13	421 ± 213	182 ± 94	197 ± 77	11
Q^c	0.40 ± 0.13	151 ± 69	60 ± 15	69 ± 22	8
Q and Sart	0.33 ± 0.07	51 ± 17	36 ± 13	28 ± 17	5

Figure 2. Modulation of Ia IN firing during fictive locomotion and scratch Mean firing frequency during locomotion (A) and scratch (B and C) for grouped INs after normalising the cycles. The mean firing frequency is shown during extension (open box, in bins 1–5 during locomotion and 1–4 during scratch) and flexion (filled box, in bins 6–10 during locomotion and 5–10 during scratch). Filled circles represent data from cats with intact spinal cords and open squares represent data from cats spinalised at the C1 level prior to evoking fictive motor activity. Note the similar modulation of IN firing with or without intact spinal cords.

During fictive scratch, the activity of five flexor-coupled Ia INs (input from Sart, Fig. 2B) was reduced during extension for both INs recorded in intact spinal and in spinalised preparations. Four Ia INs with input from Q showed more variable modulation of firing than those with input from flexors (Fig. 2C). Although, our sample size is small, the firing frequency of the INs with Q input was markedly higher in preparations spinalised at the C1 level (squares) than firing frequencies found in intact cats (circles).

Facilitation of Ia IPSPs in MNs in their inactive phase during fictive locomotion in cats without brainstem connections

To investigate the modulation of transmission in the Ia inhibitory pathway, we recorded intracellularly from hindlimb MNs whilst using the combination of a weak and a strong stimulation of the antagonist nerve. We found that the IPSPs evoked by the strong stimulus were modulated differently from those evoked by the weak stimulation. As illustrated in Fig. 3A, in a Pl MN, the Ia IPSPs evoked by weak stimulation of the TAEDL nerve (1.1 ×T, filled arrow) were larger during flexion (i.e. the inactive phase of this Pl MN), while those evoked by strong stimulation (2.2 ×T, open arrow) were larger during the extensor phase (i.e. the active phase of this Pl MN). After averaging the IPSPs separately in the two phases, the modulation of the IPSPs evoked by the strong stimulation clearly followed the membrane potential change, being largest during the active phase (depolarized membrane) (Fig. 3B), while IPSPs evoked by the weak stimuli were increased in amplitude during the inactive phase when the MN was hyperpolarised. A possible interpretation of this result would be that the strong stimulation managed to recruit (almost) all INs in the pool, irrespective of the step phase, and the change in IPSP amplitude would then fully depend on the membrane potential (distance from the IPSP equilibrium potential). The weak stimulation would leave most of the interneuronal pool in the subliminal fringe (see Methods) leaving the opportunity for a spatial facilitation when the locomotor circuits are exciting the interneuronal pool. In this case, the increased recruitment of INs by the weak test stimulus would thus reduce the opposite effect by the variation in membrane potential.

Figure 3. Different modulation of IPSPs evoked by weak and strong stimulation during l-DOPA locomotion A, the Ia IPSPs in an extensor plantaris (Pl) MN following weak, 1.1 ×T (filled arrow) and strong, 2 ×T (open arrow) stimulation of the antagonist (TAEDL) nerve. Intracellular recording of the membrane potential (MN i.c.) of this Pl MN during l-DOPA locomotion is displayed above the raw ENGs from the extensors (Pl) and a flexor (Sart) nerves, respectively. B, rectified and averaged ENGs from Sart and Pl and averaged MN i.c. records from 12 cycles. Data from 100 ms long bins (6 during flexion and 4 during extension) of the averaged cycles show the size of the IPSPs evoked by the different intensity stimuli: weak (circles) and strong (triangles) intensity. The time points are aligned with the panels above and correspond to the level of the membrane potential indicated by the dotted lines. Note that the IPSP evoked by the weak stimulus became smaller as the MN was depolarized while the IPSP evoked by the strong stimulus increased with depolarization. C, size of averaged IPSPs evoked by weak stimulation displayed as a percentage of the maximal IPSP in the active and inactive phase of all 7 MNs recorded during DOPA locomotion in spinalised cats. The white dots are data from extensor MNs and the black triangles are data from flexor MNs.

Both the small and large IPSPs should be influenced similarly by the changes in membrane potential, while the weak test stimulus would favour the possibility to increase the number of recruited INs by leaving a large subliminal fringe. In order to estimate the modulation of Ia IPSPs with respect to the output of the Ia IN pool, while minimising the influence of the changes in membrane potential, we therefore normalised the IPSPs evoked by the weak stimuli to the size of IPSPs evoked by the strong stimuli (i.e. the size of the IPSP evoked by the weak stimulus as a percentage of IPSP evoked by supramaximal strength). The results based on the normalised IPSPs show that inhibition evoked by the weak stimulus (designed to test the excitability changes in the Ia IN pool) was in fact larger in the inactive phase (i.e. in which the MN is hyperpolarised) in all tested MNs (n= 7; both flexor and extensor MNs) examined during fictive locomotion (Fig. 3C).

Facilitation of Ia IPSCs in MNs in their inactive phase during fictive locomotion and scratch is similar in cats with or without intact brainstem connections

In order to examine the modulation of IPSPs during fictive locomotion without any biasing from the large changes in membrane potential, we also recorded Ia IPSCs in voltage-clamp mode. During fictive locomotion evoked by l-DOPA in spinalised cats the IPSCs increased during the inactive phase (i.e. in the phase when they would have been hyperpolarised without voltage-clamp) as illustrated by the example in Fig. 4A and B from a Pl MN. All MNs (n= 9) examined showed such pattern (Fig. 4C). The amplitude of IPSCs in the inactive phase of locomotion was significantly larger than in the active phase (mean ±s.e.m.: 3.2 ± 0.3 vs. 5.2 ± 0.5 nA, for active and inactive phases, respectively, P= 0.001).

Figure 4. Modulation of disynaptic Ia IPSCs during fictive locomotion A, the Ia IPSCs (i.e. injected current required to clamp the membrane potential) during a short time window (12 ms) in a Pl MN following stimulation of the antagonist TAEDL nerve at 1.1 ×T (dotted line) are shown on the top and they are oriented vertically. Below, the intracellular recording of the membrane potential from this Pl MN (MN i.c.); the injected current and the ENGs from Pl and Sart are shown in a horizontal orientation during a 10 s long bout of fictive locomotor activity evoked by l-DOPA in a spinalised cat. Note that the IPSCs are smaller during Pl ENG bursts (grey columns), i.e. when the MN is active. B, averaged IPSCs in the Pl MN shown above during the extension or active phase (continuous line) and the flexion or inactive (dotted line) phase, separately (n= 50 and 66). Note that the first deflection in the CDP (open arrow) is a stimulus artifact. C, amplitude of averaged IPSCs (nA) during the active and inactive phases of all examined MNs during fictive locomotion evoked by l-DOPA in spinalised cats. D, amplitude of averaged IPSCs (nA) during the active and inactive phases of all examined MNs during fictive locomotion evoked spontaneously or by electrical stimulation of the mesencephalic locomotor region in cats with intact spinal cords.

This significant increase in IPSC amplitude in the inactive phase was also confirmed in cats with intact spinal cords in which locomotion was evoked either spontaneously or by MLR stimulation (Fig. 4D; mean ±s.e.m.: 4.7 ± 1.5 vs. 7.3 ± 2.2 nA for active and inactive phases, respectively, n= 5, P= 0.035).

The phasic modulation of disynaptic IPSCs was also examined during fictive scratch. In flexor MNs there was a reduction of IPSC amplitude during tonic flexion with respect to rest (i.e. in the absence of any motor activity before stimulating the pinna) which lasted throughout the flexion phase as illustrated in Fig. 5A. As an example, the changes in IPSCs in a TAEDL MN evoked by near threshold stimulation of the FDHL nerve are illustrated in Fig. 5B. All other flexor MNs showed a similar pattern of Ia IPSC modulation during scratch (Fig. 5C). In extensor MNs, IPSCs increased in amplitude during the tonic flexor activity when compared to rest (Fig. 5D). The overall comparison of IPSCs in the active and the inactive phases showed significant differences (mean ±s.e.m.: 2.4 ± 0.3 vs. 4.4 ± 0.6 nA for active and inactive phases, respectively, n= 11, P= 0.001). Changes in IPSCs were found to be similar in cats with intact spinal cords (Fig. 5C and D continuous lines) and in spinalised cats (Fig. 5D dotted lines).

Figure 5. Modulation of IPSCs during tonic flexion and rhythmic scratch activity A, the injected current needed to clamp the TAEDL MN membrane potential and ENGs from TAEDL and GS before and during rhythmic scratch. Note the tonic flexor period preceding the rhythmic scratch and the rhythmic depolarization (grey dotted circle) alternating with hyperpolarisation (black dotted circle) following the active phase of TAEDL and GS ENG activity, respectively. B, averaged IPSCs following stimulation at 1.1 ×T of the FDHL nerve before (thick black) and during (grey continuous) the tonic flexion phase preceding scratch and during the flexor phase (grey dotted) and extensor phase (black dotted) of rhythmic scratch. C and D, the size of the IPSCs in individual flexor (C) and extensor (D) MNs at rest, during tonic flexion and during the active and inactive phase of fictive scratch after stimulation of an antagonist nerve (FDHL, TAEDL or Q) at 1.1–1.2 ×T. The type of MN is represented by the different symbols: TAEDL: filled diamond; unknown flexor: filled square; FDHL: open diamond; SmAB: open square; GS and Pl: open triangle. The lines connect measurements made in the same MN under different conditions. The continuous lines represent recordings made in cats with intact spinal cord and the dotted lines represent data from spinalised (at C1) animals. As shown in the example in A, all flexor MNs had larger Ia IPSCs in the inactive phase (black rectangle) compared to the active phase (grey rectangle).

Discussion

Using direct recording from a sample of Ia inhibitory INs and the indirect technique of spatial facilitation of Ia IPSPs and IPSCs in MNs, we have demonstrated that Ia inhibitory INs are rhythmically excited during fictive locomotion and scratching. The rhythmic activity of Ia INs during locomotion and scratching has been reported several times before, but this is the first time that their activity has been recorded during scratch (evoked from the pinna) and locomotion (evoked by l-DOPA) in cats following a spinal transection at C1 level. The spinal rhythm generating networks are thus organised to rhythmically activate the Ia inhibitory INs in parallel with their corresponding MNs. The use of the voltage-clamp mode for recording changes in the Ia IPSCs in the MNs has for the first time resulted in conclusive information on the excitability changes in the whole pool of the Ia inhibitory INs during locomotion and scratching.

Evidence that Ia INs contribute to MN hyperpolarisation during fictive motor activity

The peak firing frequencies of the Ia inhibitory INs in the present experiments ranged from 50 to 300 Hz. Similar frequencies have been reported by Feldman & Orlovsky (1975) for MLR-evoked locomotion and by Deliagina & Orlovsky (1980) for scratching. The modulation of Ia IN activity that we report in this study is consistent with previous results, but since we obtained several simultaneous recordings from INs and their target MNs, we have provided more direct evidence for the INs being most active when their target MNs were hyperpolarised during both locomotion and scratch. This is particularly important in the case of Ia inhibitory INs excited from Sart and Q as they may connect to either hip extensor SmAB MNs and/or knee flexor PBSt MNs (Eccles & Lundberg, 1958). In the case of flexor-coupled (Sart, TAEDL) Ia INs, the activity profile paralleled the integrated ENG-activity in the motor axons in the nerve from which the Ia afferents monosynaptically activated the Ia inhibitory IN. This has been reported previously during fictive scratch (Deliagina & Orlovsky, 1980), but this is the first observation in relation to fictive spinal locomotion.

The situation was more variable for the Q-coupled Ia INs, which is similar to the finding of Pratt & Jordan (1987) during MLR-evoked fictive locomotion. However, earlier results by Feldman & Orlovsky (1975) from the same preparation showed that 22/23 Q-coupled Ia INs were maximally active in the stance phase. It is likely that this variability is related to the fact that the Q-nerve was composed of both the pure extensors (Vasti) and the combined knee extensor/hip flexor RF. Also, the Q-coupled Ia INs not only project to knee flexors (PB and St; but note that PBSt is often activated in the extensor phase; Berkinblit et al. 1980; Perret & Cabelguen, 1980) and to hip extensors (Sm, AB and St), but also to hip flexor Sart MNs (Eccles & Lundberg, 1958). It is not known whether the variation in the firing profiles of Q-coupled Ia INs reflects specific subsets of INs that may be defined by their target MNs. In the case of Q Ia INs, their target motor nuclei may be more important for their locomotor-related activity but all attempts to this date have been based on the analysis of the origin of Ia input to these INs.

The maximal activity of Ia INs during the inactive phase of their target MNs is fully supported by the increase of the Ia IPSCs in the MNs during their inactive phase (i.e. when they would have been hyperpolarised). Although the amplitude of Ia IPSPs depends on the membrane potential (distance from the IPSP equilibrium potential; Coombs et al. 1955), it turned out to be possible to estimate the excitatory convergence onto the Ia IN pool by comparing the effects on small and large test Ia IPSPs. Both are equally affected by membrane potential, while the spatial facilitation is optimised by a weak Ia input to the Ia IN pool, leaving a large subliminal fringe. When the small test Ia IPSP was normalised to the size of IPSPs evoked by the strong stimuli (i.e. the size of the IPSP evoked by the weak stimulus as a percentage of IPSP evoked by supramaximal strength), we always found an increase in the amplitude of the test Ia IPSP during the inactive (hyperpolarised) phase. Thus overall, our results clearly support the hypothesis of the parallel activation of ‘corresponding MNs and Ia inhibitory INs (i.e. those linked by the same Ia input) (Lundberg, 1970).

During MLR-evoked locomotion in the decerebrate preparation, there is phasic activity in rubro- (in high decerebrate cats; Orlovsky, 1972a), reticulo- (Perreault et al. 1993) and vestibulo-spinal (Orlovsky, 1972b) pathways. This rhythmic activity is dependent on an intact cerebellum and appears to be mediated by spinocerebellar pathways conveying information on the activity of the locomotor-generating spinal networks (Arshavsky et al. 1983, 1986). Similar results on movement-related phasic activity in rubro- and vestibulo-spinal neurones have also been described with fictive scratching (Arshavsky et al. 1978a,b, 1988). Thus it would be conceivable that the rhythmic movement-related activity of the Ia inhibitory INs indeed originates from supraspinal centres. It is therefore important that we have now demonstrated that the excitability of the Ia inhibitory INs also increases in the phase in which the receiving MNs are silent following a C1 transection. Both the direct Ia IN recordings and IPSCs in MNs showed the same modulation patterns in preparations with intact connections between the brainstem and spinal cord as in those preparations with the spinal cord transected at the C1 level. We therefore conclude that the output from the spinal networks underlying locomotion and scratching are themselves organised to produce a parallel excitation of MNs and Ia inhibitory INs. Finally, our results taken together provide evidence that Ia INs do indeed contribute to MN hyperpolarisation during fictive locomotion and scratch. This was also acknowledged earlier by Pratt & Jordan (1987) after the examination of the phase relationship between Ia INs and MNs. Several lines of evidence suggest that the rhythmic hyperpolarisation of MNs is indeed a result of postsynaptic inhibitory events (Edgerton et al. 1976; Perret, 1983; Shefchyk & Jordan, 1985a) during fictive locomotion and also during fictive scratch (Perreault, 2002). The Ia INs are prime candidates for being a source of this inhibition.

A quantitative estimation of the contribution by Ia inhibitory INs to the hyperpolarisation of MNs during the inactive phase of locomotion

Active inhibition during the inactive phase of locomotor-related MN activity has been demonstrated by the use of hyperpolarising currents and chloride ion injection which virtually ‘removed’ the hyperpolarisation between the bursts during MLR-evoked locomotion (Shefchyk & Jordan, 1985a). The inter-burst hyperpolarisation was even reversed with hyperpolarisation and chloride injection during fictive scratching in the cat (Perreault, 2002) as well as in the turtle (Robertson & Stein, 1988). Small doses of strychnine that did not change the phasic bursting during fictive locomotion abolished the inter-burst hyperpolarising phase (Pratt & Jordan, 1987). The phasic firing of identified INs mediating reciprocal Ia inhibition is also evidence of active inhibition during the inactive phase of the MNs. It is difficult to obtain a precise estimate of the contribution of Ia INs to MN hyperpolarisation based on the experimental data, but if some basic assumptions are made a rough estimate may be reached as follows.

With the amplitude of the compound IPSP resulting from a simultaneous activation of the whole Ia IN pool (P₀), the time constant of the Ia IPSPs (τ) and the firing frequency of the INs (f) it is possible to estimate the expected hyperpolarisation caused by the Ia inhibitory INs. In its simplest form, we could assume that the area of a single compound Ia IPSP is P₀ (mV) ×τ (ms). With a firing frequency of f (imp s⁻¹) the steady state hyperpolarisation (P_t) would become: P_t (mV) =P₀ (mV) ×τ (ms) ×f (imp s⁻¹).

Let us take the Ia INs with input from the hip flexor Sart and assume that these INs (n= 6, Table 1) are representative for the whole pool of Ia inhibitory INs. The approximate steady state firing of the INs based on the mean discharge rate observed in our sample (taking both locomotion and scratch together) ranged between 34 and 281 Hz. The time constant of the IPSP decay, τ, has been shown to equal the membrane time constant (Curtis & Eccles, 1959), which for MNs is around 5 ms (Burke, 1967, 1968; Burke & ten Bruggencate, 1971). As already discussed, P₀ varies more or less linearly with the distance from the equilibrium potential of the IPSP (Coombs et al. 1955) and thus is critically dependent on the present membrane potential. With resting membrane potentials between 50 and 70 mV and depolarizing ‘drive potentials’ during the active phase amounting to 5–20 mV (see Figs 1 and 3), the maximal Ia IPSPs observed in our sample of SmAB MNs, which were recorded simultaneously with the Ia INs (i.e. those with predominant inhibition from Ia INs with Sart input), were in the range of 1.1–4.7 mV (n= 5). Taking the minimum and the maximum values of our samples and 5 ms for τ: P_t= 1.1/4.7 (mV) × 5 (ms) × 0.034/0.281 (imp ms⁻¹). Thus the estimated steady state hyperpolarisation (P_t) evoked by the Ia inhibitory INs could range from 0.19 mV to 6.6 mV based on our results.

In this simplified calculation, we have not taken into account possible changes in the input resistance of the MNs. However, the input resistance (R_in) changes only marginally during MLR-evoked fictive locomotion (Shefchyk & Jordan, 1985b), while R_in, and thus τ_m, was reduced to about half during scratch (Perreault, 2002). During l-DOPA-evoked locomotion, there is only a marginal change in input resistance (H. Hultborn & K. Stecina, unpublished observation). This reduction of course reduces the time constant of the IPSP during scratching, and consequently the estimated tonic hyperpolarisation. It should be noted that the reduction of R_in by approximately one-half during scratch in the cat is somewhat lower, but in the same order of magnitude as in the turtle as described by Robertson & Stein (1988) and Alaburda et al. (2005).

We believe that our above estimation of the hyperpolarisation caused by the Ia inhibitory INs may be an under-estimation, since we have not taken into account non-linear membrane properties such as voltage-dependent persistent inward currents. These may amplify the effect of synaptic excitatory inputs (such as the locomotor drive potential) to the MNs by a factor of at least 3 – an amplification which may effectively be prevented/reduced by synaptic inhibition (Hultborn et al. 2003). The efficiency of Ia inhibition may thus not be fully reflected in the size of the evoked IPSP, but much more so in its ability to switch off plateau potentials induced by the locomotor drive potential in the previous depolarizing phase of the MN.

Implications for results obtained in humans

Our results obtained in the feline preparation during fictive motor activity are consistent with results obtained by indirect measures of reciprocal inhibition during actual movements in humans (Tanaka, 1974; Crone et al. 1987; Crone & Nielsen, 1989; Nielsen & Kagamihara, 1992). In these experiments, reciprocal Ia inhibition of plantar flexor MNs was evoked by stimulation of the peroneal nerve, and the amount of resulting inhibition was assessed by a monosynaptic reflex of the soleus muscle (the H-reflex).

One complication in these human experiments is that the sensitivity of the H-reflex to monitor the inhibition depends on the size of the H-reflex, which changes significantly during the investigated movements (Crone et al. 1990). To overcome this problem, the strength of the stimulation evoking the reflex has to be adjusted to obtain the same size of the unconditioned H-reflex during the various movements. Using this technique it has been shown that transmission in the inhibitory pathway from dorsiflexors to plantar flexors is reduced in the stance phase (i.e. when the target MNs of the Ia inhibitory INs are active) and increased in the swing phase of walking (i.e. when their target MNs are silent) (Petersen et al. 1999). During bicycling, inhibition from dorsiflexors to plantar flexors is also reduced during downstroke (during plantar flexor activity) and increased during upstroke (during dorsiflexor activity) (Pyndt et al. 2003).

We hypothesise that the adjustment of stimulation strength to attain the same size of the H-reflex independently of the phase of stepping (or voluntary dorsi- or plantar flexion) is likely to reflect that the membrane potentials of human MNs at the peak of the Ia EPSP reach the same level at the different phases during the movements. Thus the Ia IPSPs (which are timed to occur on the peak of monosynaptic Ia EPSPs) would be assessed at the same membrane potential in the various movements, as if the membrane potential of the MNs would be clamped, just like in the feline MNs that we have studied here. This ‘artificial voltage-clamping’ might be the reason why non-invasive studies in humans have shown such robust results on the parallel activation of Ia inhibitory INs and MNs during movement.

In conclusion, we have demonstrated in this study that the discharge of Ia inhibitory INs is regulated during rhythmic activity such as locomotion and scratching to produce the largest inhibition in the inactive phase of their target MNs and we have for the first time provided evidence that the resulting inhibition contributes significantly to the hyperpolarisation of the feline MNs.

Appendix

Author contributions

The experimental work was performed in the Neural Control of Movement Copenhagen laboratory at the Panum Institute, the University of Copenhagen, Copenhagen, Denmark. All authors contributed to the conception and design of the experiments, as well as to the collection, analysis and interpretation of data. S.S.G., K.S. and H.H. drafted the manuscript and all authors critically revised the manuscript and approved the final version for publication. The first and second authors (S.S.G. and K.S.) contributed equally.

Acknowledgements

This work was supported by grants from the Danish Medical Research Council, the Novo Nordisk Foundation and the Elsass Foundation. Katinka Stecina and Claire F. Meehan were supported by EU Marie-Curie intra-European fellowships. We would like to thank L. Grøndahl for expert technical assistance and J.-P. Gossard and K. Kyhl for participating in some of the preliminary experiments.