Brain Research 899 (2001) 66–81 www.elsevier.com / locate / bres Research report Reciprocal and Renshaw (recurrent) inhibition are functional in man at birth S.M. Mc Donough, G.J. Clowry, S. Miller, J.A. Eyre* Developmental Neuroscience, Department of Child Health, University of Newcastle upon Tyne, Newcastle upon Tyne NE1 4 HH, UK Accepted 2 January 2001 Abstract The aims were (1) to determine when in human postnatal development Group Ia reciprocal and Renshaw inhibition can be demonstrated; (2) to explore the relationship between the expression reciprocal inhibition and the disappearance of Group Ia excitatory reflexes between agonist and antagonist muscles. Studies were performed on 99 subjects, aged 1 day to 31 years, of whom 53 were neonates. A longitudinal study was also performed on 29 subjects recruited at birth and studied 3 monthly until 12 months of age. Reciprocal inhibitory and excitatory reflexes were recorded in the surface EMG of contracting biceps brachii (Bi), evoked by taps applied to the tendon of triceps brachii (Tri). Reciprocal excitatory reflexes were recorded in all but one neonate. Reciprocal inhibition was observed in 25% of neonates; evidence is provided that it was likely to have been masked by low threshold reciprocal excitation in the remaining neonates. Reciprocal inhibition was demonstrated in all subjects after 9 months of age. In four neonates there was depression of inhibition of Bi during co-contraction of Bi and Tri implying that Group Ia interneurones may be under segmental and suprasegmental control at birth. Renshaw cells, identified in human postmortem cervical spinal cord by their morphology, location and calbindin D28K immunoreactivity, were present at 11 weeks post-conceptional age (PCA) and by 35 weeks PCA had mature morphological characteristics. In four neonates reciprocal inhibitory responses in Bi disappeared when the tap to Tri evoked its own homonymous phasic stretch reflex, providing neurophysiological evidence for Renshaw inhibition of Group Ia inhibitory interneurones.  2001 Elsevier Science B.V. All rights reserved. Theme: Development Topic: Motor systems Keywords: Reciprocal inhibition; Renshaw inhibition; Recurrent inhibition; Neonate; Development; Phasic stretch reflex; Human 1. Introduction During spontaneous movement human neonates show a predominant pattern of co-contraction of agonist / antagonist muscle pairs in the limbs, with distinct alternating patterns of activation of agonist / antagonist pairs not being achieved until several years later in childhood [17–19]. Abbreviations: Bi, biceps brachii; C, condition; CMCD, central motor conduction delay; MVC, maximum voluntary contraction; PCA, post conceptional age; PRBS, pseudo-random binary sequence; PMCD, peripheral motor conduction delay; R, cross-correlation function; t, test; T, threshold; Tri, triceps brachii; TMCD total conduction delay; V, cross-covariance *Corresponding author. Department of Child Health, The Royal Victoria Infirmary, Queen Victoria Road, Newcastle upon Tyne NE1 4LP, UK. Tel.: 144-191-2023-013; fax: 144-191-2023-022. E-mail address: j.a.eyre@ncl.ac.uk (J.A. Eyre). The degree of co-contraction expressed by agonist / antagonist muscle pairs during limb movement depends upon many interrelated factors, in particular the descending commands to each motoneuronal pool and the balance between reciprocal excitation [32,36,37] and reciprocal inhibition [4,30]. The magnitude of reciprocal inhibition is fine-tuned by recurrent inhibition [4]. Descending commands and segmental afferents modulate the excitability of Group Ia inhibitory interneurones and Renshaw cells [26] and there is also a pattern of mutual agonist / antagonist inhibition such that the Group Ia inhibitory interneurones inhibit the Group Ia inhibitory interneurones of the antagonist muscle [4,22] and Renshaw cells excited by agonists inhibit those excited by antagonists and vice versa [40]. Thus a-motoneurones, Group Ia inhibitory interneurones and Renshaw cells together with Group Ia primary spindle afferent input form part of an ‘output stage’ spinal 0006-8993 / 01 / $ – see front matter  2001 Elsevier Science B.V. All rights reserved. PII: S0006-8993( 01 )02151-5 S.M. Mc Donough et al. / Brain Research 899 (2001) 66 – 81 network, which interacts with descending pathways to mediate the final relative output of agonist / antagonist motoneurones [26]. The inappropriate degree of co-contraction observed early in development is likely to reflect immaturity both of this spinal neural network and of supraspinal descending commands. Moreover, the rapid improvement of co-ordination, smoothness of limb movements and dexterity, which occurs in the first 24 months after birth, is related to the progressive development and refinement of both systems. The present study forms part of an investigation to ascertain the time of expression of essential components of the ‘output stage’ spinal network in man and which factors are important in shaping the functional relationships of the mature network. It addresses when in postnatal development Group Ia reciprocal inhibition and Renshaw inhibition can be demonstrated and the relationship between the functional expression of these reflexes and the restriction of heteronymous Group Ia excitatory projections between agonist and antagonist muscles, which occurs in the first four years after birth [36,37]. 2. Methods 2.1. Subjects A cross sectional study was performed on 99 subjects, 63 females, 36 males, aged from 1 day to 31 years; of whom 53 were term neonates, aged 1–4 days. A longitudinal study was performed on a further 29 subjects, 20 females and nine males, recruited at birth and studied at 3 monthly intervals until 12 months of age. Informed, written consent was obtained from the adult subjects and the parent(s) of the babies and children. The study received approval from the Newcastle Health Authority and University of Newcastle upon Tyne Joint Ethics Committee. 2.2. Electromyograms EMGs in biceps brachii (Bi) and triceps brachii (Tri) were recorded using Ag–AgCl skin-mounted, standard EEG electrodes, 5 mm in diameter, with centres separated by 15 mm for babies and children up to 9 years and 20 mm for older children and adults. To promote integration of motor unit activities in the surface EMG the electrodes were orientated in alignment with the main direction of the underlying muscle fibres. The electrodes were placed on the muscle belly for Bi and over the belly of the lateral head for Tri, in positions approximately midway between the motor point and the tendon. The EMG was recorded unrectified [31] and was fed to differential preamplifiers of 60 dB voltage gain, 100 dB at 50 Hz common mode rejection ratio, 100 MV input impedance, 50 V output impedance and 23 dB frequency bandpass of 10–1000 Hz. The impedance of the electrodes was maintained at ,5 67 kV. The preamplifiers were a.c. coupled, so that the reflexes recorded do not show a positive d.c. shift corresponding to background muscle activity. The signals were digitized using an intelligent interface (Cambridge Electronic Design, Cambridge, UK, type 1401 plus), sampled at 5 kHz, and stored on a computer for analysis. 2.3. Positioning of subjects and generation of background muscle activity Neonates and babies aged less than 6 months were studied lying quietly in a cot. An evacuable plastic bag filled with polystyrene beads was used to stabilize the head and trunk in the mid-line anatomical position. The left arm was adducted with the elbow held in 458 of flexion from full extension. Subjects older than 6 months were seated either on a parent’s knee or on a chair, with the head, trunk and arm positioned as described above. Babies and young children are not able to maintain a steady level of contraction or perform a maximum voluntary contraction (MVC) on request. Reflexes requiring muscle contraction were therefore obtained in subjects aged less than 12 months during spontaneous periods of contraction. For contraction of Bi, children over the age of 18 months held a weight in the half supinated hand with the flexed arm supported at the elbow; the weight varied between 100 and 200 g, as judged appropriate to age and stature. To contract Tri children were encouraged to push the pronated arm onto a horizontal support, while an examiner placed a hand over the child’s upper forearm to resist extension at the elbow. Adults sat in a chair with the head and trunk in the mid-line anatomical position and the left forearm resting on a support. As with babies and younger children, the left arm was adducted with the elbow held in 458 of flexion from full extension. Contraction of Bi was obtained with the arm half supinated and a strap was placed over the wrist to resist elbow flexion. For contraction of Tri the forearm was pronated and a strap placed across the upper forearm to resist elbow extension. For each muscle the EMG was also rectified and integrated (time constant 1 s) and displayed to the subject as a horizontal line on an oscilloscope. The MVC of each muscle was recorded as the greatest level of rectified, integrated EMG, obtained by the subject in three attempts separated by 1 min rest intervals. During reflex testing the subject contracted the target muscle to raise the EMG display to meet a target line set at 10% MVC. 2.4. Electromechanical tapper The reflexes were elicited in Bi or Tri with a handheld electromechanical tapper (Ling Altec, Royston, UK, Model 101) delivering a sequence of taps (cross-correlation method) or a single tap (spatial summation procedure), in each case of 5 ms duration, comprising rise and decay 68 S.M. Mc Donough et al. / Brain Research 899 (2001) 66 – 81 times of 2.5 ms. The stylus of the tapper, a Perspex round disc (diameter 8 mm for babies, 10 mm for adults) was applied to the skin overlying the tendon of the muscle. The peak force (N) of the tap was measured with a transducer placed in series with the stylus. The tapper was driven by a constant gain power amplifier. The peak voltage of the pulse feeding the amplifier correlated linearly with the peak force delivered over a range of 0–15 N (r50.998). In practice the reflexes were obtained with peak forces in the range 0.5–10.0 N [31]. 2.5. Cross-correlation method for recording spinal reflexes Spinal myotatic reflexes were elicited by a pseudorandom binary sequence (PRBS) of taps applied to the muscle tendon and were recorded in the surface EMG (Fig. 1A). The characteristics of the reflexes were measured from the functions obtained by cross-correlation of the PRBS with the EMG. The cross-covariance function V estimates the magnitude of the reflex and the cross-correla- Fig. 1. Cross-correlation method. (A) The PRBS signal (upper trace) and corresponding sequence of taps (middle trace). The lower trace shows surface EMG of Bi contracting at 10% MVC in response to taps applied over the tendon at an intensity of 1.2 T for the phasic stretch reflex. The time base is indicated on the abscissa. (B–I) Excitatory and inhibitory responses. Each response represents the average of six consecutive sequence responses (defined as a Trial). Each sequence response was defined as the cross-covariance function of the EMG and PRBS following the application of the 64 taps in a single PRBS signal lasting 1.28 s. (B–E) Adult. (F–I) Neonate. (B & F) Tri stretch reflex, homonymous phasic stretch reflex in contracting Tri at 1.2 T. (C & G) Tri→Bi heteronymous excitation, excitatory response in contracting Bi following tap to the tendon of relaxed Tri. (D & H) Bi stretch reflex, homonymous phasic stretch reflex in contracting Bi at 1.2 T. (E & I) Tri→Bi inhibition; inhibitory response in contracting Bi following a tap to the tendon of relaxed Tri at 0.8 T. Horizontal scale bar gives the time base for responses (B–I). Vertical scale bar: (B–E & I) 100 mV, (F & H) 200 mV, (G) 300 mV. S.M. Mc Donough et al. / Brain Research 899 (2001) 66 – 81 tion function R estimates the degree of correlation between input PRBS and output EMG [31]. The authors showed that the onset latencies of cross-correlation functions of the reflexes are advanced in time by 10 ms, the clock period of the PRBS, and 10 ms has therefore been added to all measures of latency [31]. A sequence response was defined as the cross-covariance function of the EMG and PRBS following the application of the 64 taps in a single PRBS signal lasting 1.28 s. A trial comprised the average of six sequence responses obtained with consecutively delivered PRBS signals [31]. 69 both to soft tissue near to the tendon of triceps and also to the lateral humeral epicondyle. If these taps evoked a reflex response in Tri or Bi the subjects were excluded from the study. Two neonates were excluded on this basis. 2.8. Control for volume conduction of the EMG If in phase or out of phase deflections in the crosscovariance functions were observed at any stage in Bi and Tri, all the data from these muscles were excluded from the analysis. In fact no subject was excluded on this basis. 2.9. Reflex threshold 2.6. Reflex response A response was defined as present if it was discernible by eye in the trial average of six cross-covariance functions and occurred with a consistent onset latency (Fig. 1B–I). In each experimental session the polarity of the initial phase of the homonymous phasic stretch reflex in biceps was determined. If it was positive the polarity of the recording electrodes was appropriate. If it was negative the polarity of the differential electrodes was reversed so that the homonymous phasic stretch reflex showed an initial positive deflection. Heteronymous reflexes evoked in biceps by a tap to triceps were then defined as excitatory if the initial component was positive going (Fig. 1B–D and Fig. 1F–H) and as inhibitory if the initial component was negative going (Fig. 1E and I). This convention was validated by corresponding changes in torque about the elbow. Thus initially upward going responses in the EMG of biceps, designated as excitatory, have been shown to correspond to an increase in flexor contractile force, while downward going responses in the EMG of biceps, designated as inhibitory, to correspond to a decrease in flexor contractile force [31]. 2.7. Controls for mechanical spread of the tap stimulus In 10 neonates and 10 adults the onset of the mechanical pulse induced by the tap to triceps was measured using an accelerometer placed over the site of the recording electrodes on biceps (Table 1). In all subjects, controls were made for reflex effects induced by the direct mechanical spread of the tap to biceps [36,37]. Taps of 1.5 T for the homonymous phasic stretch reflex in Tri, were applied Table 1 Onset of mechanical impulse in Bi following taps to the tendon of Tri Neonates Adults n Mean6S.E.M. (ms) 10 10 0.0560.08 1.6360.48 In determining reflex threshold (T), the tap magnitude was first set suprathreshold and then reduced stepwise until reflexes were identified in three out of the six sequences in a trial (a 50% response rate). 2.10. Reflexes studied 2.10.1. Homonymous phasic stretch reflexes Homonymous phasic stretch reflexes were elicited in contracting Bi and Tri. The thresholds and onset latencies of the trial averages were recorded (Fig. 1B, D, F and H). 2.10.2. Short latency heteronymous excitatory and reciprocal inhibitory responses from Tri to Bi Once the threshold for the homonymous phasic stretch reflex in Tri had been determined (T), the force of the tap was set at a level between 0.2 and 0.5 T and applied to relaxed Tri. The responses were recorded in contracting Bi. The force delivered was increased stepwise in units of 0.1 T up to a maximum of 1.5 T. At each of these stimulus levels the trial average was inspected for evidence of excitatory heteronymous responses (Fig. 1C and G) or reciprocal inhibitory responses (Fig. 1E and I). The thresholds, onset latencies and peak to peak amplitudes of the responses were recorded. 2.11. Spatial summation experiment: corticospinal projection to Group Ia inhibitory interneurones ( Fig. 4) This component of the corticospinal projection was investigated by spatial summation at presumed Tri Group Ia inhibitory interneurones in two neonates and two adults. Summation was tested between a Tri Group Ia afferent volley, subthreshold for evoking a heteronymous excitatory response or a reciprocal inhibitory response in Bi, and a corticospinal volley, subthreshold for evoking a response in Bi or Tri. 2.11.1. Excitation of a corticospinal volley Transcranial magnetic stimulation (TMS) using a Magstim 200 stimulator, (MagStim, Whitland, UK), with a 90 70 S.M. Mc Donough et al. / Brain Research 899 (2001) 66 – 81 mm circular coil placed tangentially above the contralateral motor cortex, excited corticospinal neurones [9,12–14]. The optimal position for exciting corticospinal axons and the conduction delay from TMS to the arrival of the corticospinal volley at Tri Group Ia interneurones was assumed to be the same as for corticospinal axons projecting to Tri a-motoneurones. The optimal coil position for evoking EMG responses in Tri was determined using suprathreshold stimuli. The intensity of TMS was then decreased until responses were evoked in contracting Tri in 50% of trials, defined as threshold (T). The shortest onset latency of 20 responses to TMS at 1.2 T was obtained to give the total motor conduction delay (TMCD) to Tri muscle. Magnetic stimulation, with a 5 cm circular coil, placed in the coronal plane overlying cervical spines 5–8, was used to excite spinal motor roots at the point of emergence from the intervertebral foramina [39]. The peripheral motor conduction delay (PMCD) to the Tri muscle was estimated from the longest onset latency obtained from at least 20 responses [13]. The subtraction of PMCD from TMCD was used to estimate the corticospinal conduction and trans-synaptic delay to Tri amotoneurones. measured as the peak of the short latency inhibitory response and expressed as a percentage of the RMS value of the background EMG in Bi during each set of repeats. 2.12. Renshaw (recurrent) inhibition 2.11.2. The conduction and trans-synaptic delay from Tri Group Ia afferents to Tri Group Ia inhibitory interneurones The conduction and trans-synaptic delay from Tri Group Ia afferents to Tri Group Ia inhibitory interneurones was assumed to be the same as that to Tri a-motoneurones. It was estimated by subtracting the PMCD to Tri muscle from the shortest onset latency of the homonymous phasic stretch reflex in Tri obtained from responses to at least 20 single taps delivered at 1.2 T to the tendon of Tri. 2.12.1. Anatomical study ( Fig. 6) The characteristic morphology and location and calbindin D28K (calbindin) immunoreactivity were used to identify Renshaw cells in human postmortem cervical spinal cord obtained from a surgically aborted fetus of 11 weeks post-conceptional age (PCA) and nine preterm babies aged 26–35 weeks PCA, who died from nonneurological causes (Fig. 5). The spinal cords were immersion fixed in buffered 4% paraformaldehyde solution for 24–72 h and then sectioned into segments and immersed in buffered 30% sucrose solution for at least 24 h. Frozen 50 mm sections were cut on a freezing microtome and then immunostained free floating using standard techniques. A well characterized monoclonal antibody to calbindin (Sigma), was employed as a primary antibody at a dilution of 0.05% in phosphate buffered saline containing 0.1% Triton X-100 (Sigma) and 3% normal horse serum (Vector Labs.). The sections were incubated overnight with gentle agitation. The immuno-localization of calbindin was visualized by incubation with biotinylated horse antimouse antibodies (0.5%, Vector Labs.), followed by streptavidin–horseradish peroxidase (HRP) conjugate (0.5%, Vector Labs.) in phosphate-buffered saline between and after incubations. The sections were reacted with diaminobenzidine (0.05%, Sigma) in the presence of hydrogen peroxide (0.003%) in phosphate-buffered saline and washed and mounted on slides. 2.11.3. Spatial summation trials Subthreshold tap to Tri (Fig. 4C), TMS at 0.8 T (Fig. 4D) and the two stimuli together at the interstimulus interval of expected convergence of both volleys at Group Ia interneurones (Fig. 4E), were delivered in an interleaved sequence. Five repeats of this sequence were performed three times on each neonate and adult subject. To define the time course of spatial summation in the two adults five repeats of the sequence were used, conducted over a range of interstimulus intervals such that the corticospinal EPSP arrived at biceps a-motoneurones from 2 ms before to 10 ms after the estimated arrival of the Group Ia volley (Fig. 5). The condition to test interval (C–t), the interval of expected convergence, was defined as C–t interval zero and all other C–t intervals were defined relative to this. Thus negative C–t intervals indicated the Group Ia volley arriving before the corticospinal volley and positive C–t intervals indicated the corticospinal volley arriving before the Group Ia volley. The magnitude of the inhibition was 2.12.2. Neurophysiological study ( Fig. 7) Group Ia inhibitory interneurones are inhibited by Renshaw cells [4,20]. Evidence was therefore sought for depression of reciprocal inhibition between Tri and Bi by recurrent activity in Tri motoneurone axon collaterals. The cross-correlation method for spinal reflexes was used. Four neonates were chosen, who had clear inhibitory responses in Bi following a tap to relaxed Tri with a low threshold, i.e., ,0.7 T, and who also had high thresholds for heteronymous excitatory responses in Bi, i.e., .1.3 T. Threshold taps to Tri (1.0 T) were applied to evoke a homonymous phasic stretch reflex in contracting Tri in 50% of sequence responses. These taps were suprathreshold for evoking reciprocal inhibition in Bi, but substantially below the threshold for evoking a heteronymous excitatory response in Bi. Evidence was sought for consistent absence of reciprocal inhibition of Bi in those trials when a homonymous phasic stretch reflex in Tri was simultaneously evoked [30] (Fig. 7). S.M. Mc Donough et al. / Brain Research 899 (2001) 66 – 81 2.13. Relationship between level of contraction in Tri and the probability and magnitude of the inhibition evoked in contracting Bi during co-contraction of Bi and Tri ( Fig. 8) In the same four neonates studied above, taps at 0.8 T were applied to the tendon of Tri, so that the tap would be subthreshold for either a homonymous phasic stretch reflex in Tri or a heteronymous excitatory response in Bi, but suprathreshold for an inhibitory response in Bi. The magnitude of the inhibitory response in Bi was determined (1) during periods of isolated activity in Bi and (2) during spontaneous co-contraction of Bi and Tri when the magnitude of the inhibitory response evoked in Bi was related to the level of contraction in Tri. For each trial the levels of contraction in Tri and Bi were determined as the root mean square (RMS) of their respective EMGs. The magnitude of the inhibition in each trial was determined to be the difference between the baseline to peak values of the inhibitory response in the trial average expressed as a percentage of the RMS value of the EMG in Bi during that trial. The amplitudes of inhibitory responses had a skewed distribution and underwent logarithmic transformation. It is not possible to measure MVC in neonates and therefore the magnitude of the EMG in individual subjects could not be normalized by reference to EMG values at MVC. To remove between child variation in the magnitudes of EMG in Bi and Tri, the mean values across all trials for the level of contraction of Tri and the magnitude of the inhibition in Bi were determined for each neonate. The magnitudes of contraction in Tri or inhibition in Bi in each trial were then expressed in terms of their difference from the mean value (called a residual value). Thus negative values indicate magnitudes which were less than the mean for that subject and positive values those which were greater than the mean (Fig. 8). 3. Results Unless otherwise means6S.E.M. stated all data are given 71 of the homonymous phasic stretch reflex in Bi (Table 2, P,0.0001), with a mean difference of 1.8 ms60.1 ms (Fig. 2B and C and Fig. 3A). When reciprocal inhibitory responses from Tri to Bi were observed in neonates, the onset latencies also occurred significantly later than the homonymous phasic stretch reflex in Bi (Table 2: cross-sectional study, P, 0.0001; longitudinal study, P,0.001), with a mean difference in onset latency of 1.860.3 ms for the cross-sectional study and 1.060.4 ms for the longitudinal study (Fig. 2B and C and Fig. 3B). The difference in onset latency between reciprocal inhibition from Tri to Bi and the homonymous phasic stretch reflex in Bi, remained remarkably consistent with age, the mean varying within the range 11.0 to 11.8 ms for both the cross sectional and longitudinal study groups (Fig. 2B and C). 3.2. Group Ia heteronymous excitatory responses from Tri to Bi ( Figs. 2 and 3) Heteronymous excitatory responses [36,37] were observed in 52 / 53 neonates (98%) in the cross-sectional study and 29 / 29 neonates (100%) in the longitudinal study (Fig. 2E). The probability of demonstrating a heteronymous excitatory response in Bi decreased slowly with age. At 12 months of age 38 / 53 (72%) of cross-sectional subjects and 20 / 29 (71%) of longitudinal subjects still demonstrated heteronymous excitatory responses, but in adult subjects these responses were only present in two subjects (10%; Fig. 2E). For neonates there were no significant differences between the onsets of homonymous phasic stretch reflexes in Bi and the heteronymous excitatory response from Tri to Bi (Table 2, Fig. 2F and G and Fig. 3B). However, the difference in the onset latencies of the heteronymous excitatory responses and the homonymous phasic stretch reflex in Bi increased progressively with age (Fig. 2F and G), so that by adulthood the onsets of the heteronymous excitatory responses were 11 ms and 12 ms longer than the onset of the homonymous phasic stretch reflex. as 3.3. Thresholds of heteronymous excitatory and reciprocal inhibitory responses 3.1. Reciprocal inhibition from Tri to Bi ( Figs. 2 and 3) Short latency reciprocal inhibition from Tri to Bi was demonstrated in neonatal subjects in 15 / 53 (28%) of the cross sectional and 6 / 29 (21%) of the longitudinal study. The probability of demonstrating short latency inhibition however increased rapidly postnatally so that after 9 months postnatal age all subjects in both the cross-sectional and longitudinal studies demonstrated short latency reciprocal inhibition (Fig. 2A). For the adult subjects the onset latency of reciprocal inhibition from Tri to Bi was significantly longer than that The thresholds for reciprocal inhibitory and heteronymous excitatory responses in Bi (Fig. 2D and H and Fig. 3 and D) were normalised by reference to the threshold (T) of the homonymous phasic stretch reflex in contracting Tri. In adult subjects the threshold of reciprocal inhibitory responses in Bi were significantly lower than those of the homonymous phasic stretch reflex in Tri (Table 2, P, 0.02, Fig. 3C) and of the heteronymous excitatory responses in the two adult subjects (Table 2, Fig. 3C). Similarly, when reciprocal inhibitory responses in Bi were demonstrated in neonates, the thresholds were sig- 72 S.M. Mc Donough et al. / Brain Research 899 (2001) 66 – 81 Fig. 2. Reciprocal inhibitory (A–D) and heteronymous excitatory responses (E–F) in Bi following taps to Tri. The symbols represent the means and the vertical lines the 95% confidence limits of the means. Filled circles: data from the cross-sectional study, numbers above the abscissa give subjects in each age group. Open circles: data from 29 children studied longitudinally. (A and E) The percentage of subjects demonstrating a response in relation to age. (B and F) Onset latencies of the trial average responses in relation to age. The onset latencies of homonymous phasic reflexes in Bi are also plotted as triangles. Filled triangles: cross-sectional data; open triangles: longitudinal data. (C and G) Relative onset latencies of responses, defined as the onset latency of responses minus the onset latency of the homonymous phasic stretch reflex in Bi. The dashed horizontal line through 0 represents the onset of the homonymous phasic stretch reflex in Bi. (D and H) Threshold of responses. Thresholds are defined as percentages of the threshold of the homonymous phasic stretch reflex in contracting Tri. The horizontal dashed line at 100% represents the threshold of the homonymous phasic stretch reflex of Tri. The horizontal dotted line represents the mean threshold of the reciprocal inhibitory responses in adults. nificantly lower than those for the homonymous phasic stretch reflexes in Tri (Table 2: cross-sectional study, P,0.02; longitudinal study, P,0.02, Fig. 3D) and lower than the threshold of the heteronymous excitatory re- sponses in Bi (Table 2: cross-sectional study, P.0.05, longitudinal study, P,0.01; Fig. 3D). In contrast, the neonates, who did not demonstrate reciprocal inhibition, had heteronymous excitatory re- S.M. Mc Donough et al. / Brain Research 899 (2001) 66 – 81 73 Table 2 Onset and threshold of reflexes Onset latency Adults Neonates: Cross-sectional Longitudinal Bi homonymous phasic stretch reflex Bi→Bi Reciprocal inhibition Tri→Bi Heteronymous excitatory reflexes Tri→Bi n Mean6S.E.M. (ms) n Mean6S.E.M. (ms) n 20 18.560.4 20 20.360.5 2 53 29 13.960.2 14.060.3 15 6 15.760.5 15.060.5 52 28 Threshold Adults Neonates with inhibition: Cross-sectional Longitudinal Neonates not demonstrating inhibition: Cross-sectional Longitudinal sponses in Bi with thresholds that were significantly lower than the threshold of their homonymous phasic stretch reflex in Tri (Table 2: cross-sectional study, P,0.001; longitudinal study, P,0.001; Fig. 3D) and were also significantly lower than the thresholds of the heteronymous excitatory responses observed in those neonates in whom reciprocal inhibition was demonstrated (Table 2: crosssectional study, P,0.01; longitudinal study, P,0.001; Fig. 3D). The thresholds of the heteronymous excitatory responses in the neonates without reciprocal inhibition were similar to the thresholds of the inhibitory responses observed in the neonates with reciprocal inhibition (Fig. 3D). The threshold for reciprocal inhibition relative to that of the homonymous phasic stretch reflex in Tri remained remarkably consistent with age with the mean lying between 0.65 T and 0.8 T at all ages in both the longitudinal and cross-sectional studies (Fig. 2D). In contrast, the threshold for heteronymous excitatory responses increased progressively with age relative to that of the homonymous phasic stretch reflex in Tri. By 9 months the mean threshold of the heteronymous responses was greater than that of reciprocal inhibition, and by 12 months the mean threshold of the heteronymous responses was greater than that of the homonymous phasic stretch reflex in Tri (Fig. 2H). 3.4. Corticospinal projection to Group Ia inhibitory interneurones ( Fig. 4) Spatial summation of a Tri Group Ia afferent volley subthreshold for evoking a heteronymous excitatory or reciprocal inhibitory response in Bi and a corticospinal volley subthreshold for evoking a response in either Bi or Mean6S.E.M. (ms) Mean6S.E.M. (T) 31, 33 13.760.3 13.960.3 Mean6S.E.M. (T) 20 0.860.10 2 1.35, 1.20 15 6 0.760.09 0.660.17 14 6 1.0060.09 1.0960.05 38 23 0.6060.06 0.5660.06 Tri occurred at the estimated coincidence of volleys at Tri Group Ia inhibitory interneurones in the two neonates and the two adults studied. In all subjects spatial summation resulted in inhibition of Bi EMG with onset latencies 1–2 ms longer than EMG responses evoked in Bi by suprathreshold TMS (Fig. 4E). In the neonatal subjects the EMG was depressed by a mean of 32.365.2% (P,0.01) and 29.366.1% (P,0.01) and in the adults by 1563.2% (P,0.01) and 3864.1% (P,0.01), respectively. The time course of the spatial summation was defined in the two adult subjects (Fig. 5). Spatial summation occurred over C–t intervals of 0 to 14 ms and 0 to 17 ms, with peak depressions of the EMG of 32.564% and 55.6613% respectively (Fig. 5A). 3.5. Renshaw (recurrent) inhibition 3.5.1. Anatomical study ( Fig. 6) In the human post-mortem material calbindin positive cells were observed in the appropriate location for Renshaw cells at all the ages examined [2,3] (Fig. 6). At 11 weeks PCA most putative Renshaw cells had a strongly immunoreactive cell nucleus with only weakly immunoreactive cytoplasm (Fig. 6A and B). Every cell exhibited weak immunoreactivity in two or three dendritic processes but only close to the cell body. In the motoneuronal pools the occasional cell resembling a motoneurone was calbindin positive, but no calbindin positive axons were observed either in the grey matter or in the white matter. By 23 weeks PCA (Fig. 6C and D) calbindin immunoreactivity was much stronger in the cytoplasm of cells with the characteristic morphology and location of Renshaw cells. Each cell had two or three immuno-positive proximal dendrites that were thin and gave rise to even thinner 74 S.M. Mc Donough et al. / Brain Research 899 (2001) 66 – 81 Fig. 3. Onset latencies and thresholds of trial average responses. (A and C) Adults; (B and D) neonates. The subjects are divided into those in whom reciprocal inhibition was not evoked (inhibition absent) and those in whom it was (inhibition present). The circles represent means and the vertical lines 95% confidence limits of the means. For neonates filled circles represent data from the cross-sectional study, open circles from the longitudinal study. (A and B) Onset latencies. The horizontal dotted line defines the mean onset latency of the homonymous phasic stretch reflex in Bi in adults and neonates, respectively. (C and D) Threshold. Thresholds are defined as percentages of the threshold of the homonymous phasic stretch reflex in contracting Tri. The horizontal dotted line at 100% represents the threshold of the homonymous phasic stretch reflex of Tri. Bi, SR homonymous stretch reflex in Bi; Tri→Bi EXCIT, heteronymous excitatory responses from Tri to Bi; Tri→Bi INHIB, inhibition from Tri to Bi; the numbers above the absicca indicate the number of subjects; NA, not applicable. branches. Many calbindin positive axons could be seen towards the inner boundary of the white matter all round the ventral horn. No axons were present in the grey matter of the motoneuronal pools and no motoneurones were calbindin positive at this age. Over the next 11 weeks of development the appearance of presumed Renshaw cells changed, so that by 35 weeks post-conceptional age (Fig. 6E, F and G) the proximal dendrites and their branches had thickened considerably and could be traced for several hundred microns. They often appeared to run in parallel bundles along the medial edge of the motoneuronal pools or along the grey / white matter boundaries. Many calbindin immunoreactive varicose axons were present in the motoneuronal pools and some appeared to make contact with motoneurone cell bodies. 3.5.2. Neurophysiological study ( Fig. 7) In all four neonates studied it was a consistent finding that when a tap of 1.0 T evoked a homonymous phasic stretch reflex in contracting Tri, inhibition was not observed in contracting Bi (Fig. 7D). 3.6. Relationship between level of co-contraction in Tri and Bi and the probability and magnitude of the inhibition evoked in contracting Bi ( Fig. 8) One hundred and forty-five informative trial averages were obtained in the four neonates studied: minimum number of trial averages per subject530 (representing 180 sequences), maximum543 (representing 258 sequences). The mean RMS values for EMG in Tri varied from 12.3 to 14.9 mV and the mean values for the amplitude of inhibitory responses in Bi, expressed as the percentage reduction in the background level of contraction in Bi, varied from 38 to 64%, respectively. Reciprocal inhibitory responses were observed in 98 / 145 (68%) trials. Reciprocal inhibition was most frequently observed and had its maximum amplitude when there was no contraction in Tri (Fig. 8; frequency, Tri not active, S.M. Mc Donough et al. / Brain Research 899 (2001) 66 – 81 75 Fig. 4. Spatial summation of subthreshold corticospinal volley and subthreshold Tri Group Ia volley in a neonatal subject. Open arrows indicate onset of tap to Tri. Filled arrows indicate onset of TMS. (A and B) Taps delivered to the tendon of Tri and responses recorded in Tri (A) and Bi (B–E). The intensity of the tap was decreased until it was subthreshold for evoking excitation or inhibition in Bi. (Note: heteronymous excitation of Bi with a threshold of 0.4 T was observed but reciprocal inhibition was not demonstrated). (C) Subthreshold tap to Tri. (D) TMS subthreshold for evoking a response in Bi or Tri (0.8 T). (E) The two stimuli together at the interstimulus interval of estimated convergence at Tri Group Ia interneurones resulting in inhibition of contracting Bi. T, Threshold; TMS, transcranial magnetic stimulation. 2 12 / 13 trials, 93%; Tri active, 86 / 132 trials, 65%; x 598 P,0.001; amplitude of reciprocal inhibition; Tri not active, n512, 2.0360.14 l n %; Tri active, n586, 0.1060.09 l n %; P,0.0001). When reciprocal inhibition in Bi was evoked during co-contraction of Tri and Bi, there was a significant negative correlation between the level of contraction in Tri during the trial and the magnitude of the inhibitory response in Bi (r 2 50.22, P,0.001; Fig. 8). 4. Discussion Short latency inhibition and excitation of Bi EMG was demonstrated in neonates and adults following a tap to the tendon of relaxed Tri. Direct volume conduction of reflexes evoked in Tri is excluded since the tap was subthreshold for reflexes in relaxed Tri. It is possible that these heteronymous responses could have arisen from Fig. 5. Inhibition of biceps following spatial summation of a subthreshold triceps Group Ia afferent volley and a subthreshold corticospinal volley at triceps Group Ia inhibitory interneurones. (A) Time course and magnitude of spatial summation, open symbols represent the mean for each subject and vertical lines the S.E.M. of the peak of the depression of the EMG expressed as a percentage of the background EMG. Time base C–t interval 0: the interstimulus interval of expected spatial convergence of a corticospinal and Group Ia volley at triceps Group Ia inhibitory interneurones. Negative C–t intervals: Group Ia before corticospinal volley, positive C–t intervals: corticospinal before Group Ia volley. Stars indicate statistically significant (P#0.05) differences between conditioned and test reflexes. (B) EMG of biceps when the subthreshold MAG and subthreshold tap to Tri are delivered together at the defined interstimulus (C–t) intervals for the adult subject illustrated by s in (A). Open arrows indicate onset of tap to Tri. Filled arrows indicate onset of TMS. mechanical transmission of the tap through the arm exciting spindle afferents of Bi (Table 1). However, a tap of up to twice the force delivered to Tri’s tendon failed to evoke reflexes in Bi when applied to soft tissue near Tri’s tendon or to the lateral epicondyle. This indicates that if 76 S.M. Mc Donough et al. / Brain Research 899 (2001) 66 – 81 Fig. 6. Calbindin D28K immunoreactivity in the ventral horn of human fetal spinal cord at C 6 segmental level. At all ages a strongly stained group of neurones is located where the ventral roots coalesce before traversing the ventral white matter. These are putative Renshaw cells (R). (A) Calbindin D28K positive neurones are present as early as 11 weeks post-conceptional age which include putative Renshaw cells. (B) A higher power image at 11 weeks post-conceptional age shows the nucleus (arrow) to be more intensely stained than the cytoplasm. Some staining of proximal dendrites can also be seen. Most motoneurones (M) appear weakly immunoreactive with occasional motoneurone-like cell strongly positive (arrow in A). There is no axonal staining in either the grey or white matter. (C and D) By 23 weeks post-conceptional age all motoneurones have lost calbindin D28K immunoreactivity, and there is no immunoreactivity in the surrounding neuropil, although calbindin D28K positive axons are present in the white matter. The Renshaw cell group (R) is again prominent. At higher magnification the cytoplasm is seen to be heavily immunostained and thin dendrites can be clearly seen. (E, F and G) By 35 weeks post-conceptional age axonal staining is strong both in the ventral white matter and in the neuropil around motoneurones (M in E and G). The Renshaw cells have thick dendrites that can be arranged in bundles (arrow in F). Scale bars: A, C and E, 250 mm; D, F and G, 50 mm; B, 25 mm. S.M. Mc Donough et al. / Brain Research 899 (2001) 66 – 81 Fig. 7. Trial averages recorded in the EMGs of contracting Tri and Bi in response to PRBS taps to the tendon of Tri in a neonate. The tap intensity is expressed as multiples of the threshold (T) of the homonymous phasic stretch reflex in contracting Tri. (A–C) Inhibitory responses are observed in Bi in response to taps of Tri at 0.5, 0.9 and 1.0 T in the absence of a homonymous phasic stretch reflex in Tri. D. Inhibition of Bi is not present when a homonymous phasic stretch reflex is evoked in Tri by a tap of 1.0 T. mechanical spread of the stimulus did activate spindles in Bi, the activity was subthreshold for evoking reflex responses. Since activation of Bi spindle afferents excites Bi motoneurones, mechanical transmission of the tap is likely to have reduced the probability of demonstrating reciprocal inhibition and enhanced the probability of demonstrating heteronymous excitation. Furthermore, to demonstrate inhibition of Bi, using the paradigm of the present study, background contraction was required. Voluntary contraction of one muscle in an agonist / antagonist pair leads to descending excitation of its Group Ia inhibitory interneurones [10,21,35,44]. Thus, isolated contraction of Bi, would have led to inhibition of the antagonists (Tri) Group Ia inhibitory projections to Bi [21]. Overall therefore the paradigm used in the present study will have reduced the probability of demonstrating reciprocal inhibition, but will have enhanced the ability to detect heteronymous excitation. 4.1. Reciprocal inhibition A major finding of this study was that short latency reciprocal inhibition from Tri to its antagonist Bi was 77 Fig. 8. (A) The magnitude of the inhibitory responses in Bi in relation the level of contraction in Tri in four neonates. To remove between subject variability in the absolute value of the EMG in Bi and Tri, the mean value for the level of contraction of Tri and the magnitude of the inhibition in Bi, was determined for each neonate. The magnitudes of contraction in Tri or inhibition in Bi were then expressed as a residual from their respective mean. Thus negative values indicate magnitudes which were less than the mean for that subject and positive values those which were greater than the mean. The amplitudes of inhibitory responses had a skewed distribution and underwent logarithmic transformation. NI, no inhibition of Bi; NC, no contraction in Tri. present in approximately 25% of neonates, with thresholds and central synaptic delays similar to those of adults [30]. 4.1.1. Group Ia afferents mediate short latency reciprocal inhibition in adults and neonates and throughout postnatal development Group Ia inhibitory interneurones have been shown to mediate short latency reciprocal inhibition between antagonist muscles in the cat [28]. The threshold for evoking inhibitory responses in Bi lay significantly below that of the homonymous phasic stretch reflex in Tri (Fig. 2D, Table 2) implying that the inhibition was mediated by low threshold afferents. A brief, low intensity tap to relaxed Tri was chosen to evoke the reflexes since it excites predominantly Group Ia afferents, although a small number of low threshold cutaneous afferents are also excited [6,38]. The inhibition mediated by cutaneous afferents is not present in neonates and does not appear until 18 months of age [25]. Furthermore, in adults cutaneous inhibition is 78 S.M. Mc Donough et al. / Brain Research 899 (2001) 66 – 81 part of a triphasic reflex where short latency inhibition (I 1 ) is preceded and followed by early (E 1 ) and late (E 2 ) excitation, respectively. Thus the inhibition observed in the present study is unlikely to have been mediated by cutaneous afferents. 4.1.2. Short latency reciprocal inhibition has a disynaptic linkage from birth Group Ia afferents project monosynaptically to Group Ia interneurones leading to a disynaptic inhibition of antagonist muscles [1,28]. The estimated additional synaptic delay above that for a monosynaptic reflex, for the subjects in the present study was 1.0–1.8 ms, consistent with a disynaptic linkage. This additional synaptic delay is longer than that previously estimated for reciprocal inhibition in adults [10,30], when electrical stimulation of the afferent fibres from the antagonist muscle was used to evoke reciprocal inhibition. The rise time of motoneuronal EPSPs following tendon tap is longer than that following electrical stimulation to the peripheral nerve [5] and together with the time required for transduction of the tap at muscle spindles, would be more than sufficient to account for the longer central delays of the present study. Katz et al. [30] also included a tendon tap to the antagonist muscle as part of the protocol of their study, and demonstrated a 12 ms discrepancy in the onset of the inhibitory responses in comparison to electrical stimulation, which they also attributed to the temporal dispersion of the volley evoked by the tendon tap. 4.1.3. Direct corticospinal projections to Group Ia inhibitory interneurones In the monkey Group Ia interneurones receive a monosynaptic projection from Group Ia afferents and from corticospinal axons [29]. In the present study we demonstrated inhibition of Bi EMG following spatial summation of a subthreshold Tri Group Ia afferent volley and a subthreshold corticospinal volley at their expected time of convergence on Group Ia interneurones. This spatial summation establishes a monosynaptic corticospinal projection to Group Ia inhibitory interneurones is also present in human neonates and adults. 4.2. Heteronymous excitatory responses Heteronymous excitatory responses evoked from Tri to Bi were observed in all but one of the neonates (Fig. 1F). Heteronymous excitatory reflexes to non-synergistic motoneurones, including those of the antagonist, have been reported previously in immature animals [11,41,43] and by ourselves in babies [32,36,37]. We confirmed our previous observations that the initial component of the heteronymous excitatory reflexes is monosynaptically linked in the newborn, since heteronymous excitatory reflexes in Bi had similar onsets to that of the homonymous phasic stretch reflex [36,37]. The present study however was the first to demonstrate that the threshold for evoking heteronymous excitatory responses is significantly lower in neonates than that for evoking homonymous phasic stretch reflexes (see Fig. 4 and Figs. 2H and 3D). Heteronymous excitatory responses occurred less frequently with age and their thresholds and onset latencies increased. Thus with development heteronymous excitatory responses may require more temporal summation and / or be mediated by afferents with slower conduction velocities. The homonymous phasic stretch reflex evoked by a tap produces oligosynaptic as well as monosynaptic excitation of the motoneurones [5]. Such oligosynaptic projections are also likely also to be involved in heteronymous excitation of Bi following a tap to Tri. The later onset of heteronymous excitation of Bi in adults may therefore reflect withdrawal of the monosynaptic projection during development leaving the onset of the reflex dependent upon an oligosynaptic linkage. Neonates and infants who demonstrated reciprocal inhibition, had significantly higher thresholds for their heteronymous excitatory responses than those not demonstrating reciprocal inhibition (Fig. 3D). We suggest two possible explanations for this observation. First, the restriction of heteronymous excitation between agonist and antagonists may occur because of the establishment of short latency reciprocal inhibition during development. Thus the establishment of short latency reciprocal inhibition leads to higher thresholds for heteronymous reciprocal excitation. If this is so then the establishment during development of reciprocal inhibition is not the only, nor the most significant process involved in the restriction of heteronymous reciprocal excitation. From 9 months of age all subjects demonstrate reciprocal inhibition but 70% still had heteronymous reciprocal excitation (Fig. 2A, E, F and G). An alternative explanation may be that low threshold and short onset heteronymous excitatory responses in neonates and infants mask reciprocal inhibitory responses which have at that stage both a higher threshold and a later onset. Thus, as the threshold for heteronymous reciprocal excitation increases with development to be greater than that for reciprocal inhibition, reciprocal inhibition can be demonstrated. The latter argument receives strong support from an observation in the present study. The neonates in whom the corticospinal projection to Group Ia inhibitory interneurones was investigated had, by chance, very low thresholds for heteronymous excitation between Tri and Bi. Neither demonstrated short latency reciprocal inhibition from Tri to Bi when a tap was delivered to Tri (Fig. 4B). However, spatial summation of a Tri Group Ia afferent volley (subthreshold for modulating the activity of Bi motoneurones) with a subthreshold corticospinal volley, revealed short latency inhibition of Bi in both subjects (Fig. 4E). If, as is likely from this observation, that reciprocal inhibition from Tri to Bi is masked by low threshold heteronymous excitatory reflexes, it is possible that reciprocal inhibition may be present in all neonates. S.M. Mc Donough et al. / Brain Research 899 (2001) 66 – 81 4.3. Renshaw (recurrent) inhibition 4.3.1. Anatomical study Calbindin immunoreactive neurones found in the ventral horn of both rat [2] and monkey [3] have the same morphology and location as electrophysiologically identified and intracellularly labelled Renshaw cells in the cat [27]. Furthermore, in both rat and cat, calbindin positive neurones in this location express characteristic glycine receptor complex clusters of identified Renshaw cells [3,7,8]. Injury to sciatic motor axons in rat down-regulates calbindin expression in putative Renshaw cells of the lumbar cord, suggesting a functional link between sciatic motoneurones and these cells [42]. In rat cervical cord, using double immunofluorescence techniques, putative calbindin positive Renshaw cells have been shown to be richly innervated by motor axon collateral terminals, labelled by injection of cholera toxin B into muscle [15,16]. Taken together, these observations make a strong case for using calbindin immunoreactivity as a marker for Renshaw cells in human post-mortem tissue, particularly as the immunostaining is reliable even in less than optimally fixed tissue. In human spinal cord putative Renshaw cells were identified from as early as 11 weeks post-conceptional age and by 35 weeks Renshaw cells were identified which were morphologically mature. In addition, calbindin immunoreactive varicose axons were present in the motoneuronal pools, some of which appeared to make contact with motoneurone cell bodies. Without formally identifying them as Renshaw cells, Ince et al. [24] described a population of calbindin positive neurones in a ventromedial location in adult human spinal cord. They also noted the presence of calbindin positive axonal varicosities in the motoneuronal pools indicating that this is a feature of mature spinal cord. This pattern of axonal staining has also been observed in the lumbar cord of the adult macaque monkey [3] where it was suggested that most, if not all, calbindin positive varicose axons around the motoneurones derive from Renshaw cell axons. 4.3.2. Neurophysiological study In the four neonates investigated, reciprocal inhibition of Bi was consistently absent when an homonymous phasic stretch reflex was evoked in Tri (Fig. 7D). The intensity of the tap to Tri was chosen to be at threshold so that a homonymous phasic stretch reflex was evoked in 50% of responses. Inhibition of Bi occurred however when Tri was contracting, but a tap to the tendon of the same intensity did not evoke the homonymous phasic stretch reflex (Fig. 7C). This observation excludes mechanical spread of the stimulus to the Group Ia afferents of Bi and / or excitation of heteronymous excitatory reflexes to Bi [36,37] counterbalancing the inhibitory reflex, since both these excitatory inputs to Bi motoneurones would be the same irrespective of whether a homonymous phasic stretch reflex in Tri was 79 evoked or not. Similarly descending inhibition of the respective pools of reciprocal inhibitory interneurones during spontaneous co-contraction of Tri and Bi [34] will also remain the same, irrespective of whether a homonymous phasic stretch reflex in Tri was evoked or not. It is concluded, therefore, that the suppression of reciprocal inhibition in Bi arose from Renshaw inhibition evoked by recurrent activity in Tri motoneurone axon collaterals, in a similar manner to that described by Katz et al. [30] in adult subjects. In that study a suprathreshold tap to Tri had to be delivered at least 1.5 ms before the electrical conditioning stimulus to abolish the inhibition in Bi. In the present study a series of taps was delivered rapidly to Tri over a period of 7.68 s. Thus activation of Tri motoneurones and their coupled Renshaw cells could be evident in the crosscovariance functions and co-incidentally collected EMG from Bi. The combined anatomical and physiological data of the present study provide the first evidence for Renshaw cells and Renshaw inhibition in human neonates. On the basis of intracellular recordings Naka [33] concluded that Renshaw inhibition was well developed in the kitten 1 week prior to birth, indicating that Renshaw inhibition is likely to be established early in development. 4.3.3. Central inhibition of Group Ia interneurones is present in the newborn During isolated contraction of Bi there was a higher probability of evoking inhibition which had a greater magnitude than when there was co-contraction of Bi and Tri (Fig. 8). In addition a significant negative correlation was found between increasing levels of contraction of Tri and the magnitude of reciprocal inhibition, in Bi. Nielsen and Kagamihara [34] have demonstrated central depression of Group Ia inhibition during co-contraction of agonist and antagonist muscles. As proposed by Nielsen and Kagamihara [34], it is possible that the depression of Group Ia interneurones is also produced indirectly by facilitation of recurrent inhibition, both by increasing levels of contraction in Tri and also by descending pathways. A further mechanism which may underlie the depression of reciprocal inhibition in the present study is presynaptic inhibition of Group Ia afferents, which has been shown to be modulated during various voluntary movements in adults [23]. The observation that the reciprocal inhibition in the neonate is depressed during co-contraction between Tri and Bi implies that Group Ia inhibitory interneurones may be under segmental and suprasegmental control. 4.4. Development of inhibitory and excitatory spinal reflexes Around the time of birth in several animal species, including man, spinal motor centres are undergoing extensive development and refinement of connections [33,36,37,41,43]. The present study has demonstrated that 80 S.M. Mc Donough et al. / Brain Research 899 (2001) 66 – 81 reciprocal inhibition is present in approximately 25% of babies at birth and is established in all subjects after 9 months of age, but co-contraction underlies limb movements until at least 18 months of age [18,17]. The discrepancy between the times of appearance of reciprocal inhibition and independent agonist / antagonist muscle control may lie in the gradual maturation and integration of spinal interneuronal reflex systems and the maturation of their differential control by supraspinal centres. Acknowledgements The support of Scope, which provided a research training scholarship for S.M.M. and of the Wellcome Trust is gratefully acknowledged. Technical support from Mr. S. Kelly, Mr. J. Clarke and Mr. G. Arnott is warmly acknowledged. References [1] T. Akaki, J.C. Eccles, M. Ito, Correlation of inhibitory postsynaptic potential of motoneurones with the latency and time course of inhibition of monosynaptic reflexes, J. Physiol. 154 (1960) 354– 377. [2] M. Antal, T.F. Freund, E. Polgar, Calcium-binding proteins, parvalbumin and calbindin D28K-immunoreactive neurons in the rat spinal cord and dorsal root ganglia: a light and electron microscopic study, J. Comp. Neurol. 295 (1990) 467–484. [3] U. Arvidsson, B. Ulfhake, S. Cullheim, V. Ramirez, S. Shupliakov, T. Hockfelt, Distribution of calbindin D28K-like immunoreactivity (L1) in the monkey ventral horn: do Renshaw cells contain calbindin D28K-L1?, J. Neurosci. 12 (1992) 718–728. [4] F. Baldissera, H. Hultborn, M. Illert, Integration in spinal neuronal systems, in: V.B. Brooks (Ed.), Handbook of Physiology, Section 1. The Nervous System, Motor Control, Vol. 2, American Physiological Society, Bethesda, MD, 1981, pp. 509–595. [5] D. Burke, S.C. Gandevia, B. Mckeon, The afferent volleys responsible for spinal proprioceptive reflexes in man, J. Physiol. 339 (1983) 535–552. ¨ [6] D. Burke, K.E. Hagbarth, L. Lofstedt, B.G. Wallin, The responses of muscle spindle endings to vibration of non-contracting muscles, J. Physiol. 261 (1976) 673–693. [7] P.A. Carr, F.J. Alvarez, E.A. Leman, R.E.W. Fyffe, Calbindin D28K expression in immunohistochemically identified Renshaw cells, Neuroreport 9 (1998) 2657–2661. [8] F.J. Alvarez, D.E. Dewey, D.A. Harrington, R.E.W. Fyffe, Cell-type specific organisation of glycine receptor clusters in mammalian spinal cord, J. Comp. Neurol. 379 (1997) 150–170. [9] E.A. Conway, The Development of Corticospinal Projections in Man, Ph.D. Thesis, University of Newcastle upon Tyne, Newcastle, 1995. [10] B.L. Day, C.D. Marsden, J.A. Obeso, J.C. Rothwell, Reciprocal inhibition between the muscles of the human forearm, J. Physiol. 349 (1984) 519–534. [11] J.C. Eccles, C.N. Shealy, W.D. Willis, Patterns of innervation of kitten motoneurones, J. Physiol. 165 (1963) 392–402. [12] S.A. Edgley, J.A. Eyre, R. Lemon, S. Miller, Comparison of activation of corticospinal neurones and spinal motoneurones by magnetic and electrical stimulation in the monkey, Brain 120 (1997) 839–853. [13] J.A. Eyre, S. Miller, V. Ramesh, Constancy of central conduction delays during development in man: investigations of motor and somatosensory pathways, J. Physiol. 434 (1991) 441–452. [14] J.A. Eyre, S. Miller, G.J. Clowry, E.A. Conway, C. Watts, Functional corticospinal projections are established prenatally in the human foetus permitting involvement in the development of spinal motor centres, Brain 123 (2000) 51–64. [15] Z. Fallah, Development and Plasticity of Markers for Inhibitory Neurotransmission in the Spinal Cord, Ph.D. Thesis, University of Newcastle upon Tyne, Newcastle, 1998. [16] Z. Fallah, G.J. Clowry, The effect of peripheral nerve lesion on calbindin D28K immunoreactivity in the cervical ventral horn of developing and adult rats, Exp. Neurol. 156 (1999) 111–120. [17] H. Forssberg, Ontogeny of human locomotor control. I. Infant stepping, supported locomotion and transition to independent locomotion, Exp. Brain Res. 57 (1985) 480–493. [18] V. Gatev, Role of inhibition in the development of motor coordination in early childhood, Dev. Med. Child Neurol. 14 (1972) 336–341. [19] M. Hadders-Algra, L.A. Van Eykern, A.W.J. Klip Van Den Nieuwendijk, H.F.R. Prechtl, Developmental course of general movements in early infancy. II. EMG correlates, Early Hum. Dev. 28 (1992) 231–251. ¨ Recurrent inhibition from [20] H. Hultborn, E. Jankowska, S. Lindstrom, motor axon collaterals of transmission in the Ia inhibitory pathway to motoneurones, J. Physiol. 215 (1971) 591–612. [21] H. Hultborn, Convergence on interneurones in the reciprocal Ia inhibitory pathway to motoneurones, Acta Physiol. Scand. Suppl. 375 (1972) 1–42. [22] H. Hultborn, M. Illert, M. Santini, Convergence on interneurones mediating the reciprocal Ia inhibition of motoneurones. I. Disynaptic Ia inhibition of Ia inhibitory interneurones, Acta Physiol. Scand. 96 (1976) 193–201. [23] H. Hultborn, S. Meunier, E. Pierrot-Deseilligny, M. Shindo, Changes in presynaptic inhibition of Ia fibres at the onset of voluntary contraction in man, J. Physiol. 389 (1987) 757–772. [24] P. Ince, N. Stout, P. Shaw, J. Slade, W. Hunzicker, C.W. Heizmann, K.G. Bainbridge, Parvalbumin and calbindin D28k in the human motor system and in human motor neuron disease, Neuropathol. Appl. Neurobiol. 19 (1993) 291–299. [25] H. Issler, J.A. Stephens, The maturation of cutaneous reflexes studied in the upper limb in man, J. Physiol. 335 (1983) 643–654. [26] E. Jankowska, Interneuronal relay in spinal pathways from proprioceptors, Prog. Neurobiol. 38 (1992) 335–378. ¨ [27] E. Jankowska, S. Lindstrom, Morphological identification of Renshaw cells, Acta Physiol. Scand. 81 (1971) 428–430. [28] E. Jankowska, W.J. Roberts, Synaptic actions of single interneurones mediating reciprocal Ia inhibition of motoneurones, J. Physiol. 222 (1972) 623–642. [29] E. Jankowska, Y. Padel, R. Tanaka, Disynaptic inhibition of spinal motoneurons from the motor cortex of the monkey, J. Physiol. 258 (1976) 467–487. ´ [30] R. Katz, A. Penicaud, A. Rossi, Reciprocal 1a inhibition between elbow flexors and extensors in the human, J. Physiol. 437 (1991) 269–286. [31] S. Miller, J. Clarke, J.A. Eyre, S. Kelly, E. Lim, V.M. McClelland, S. Mc Donough, A.V. Metcalfe, Comparison of spinal myotatic reflexes in human adults investigated with cross-correlation and signal averaging methods, Brain Res. 899 (2001) 47–65. [32] B.M. Myklebust, G.L. Gottlieb, R.D. Penn, G.C. Agarwal, Reciprocal excitation of antagonist muscles as a differentiating feature in spasticity, Ann. Neurol. 12 (1982) 367–374. [33] K. Naka, Electrophysiology of the fetal spinal cord. II. Interaction among peripheral inputs and recurrent inhibition, J. Gen. Physiol. 47 (1964) 1023–1037. [34] J. Nielsen, Y. Kagamihara, The regulation of disynaptic reciprocal inhibition during co-contraction of antagonist muscles in man, J. Physiol. 456 (1992) 373–394. S.M. Mc Donough et al. / Brain Research 899 (2001) 66 – 81 [35] J. Nielsen, Y. Kagamihara, C. Crone, H. Hultborn, Central facilitation of Ia inhibition during tonic ankle dorsiflexion revealed after blockade of peripheral feedback, Exp. Brain Res. 88 (1992) 651– 656. [36] M.C. O’Sullivan, J.A. Eyre, S. Miller, Radiation of the phasic stretch reflex in biceps to muscles of the upper limb in man and its restriction during development, J. Physiol. 439 (1991) 529–543. [37] M.C. O’Sullivan, S. Miller, V. Ramesh, E.A. Conway, K. Gilfillan, S. Mc Donough, J.A. Eyre, Abnormal development of biceps brachii phasic stretch reflex and persistence of short latency heteronymous excitatory responses to triceps brachii in spastic cerebral palsy, Brain 121 (1998) 2381–2395. [38] E. Pierrot-Deseilligny, C. Morin, C. Bergego, N. Tankov, Pattern of group I projections from ankle flexors and extensors in man, Exp. Brain Res. 42 (1981) 337–350. 81 [39] B.L. Plassman, S.C. Gandevia, High voltage stimulation over human spinal cord:sources of variation, J. Neurol. Neurosurg. Psychiatry 52 (1989) 213–217. [40] R.W. Ryall, Patterns of recurrent excitation and mutual inhibition of cat Renshaw cells, J. Physiol. 316 (1981) 439–452. [41] K. Saito, Development of spinal reflexes in the rat fetus, J. Physiol. 294 (1979) 581–594. [42] P. Sanna, M.R. Celio, F.L. Bloom, M. Rende, Presumptive Renshaw cells contain decreased calbindin during recovery from sciatic nerve lesions, Proc. Natl. Acad. Sci. 90 (1993) 3048–3052. [43] B.S. Seebach, L. Ziskind-Conhaim, Formation of transient inappropriate sensorimotor synapses in developing rat spinal cords, J. Neurosci. 17 (1994) 4520–4528. [44] M. Simoyama, R. Tanaka, Reciprocal Ia inhibition at the onset of voluntary movements in man, Brain Res. 82 (1974) 334–337.