Sobell Department of Neurophysiology, Institute of Neurology, University College London, London WC1N 3BG, United Kingdom
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Olivier, E., S. N. Baker, K. Nakajima, T. Brochier, and R. N. Lemon. Investigation Into Non-Monosynaptic Corticospinal Excitation of Macaque Upper Limb Single Motor Units. J. Neurophysiol. 86: 1573-1586, 2001. There has been considerable recent debate as to relative importance, in the primate, of propriospinal transmission of corticospinal excitation to upper limb motoneurons. Previous studies in the anesthetized macaque monkey suggested that, compared with the cat, the transmission of such excitation via a system of C3-C4 propriospinal neurons may be relatively weak. However, it is possible that in the anesthetized preparation, propriospinal transmission of cortical inputs to motoneurons may be depressed. To address this issue, the current study investigated the responses of single motor units (SMUs) to corticospinal inputs in either awake (n = 1) or lightly sedated (n = 3) macaque monkeys. Recordings in the awake state were made during performance of a precision grip task. The responses of spontaneously discharging SMUs to electrical stimulation of the pyramidal tract (PT) via chronically implanted electrodes were examined for evidence of non-monosynaptic, presumed propriospinal, effects. Single PT stimuli (up to 250 µA; duration, 0.2 ms, 2 Hz) were delivered during steady discharge of the SMU (10-30 imp/s). SMUs were recorded from muscles acting on the thumb (adductor pollicis and abductor pollicis brevis, n = 18), wrist (extensor carpi radialis, n = 29) and elbow (biceps, n = 9). In all SMUs, the poststimulus time histograms to PT stimulation consisted of a single peak at a fixed latency and with a brief duration [0.74 ± 0.25 (SD) ms, n = 56], consistent with the responses being mediated by monosynaptic action of cortico-motoneuronal (CM) impulses. Later peaks, indicating non-monosynaptic action, were not present even when the probability of the initial peak response was low and when there was no evidence for suppression of ongoing SMU activity following this peak (n = 20 SMUs). Even when repetitive (double-pulse) PT stimuli were used to facilitate transmission through oligosynaptic linkages, no later peaks were observed (16 SMUs). In some thumb muscle SMUs (n = 8), responses to PT stimulation were compared with those evoked by transcranial magnetic stimulation, using a figure-eight coil held over the motor cortex. Responses varied according the orientation of the coil: in the latero-medial position, single peak responses similar to those from the PT were obtained; their latencies confirmed direct excitation of CM cells, and there were no later peaks. In the posterio-anterior orientation, responses had longer latencies and consisted of two to three subpeaks. At least under the conditions that we have tested, the results provide no positive evidence for transmission of cortical excitation to upper limb motoneurons by non-monosynaptic pathways in the macaque monkey.
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The corticospinal tract
represents the major output pathway linking the motor areas of the
cerebral cortex to the spinal cord. The tract makes widespread
terminations throughout the spinal gray matter, including direct
cortico-motoneuronal (CM) connections to spinal motoneurons
(Armand 1982; Armand et al. 1997
;
Dum and Strick 1996
; Kuypers 1981
). CM
connections appear to be a feature unique to primates and are
particularly well developed in the more dextrous species
(Bortoff and Strick 1993
; Maier et al.
1997
; Nakajima et al. 2000
; Porter and
Lemon 1993
). However, the termination of many corticospinal
fibers in the spinal gray matter beyond the motor nuclei suggests that
the corticospinal tract may control motoneurons through other,
non-monosynaptic pathways operating in parallel with the CM pathway
(Lundberg 1992
).
One such possible pathway is the
C3-C4 propriospinal
system, which has been extensively studied in the cat and which has
been shown to transmit corticospinal excitation to forelimb motoneurons (Illert et al. 1978; see Alstermark and Lundberg
1992
). The basic organization of the CM and propriospinal
systems is shown in Fig. 1. In the cat,
the C3-C4 system provides
an important integrative system for motor commands originating from the
cortex, superior colliculus, red nucleus, and reticular formation
(Alstermark and Lundberg 1992
; Baldissera et al.
1981
).
|
Recent investigations by Maier et al. (1998) and
Nakajima et al. (2000)
found that transmission of
corticospinal excitation to upper limb motoneurons via
C3-C4 propriospinal
neurons in the chloralose-anesthetized macaque monkey was rather
uncommon. These studies emphasized major differences between species in
the organization of the corticospinal system: we pointed out that in
the macaque monkey, with a well-developed CM system,
C3-C4 propriospinal
excitation of motoneurons was uncommon, whereas in the squirrel monkey,
with rather a weak CM projection, such excitation was stronger and more
common. We also speculated that in humans, particularly strong and
ubiquitous CM connections may have subsumed some of the key functions
carried out by the propriospinal system in other species such as the
cat. However, the situation remains uncertain because the animal
studies were all carried out under general anesthesia, which may
selectively enhance transmission through direct pathways while
oligosynaptic transmission may be depressed. Indeed one explanation for
the paucity of propriospinal-like effects in the chloralose-anesthetized macaque has been that the relevant
C3-C4 interneurons are
under some form of inhibition in this preparation (Alstermark et
al. 1999
).
In the experiments reported here, we have recorded the responses of
single motor units (SMUs) in hand and arm muscles to single stimuli
delivered directly to the corticospinal tract via electrodes chronically implanted in the medullary pyramidal tract (PT; see Fig.
1). Because recordings were made from monkeys in either the awake or
lightly sedated state, any effects of general anesthesia were avoided.
To enhance transmission through oligosynaptic pathways, we also tested
double PT stimuli. In addition, we compared responses of the same SMU
to PT stimulation and to transcranial magnetic stimulation (TMS) over
motor cortex, using either a latero-medial coil orientation, which
directly excites corticospinal neurons or a postero-anterior
orientation, which excites them indirectly (di Lazzaro et al.
2001; Werhahn et al. 1994
). These results can be
compared directly with studies carried out in human subjects (de
Noordhout et al. 1999
; Pierrot-Desilligny 1996
).
All SMUs tested showed single, brief response peaks to both PT
stimulation and TMS in the lateromedial orientation. Later peaks,
possibly resulting from non-monosynaptic actions were not seen even
with double PT stimuli. Thus it would appear unlikely that the lack of
non-monosynaptic excitation revealed in our previous studies of the
macaque monkey (Maier et al. 1998; Nakajima et al. 2000
) was the result of depression by general anesthesia. The findings offer no positive evidence for the transmission of cortical excitation by such pathways in the macaque and indicate that
this form of excitation may play only a minor role in the cortical
control of upper limb motoneurons, at least under the conditions we
have been able to test.
A preliminary account of these results has been published (Lemon
et al. 2000).
![]() |
METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Monkeys
Recordings were taken from three purpose-bred female macaque monkeys. In one Macaca nemestrina (monkey 29, 6.0 kg body wt) recordings were made from intrinsic hand muscle SMUs during performance of a precision grip task. Further recordings were made from two M. mulatta (monkey 32, 5.3 kg and monkey 35, 5.0 kg) under light sedation with ketamine (5 mg/kg im).
Behavioral task
The precision grip task (Lemon et al. 1986)
required the monkey to position its thumb and index finger on two fixed
levers about 25 mm apart and to exert a steady isometric force of
around 1.0 N on each of them. The forces exerted were registered by
sensitive strain gauges attached to the levers. A tone indicated to the monkey when both finger and thumb forces were in target; successful trials were rewarded with small pieces of fruit. In some sessions, thenar muscle surface electromyographic (EMG) activity was rectified and smoothed and used as a biofeedback signal to control the "in target" tone signal; the levels were then set so as to encourage the
monkey to adopt a contraction level appropriate for steady discharge of
the sampled SMU.
Implant of PT-stimulating electrodes
Each monkey underwent an operation for implantation of a
stainless steel headpiece (for head fixation) and pyramidal tract stimulating electrodes. Surgery was carried out after induction with 10 mg/kg ketamine intramuscularly and maintained with 2-2.5% isoflurane
in a 50:50 mixture of
O2:N2O. Two fine
epoxy-insulated tungsten wire electrodes (tip impedance, 10-20 k at
1 kHz) were positioned in the medullary pyramids, at stereotaxic
coordinates A3 and P2 and 1.0 mm from the midline. Their location was
confirmed during the implant by stimulating and recording the
antidromic field potential from the surface of the ipsilateral motor
cortex. The final position of each electrode was adjusted to obtain the lowest possible threshold for the antidromic volley (range, 20-30 µA). After surgery, antibiotics (20 mg/kg im terramycin LA, Pfizer, Sandwich, Kent, UK) and an analgesic (buprenorphine hydrochloride 5-10
µg/kg im Vetergesic, Reckitt and Colman, York, UK) were
administered. Monkey 29 underwent a terminal experiment
under deep general anesthesia (see Maier et al. 1998
for
details) during which it was confirmed that stimulation through the PT
electrodes evoked a large orthodromic volley in the corticospinal
tract, recorded from the surface of the dorsolateral funiculus. At the
end of the experiment, monkeys were killed by an overdose of general
anesthetic (pentobarbital sodium, Nembutal, 80 mg/kg ip) and perfused
through the heart. The location of both electrode tips in the PT was
confirmed histologically in all three monkeys. All procedures were
carried out under appropriate licences from the UK Home Office.
Experimental protocol
AWAKE MONKEY.
EMG and SMU recording.
A pair of surface EMG electrodes was placed over the thenar muscles.
For SMU recordings from adductor pollicis (AdP), the method described
by Lemon et al. (1990) was adopted. In brief, pairs of
fine (25 µm) nylon-insulated microwires (Karma wire, Californian Fine
Wire) were cut across at the tip to produce a selective recording
electrode (tip impedance typically 1 M
at 1 kHz) and introduced into
the AdP muscle using a single sterile 27 gauge needle, which was then
immediately removed. Signals were amplified and high pass filtered at 1 kHz using Neurolog modules (Digitimer, Welwyn Garden City, UK) and
sampled at 10 kHz using a CED 1401plus interface (CED, Cambridge, UK).
In most experimental sessions, it was possible to record stable SMU
action potentials during the hold period of the task (Lemon et
al. 1990
). The precise orientation of the manipulandum was
adjusted until steady discharge (20-30 imp/s) of one clearly
discriminable SMU was obtained.
SEDATED MONKEYS.
SMU recordings.
These were made from muscles acting on the thumb (adductor pollicis,
AdP and abductor pollicis brevis, AbPB), wrist (extensor carpi ulnaris
radialis, ECR) and elbow (biceps), using a 26-gauge needle bipolar
needle electrode containing two 50-µm insulated stainless steel
wires, each having an impedance of around 0.5-1 M at 1 kHz
(Bawa and Lemon 1993
). With the low dose of ketamine given, there was considerable muscle tone present, and the steady firing rate of the selected SMU could be maintained by either cutaneous
or proprioceptive manipulation (brushing or squeezing the skin, bending
joints, etc). Throughout each experimental run, the precise form and
amplitude of the action potential was carefully monitored using a
digital storage oscilloscope and audio amplifier to ensure that records
were made from the same SMU.
Analysis
The difficulty of obtaining stable recordings from SMUs is well
recognized, especially in awake behaving animals. We took particular
care to ensure that our analysis was based on reliable recordings from
only one SMU. We used custom-written spike software (Getspike, SN
Baker) to discriminate the potentials generated by a single unit. This
process first displayed five sets of superimposed potentials,
discriminated by crossing a preset voltage/time threshold and selected
at random from five different sections of the total spike train (Fig.
2A). The similarity of the
discriminated waveforms in this display was then confirmed using
principle component analysis and cluster cutting techniques
(Eggermont 1990) to select the population of waveforms
most likely to belong to a given SMU (Fig. 2B). We then
examined the interval histogram of the discriminated events that this
analysis generated (Fig. 2C) to confirm that there were no
physiologically implausible short intervals: the presence of such
intervals is a clear indication that events from more than one SMU have
been discriminated. Peri-stimulus time histograms (PSTHs) of SMU
discharge were constructed for TMS and PT stimuli, using a 0.2-ms
binwidth (Fig. 2E). Each sweep was also checked by hand to
ensure that the SMU of interest could be clearly identified (Fig.
2D). We included in the analysis only those sweeps in which
there was steady discharge of the unit for the last 200 ms before
stimulus delivery (see Fig. 3A); this ensured that the
excitability of the motoneuron was comparable across all stimulation
protocols. The discharge times of all identified SMU discharges in the
period 50 ms before and after the stimulus were measured either by hand
or using Spike2 software (Cambridge Electronic Design). We measured the
entire duration of the response peak in the PSTH (not its half-width);
the onset and offset of the peak were determined from the cumulative
sum (CUSUM) (Ellaway 1978
) (see Figs. 3-5).
Mean latencies and response peak durations are given together with ± SD.
|
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Responses to PT stimulation were recorded in 56 SMUs that met all the strict criteria for well-discriminated SMUs. Nine AdP SMUs were investigated in the awake monkey (monkey 29); eight of these were also tested with TMS (7 responded to LM and 6 to PA orientation). In the sedated monkeys, a further 9 SMUs were recorded from either AdP (6) or AbPB (3) in monkey 32, as well as 13 SMUs from ECR and 9 from biceps. In monkey 35, 16 SMUs were recorded from ECR.
Responses to PT stimulation
COMPARISON BETWEEN THE AWAKE AND SEDATED CONDITIONS. Figure 3A illustrates a single-sweep recording from the awake monkey. It shows delivery of a single PT stimulus during steady discharge of an AdP SMU at around 20 imp/s. The PSTH (binwidth of 0.2 ms) and CUSUM of this unit's response to 104 PT stimuli are given in Fig. 3B. The inset shows five superimposed sweeps of the responding SMU to illustrate the stability of the recording. The stimulus intensity (100 µA, 0.2 ms) was adjusted so as to produce a low response probability (P = 0.16). The response consisted of a single, brief peak with an onset latency of 12.0 ms. The duration of the peak was only 0.8 ms. There was no evidence of later excitatory peaks. The CUSUM shows that there was no suppression after the initial peak that might have masked any later peaks.
|
SMU RESPONSES IN DIFFERENT MUSCLES. Examples of SMU recordings taken from biceps, ECR and AbPB are shown in Fig. 4, A-C, respectively. All were recorded from the same monkey (32) under sedation. The PT stimulation intensity (50-55 µA) was again adjusted to give a low response probability (~0.1). The responses all have a very similar form: a single, brief peak at a fixed latency. In both the biceps and ECR units, the responses peak was followed immediately by a period of suppression. The mean duration of the response peak for all thenar SMUs (0.91 ± 0.32 ms, n = 17) was significantly longer than that for ECR + biceps (0.66 ± 0.17 ms, n = 35, t-test, P < 0.001). All SMUs had a similar low response probability (mean: 0.13 ± 0.01, n = 56). Mean onset latencies were 12.0 ± 2.0 ms for thenar muscles (n = 18), 7.8 ± 1.9 ms for ECR (n = 29), and 6.2 ± 0.9 ms for biceps (n = 9). These mainly reflect differences in peripheral conduction distance from spinal cord to the sampled muscles.
|
EFFECT OF PT STIMULATION INTENSITY. Figure 5 shows responses of an AbPB SMU recorded in a sedated monkey in which the effects of different intensities of PT stimulation were tested. The background discharge rate of the SMU was around 12 imp/s; the threshold for excitation of this SMU lay somewhere between 40 µA (Fig. 5A, no clear response) and 50 µA (Fig. 5B, response present). The response was stronger at the higher intensities used (55 and 75 µA; Fig. 5, C and D, respectively), but the duration of the peak was unchanged. Weak suppression of SMU discharge was already present at 40 µA (Fig. 5A) and became steadily more prolonged with higher intensities. Similar results were obtained in a further 14 SMUs tested with two or more intensities.
|
RESPONSES TO PAIRED PT STIMULI.
In the anesthetized preparation, it is necessary to use repetitive PT
stimuli to ensure transmission of corticospinal excitation to
motoneurons through the oligosynaptic propriospinal pathway (Illert et al. 1978; Maier et al. 1998
;
Nakajima et al. 2000
). We reasoned that even without the
depressive effect of general anesthesia, single PT stimuli might be
ineffective in bringing significant numbers of PNs to discharge, in
which case no PN-mediated late excitation of SMUs would be seen. We
therefore tested 16 ECR SMUs recorded in monkey 35 with
pairs of PT stimuli, using an inter-stimulus interval of 3 ms. PSTHs
from two SMUs are shown in Fig. 6,
A1 and A2, and in Fig. 6, B1 and
B2, respectively. A number of interesting features can be
discerned: the SMUs showed a clear response to both the first and
second stimulus with identical latency. There was no evidence for
either broader or additional peaks after the second stimulus: the
duration of the peaks was identical. Finally, the response to the
second stimulus had a higher probability than the first. This occurred
despite the postpeak suppression of discharge that occurred after a
single stimulus; this is particularly noticeable for the SMU shown in
Fig. 6B. For the 16 ECR SMUs tested with double PT stimuli,
there was no significant difference between the duration of the first
and second response peaks (means 0.61 ± 0.2 and 0.65 ± 0.19, respectively, paired t-test, P = 0.57). The distribution of second peak durations is shown in Fig.
7C. The mean probability of
the first and second peaks was 0.12 ± 0.06 and 0.13 ± 0.05, respectively.
|
|
Responses to TMS
Figure 8 shows the responses of an AdP unit recorded in the awake monkey during performance of the precision grip task. This SMU was tested with several different intensities of TMS with the coil in the LM orientation. The SMU fired at a steady rate of about 30 imp/s for several hundred milliseconds before delivery of TMS. The PSTHs all show a short single response peak at just under 12 ms, followed by a suppression of spontaneous SMU discharge. At the lowest intensity tested (14% maximum stimulator output, MSO), response probability was low (P = 0.2), but a clear peak with a duration of 1 ms can be seen (Fig. 8A). Raising the intensity steadily increased the response probability (Fig. 8D), which rose to 0.55 at 16% (Fig. 8B) and to 0.78 at 20% MSO (Fig. 8C). There were no changes in either latency or duration of the response over the range 14-25% MSO (Fig. 8, E and F). This result was repeated for all five SMUs tested with different intensities of TMS.
|
Comparison of responses to TMS and PT stimulation
Figure 9 compares responses of the same AdP unit to TMS and to direct stimulation of the PT (100 µA). The latter produced a single, brief response peak with a sharp onset latency of 10.9 ms and short duration of 0.6 ms (Fig. 9A). The response probability of this peak was low (0.15), and there was no evidence for later excitatory peaks nor of any obvious postpeak suppression. TMS in the LM orientation also produced a single, brief response with a duration of 1.0 ms (Fig. 9B). This response had an onset latency of 11.2 ms, 0.3 ms longer than that from the PT (dashed vertical line). Suppression was evident after this initial peak. With the coil in the PA orientation (Fig. 9C), a quite different pattern of response was observed, with three separate peaks, each having a duration of around 1-1.4 ms. The first peak had an onset latency of 14 ms i.e., 2.8 ms later than that in the LM orientation. The third peak was followed by suppression of discharge.
|
Figure 9D summarizes the findings for those AdP units tested
with both TMS and PT. To remove variation in latency across SMUs due to
differences in the location of the intramuscular electrode (see
Lemon et al. 1990), the timing of the responses was
normalized to that obtained with PT stimulation, by subtracting the
onset latency to PT stimulation from the TMS response latency. Thus the
PT response would appear at 0 ms in Fig. 9D. All SMUs
responding to TMS (n = 7) showed a single brief peak
with the LM orientation, and there was little difference in the
duration of the excitatory response compared with that from the PT (PT:
0.93 ± 0.4 ms; TMS: 1.05 ± 0.34 ms). The TMS responses
occurred, on average, 0.5 ± 0.2 ms later than that from the PT.
In contrast, the PA orientation (n = 6 responding SMUs)
usually produced a much more variable response pattern, with 1, 2, or 3 response peaks, which had longer latencies and total duration than the
LM response. For the PA orientation, the mean latency difference from
the PT response was 3.9 ± 1.1 ms. The interval between the
different subpeaks, when present, was around 1.5 ms. The threshold for
evoking responses for the PA orientation was always higher than that
for the LM response; for example, Fig. 9 shows that an intensity of
16% in the LM orientation produced a similar level of response
probability to 40% in the PA orientation.
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The important conclusion resulting from this study is
that macaque upper limb SMUs show brief response peaks to both PT
stimulation and TMS (LM orientation) consistent with monosynaptic,
cortico-motoneuronal activation of the parent motoneuron. These SMUs
showed no sign of any later peaks that might be expected if there were
significant transmission of corticospinal excitation of motoneurons
through non-monosynaptic pathways. Our sample of SMUs offers the unique opportunity of comparing responses to TMS with those evoked from direct
stimulation of the PT. Interpretation of the motor effects produced by
TMS is often complicated by its repetitive activation of the
corticospinal system. It generates a succession of direct and indirect
waves (Day et al. 1987; di Lazzaro et al.
1998
; Edgley et al. 1997
; Rothwell et al.
1991
), and it is therefore difficult to be certain about the
origin of motor responses or reflex facilitation that occur after the
initial, presumably monosynaptic component. Such later responses could
result from the later arrival of I waves from the corticospinal tract,
as has been widely assumed (Day et al. 1989
;
Olivier et al. 1995
), but could also be due to
non-monosynaptic excitatory action mediated by propriospinal neurons
(see Pauvert et al. 1998
).
In this study, we found SMU responses were highly consistent from one SMU to another, and the response characteristics of units in thumb muscles, a wrist extensor (ECR), and elbow flexor (biceps) to be remarkably similar. Responses to TMS varied between simple, single peaks similar to those evoked from the PT, to more complex responses consisting of several subpeaks. We argue that all of these responses are probably due to CM action.
Responses to PT stimulation
Direct stimulation of the PT produces a relatively simple
corticospinal volley, discharging each corticospinal axon once only, compared with the repetitive discharge evoked by surface stimulation of
the cortex, intracortical microstimulation or noninvasively with
transcranial magnetic (TMS) or electrical (TES) stimulation (Burke et al. 1990; Day et al. 1987
;
Edgley et al. 1990
,1997
; Jankowska et al.
1975
; Kernell and Wu 1967
). Thus PT stimulation allows us to examine the response of upper limb motoneurons to a much
less complex input than is generated by TMS or TES; this can in turn
illuminate the likely origin of responses to noninvasive stimulation.
The most plausible explanation for the single peaks of short duration
evoked from the PT is that they result from monosynaptic excitatory
postsynaptic potentials (EPSPs) evoked in the motoneurons by CM
connections. A high proportion of macaque upper limb motoneurons receive a monosynaptic EPSP from single stimuli delivered to the PT,
and these are particularly common and large for intrinsic hand muscle
motoneurons (Fritz et al. 1985; Maier et al.
1998
; see Porter and Lemon 1993
). Maier
et al. (1998)
found that in the chloralose anesthetized macaque
monkey, 73% of upper limb motoneurons showed such monosynaptic EPSPs;
their sample included motoneurons supplying elbow, wrist, and hand
muscles. The lack of variation in the SMU response latency and its
brief duration (less than 1.1 ms in 52/56 SMUs, mean: 0.74 ms) are both
consistent with a monosynaptic origin (Fetz and Gustafsson
1983
; Kirkwood 1995
). The duration of these
peaks probably reflects the brief rise time of CM EPSPs in hand and
forearm motoneurons (0.93 ± 0.18 ms) (Maier et al.
1998
). It is also similar to or less than the duration of peaks
observed in cross-correlations between CM cells and SMUs (Lemon
and Mantel 1989
; Mantel and Lemon 1987
; see
Porter and Lemon 1993
).
The PSTHs obtained provide no evidence for later non-monosynaptic
excitation from the PT, at least under the present experimental conditions. It would be expected that non-monosynaptic inputs, mediated
either through a C3-C4
propriospinal system or through segmental excitatory interneurons,
would either broaden the initial response peak or even result in later
separate subpeaks of activity. Maier et al. (1998) did
find a small proportion of motoneurons that showed non-monosynaptic
EPSPs after repetitive PT stimulation. These EPSPs had latencies
0.7-2.5 ms longer than the earliest CM monosynaptic EPSPs (see also
Alstermark et al. 1999
). Therefore it could be predicted
that any SMU discharges due to such non-monosynaptic effects should be
clearly discriminable from the early, brief monosynaptic peak.
There are a number of reasons why non-monosynaptic actions may have
been missed in the present experiments. First, it could be argued that
the sample of motor units tested was dominated by those from more
distal muscles and that we may have missed propriospinal-mediated
excitation in more proximal muscles. There is evidence in the cat that
such excitation may be particularly directed to more proximal muscles
concerned in the act of reaching rather than to more distal muscle
groups involved in movements such as food-taking (see Alstermark
and Lundberg 1992). However, it is also essential to note that
although behavioral and other evidence has implicated the
C3-C4 system in reaching
all cat forelimb motoneurons exhibit corticospinal
excitation via C3-C4 PNs,
including those supplying distal muscles. In support of this,
stimulation of the lateral reticular nucleus (which has been shown to
activate ascending axons of
C3-C4 PNs; see Fig. 1)
evokes large EPSPs in most cat forelimb motoneurons, and some of the
largest effects were in ECR motoneurons (Alstermark and Sasaki
1986
). Thus if a
C3-C4 propriospinal system
is similarly organized in the macaque monkey, we should expect
PN-mediated effects in most motoneurons.
Second, it is possible that neither single- nor double-pulse PT
stimulation is capable of bringing
C3-C4 PNs to discharge, preventing transmission through to the tested motoneurons. In the
deeply anesthetized preparation, repetitive stimulation with up to
three or four shocks is sometimes necessary to obtain
C3-C4 transmission.
However, in an awake or lightly sedated animal, one would expect the
PNs to be more excitable and therefore able to transmit effects evoked
by even single PT shocks. We should add that in both the
chloralose-anesthetized cat and squirrel monkey, transmission through
PNs was clearly evident after double shocks (see Illert et al.
1976, Fig. 2; Nakajima et al. 2000
, respectively).
Third, there is the issue of whether the fast monosynaptic CM-EPSPs
dominated the response of the motoneuron and that discharges resulting
from these EPSPs prevented any response to later EPSPs mediated by
non-monosynaptic pathways. To avoid this, we specifically selected a
rather weak PT stimulation intensity so that the response probability
was low (Figs. 3-6). Under these circumstances, there should be a
disproportionately large number of sweeps (80-90%) in which the early
CM-EPSP fails to fire the motoneuron but would summate with any later
oligosynaptic EPSPs to discharge it at a longer latency (cf.
Malmgren and Pierrot-Deseilligny 1988). We should add
that primate PN-mediated EPSPs, when they do occur, have rise times
similar to those in the cat (Maier et al. 1998
; Nakajima et al. 2000
).
Finally, it is possible that PT-evoked inhibition could mask
non-monosynaptic excitation (see Maier et al. 1998).
Such inhibition could arise either from local, segmental inhibitory
interneurons (Jankowska et al. 1976
; Perlmutter
et al. 1998
) or indeed from propriospinal inhibitory neurons
(Alstermark et al. 1999
) with corticospinal input. In
this respect, the results from the 20 SMUs in which there was no clear
postpeak suppression is of critical importance: examples of such
responses are shown in Figs. 3, B and C,
5B, and 6A. In these cases, any significant
non-monosynaptic effects should have been detectable, but none was observed.
In other SMUs we did observe suppression of ongoing SMU activity shortly after the initial excitatory peak (Figs. 2E, 4, A and B, 5, C and D, and 6C). However, in some of these SMUs, CUSUM analysis indicated that suppression did not begin until 1-4 ms after the peak (e.g., Fig. 4C), i.e., it was not present during the 0.7- to 2.5-ms period after the initial peak in which non-monosynaptic effects would be expected and therefore could not have masked any later peaks. Further, robust responses to double PT stimuli proved that the inhibitory postsynaptic potential (IPSP) induced by the first volley was not sufficiently large to obscure responses to the second (Fig. 6); however, this might not apply to rather a small and weak oligosynaptic input.
The present results are in keeping with recent investigations by
Maier et al. (1998), who, in chloralose-anesthetized
monkeys, made intracellular recordings from a large number of upper
limb motoneurons in response to PT stimulation. Late EPSPs reflecting oligosynaptic transmission were rarely observed. Repetitive PT stimulation was used to facilitate transmission through
non-monosynaptic pathways but only 19% of motoneurons showed late
EPSPs and only 3% had EPSPs with a disynaptic latency, appropriate for
a C3-C4 propriospinal
linkage (see Alstermark and Lundberg 1992
). A
C5 lesion was made to reduce the corticospinal
input to the lower cervical segments, with the aim of reducing
segmental excitatory and inhibitory actions and thereby revealing any
propriospinal actions. However, even after such a lesion, the
proportion of disynaptic EPSPs was still small (14%). Maier et
al. (1998)
provided a number of arguments to show that their
results were not due to selective depression of non-monosynaptic
pathways by general anesthesia. The absence of late peaks in both the
awake and lightly sedated monkey reinforces that conclusion. The use of
repetitive (double-pulse) PT stimulation, which should have enhanced
propriospinal transmission, also failed to produce any evidence of
later peaks (Fig. 6). If the
C3-C4 system is indeed
under some form of inhibition, as suggested by recent experiments by
Alstermark et al. (1999)
, then our data indicate that
transmission through this system remains suppressed even in the awake state.
Our results raise fundamental questions as to the functional importance
of C3-C4 transmission
under natural conditions. Of course we cannot exclude the possibility
that such transmission might occur under different experimental
conditions than we have tested here. There are plentiful examples in
the literature of pathways that are patent only under particular
behavioral conditions. For instance, disynaptic excitation of
external intercostals motoneurons by muscle spindle afferents during
inspiration (Kirkwood and Sears 1982), disynaptic
excitation of flexor and extensor motoneurons during fictive scratching
(Degtyarenko et al. 1998
), or during particular phases
of locomotion (McCrea 1998
; McCrea et al.
1995
; Quevedo et al. 2000
). Feed-forward
corticospinal inhibition of the PN system is likely to be enhanced by
strong stimulation (Nicolas et al. 2001
); note however
that all of our experiments were done with rather weak PT stimuli,
which evoked low response probabilities, and that PN-type effects were
absent even at the weakest intensities (Fig. 5).
The absence of non-monosynaptic excitation is perhaps surprising given
the widespread termination of the corticospinal tract in regions beyond
the motor nuclei, i.e., in laminae VI-VIII (Armand et al.
1997). Many of these terminations are probably concerned with
pre- and postsynaptic control of reflex inputs; many excite spinal
interneurons with inhibitory actions on upper limb motoneurons (Baldissera et al. 1981
; Cheney et al.
1985
; Jankowska et al. 1976
; Rothwell et
al. 1984
) (see Fig. 1). Inhibitory effects from the PT are
widespread: Maier et al. (1998)
found that 65% of
sampled motoneurons had IPSPs within the disynaptic range. These IPSPs probably explain the early suppression by PT stimulation of ongoing activity in some SMUs. The presence of this suppression demonstrates that non-monosynaptic pathways other than the propriospinal one were
operating effectively under the conditions of our experiments. Recent
work in the awake behaving monkey has demonstrated the presence of many
segmental interneurons which facilitate upper limb motoneurons
(Perlmutter et al. 1998
; Prut and Fetz
1999
); whether or not these interneurons receive a significant
corticospinal input is as yet unresolved. Clearly this approach would
be able to clarify the issue of
C3-C4 propriospinal
transmission still further.
Responses to TMS
The responses of AdP units to TMS in the LM orientation consisted
of single brief peaks, identical to those evoked from the PT, but with
a slightly longer onset latency (mean: 0.5 ± 0.2 ms). This is
similar to the conduction delay from cortex to medullary pyramid in the
fastest corticospinal fibers (Baker et al. 1994, 1995
)
and is therefore consistent with direct activation of corticospinal neurons at or close to their initial segment (Edgley et al.
1997
). The SMU responses are most likely the result of the
monosynaptic action of impulses in the direct or D wave that can be
recorded from the corticospinal pathway (Baker et al. 1994
,
1995
; Edgley et al. 1990
, 1997
).
Several earlier studies have shown that the direction of the induced
cortical current flow can markedly alter the type of corticospinal
activation (Amassian et al. 1990) and the latency and
form of the EMG and SMU responses (Amassian et al. 1989
;
Davey et al. 1994
). Werhahn et al. (1994)
and di Lazzarro et al. (2001)
have shown that when
current is induced in the LM direction, there is direct activation of
corticospinal neurons and significantly shorter EMG response latencies
than for currents flowing in the PA direction, which induces mainly
indirect, trans-synaptic activation. The precise reason for
this difference probably depends on the orientation the corticospinal
cells' soma, dendrites and axon initial segment (Hern et al.
1962
) and on the trajectory of their axons within the
subcortical white matter with respect to the induced current flow
(Amassian et al. 1992
).
The similarity in the duration and form of the response peaks after PT
stimulation and after TMS (LM orientation) suggests that both activate
a rather similar population of the CM cells projecting to the sampled
SMU. The absence of later peaks in the response to TMS in the LM
orientation means that its principal effect on these cells, at least in
the monkey, is D-wave activation; our earlier work suggested relatively
little I-wave activity using this coil orientation (Baker et al.
1995). In contrast, the responses to TMS in the PA orientation
had longer latencies and were composed of multiple peaks with interpeak
intervals of 1.2-1.5 ms, similar to those reported in many earlier
human studies (Day et al. 1989
; Olivier et al.
1995
; Werhahn et al. 1994
) and primate studies (Edgley et al. 1990
, 1997
; Maier et al.
1997
). This interval corresponds exactly with that between
successive I waves in epidural recordings from the spinal cord in man
(Burke et al. 1993
; di Lazzaro et al.
1998
) and from recordings of single corticospinal axons in the
monkey (Edgley et al. 1997
). Thus it is probable that
each of the peaks in the PSTH reflects monosynaptic activation of the motoneuron by successive volleys of corticospinal I-wave activity rather than by non-monosynaptic inputs. The presence of later peaks
after TMS in the PA orientation demonstrates that the motoneurons remain excitable in this period and again argues against masking of
non-monosynaptic excitation by inhibition. For example the probability
of discharge in the first subpeak in Fig. 9C
(P = 0.08) is not very different to that evoked from
the PT in Fig. 9A (P = 0.15): with TMS,
there were subsequent subpeaks, but with the single volley set up by PT
stimulation, no later peaks were seen. The experiments with double PT
stimuli also confirmed the excitability of SMUs in the postpeak period
(see preceding text).
Propriospinal transmission of corticospinal excitation in humans?
Compared with TMS, low-intensity anodal electrical stimulation
(TES) of the motor cortex in human subjects produces a much simpler
pattern of descending activity that is dominated by direct activation
of corticospinal axons (Burke et al. 1990; Day et
al. 1987
). Recent studies by de Noordhout et al.
(1999)
using this type of stimulation demonstrated that SMUs
recorded from a wide range of human upper limb muscles responded with
brief single peaks almost identical to those reported here (mean
duration: 0.97 ms), again with no sign of later excitation. This
suggests that in man, as in the macaque monkey, direct stimulation of
the corticospinal tract does not lead to activation of motoneurons via
oligosynaptic pathways. Responses to TES are followed by profound suppression of SMU discharge, and one criticism of the results obtained
by de Noordhout et al. (1999)
is that propriospinal
excitation may have been masked by this suppression. The results from
our study showing absence of any clear PN-mediated effects in those SMUs without such suppression suggests that such objections may be unfounded.
The conclusion that propriospinal transmission of cortical excitation
is not present in humans is in conflict with the long series of
experiments on human subjects carried out by Pierrot-Deseilligny and
his colleagues indicating that a significant proportion of cortical
excitation may be transmitted through a propriospinal-like system,
organized along similar lines to the
C3-C4 system in the cat
(Burke et al. 1994; Gracies et al. 1994
;
Marchand-Pauvert et al. 2000
; Pauvert et al.
1998
; Pierrot-Deseilligny 1996
). Using cortical
TMS alone, these authors have also not reported any longer-latency effects in SMUs that could be attributed to non-monosynaptic pathways. Rather evidence for non-monosynaptic transmission has been obtained by
interacting the descending effects set up by TMS with activation of
non-monosynaptic reflex inputs to upper limb motoneurons. These inputs
have been generated by stimulating particular peripheral nerves (those
without any monosynaptic connections to motoneurons of the tested
muscle) or using particular stimulus intensities at which
non-monosynaptic reflex effects are present (Pauvert et al.
1998
; Pierrot-Deseilligny 1996
).
Thus it seems possible that in humans, as in macaque monkeys, propriospinal transmission from the fast-conducting corticospinal system that is activated by TMS, TES, or PT is weak and is not revealed in either EMG or SMU recordings unless the propriospinal neurons are facilitated by additional excitatory peripheral inputs. We should point out that such inputs were probably present in our study, both in the awake monkey performing the precision grip task, and even in the sedated animals, since natural proprioceptive and cutaneous stimuli were applied to sustain steady SMU discharge. Thus facilitation from these sources should have been present. We did not attempt spatial facilitation with electrical stimulation of peripheral nerves; a pathway that transmits only when highly synchronous corticospinal and peripheral inputs generated by electrical stimulation are presented simultaneously probably reflects a rather weak transmission system, which may have little functional relevance for transmission of asynchronous inputs present under natural conditions.
Conclusion
Our experiments do not provide any evidence in favor of
non-monosynaptic excitation of macaque upper limb motoneurons from the
corticospinal tract but rather serve to emphasize the strength of the
CM input. Our results in the macaque, like those of de Noordhout
et al. (1999) in humans, do not discount the existence of a
C3-C4 propriospinal
pathway but rather suggest that this pathway may be too weak to be
responsible for transmission of a significant proportion of
corticospinal input to motoneurons.
![]() |
ACKNOWLEDGMENTS |
---|
We thank Dr. Marc Maier for help. H. Lewis, R. Spinks, N. Philbin, and Dr. Chris Seers provided expert technical support. We are grateful to Drs. Peter Kirkwood and Marc Maier for comments on the manuscript.
This work was supported by grants from the Wellcome Trust and from the Medical Research Council.
![]() |
FOOTNOTES |
---|
Address for reprint requests: R. N. Lemon, Sobell Dept. of Neurophysiology, Institute of Neurology, Queen Square, London WC1N 3BG, UK (E-mail: rlemon{at}ion.ucl.ac.uk).
Received 20 February 2001; accepted in final form 4 June 2001.
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|