Cerebral Cortex, Vol. 12, No. 3, 281-296,
March 2002
© 2002 Oxford University Press
Differences in the Corticospinal Projection from Primary Motor Cortex and Supplementary Motor Area to Macaque Upper Limb Motoneurons: An Anatomical and Electrophysiological Study
Sobell Department of Neurophysiology, Institute of Neurology, University College London, London, UK, , 1 INSERM U. 483, Université Pierre et Marie Curie, 75005 Paris, France, , 2 CNRS & Université de la Mediterranée, Faculté des Sciences du Sport, Marseille, France and , 3 Department of Life Sciences, Chung Shan Medical and Dental College, Taichung 402, Taiwan
Address correspondence to Professor Roger Lemon, Sobell Department of Neurophysiology, Institute of Neurology, University College London, Queen Square, London WC1N 3BG, UK. Email rlemon{at}ion.ucl.ac.uk.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionNotesReferences |
|---|
To further our understanding of the functional roles of different motor cortical areas, we made a quantitative comparison of the density of corticospinal projections from primary motor cortex (M1) and supplementary motor area (SMA) to spinal motor nuclei supplying hand and finger muscles in four macaque monkeys. We also compared the action of corticospinal outputs excited by electrical stimulation of these two areas on upper limb motoneurons recorded in three anaesthetized macaques. The hand representations of SMA and M1 were first identified using structural magnetic resonance imaging scans and intracortical microstimulation. In the anatomical study we then made focal injections of wheatgerm agglutinin– horseradish peroxidase into these representations, which were subsequently confirmed by analysis of retrograde cortical labelling. Densitometric analysis showed that corticospinal projections from M1 were denser and occupied a greater proportion of the hand muscle motor nuclei than did projections from SMA. In caudal Th1 the densest projections from M1 occupied 81% of this motoneuronal area, compared with only 6% from SMA. In the electrophysiological study, bipolar intracortical stimulation of the hand representation of M1 and SMA evoked direct (D) and indirect (I) corticospinal volleys. Volleys elicited by M1 stimulation had larger amplitudes and faster conduction velocities than those evoked from the SMA. Intracellular recordings were made from 84 contralateral upper limb motoneurons. M1 and SMA stimulation evoked markedly different responses in tested motoneurons: EPSPs were larger and more common from M1 (88% of motoneurons) than from SMA (48%). Some motoneurons (16/84) showed evidence of excitatory postsynaptic potentials mediated by monosynaptic action of the D-wave evoked from M1; these early effects were not observed from the SMA. In most motoneurons (74/84) EPSPs had segmental latencies indicating that they were due to monosynaptic action of the I-wave. The results are consistent with cortico-motoneuronal (CM) connections originating from both SMA and M1 converging upon single motoneurons, but those from M1 are far more numerous and exert stronger excitatory effects than from the SMA. Thus although they may function in parallel, the two CM projections probably make different contributions to upper limb motor control.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionNotesReferences |
|---|
It is well-established that many different areas of the cerebral cortex contribute to the corticospinal projection. In the frontal lobe, projections arise from primary motor cortex (M1), dorsal and ventral premotor areas, from a number of cingulate motor areas and from the supplementary motor area (SMA) (Dum and Strick, 1991
,1996; He et al., 1993; Galea and Darian-Smith, 1994; Picard and Strick, 1996; Wise, 1996). The function of these different projections is the subject of an ongoing debate: are specific functions carried out by individual structures that together go to make up the motor network, or are they distributed throughout the network (Fetz, 1992; Tanji, 1994; Dum and Strick, 1996)? Single neuron recordings in monkeys and brain-imaging studies in humans have stressed that both M1 and SMA are co-activated during many motor tasks (Alexander and Crutcher, 1990; Crutcher and Alexander, 1990; Dettmers et al., 1995; Fink et al., 1997; Kazennikov et al., 1999) and the ‘executive’ functions of both areas have been emphasized (Matsumura et al., 1991; Boecker et al., 1998; Jenkins et al., 2000; Crammond and Kalaska, 2000). However, characteristic functional differences between the two areas have also been identified in relation to bimanual activity, to external vs. internal guidance of movement, and to the planning of complex movement sequences (Tanji 1994; Jahanshahi et al., 1995; Picard and Strick, 1996; Sadato et al., 1997).
Insight into the functional specificity of the different corticospinal projections can be gained by comparing their pattern of spinal connections, since these connections ultimately determine the influence a cortical motor area can exert on the spinal machinery for movement. Of particular interest is whether each cortical motor area gives rise to direct cortico-motoneuronal (CM) connections. CM connections are a distinctive feature of the primate motor system and are known to be important for the capacity to perform independent finger movements (Porter and Lemon, 1993; Bortoff and Strick, 1993; Lemon, 1993; Maier et al., 1997b; Nakajima et al., 2000).
Dum and Strick (Dum and Strick, 1996) made a systematic study of corticospinal projections and showed that there was a high degree of similarity between the corticospinal projections from the hand/arm regions of M1, SMA and the different cingulate motor areas. They emphasized that all of these regions gave rise to corticospinal projections to lamina IX throughout the cervical enlargement, and they suggested that these different corticospinal outputs may function in a parallel fashion. Rouiller et al. (Rouiller et al., 1996) also demonstrated terminals from SMA and from M1 close to retrogradely labelled motoneurons supplying hand muscles. However, in neither study was it possible to make a quantitative comparison between the density of projections from M1 vs. those from SMA. Without this information it is difficult to gauge to what extent the M1 and SMA have similar potential to generate and control movements.
While there is extensive electrophysiological evidence of direct CM connections from M1: motoneuron recording (Landgren et al., 1962; Shapovalov, 1975; Maier et al., 1998); spike-triggered averaging of EMG (Fetz and Cheney, 1980; Lemon et al., 1986; Cheney et al., 1991; McKiernan et al., 1998), there is far less information about CM connections from SMA. Early work demonstrated that single pulse intracortical microstimulation (ICMS) in SMA could evoke short-latency EMG responses (Hummelsheim et al., 1986), but there has been no direct demonstration of CM connections from SMA, nor has a systematic comparison with CM effects from M1 been made.
The objectives of the current study were twofold. First, we made a quantitative comparison of the density of anterogradely labelled corticospinal projections from the hand representations of M1 and SMA to the hand muscle motor nuclei in the lower cervical cord. Second, to assess the strength of functional CM connections from M1 and SMA, we made intracellular recordings from motoneurons innervating hand and arm muscles while stimulating these cortical areas. The results establish that both areas give rise to CM connections which converge on single motoneurons. They also demonstrate that M1 exerts larger and more widespread CM excitation than does SMA, which has important implications for the respective roles of the two structures in movement control.
Preliminary accounts have been published previously (Maier et al., 1997a).
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionNotesReferences |
|---|
The study comprised parallel anatomical and electrophysiological experiments in seven adult macaque monkeys (see Table 1
). Anatomical data were obtained from four animals (one Macacca mulatta and three M. fascicularis); some results from one monkey (case D3) have been reported previously (Armand et al., 1997). The electrophysiological study was based on three monkeys (one M. mulatta and two M. fascicularis). All animals were captive bred for research purposes. Animal care and use was in accordance with the UK Animals (Scientific Procedures) Act 1986.
|
Identification of the SMA and M1 Hand/Arm Representations
Two approaches were used to locate accurately these representations: magnetic resonance imaging (MRI) and intracortical microstimulation (ICMS).
MRI
MRI was used to determine the precise sulcal geometry of the relevant cortical regions and allowed accurate placement of recording chambers, the angle of microelectrode penetrations for ICMS mapping, bipolar stimulating electrodes and of needle tracks for wheatgerm agglutinin– horseradish peroxidase (WGA–HRP) injection. Scans were carried out under deep general anaesthesia [for detailed methods see Baker et al. (Baker et al., 1999)].
ICMS Mapping
After an initial injection of ketamine (10 mg/kg i.m.), the monkey was intubated and anaesthetized with isoflurane (2–2.5% in a 1:1 O2/N2O mixture). A craniotomy was made and stainless steel chambers (i.d. 18 mm) were mounted over the SMA and, for monkeys CS5 and CS8, over lateral M1. The final location and orientation of the chambers was guided by the MRI scans. Following the surgery, an antibiotic (terramycin LA 20 mg/kg–1 i.m., Pfizer, Sandwich, Kent, UK) and an analgesic (buprenorphine hydrochloride 5–10 µg/kg i.m. Vetergesic, Reckitt & Colman, York, UK) were administered. Over the next 2–3 weeks, repetitive-pulse ICMS (rICMS) was used to map the motor representation of the SMA and M1. For this, the monkey was lightly sedated with ketamine (initial dose 10 mg/kg, subsequent doses of 10 mg/kg/h), so that there was clear muscle tone and a small amount of spontaneous movement. rICMS was delivered through a glass insulated platinum–iridium microelectrode with a low tip impedance (0.3–0.5 M
at 1 kHz). The coordinates of each penetration were selected according to the MRI and stereotaxic measurements of the chamber. Trains of rICMS pulses (20 0.2 ms pulses at 300 Hz delivered at rates of 0.5–1 Hz, search intensity up to 50 µA) were tested every 250 µm along the penetration, up to 7.5 mm from the point of electrode entry. The location and type of the lowest threshold evoked movements were noted. Mapping included the full extent of the SMA motor representation, including face, hand/arm, trunk, leg and tail areas (see Fig. 1). ICMS tracks were reconstructed from histological sections.
|
Anatomical Study
Injection of WGA-HRP
In three of the monkeys our aim was to make a large injection of WGA–HRP to include the hand representation of either M1 (D3) or SMA (CS3 and CS6). In the fourth (CS8), the objective was to make a smaller, focused injection centred on the hand representation of M1 in one hemisphere, and of the SMA in the opposite hemisphere. All injections were carried out under deep general anaesthesia [for details see Armand et al. (Armand et al., 1997)]. 0.1 µl of 10% WGA–HRP (Sigma) in 0.15 M saline was pressure injected via a 29 gauge stainless steel needle at 1 mm intervals along each track. One minute was allowed to elapse between injections at successive depths. The volume injected in each case is given in Table 1
. The survival period was 72–90 h.
Histochemical Processing
At the end of the survival period, the animal was deeply anaesthetized with Nembutal (30 mg/kg i.v.) and perfused through the heart with a vascular rinse [0.9% NaCl, 10 mM NaNO2, 5% polyvinyl pyrrolidone (PVP40), 5000 IU heparin at 36°C] followed by fixative (1% paraformaldehyde, 2.5% glutaraldehyde in 0.1 M phosphate buffer pH 7.3 at 4°C). After 30 min, the perfusion with fixative was stopped and continued with 0.1 M phosphate buffer (at 4°C) containing 10% sucrose for a further 15 min, and 20% sucrose for a final 15 min. After perfusion, blocks of tissue containing the injection site(s) were cut in the stereotaxic plane. After laminectomy, each spinal segment was carefully identified and the cord cut into blocks. Frozen sections were cut at 50 µm. The cortical sections were collected in two series; the first to visualize HRP using the conventional nitroprusside–tetramethyl benzidine (TMB) method (Mesulam, 1982) and the second for cytoarchitectonic analysis (stained with cresyl violet). Alternate spinal cord sections were processed with nitroprusside–TMB and paratungstate–TMB (Weinberg and Van Eyck, 1991) respectively; the latter sections were counterstained with neutral red to identify the motor nuclei.
Reconstruction of the Cortical Injection Site
Drawings of nitroprusside–TMB-reacted sections and cresyl violet stained sections were superimposed and the boundary of the injection site determined from the zones of reaction product density (Armand et al., 1997). Cytoarchitectonic areas in the region of the M1 and SMA injection sites were delineated according to previously established criteria (Jones et al., 1978; Luppino et al., 1991; Dum and Strick, 1991, 1996).
Densitometric Analysis of Corticospinal Projections
Our aim was to obtain a quantitative analysis of the relative density and distribution of reaction product in different parts of the spinal grey matter and at different segmental levels, so that we could carry out a detailed comparison of the corticospinal projection labelled from M1 and from SMA. The reaction product in the nitroprusside–TMB-reacted sections was photographed with polarizing filters and a dark-field effect; all sections were digitized with the same standard level of illumination (Armand et al., 1997). We chose this photographic method to enhance the resolution of the images. We carefully controlled for possible sources of error: quality of film, development, illumination, exposure and saturation effects [for details see Armand et al. (Armand et al., 1997)]. Densitometric analysis was carried out on sections from five half segments: caudal C7 to caudal Th1; 6–12 sections per half-segment were analysed. A ‘window of termination’ was selected which excluded any false labelling due to background or other artefacts (Armand et al., 1997). This range of labelling within the spinal grey matter was divided into five equal ranges (i.e. 1–20%, 21–40%, etc).
Density of Corticospinal Projection to Hand Muscle Motor Nuclei
Using an xy-plotter the locations of all motoneurons were determined (i) in each nitroprusside–TMB-reacted section with phase contrast and (ii) in the consecutive paratungstate–TMB-reacted section (counterstained with neutral red) with bright field. This allowed direct and precise delineation of the motor nuclei supplying the hand and finger muscles (Jenny and Inukai, 1983; Armand et al., 1997). On each nitroprusside-reacted section from caudal C7 to caudal Th1, the region of lamina IX containing these motor nuclei was selected and its area measured. The proportion of this area that was occupied by labelled projections in the different density ranges was determined (Armand et al., 1997).
In case CS8 we were able to make a direct comparison of corticospinal projections to the hand motor nuclei from M1 hand area (injected on the left side; Fig. 1A) and SMA (injected on the right side; Fig. 1B) because there are few, if any, ipsilateral projections from either SMA or M1 to the dorsolateral hand motor nuclei (Kuypers, 1981; Armand, 1982; Dum and Strick, 1996; Armand et al., 1997). Therefore any labelling within lamina IX must have arisen from the contralateral injection site. This approach avoided any possible errors due to variations in survival time, anterograde transport, strength of the histochemical reaction, etc.
Electrophysiological Study
Terminal experiment
When the ICMS mapping of SMA and M1 was complete, a terminal electrophysiological experiment was carried out. All preparatory surgery was carried out under isoflurane anaesthesia [as detailed above and by Maier et al. (Maier et al., 1998)]. Cuff electrodes were mounted on the median, ulnar and radial nerves at the axilla (Ma, Ua and Ra respectively), median and ulnar nerves at the wrist (Mw and Uw) and the deep radial nerve (DR). A laminectomy over spinal segments C3 to Th1 and an occipital craniotomy were carried out. When surgery was complete, isoflurane was discontinued and 
-chloralose was given up to a maximum of 80 mg/kg i.v. (range 50–80 mg.kg i.v.). The animal was mounted in a spinal frame and headholder, with clamps on the vertebral column at Th3 and in the lumbar region, and then paralysed with pancuronium bromide (Pavulon, Oregon-Technika, Cambridge UK) at a dose of 0.3 mg/kg/hr i.v. and artificially ventilated at a rate of 45 cycles/min. Adequacy of the anaesthesia was continuously assessed by reference to the blood pressure, heart rate and pupillary reflexes. Small doses (2–4 mg/kg i.v.) of pentobarbitone (Sagatal, Rhone Merieux, Harlow, UK) were administered when necessary. Body temperature was carefully maintained at 37–39°C. Fluid balance and blood gases were monitored and maintained; each animal remained in good physiological condition throughout the recording. Mean blood pressure was maintained above 80 mmHg.
Stimulation of the Pyramidal Tract (PT)
An electrode (varnish-insulated tungsten, tip impedance 20–30 k at 1 kHz) was inserted just rostral to obex and 0.5–1.5 mm to the right of the midline. Corticospinal volleys excited from the PT were recorded from the surface of the dorsolateral funiculus (DLF) at a rostral site (usually C3 or C4) and at a caudal site close to the region from which motoneuron recordings were made (C7, C8 or Th1). Conventional electrophysiological criteria were used for final electrode positioning (Maier et al., 1998). Stimuli of 10–200 µA (0.1 ms) were used.
Electrode Configuration for Stimulation of M1 and SMA
In pilot tests we found that bipolar intracortical stimulation, with the anode positioned superficially and the cathode deeper, was by far the most effective in producing direct excitation of corticospinal neurons. This was monitored by the size of the direct corticospinal volley recorded from the surface of the dorsolateral funiculus (see Figs 5 and 6). This electrode configuration was used for all three experiments. After removal of the ICMS mapping chamber and opening of the dura, bipolar electrodes were positioned within the hand representation of M1 and hand representation of the SMA (see Fig. 4). Varnish-insulated tungsten or elgiloy electrodes, with a tip separation of 2–3 mm and impedance <30 k at 1 kHz were used (Mitz and Wise, 1987). On the basis of the MRI scans, the electrodes were angled in an attempt to locate their tips in the deep (cathode) and superficial (anode) cortical layers (Fig. 4). Single cortical stimuli of up to 400 µA (duration 0.2 ms) were used.
|
|
|
Intracellular Recordings from Motoneurons
These recordings were made with glass microelectrodes filled with 3 M potassium acetate and having a DC resistance of 2–5 M. A small pressure foot was used to reduce movement of the spinal cord. All motoneurons were identified antidromically from the forelimb nerves. Intracellular and cord surface recordings were digitized directly at 10 kHz using a 1401plus interface (CED, Cambridge, UK). Membrane potential was monitored throughout the recording, and only data from stable periods of recording used for analysis (membrane potential <–50 mV). Latency and amplitude measurements where possible were derived from sets of measurements made from a number of single traces.
At the end of the experiment, small electrolytic lesions were placed at the cortical stimulation sites and the animal was killed with an intravenous overdose of barbiturate, and perfused through the heart with formal saline. Sites of stimulating electrodes were confirmed histologically (Suzuki and Azuma, 1976).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionNotesReferences |
|---|
Identification of the M1 and SMA Hand Representations
Of fundamental importance to this study was the accurate identification of the hand representation in M1 and SMA. Detailed ICMS mapping was carried out to guide both the WGA–HRP injections in the anatomical study, and the location of cortical stimulating electrodes in the electrophysiological experiments. Figure 1 shows reconstructions from case CS8 of ICMS tracks. In M1 (Fig. 1A) and SMA (Fig. 1B) hand/digit movements were evoked at low threshold (5–15 µA in M1, 12–15 µA in SMA). As noted by previous investigators, we found that effects from SMA had higher thresholds than those from M1, and that whereas tracks made in the centre of the M1 hand representation tended to evoke only hand or digit movements, the representation in SMA was far more intermingled, with the lowest threshold hand/digit loci being close to points from which elbow or shoulder movements could be evoked (Fig. 1) (Brinkman and Porter, 1979; Macpherson et al., 1982; Hummelsheim et al., 1986; Mitz and Wise, 1987; Luppino et al., 1991).
Anatomical Study
Injection Sites
M1. In case D3 a large injection (8.7 µl) of WGA–HRP was made in M1 hand region. Details of the injection site are given in Armand et al. (Armand et al., 1997) (Fig. 5, adult case 3). In case CS8 we made a direct comparison of corticospinal projections to the hand motor nuclei from M1 hand area and SMA (see Materials and Methods). The M1 injection (Fig. 1A; total volume injected 1.2 µl) was made in the immediate vicinity of the ICMS tracks that yielded low threshold digit movements (see inset). The injection site was restricted to the portion of area 4 located on the lip and the rostral bank of the central sulcus, and extended rostro-caudally ~3.5 mm.
SMA.
In cases CS3 and CS6, large injections were made in order to label the great majority of corticospinal projections arising from SMA and terminating in the cervical enlargement. A total of 4 µl was deposited at 40 different sites. In these two cases, the injection site extended outside the cortical area from which hand and forearm movements were elicited by ICMS. Thus in CS3 (Fig.1C), for example, the injection site occupied a cortical surface area of ~95 mm2, and was centred on the ICMS tracks which yielded hand and arm movements at the lowest threshold, whereas the cortical surface area covered by these ICMS tracks was ~35 mm2. This injection site involved the rostral three-quarters of area F3 (SMA proper) located in the medial wall of the hemisphere, as well as the adjoining cortex extending on to the convexity of the hemisphere. It also involved the region of SMA in the dorsal bank of the cingulate sulcus, an area known to give rise to a dense corticospinal projection (Dum and Strick, 1996). This injection slightly encroached the F3/F2 (SMA/PMd) border, on the convexity of the hemisphere, and the F3/24d (SMA/CMAd) border in the depth of the cingulate sulcus. The caudal boundary of the injection site was 2 mm rostral to the SMA/M1 border: the M1 leg area was not injected. A very similar injection was made in case CS6.
In case CS8 the SMA injection was made in the right hemisphere (Fig. 1B); the total volume injected (1.2 µl) was the same as for the M1 site. The SMA injection site was restricted to the portion of area F3 (Luppino et al., 1991), located on the medial wall of the hemisphere, and extending dorsally on the adjoining convexity, and ventrally to the dorsal bank of the cingulate sulcus (Fig. 1B). The injection site included the sites at which low-threshold digit movement was evoked by ICMS (Fig. 1B, inset).
Cortico-cortical and Callosal Labelling
Confirmation that the injection site in different cases did include the SMA hand area was obtained by plotting the distribution of neurons retrogradely labelled from the injection site. In the ipsilateral hemisphere, there was extensive labelling in M1 (area 4), with labelled cells on the convexity of the lateral precentral gyrus (arm area) and in the rostral bank of the central sulcus (hand area). We also saw labelling in area F6 (pre-SMA) (Luppino et al., 1991, 1993) and in F2 [PMd, dorsal premotor area (Barbas and Pandya, 1987; Luppino et al., 1993)] and some scattered labelling in the most caudal part of F7. Labelling was also seen in F4 and F5 (PMv, ventral premotor area), but no labelled cells were present in the fundus of the arcuate sulcus. Labelled neurons were found rostrally in area 24c (CMAr) and caudally in area 24d (CMAd). This pattern of labelling is consistent with injection in the hand representation of the SMA (Gentilucci et al., 1988; Hutchins et al., 1988; Kurata, 1989; Dum and Strick, 1991; Luppino et al., 1991) and is in keeping with our ICMS mapping (Mitz and Wise, 1987). Further confirmation was gained by examining the pattern of labelling in the contralateral hemisphere after SMA injections; this was very similar to that described by Roullier et al. (Roullier et al., 1994).
Corticospinal Terminations in C8–Th1 from M1 and SMA
After an SMA injection, the labelling within the spinal grey matter was relatively light and uniform compared to that after a M1 injection, although the overall distribution followed a similar pattern (Dum and Strick, 1996). This was particularly clear in case CS8, with small injections targeted at the hand areas of M1 and SMA, in opposite hemispheres (Fig. 1A,B): the majority of corticospinal projections was focused in the C8–Th1 grey matter. The overall labelling was much stronger in the hemispinal grey matter contralateral to M1 than to SMA (Fig. 2A). Lamina IX labelling from the M1 injection was relatively dense and distributed in all parts of hand and finger motor nuclei, which are found mainly in the C8 and Th1 segments (Jenny and Inukai, 1983). From SMA, lamina IX labelling was very light, although somewhat denser at the dorsal fringe of these same motor nuclei.
|
Figure 2B
shows the densitometric analysis of corticospinal labelling in C8 and Th1 from cases D3 (M1) and CS3 (SMA). The most striking difference is the much lighter labelling from SMA compared with that from M1. In these segments, as in the other cervical segments, the labelling was mainly distributed contralaterally in the base of the dorsal horn (laminae V–VI) and the dorsolateral part of the intermediate zone (lamina VII). Lighter labelling was seen bilaterally in its ventromedial part (lamina VIII and adjoining part of VII). Note the much lighter labelling of the dorsolateral motor nuclei from SMA compared with M1. Similar labelling from SMA was obtained in CS6. There was a clear rostro-caudal increase in lamina IX labelling from both SMA and M1.
Figure 3 summarizes the densitometric results. The first measure illustrated is the fraction of the area occupied by the hand muscle motor nuclei that was labelled (i.e. all labelling within the area, irrespective of density). After M1 injections (left panel) this measure saturated in most of the half-segments analysed i.e. the entire region occupied by the hand muscle nuclei was labelled. In the SMA cases (right), between 17.7 ± 5.1% and 65.1 ± 11.4% of this region was covered by label, with a clear rostro-caudal gradient from rostral C8 to caudal Th1. Similar results were obtained for both large and small M1 (D3 and CS8, respectively) and SMA injections (CS3 and CS8 respectively).
|
A more sensitive measure is given by the area of the hand muscle motor nuclei occupied by the densest labelling (i.e. in the range from 61–100% density; yellow and red pixels in Fig. 2B
). Using this measure, it is clear that at all levels the labelling was much more intense from M1 than from SMA (Fig. 3, hatched columns). In caudal Th1, for example, the area covered from the SMA projections in case CS3 was only 6.4 ± 2.4% compared to 81.0 ± 1.4% from M1 (case D3), a ratio of 12.7. This M1/SMA area ratio for these two cases showed even larger differences in Th1 rostral (26.9) and in caudal and rostral C8. Once again there was a good correspondence between the analysis for large and small M1 injections (D3 and CS8, respectively) and large and small SMA injections (CS3 and CS8, respectively). For case CS8 the M1/SMA ratio based on the 40% densest projections varied from 13.4 in caudal Th1 to 31.2 in caudal C8. Electrophysiological Study
Cortical Stimulation Sites
The cortical stimulation sites for the terminal experiment were positioned within the hand representation of SMA and M1 identified by previous ICMS mapping. In Figure 4A–C, penetrations yielding low-threshold arm or hand responses are marked by diamonds (
). The entry points of the final anode (+) and cathode (–) tracks in each of the three cases are shown in Figure 4A–C; several other exploratory tracks were made (+). Examples of electrode tracks in M1 are shown in Figure 4D,E (anterior bank of central sulcus) and in SMA (Fig. 4F). Electrode orientation was guided by previous ICMS tracking and MRI. Their final position and depth were adjusted to obtain the maximal corticospinal direct volley (see below). In the M1 case, the anode lay amongst the large layer V pyramidal neurons (Fig. 4D), and the cathode lay deeper (Fig. 4E). In the SMA case (Fig. 4F) the anode lay in lamina V, and the cathode in the white matter deep to it.
Corticospinal Volleys Recorded from the Contralateral DLF
Volleys from PT and from M1 and SMA cortex.
A single 200 µA shock to the PT (Fig. 5A) evoked a simple, biphasic volley followed by a longer-lasting positivity, which probably reflects postsynaptic activation (Maier et al., 1998). Unlike PT stimulation, single bipolar stimulation of M1 (400 µA) elicited a complex series of volleys (Fig. 5B). These volleys were identified as corticospinal by collision from the PT (not shown) (Edgley et al., 1990). The short latency of the first volley indicated that it was a D-wave, produced by direct excitation of corticospinal neurons (Patton and Amassian, 1954; Edgley et al., 1990; Baker et al., 1995). When recorded at the C3 level, this D-wave had a latency about double that of the volley evoked from the medullary PT (compare Fig. 5A with B), consistent with the longer conduction distance from cortex compared with PT. The later volleys from M1 stimulation were I-waves, resulting from indirect stimulation of corticospinal neurons (Patton and Amassian, 1954; Edgley et al., 1990, 1997). The interval between the D-wave and the first I-wave (labelled I1 in Fig. 5B) was typically 1.0–1.4 ms, and a similar interval was observed between I1 and I2. Stimulation of the SMA also elicited D- and I-waves (Fig. 5C); they had longer latencies and smaller amplitudes than those elicited from M1.
In case CS9, the conduction velocity of the fastest volleys recorded between C3 and Th1 was highest for the PT volley (86 m/s); the M1 D-volley was a little slower at 79 m/s, and that from SMA slower still at 61 m/s (Fig. 5A–C). For a given cortical stimulation site, the I1 wave had a similar conduction velocity to the D-wave: for M1 this was 79 m/s (D-wave) and 81 m/s (I1-wave). The D–I1 interval was similar for both the faster M1 and the slower SMA volleys (Fig. 5B,C). Comparable results were obtained in the other monkeys.
The amplitudes of the volleys evoked from the cortex were much smaller than those evoked from the PT (note the size of the calibration bars in Fig. 5A vs. 5B,C) and the I-waves were usually larger than the D-wave. Comparing the effects of a 200 µA PT shock (at which intensity it was close to maximal) and a 400 µA cortical shock revealed that the amplitude of the D-wave from M1 ranged from 3.7 to 3.9% of the PT volley; the larger I-waves were typically 3.3–9.3%. For SMA stimulation the corresponding ranges were 1.1–2.8% and 2.4–4.5%.
Volleys Elicited by Different Intensities and Locations in M1 and SMA.
In CS9 the threshold for eliciting cortical volleys was ~40 µA for both M1 and SMA (Fig. 5D,E). D- and I-wave amplitude grew approximately in parallel with increasing intensity. Only at the stronger intensities (200–400 µA) could the volleys be clearly identified in single traces (Fig. 5D,E).
Evidence for Independent Activation of M1 and SMA.
That corticospinal neurons in SMA and M1 were stimulated independently of each other was demonstrated by combined stimulation. If different populations of corticospinal neurons were activated at these sites, the resulting volleys from paired stimulation should be the same as the linear sum of the volleys evoked by stimulation of each site alone. Any occlusion of the D-wave would indicate current spread and activation of fibres from both sites; occlusion of the I-wave would indicate trans-synaptic convergence onto the same corticospinal neurons (e.g. via long range cortico-cortical connections). Figure 6A,B shows averaged volleys after separate stimulation of M1 and SMA, while Figure 6C shows the effect of combined stimulation. The D-volley produced by combined stimulation (Fig. 6C, thick line) is exactly the same as the linear sum of the separate D-volleys (thin line), while I1-volleys exhibited almost linear summation.
Responses in Motoneurons
Intracellular recordings were made from a total of 84 antidromically identified motoneurons (33 from CS2, 37 from CS5 and 14 from CS9) recorded in spinal segments C7, C8 and Th1. Most of these motoneurons innervated forearm finger or wrist flexors (Ua: 25, Ma: 12) or extensors (DR: 14) or intrinsic hand muscles (Uw: 15, Mw: 1); 17 motoneurons were identified from the whole radial nerve (Ra: 17) but not from the DR. Effects from PT and M1 stimulation were tested in all 84 motoneurons and from SMA stimulation in 75. A variety of synaptic potentials were observed. Figures 7–9 illustrated the range of responses and their classification.
|
|
|
Excitatory Postsynaptic Potentials (EPSPs). Two examples are shown in Figure 7
, both from radial motoneurons. In the first case (A–D), PT stimulation evoked an early EPSP, followed by an IPSP. The segmental latency of the EPSP (calculated as shown in Fig. 7B as the time from the positive peak of the corticospinal volley recorded from the same segment to the onset of the EPSP) was 1.0 ms, within the monosynaptic range (Jankowska et al., 1975b; Maier et al., 1998). Responses to cortical stimulation were always more complex than from the PT. A single 400 µA shock to M1 (Fig. 7C) evoked two EPSPs, the second following shortly after the first, and both having a size similar to the PT-evoked EPSP. The segmental latency of the first EPSP from the D-wave was 1.1 ms (see left-hand set of dotted lines in Fig. 7C), i.e. a value similar to that from the PT. These EPSPs were therefore termed direct-wave EPSPs because their segmental latencies from the D-wave were consistent with monosynaptic action of corticospinal impulses in that wave. Note that the absolute latency of the EPSP measured from the shock artefact was 2.3 ms, which is longer than that from the PT (1.4 ms), as would be expected from the longer conduction time. Direct-wave EPSPs were only seen in a minority of motoneurons, and were never seen after SMA stimulation.
The second, later EPSP in Figure 7C had a segmental latency from the D-wave of 2.3 ms, which places it beyond the monosynaptic range of the D-wave. The absence of any late excitatory effects from the single corticospinal volley generated by PT stimulation (Fig. 7B) makes it unlikely that it was due to non-monosynaptic effects of the D-wave (Maier et al., 1998). Rather, it was probably due to monosynaptic action of impulses in the I1-wave; the segmental delay after this wave was 1.1 ms (see right-hand set of dotted lines in Fig. 7C), identical to the value for monosynaptic action from the D-wave. Accordingly, EPSPs with segmental latencies from the first I-wave that were within the monosyaptic range are referred to as indirect-wave EPSPs. In this motoneuron no effects were observed from the SMA electrodes (Fig. 7D).
In the second motoneuron (Fig. 7E–H), PT stimulation also evoked a monosynaptic EPSP (segmental delay 0.9 ms), again followed by an IPSP (Fig. 7F). M1 stimulation (400 µA) produced a clear EPSP (Fig. 7G) with a segmental latency of 2.3 ms after the D-wave, but only 1.1 ms after the I1 wave. This EPSP was therefore also classified as an indirect-wave EPSP. Note that the form and size of the EPSP is practically identical to that from the PT, and that there was very little jitter in EPSP onset latency. In this motoneuron, SMA stimulation evoked an EPSP that was smaller than that from M1, but with a similar form; it had a segmental latency of 1.2 ms from the I1-wave. This demonstrates convergence of excitation from both cortical areas upon the same motoneuron.
Effect of Stimulus Intensity.
In many motoneurons, increasing the intensity of cortical stimulation evoked successive EPSPs associated with the augmented D- and I-waves; temporal summation between these effects often led to motoneuron discharge. In most cases I-wave EPSPs had lower thresholds than D-wave EPSPs. The intrinsic hand muscle motoneuron shown in Figure 8 showed a monosynaptic EPSP after PT stimulation (Fig. 8A: segmental latency 0.8 ms). The lowest intensity of M1 stimulation tested (50 µA), evoked an indirect wave-EPSP with an absolute latency of 3.0 ms and a segmental latency of 0.7 ms after the I1 wave (Fig. 8D). It had an amplitude of 1.4 mV. Increasing the intensity to 100 µA (Fig. 8C) evoked an earlier, presumed direct wave-EPSP with an onset latency of ~2.0 ms and a segmental latency of 0.6–0.7 ms after the D-wave. The amplitude of the compound EPSP grew to ~3.0 mV. With a 200 µA shock (Fig. 8B), the early direct-wave EPSP was much larger; this EPSP was followed by two successive indirect-wave EPSPs: the first of these had a segmental latency from the I1 wave of 0.7 ms, the second began 1.1 ms later, and may have been due to an I2 wave, although none was obvious in the cord surface recording. The compound EPSP had an total amplitude of 6.3 mV, substantially larger than the single EPSP evoked by suprathreshold activation of the entire PT (Fig. 8A: ~2.0 mV).
Inhibitory Postsynaptice Potentials (IPSPs).
In some motoneurons cortical stimulation produced ‘pure’ IPSPs uncontaminated by a preceding EPSP. Figure 9 shows a Ma motoneuron with a monosyaptic EPSP followed by a large IPSP from the PT (Fig. 9B). The IPSP had a segmental latency of 1.5 ms (dotted line in Fig. 9B), within the disynaptic range of 1.3–1.9 ms (Maier et al., 1998). Stimulation of M1 evoked only IPSPs: a small IPSP after the D-volley and a much larger one after the I-volley (Fig. 9C). In both cases, the segmental delay (1.6 and 1.7 ms) was within the disynaptic range. A disynaptic IPSP after the I-volley was also seen after SMA stimulation (Fig. 9D).
Segmental Latency of Motoneuron Responses
In Figure 10A–D, segmental latencies of EPSPs and IPSPs have been plotted with respect to the arrival of the D-volley from the appropriate area (M1 or SMA) at the appropriate segment, as shown by the arrow at far left of Figure. 10. This normalization procedure took account of the different corticospinal conduction velocities evoked from M1 and SMA (Fig. 5).
|
M1 EPSPs. EPSPs evoked from M1 fell into three clear groups. The first group (16 motoneurons) had latencies of 0.6–1.3 ms after the D-wave (Fig. 10A
, hatched columns) and were categorized as direct-wave EPSPs. For 14 of these motoneurons, the range of delays corresponded exactly with that of monosynaptic EPSPs evoked from the PT (0.5–1.1 ms; horizontal arrow in Fig. 10A) and the mean delay for this group of EPSPs, 0.95 ± 0.2 ms (n = 16), was only slightly longer than the segmental latency of the monosynaptic EPSP to PT stimulation (0.8 ± 0.1 ms, n = 78).
A second, larger (n = 74) group of EPSPs had segmental latencies of 1.6–3.0 ms from the D-volley (filled columns in Fig. 10A). These EPSPs were categorized as indirect-wave EPSPs due to monosynaptic action from the I-wave. If this classification is correct, one could predict that the latencies might be rather dispersed, because of the differences in D–I1 interval in the three different monkeys (see arrows above Fig. 10A). Indeed, when the segmental latencies of these EPSPs are replotted, as in Figure 10E, to normalize the D-I1 interval to 1.1 ms (arrow in Fig. 10E), the distribution of latencies became considerably narrower. The segmental delays of these EPSPs relative to the I1 volley ranged from 0.5 to 1.4 ms (mean 0.86 ± 0.24 ms, n = 74). A final group of late EPSPs had longer latencies (>2.7 ms after the D-wave); these late effects occurred just after the second I-wave.
SMA EPSPs.
No direct-wave EPSPs were observed and EPSPs with segmental latencies of 2.0–3.0 ms were considered to be indirect-wave EPSPs from the I1-volley. The segmental delays of these EPSPs, relative to the normalized I1 volley (Fig. 10G), ranged from 0.6 to 1.5 ms (mean 1.08 ± 0.21 ms, n = 32).
IPSPs.
Eight motoneurons showed IPSPs in response to the D-wave (hatched columns in Fig. 10B), with segmental latencies of 1.8–2.0 ms from the D-volley (mean 1.9 ± 0.1 ms). These delays were similar to those of disynaptic IPSPs evoked from the PT (1.4–2.2 ms, horizontal arrow in Fig. 10B). No such early IPSPs were seen from SMA (Fig. 10D). M1 and SMA stimulation also evoked IPSPs with longer latencies (filled columns in Fig. 10B,D) beginning at least 2.5 ms after the D-volley; many of these effects had segmental latencies with respect to the I1-volley that were within the disynaptic range [for M1, range 1.4–2.3 ms, mean 2.0 ± 0.2 ms (n = 34; Fig. 10F) and for SMA range 1.7–2.3 ms, mean 2.06 ± 0.16 ms (n = 18; Fig. 10H)].
Summary of Responses Evoked from PT, M1 and SMA
Table 2 lists the number of motoneurons in each type of response evoked by stimulation of the PT, M1 and SMA with a single shock, while Table 3 lists the number of response types observed.
|
|
PT stimulation (200 µA) usually produced a monosynaptic EPSP followed by a disynaptic IPSP (61/84 motoneurons; 73%). ‘Pure’ monosynaptic EPSPs (uncontaminated by later IPSPs) were found in 15 (18%) motoneurons, while IPSPs with no preceding EPSP were seen in five (6%) motoneurons. Longer latency EPSPs were seen in only three cases (Maier et al., 1998
).
M1 stimulation
(400 µA) produced a response in all 84 tested motoneurons. Direct-wave EPSPs were found in 16 cases (19%; see Fig. 11A), all 16 also showed later I-wave related effects. Indirect-wave EPSPs were found in 74 motoneurons (16 + 58; 88%); the remainder (12%) had only IPSPs. In any given motoneuron, there was a close parallel between excitatory effects from PT and M1: thus 88% of the 16 motoneurons with a ‘pure’ EPSP from the PT showed a similar response from M1. However, of the 63 motoneurons with an EPSP/IPSP sequence from the PT, only 24 (38%) showed an EPSP/IPSP from M1, while 34 (54%) showed a pure EPSP, and the remainder (8%), a pure IPSP. This suggests that cortical stimulation activates a rather different balance of corticospinal inputs to a given motoneuron than does PT stimulation.
|
SMA stimulation (400 µA) did not produce any direct-wave EPSPs (Figs 10C, 11A
). Indirect-wave EPSPs were recorded in 36 motoneurons (48%) and IPSPs in 10 (13%). Thus all of these received convergent effects from both SMA and M1. Of the 16 motoneurons with a pure EPSP from the PT, 11 (78%) also showed a pure EPSP from the SMA; of the 63 that showed an EPSP/IPSP sequence, only seven (21%) showed the same sequence from the SMA, while 21 (62%) showed a pure EPSP. Amplitudes of EPSPs
The distribution of EPSP amplitudes is shown in Figure 11B–D. PT stimulation (200 µA) evoked monosynaptic EPSPs with a broad range of amplitudes, the mean being 2.0 ± 1.1 mV (n = 53; range 0.5–5.0 mV; Fig. 11B). A similarly broad distribution was seen after M1 stimulation (400 µA), with a mean of 1.3 ± 1.1 mV (n = 104; range 0.2–4.4 mV). Direct-wave EPSPs tended to be rather small (1.2 ± 0.7 mV, n = 16) and indirect-wave EPSPs were larger (mean 1.5 ± 1.0 mV, n = 59, range 0.2–4.4 mV; Fig. 11C). This difference in amplitude was not significant (Mann–Whitney test P > 0.3). SMA stimulation (400 µA) evoked a narrower distribution with a strong skew towards very small effects (Fig. 11D). The average amplitude of only 0.5 ± 0.5 mV (n = 50) was significantly smaller than for M1 evoked EPSPs (Mann–Whitney test, P < 0.0001).
Figure 12 plots the relationship between the amplitude of EPSPs observed in a given motoneuron from PT, M1 and SMA. Motoneurons that showed a large CM EPSP in response to stimulation of the entire PT tended to show large direct-wave EPSPs after stimulation of M1 (Fig. 12A, P < 0.06, slope = 0.27, r = 0.45, n = 16). This correlation was also seen for M1 indirect-wave EPSPs (Fig. 12B, P < 0.05, slope = 0.44, r = 0.47, n = 42) and SMA indirect-wave EPSPs (Fig. 12D;P < 0.01, slope = 0.12, r = 0.54, n = 28). There was also a significant difference between the amplitude of indirect and late EPSPs evoked from M1 and those from SMA (Mann–Whitney test, P < 0.0001).
|
