Inhibitory influence of the ipsilateral motor cortex on responses to stimulation of the human cortex and pyramidal tract
Author's present address C. Gerloff: Cortical Physiology Research Group, Department of Neurology, University of Tuebingen, Hoppe-Seyler-Strasse 3, 72076 Tuebingen, Germany.
Corresponding authors L. G. Cohen: Human Cortical Physiology Unit, NINDS, NIH, Building 10, Room 5N234, 10 Centre Drive, MSC-1430, Bethesda, MD 20892-1430, USA. Email: firstname.lastname@example.org
The primary motor cortex (M1) is capable of exerting control over motor responses not only in the contralateral arm, but also in the ipsilateral arm (Colebatch & Gandevia, 1989; Jones, Donaldson & Parkin, 1989; Wassermann, Fuhr, Cohen & Hallett, 1991; Chiappa, Cros, Kiers, Triggs, Clouston & Fang, 1995). In humans, this descending ipsilateral motor control seems to be functionally relevant under physiological as well as pathophysiological conditions. It has been suggested, for example, that after a unilateral cerebral stroke the unaffected M1 ipsilateral to the paralytic limb may take over functions for the damaged M1 (Frackowiak, Weiller & Chollet, 1991; Miller-Fisher, 1992). Further, it has been speculated that a decrease in inhibition between the M1 of both hemispheres may cause mirror movements (Schott & Wyke, 1981). It is not clear, however, where along the neuroaxis this descending modulatory influence of the M1 on muscles of the ipsilateral limb is relayed. A transcallosal route has been proposed (Ferbert, Priori, Rothwell, Day, Colebatch & Marsden, 1992; Meyer, Roricht, Graefin von Einsiedel, Kruggel & Weindl, 1995; Boroojerdi, Diefenbach & Ferbert, 1996), but transcallosal connections between cortical hand motor representations in primates are sparse (Gould, Cusick, Pons & Kaas, 1986; Rouiller, Babalian, Kazennikov, Moret, Yu & Wiesendanger, 1994) and, from an anatomical point of view, various other pathways such as the reticulospinal tract (Nathan, Smith & Deacon, 1996) or spinal interneuron circuits (Shahani & Young, 1971; Roby-Brami & Bussel, 1987; Burke, Gracies, Mazevet, Meunier & Pierrot-Deseilligny, 1992; Gracies, Meunier & Pierrot-Deseilligny, 1994; Mazevet, Pierrot-Deseilligny & Rothwell, 1996) could also mediate ipsilateral effects, because they are, at least in part, bilaterally organized. Therefore, we hypothesized that the descending modulatory influence of the human M1 on muscles of the ipsilateral limb may be relayed not only through the corpus callosum but also through subcortical routes.
Cortical transcranial magnetic stimulation (cTMS) in humans has shown that the influence of the M1 on ipsilateral limb muscle activation is mostly inhibitory (Ferbert et al. 1992; Meyer et al. 1995; Brown, Ridding, Werhahn, Rothwell & Marsden, 1996). A conditioning stimulus (CS) applied over the M1 of one hemisphere induces an amplitude decrease of motor-evoked potentials (MEPs) elicited by a subsequent test stimulus (TS) over the homologous M1 of the other hemisphere (Ferbert et al. 1992). It was proposed that this particular effect might be mediated through a cortico-cortical route via the corpus callosum. If this effect is entirely transmitted through the corpus callosum, then MEPs elicited by non-cortical, direct stimulation of the pyramidal tract should not be affected by a conditioning shock applied over the homologous M1, since this type of stimulation would bypass modulatory cortico-cortical influences. Cortical TMS (cTMS) is thought to stimulate pyramidal cells largely trans-synaptically, thereby producing predominantly indirect corticospinal volleys (I-waves). These are subject to modulations of cortical excitability, including possible cortico-cortical inhibitory interactions via the corpus callosum. In contrast, transcranial electrical stimulation at the level of the pyramidal decussation (pdTES) can produce corticospinal volleys that are clearly not modulated by different levels in cortical excitability (Ugawa, Rothwell, Day, Thompson & Marsden, 1991). Cortical transcranial electrical stimulation (cTES) can produce both D-waves and I-waves, depending on the stimulus intensities applied (Rothwell, Burke, Hicks, Stephen, Woodforth & Crawford, 1994). A systematic combination of magnetic and electrical stimuli can therefore be used to differentiate where modulatory effects may be relayed (cortical or subcortical).
In the present study, we investigated the effects of magnetic cortical conditioning stimulation, applied to the M1 ipsilateral to a target muscle in the upper limb, on MEPs elicited by cTMS and pdTES, in both relaxed and preactivated muscles; further experiments included cTES and monosynaptic H-reflexes.
We studied sixteen normal volunteers (11 men and 5 women) whose mean (±s.d.) age was 36.2 ± 14.2 years. Fourteen subjects were right handed, according to the Edinburgh handedness inventory, and two were ambidextrous. The study protocol was approved by the Institutional Review Board, and all subjects gave their written informed consent for the study.
All stimulation parameters, as well as motor threshold values, and unconditioned control amplitudes are listed in Table 1 (relaxed target muscles) and Table 2 (preactivated target muscles). All values are presented as means ± 1 s.d.
Table 1. Experimental series 1 with muscles at rest: stimulation parameters, motor thresholds and control values
|Mode of stimulation TS:||cTMS||cTMS||pdTES||cTMS||cTES||cTMS||H-reflex|
|Motor threshold (right FDI) (% stimulator output)||68 ± 10||69 ± 9||60 ± 14||63 ± 12||44 ± 7||64 ± 13||5.76 ± 2.39 mA|
|TS intensity (% MT)||124 ± 10||123 ± 13||120 ± 0||131 ± 12||122 ± 4||135 ± 4||113 ± 4|
| || || ||(504 ± 127) *|| ||(404 ± 52) *|| || |
|Control response amplitude (no CS) (mV)||1.23 ± 0.89||1.88 ± 0.39||0.67 ± 0.22||1.64 ± 0.89||1.27 ± 0.35||0.75 ± 0.19||0.43 ± 0.19|
|Motor threshold (left FDI) (% stimulator output)||69 ± 9||66 ± 5||66 ± 5||63 ± 6||63 ± 6||65 ± 12||65 ± 12|
|CS intensity (%MT)||127 ± 14||143 ± 6||143 ± 6||142 ± 16||142 ± 16||143 ± 16||143 ± 16|
|CS-induced MEP amplitude (mV)||1.60 ± 0.87||2.04 ± 0.57||2.35 ± 0.72||2.20 ± 1.12||2.56 ± 0.96||1.18 ± 0.68||1.33 ± 0.74|
|Interstimulus intervals||2, 5, 10, 20, 30, 40, 50, 70, 100||10, 20||10, 20, 40, 70, 100||10, 40, 100|
Table 2. Experimental series 2 with preactivated muscles: stimulation parameters, motor thresholds and control values
|Active motor threshold (aMT)||51 ± 6||42 ± 9|
|Relaxed motor threshold (rMT) (right FDI) (% stimulator output)||58 ± 6||49 ± 8|
|(% aMT)||120 ± 7||116 ± 7|
|(% rMT)||105 ± 5||99 ± 8|
| || ||(366 ± 83) *|
|Control response amplitude (no CS) (mV)||2.46 ± 0.77||1.51 ± 1.08|
|Active motor threshold (aMT)||44 ± 8||44 ± 8|
|Relaxed motor threshold (rMT) (left FDI) (% stimulator output)||52 ± 6||52 ± 6|
|(% aMT)||137 ± 8||137 ± 8|
|(% rMT)||116 ± 14||116 ± 14|
|CS-induced MEP amplitude (mV)||2.81 ± 0.65||3.06 ± 0.62|
|Interstimulus intervals (ms)||6, 7, 10, 20||10, 20|
For paired-pulse cTMS, we used two magnetic stimulators (Cadwell Laboratories Inc., Kennewick, WA, USA) with figure of 8-shaped coils, the characteristics of which are described elsewhere (Cohen et al. 1990). Each loop of the coils measured 4.5 cm in diameter. With each coil, the intersection of the loops was placed perpendicular to the expected orientation of the central sulcus of the respective hemisphere, tangentially to the scalp surface, and then adjusted until an MEP of maximal amplitude was produced in the relaxed target muscle, at a given stimulus intensity. At the optimal orientation, the coil handle usually pointed posteriorly with an angle of 45-80 deg to the sagittal axis. Stimulus intensity was expressed as a percentage of the motor threshold of the target muscle. Motor threshold was defined as the minimal output of the stimulator capable of inducing five MEPs in the target muscle with an amplitude of ≥ 50 μV in ten trials with single stimuli applied to the optimal scalp position for activation of this muscle. This was done at rest (the relaxed motor threshold, rMT) or with the target muscle preactivated at about 5 % maximum (the active motor threshold, aMT; cf. Ferbert et al. 1992). For conditioning and test stimulation, we used intensities that were likely to produce test responses comparable to the ones reported by Ferbert et al. (1992). In relaxed muscles, we aimed at inducing stable unconditioned test MEPs of approximately 1.5 mV peak-to-peak amplitude with cTMS and cTES, and in preactivated muscle MEPs of approximately 2.5 mV peak-to-peak amplitude (only cTMS). For the CS, the intensity chosen was slightly higher, since the inhibitory effects are more pronounced as the CS intensity increases (Ferbert et al. 1992). With electrical stimulation at the level of the foramen magnum, the maximal MEP amplitudes that could be obtained at a tolerable level of discomfort and with no significant background EMG activity were of the order of 0.5-1.0 mV in relaxed muscles, and approximately 1.5 mV in preactivated muscles.
Brainstem and cortical TES.
A Digitimer D180 electrical stimulator (maximal output, 750 V; time constant, 100 μs; Digitimer, Welwyn, Garden City, UK) was used for pdTES and cTES. For pdTES, the electrodes were placed over the mastoid process bilaterally, approximately 5 cm lateral to the inion (Ugawa et al. 1991), and the polarity was chosen so that the response was larger in the right than in the left target muscle. Stimulus intensity was then increased stepwise until stable MEP responses could be elicited in at least five subsequent trials. This intensity was considered to be the motor threshold. The slightly different approach in pdTES, compared with cTMS and cTES, was chosen in order to minimize the number of stimuli applied, since pdTES is generally associated with more discomfort owing to neck muscle contraction. For cTES, the anode (cup-shaped gold electrode) was placed over the M1 contralateral to the target muscle (7 cm lateral to the Cz position of the international 10/20 system of electrode placement), and the cathode over position Fz. Thresholds and stimulus intensities were determined as described for cTMS.
The stimulator module of a Dantec Counterpoint electromyograph (Dantec Medical A/S, Skovlunde, Denmark) was used to apply a single-pulse stimulus to the median nerve at the elbow for eliciting an H-reflex in the flexor carpi radialis (FCR) muscle.
Recording technique and data pre-processing
Bipolar EMG was recorded from the first dorsal interosseous (FDI) or FCR muscle (H-reflex experiment) bilaterally. For FDI recordings, the active electrode was placed over the muscle belly and the reference electrode over the radial styloid process (approximately 5 cm interelectrode distance). For FCR recordings, the active electrode was placed over the muscle belly, approximately 6 cm distal to the elbow, and the reference electrode was positioned 5 cm distal and radial to the active one. The EMG was sampled at 5 kHz, band-pass filtered from 5 Hz to 1.5 kHz (Dantec Counterpoint electromyograph), and stored in a personal computer (IBM compatible) for off-line analysis.
To assess the conditioning effects, peak-to-peak amplitudes were measured, and averages across trials were computed for each experiment and normalized as described below. In addition, the MEP latencies were determined for cTMS, pdTES and cTES experiments.
Two series of experiments were carried out. In the first series (Table 1), the effect of conditioning stimulation on responses elicited by cTMS, pdTES or cTES, or on H-reflexes was explored in relaxed muscles. In the second series (Table 2), the conditioning effect on responses evoked by cTMS and pdTES test stimuli was studied in preactivated muscles.
For all experiments, subjects were seated comfortably in an armchair with both arms resting on a pillow. The CS was always magnetic (cTMS) and always applied to the right M1, ipsilateral to the relaxed target muscle. The interstimulus intervals (ISIs) between CS and TS were presented in a random fashion, interleaved with control trials (without CS). In experimental series 1 magnetic and electrical test stimuli were presented blockwise within the same session; in series 2 cTMS and pdTES stimuli were presented randomly within the same run of trials. Accurate triggering of CS and TS was achieved with a personal computer and a custom-made software module based on ASYST (version 4.0, Keithley Instruments, Inc., Cleveland, OH, USA). The pairs of stimuli were applied once every 4-10 s for magnetic and electrical transcranial stimulation, and every 10-20 s for the H-reflex experiments. Short breaks between stimulations were allowed as necessary to assure that the subject stayed alert and relaxed (series 1) or could maintain the required preactivation (series 2).
Series 1 (conditioning effects on MEPs in relaxed muscles)
Experiment 1 (cTMS).
Nine subjects participated in experiment 1. The TS was magnetic and applied to the left M1; the CS was also magnetic and applied to the right M1. ISIs between CS and TS were 2, 5, 10, 20, 30, 40, 50, 70 and 100 ms. EMG was recorded from FDI bilaterally (right FDI was the target muscle). Ten MEP peak-to-peak amplitudes per ISI were averaged, and for each ISI the mean MEP amplitude was expressed as a percentage of unconditioned control (100 %).
Experiment 2 (pdTES).
Nine subjects participated in experiment 2. Five subjects could not tolerate pdTES, or could not relax sufficiently during the pre-stimulus period. The final analysis was therefore based on four subjects. The ISIs were 10 and 20 ms. EMG was recorded from FDI bilaterally (right FDI was the target muscle). Five MEP peak-to-peak amplitudes per ISI were averaged. Experiment 1 was repeated for the ISIs of 10 and 20 ms within the same session.
Experiment 3 (cTES).
Six subjects participated in experiment 3. One subject had to be excluded because of failure to relax. The ISIs were 10, 20, 40, 70 and 100 ms. EMG was recorded from FDI bilaterally (right FDI was the target muscle). Five MEP peak-to-peak amplitudes per ISI were averaged. Experiment 1 was repeated for the ISIs of 10, 20, 40, 70 and 100 ms within the same session.
Experiment 4 (H-reflex).
Five subjects participated in experiment 4. In one subject, a stable H-reflex could not be obtained. To elicit an H-reflex in the FCR muscle, the right median nerve was stimulated at the elbow using a standard bar electrode. Stimulus duration was 0.7 ms. The ISIs were 10, 40 and 100 ms. EMG was recorded from FCR bilaterally (right FCR was the target muscle). Twenty H-reflex peak-to-peak amplitudes per ISI were averaged. The FCR muscle was selected because it was not possible to elicit a stable H-reflex in the FDI in a sufficient number of subjects. To ensure that the FCR behaviour was comparable with the behaviour of the FDI in the magnetic-magnetic paired-pulse protocol, we repeated experiment 1 for the ISIs of 10, 40 and 100 ms using the FCR as the target muscle in this session.
Series 2 (conditioning effects on MEPs in preactivated muscles)
Test stimulation: cTMS (cf. series 1, experiment 1) and pdTES (cf. series 1, experiment 2) randomized.
Eight subjects participated in series 2. Two subjects could not tolerate pdTES and therefore could not complete the study. In four subjects, we focused on the randomized presentation of cTMS and pdTES test stimuli. For this part, the ISIs were 6, 7, 10 and 20 ms for cTMS, and 10 and 20 ms for pdTES. A direct comparison of ISIs of 6 and 7 ms with cTMS and 10 ms with pdTES was desired in order to match the response latencies when taking into account the differences in central conduction time for the two stimulus modalities (pdTES 3-4 ms shorter than cTMS). In another subgroup of four subjects, we assessed the effects of different stimulus intensities on pdTES-evoked responses. This was done for ISIs of 10 ms, since inhibition was generally most pronounced at that latency. EMG was recorded from preactivated FDI bilaterally (right FDI was the target muscle). Five MEP peak-to-peak amplitudes per ISI were averaged for pdTES, ten for cTMS. The different number of trials averaged for pdTES and cTMS was associated with comparable variances across conditions (s.d., 0.67 mV for all cTMS-evoked responses pooled, 0.61 for all pdTES test shock responses pooled).
All statistical tests were computed on normalized (percentage) data to account for intersubject variability of absolute MEP or H-reflex amplitudes. Effects were considered significant if P < 0.05 after appropriate correction for multiple comparisons.
A factorial one-way ANOVA (main effect for INTERVAL) and Bonferroni-corrected post hoc comparisons were used to determine significant inhibition of the TS-evoked amplitudes (MEP, H-reflex) for the different ISIs in experimental series 1 and 2.
Linear regression analysis was used in series 2 to determine the effect of changing the CS intensity on the amount of inhibition of the test response elicited by pdTES.
The present results provide novel evidence that the inhibitory influence of the human M1 on ipsilateral hand muscles is to a significant extent mediated below the cortical level, and not only through cortico-cortical transcallosal connections.
Ipsilateral inhibition of cortically evoked muscle responses
Conditioning stimulation over the M1 caused pronounced inhibition of MEPs in the FDI and FCR muscles of the ipsilateral arm evoked by a magnetic test stimulus to the contralateral M1. Inhibition occurred when the conditioning stimulus was given between 6 and 50 ms before the test stimulus.
A transcallosal route has been proposed for the transmission of inhibitory interactions between the bilateral M1 hand representations (Wassermann et al. 1991; Ferbert et al. 1992; Meyer et al. 1995; Boroojerdi et al. 1996). In non-human primates, however, the transcallosal connections between motor representations of distal arm muscles are sparse, compared with the numerous connections between representations of proximal muscles (Pandya, Karol & Heilbronn, 1971; Pandya & Vignolo, 1971; Gould et al. 1986; Rouiller et al. 1994). It is not known whether the relatively few interhemispheric connections between distal limb motor representations transmit predominantly facilitatory or inhibitory commands (Matsunami & Hamada, 1984; Jones, 1993). In humans, behavioural experiments in patients with callosotomy indicated that the corpus callosum plays an important role in sensory and high-level cognitive integration (Seymour, Reuter-Lorenz & Gazzaniga, 1994; Lassonde, Sauerwein & Lepore, 1995), but that, with respect to motor performance in particular, there is little evidence for the transfer of explicit motor commands (Geffen, Jones & Geffen, 1994; Sauerwein & Lassonde, 1994). It is also important to realize that mirror movements are not the predominant feature of the ‘split-brain’ syndrome in patients with lesions of the corpus callosum (Seymour et al. 1994; Sauerwein & Lassonde, 1997), rendering it unlikely that inhibitory interactions between the output of the two primary motor cortices are solely relayed through the corpus callosum. On the other hand, there is anatomical and physiological evidence for ipsilateral projections to the spinal cord and for crossed spinal interneuron circuits, which could also mediate descending modulatory effects from the M1 upon ipsilateral limb muscles (Shahani & Young, 1971; Roby-Brami & Bussel, 1987; Delwaide & Pepin, 1991; Burke et al. 1992; Gracies et al. 1994; Mazevet et al. 1996; Nathan et al. 1996).
If the inhibitory interaction between the right and left M1 were exclusively transmitted through the corpus callosum, that is, caused by a cortico-cortically mediated decrease in the excitability of the M1 contralateral to the target muscle, then MEPs that are elicited by stimulation at subcortical levels (pdTES) should be unaffected. The present findings demonstrate that under various experimental conditions MEPs which are not evoked at the cortical level can also be modulated by conditioning stimuli applied to the ipsilateral M1. A first hint was that MEPs induced by cTES were inhibited (series 1). It has been suggested that anodal cTES can produce MEPs by stimulation of pyramidal cell axons directly (D-waves) rather than by trans-synaptical intracortical stimulation indirectly (I-waves) (Day, Thompson, Dick, Nakashima & Marsden, 1987; Rothwell et al. 1994). The stimulus intensities used for cTES in series 1 (on average >400 V) were 20-30 % above threshold and, therefore, likely to stimulate the pyramidal cell axons distal of the axon hillock evoking so-called D2-waves and possibly some D3-waves (Rothwell et al. 1994). However, at these relatively high stimulus intensities some low-threshold D1-waves and I-waves may have contributed to the EMG responses as well. The latter can be affected by cortico-cortical transcallosal inhibition mechanisms. Therefore, the cTES results alone were not conclusive, and it was necessary to study the effects of conditioning stimulation on MEPs that were elicited unequivocally below the cortical level, that is, at the level of the brainstem by pdTES.
Ipsilateral inhibition of muscle responses elicited at the level of the brainstem
The MEPs that are least likely to be subject to changes in cortical excitability and thus least likely to be sensitive to transcallosal cortico-cortical inhibition are those evoked by electrical stimulation of the pyramidal tract at the level of the brainstem (pyramidal decussation; cf. D3-waves) (Ugawa et al. 1991; Rothwell et al. 1994). In experiment 2 of series 1, pdTES-evoked MEPs were studied in relaxed FDI muscles. These responses were significantly inhibited by conditioning stimulation, similar to the cTMS-evoked responses (experiment 1 of series 1), but the inhibition effect was slightly less pronounced. This added further support to the interpretation that ipsilateral inhibitory modulation of motor responses can be mediated, or at least significantly supplemented, at the subcortical level.
A potential problem with pdTES in relaxed muscles is that very high stimulus intensities (504 ± 127 V) are necessary to elicit stable MEPs. High intensities of brain stimulation can produce more than one descending volley in the pyramidal tract, and it cannot be excluded that the reticulospinal or vestibulospinal tracts are also engaged in the production of responses thus elicited. One could speculate that inhibition directed to some of the segmental interneurons which mediate motoneuron responses to such stimuli could be in part responsible for the inhibition seen in experiment 2 (series 1). The polyphasic waveforms of some of the MEPs produced in relaxed muscles by high intensity pdTES would be consistent with this concern. To address this problem, we performed the second series of experiments with pdTES on preactivated FDI muscles. This approach allowed us to produce clean biphasic responses at substantially lower stimulation intensities compared with series 1.
Significant inhibition of pdTES-evoked responses also occurred under preactivation conditions, at both ISIs tested (10 and 20 ms). The stimulus intensities for TS and CS were substantially lower than in the experiments on relaxed muscles. The configuration of the pdTES-evoked MEPs in the preactivated FDI was simple and biphasic, not only in the control condition (no CS) but also when preceded by a conditioning shock and partially inhibited. There was no evidence that only certain parts of the MEP were inhibited. Altogether, the inhibition induced in pdTES-evoked responses under preactivation conditions is hard to explain by the action of pathways other than the large diameter corticospinal tract with oligosynaptic inputs to motoneurons. In this series, presentation of cortical magnetic and brainstem electrical test responses was also randomized, so that it was not possible to predict the type of stimulus to come. This rules out non-specific effects related to anticipation of different amounts of discomfort associated with the two types of stimulation. Finally, in a similar fashion to previous results on inhibition of cTMS-evoked MEPs (Ferbert et al. 1992), the conditioning shock intensity was positively correlated with the amount of inhibition of pdTES responses. This supports further the possibility that the conditioning effects on cortically and subcortically evoked MEPs were similar phenomena and that they can be interpreted in a comparative way.
The H-reflex experiment in series 1 reproduced the observation by Ferbert et al. (1992) that postsynaptic inhibition of the α-motoneuron pool cannot explain the results. In addition, the absence of inhibition of the H-reflex in our study may serve as a negative control, indicating that there was no technical peculiarity to our experimental set-up that would have caused a generally less specific inhibition pattern than reported previously.
Taking all observations of the present study together, there is little doubt that ipsilateral inhibition is at least in part mediated at the subcortical level, similar to what has been reported for parts of the silent period (Fuhr, Agostino & Hallett, 1991; Inghilleri, Berardelli, Cruccu & Manfredi, 1993). The tendency for pdTES responses to be somewhat less susceptible to conditioning stimulation (Figs 2 and 3) could be indicative of a combined cortical (transcallosal) and subcortical mediation of ipsilateral inhibition in the normal brain. Physiologically, this interpretation is attractive because it offers a concept of inhibitory interaction that includes considerably higher redundancy than a concept which relies exclusively on the corpus callosum. A subcortical component also offers an explanation for the frequent absence of pathological mirror movements in patients with lesions of the corpus callosum, which is rather difficult to explain if one assumes that ipsilateral inhibition is exclusively transcallosally relayed (see above, Seymour et al. 1994; Sauerwein & Lassonde, 1997).
TMS studies in patients with callosal lesions
Rothwell et al. (1991) studied a 19-year-old woman with mirror movements and agenesis of the corpus callosum. In this patient, ipsilateral inhibition, demonstrated with the same protocol as in our experiment 1, was absent. This and the presence of mirror movements was interpreted as evidence for a transcallosal route of this effect in normal subjects. However, this patient had a symptomatic cervical meningomyelocele as well, so that these findings are, in the light of the present results, equally consistent with transmission of ipsilateral inhibition at the level of the cervical spinal cord. Boroojerdi et al. (1996) used the same protocol in patients with subcortical and cortical strokes. Inhibition was not significantly different in the two patient groups, with MEP amplitude reductions to 20-67 % in the subcortical group, and more variable MEP amplitude reductions to 5-80 % in the group with cortical lesions. The authors interpreted the preserved inhibition in the subcortical stroke group as evidence for a transcallosal route, but could not explain why inhibition was also present in patients with cortical lesions. Our results offer an explanation for these findings, namely, the existence of multiple routes that relay ipsilateral inhibition, with a considerable portion of them being subcortical. These would still be efficient in patients with cortical lesions. Meyer et al. (1995) reported that in patients with radiological abnormalities of the corpus callosum, the ipsilateral silent period (iSP) was absent or altered. The authors concluded that the iSP (as another example of descending ipsilateral inhibition) was mediated via the corpus callosum. However, it is not known whether in patients with chronic malformations of the corpus callosum (7 of 10 had congenital lesions) the abnormalities are restricted to that structure. It cannot be ruled out that other candidate pathways for bilateral motor integration are altered as well. In fact, in the three patients of that study who had acquired lesions of the corpus callosum, the iSP was actually present, although with delayed onset latency. This would support the caveat that the pattern of abnormalities in patients with congenital lesions of the corpus callosum might be more complex and involve other structures as well. It is also noteworthy that the occurrence of an iSP was not dependent on a normal volume of the corpus callosum in that study (Meyer et al. 1995).
We can only speculate on underlying mechanisms and on candidate systems that may be the anatomical substrate of ipsilateral inhibition. Particularly interesting are spinal interneuron circuits, a precise physiological and anatomical identification of which was beyond the scope of the present series of experiments.
Inhibitory modulation of cortically generated motor responses at a subcortical level has been described for the SP (Fuhr et al. 1991; Inghilleri et al. 1993; Ziemann, Netz, Szelenyi & Homberg, 1993). It is therefore tempting to speculate that combining cortical and subcortical circuits to serve inhibitory fine tuning of the M1 outflow might be a more general principle in the motor system. Candidate systems include those that utilize uncrossed monosynaptic corticospinal pathways (8-10 % of the human pyramidal tract fibres; Yakolev & Rakic, 1966) and those that cross twice, once at the level of the pyramidal decussation and once more at the level of the spinal cord (Delwaide & Pepin, 1991) (e.g. related to reciprocal inhibition or polysynaptic interneuron circuits). Ipsilateral inhibition following a conditioning shock to the M1 could occur as a result of after-hyperpolarization (AHP) or recurrent (Renshaw) inhibition of the motoneuron pool. Since conditioning stimulation was not associated with ipsilateral MEPs this explanation seems unlikely. As already demonstrated by Ferbert et al. (1992) and reproduced in the present study, there is no evidence for postsynaptic inhibition of the α-motoneuron pool in this protocol (absence of H-reflex inhibition), at least with the muscles at rest. Thus, inhibition via Renshaw cells, Ia-inhibitory interneurons or Ib-inhibitory interneurons, or presynaptic inhibition of Ia afferents is unlikely under these circumstances. Another possibility is that the ipsilateral inhibitory effects are caused by disfacilitation of an excitatory drive onto the α-motoneuron. In order to explain the inhibition of pdTES-evoked MEPs in relaxed and preactivated muscles, this excitatory drive should be generated in the caudal brainstem or the spinal cord, for example by excitatory spinal interneuron pathways such as flexor reflex neurons (Shahani & Young, 1971; Roby-Brami & Bussel, 1987; Floeter, Gerloff & Hallett, 1996) or C3-C4 propriospinal interneurons (Alstermark & Kummel, 1990; Burke et al. 1992; Gracies et al. 1994; Mazevet et al. 1996). The physiological significance of these interneurons in humans, however, is largely unclear. In preactivated muscles, a conditioning stimulus may also remove some of the normal tonic voluntary facilitation of the ipsilateral motoneuron pool and this may contribute to the ipsilateral inhibition seen in series 2. The precise identification of the subcortical system (or systems) relaying ipsilateral inhibition remains to be investigated.
In conclusion, the present data point to a concept of inhibitory interaction between the two primary motor cortices which is relayed at multiple levels along the neuroaxis, thus perhaps providing a structurally redundant system which may become important in case of lesions.
The authors wish to thank Drs E. Wassermann and U. Ziemann for discussion of the results and comments on the manuscript, and Ms B. J. Hessie for skilful editing. Dr C. Gerloff was supported by the Deutsche Forschungsgemeinschaft (grant Ge 844/1-1).