Interhemispheric facilitation of the hand motor area in humans

Authors


Corresponding author Y. Ugawa: Department of Neurology, Division of Neuroscience, Graduate School of Medicine, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8655, Japan., Email: ugawa-tky@umin.ac.jp

Abstract

  • 1We investigated interhemispheric interactions between the human hand motor areas using transcranial cortical magnetic and electrical stimulation.
  • 2A magnetic test stimulus was applied over the motor cortex contralateral to the recorded muscle (test motor cortex), and an electrical or magnetic conditioning stimulus was applied over the ipsilateral hemisphere (conditioning motor cortex). We investigated the effects of the conditioning stimulus on responses to the test stimulus.
  • 3Two effects were elicited at different interstimulus intervals (ISIs): early facilitation (ISI = 4–5 ms) and late inhibition (ISI ≥ 11 ms).
  • 4The early facilitation was evoked by a magnetic or anodal electrical conditioning stimulus over the motor point in the conditioning hemisphere, which suggests that the conditioning stimulus for early facilitation directly activates corticospinal neurones.
  • 5The ISIs for early facilitation taken together with the time required for activation of corticospinal neurones by I3-waves in the test hemisphere are compatible with the interhemispheric conduction time through the corpus callosum. Early facilitation was observed in responses to I3-waves, but not in responses to D-waves nor to I1-waves. Based on these results, we conclude that early facilitation is mediated through the corpus callosum.
  • 6If the magnetic conditioning stimulus induced posteriorly directed currents, or if an anodal electrical conditioning stimulus was applied over a point 2 cm anterior to the motor point, then we observed late inhibition with no early facilitation.
  • 7Late inhibition was evoked in responses to both I1- and I3-waves, but was not evoked in responses to D-waves. The stronger the conditioning stimulus was, the greater was the amount of inhibition. These results are compatible with surround inhibition at the motor cortex.

The corpus callosum is involved in interhemispheric transfer of cognitive and sensory information. In anatomical studies on animals, callosal projection cells and terminal ramifications of their fibres were identified in the motor cortex (Jenny, 1979; Jones et al. 1979; Castman-Berrevoets et al. 1980), which indicates the presence of transcallosal outputs from and inputs to the motor cortex. The importance of transcallosal information for motor execution was also suggested in humans, for example interhemispheric facilitatory influences on the motor cortex in visuomotor tasks (Lassonde, 1986) and the gradual maturation of interhemispheric inhibition between the motor cortices in the two hemispheres during the development of hemispheric dominance (Thut et al. 1997). The transcallosal connection between the motor cortices has been studied in humans using transcranial stimulation. Transcallosally evoked potentials could be recorded from the motor cortex after electrical (Amassian & Cracco, 1987) or magnetic (Cracco et al. 1989) stimulation of the contralateral frontal cortex. Interhemispheric inhibitory connections between the motor cortices have also been observed (Ferbert et al. 1992; Meyer et al. 1995; Boroojerdi et al. 1996; Meyer et al. 1998). Ferbert et al. (1992) reported that magnetic conditioning stimuli over the motor cortex in one hemisphere reduced the size of responses to magnetic stimuli over the motor cortex in the other hemisphere. They considered that this effect was produced by interhemispheric inputs through the corpus callosum. They also described capricious interhemispheric facilitatory influences between the motor cortices in the two hemispheres. Ferbert et al. (1992) sometimes, but not always, observed the interhemispheric facilitation at shorter interstimulus intervals than those for transcallosal inhibition. In our previous report (Ugawa et al. 1993), weak facilitation was observed consistently when the conditioning and test stimuli were positioned strictly over the motor points in both hemispheres and their intensities were appropriately adjusted. Both the conditioning and test stimuli should have been just above the threshold for active muscles. Probably because this facilitatory effect on I1-waves was very weak, facilitation did not occur on responses larger than 0.2 mV (if the test stimulus was fixed at a slightly higher intensity) and this effect was masked by inhibition when using a higher intensity conditioning stimulus. In another report (Salerno & Georgesco, 1996), a facilitatory effect was evoked by a strong conditioning stimulus. In spite of these previous works, details of the facilitatory connection remain to be investigated.

In the present paper, we tried to clarify the details of interhemispheric facilitation by studying transcallosal facilitatory effects on responses preferentially produced by I1-waves (indirect-waves: descending volleys produced by indirect activation of pyramidal tract neurones via presynaptic neurones), those produced by I3-waves and those elicited by D-waves (direct waves; descending volleys produced by direct activation of pyramidal tract neurones). From differences in the effects on different descending volleys (D-, I1- and I3-waves), we can speculate where facilitation occurs (at the cortex or spinal cord). We used a previously reported method (Sakai et al. 1997) which relies on the fact that the lowest threshold I-wave depends on the direction of currents induced in the brain by magnetic stimulation. Usually, anteriorly directed currents in the brain preferentially elicit I1-waves, whereas posteriorly directed currents elicit I3-waves.

METHODS

Subjects

Eleven healthy volunteers (10 men and 1 woman, 29-45 years old) participated in this study. All the subjects were right handed, scoring more than 50 on the laterality quotient of the Edinburgh Handedness Inventory (Oldfield, 1971). Written informed consent was obtained from all subjects. The experimental procedures used here were approved by the Ethics Committee of the University of Tokyo. No side effects were noted in any of the individuals.

Electromyographic (EMG) recordings

Surface EMGs were recorded from the first dorsal interosseous muscles (FDIs) with 9 mm diameter, Ag-AgCl surface cup electrodes. The active electrode was placed over the muscle belly, and the reference electrode over the metacarpophalangeal joint of the index finger. Responses were amplified with an amplifier (Biotop, NEC Medical Systems, Japan) through filters set at 100 Hz and 3 kHz, then recorded by a computer (Signal Processor DP-1200, NEC Medical Systems) on which a randomized conditional averaging was performed. During the experiments subjects maintained a slight contraction of the right FDI (5-10 % of the maximum voluntary contraction) with the aid of an oscilloscope monitor.

Stimulation

Transcranial electrical stimulation (TES) was performed with a high voltage electrical stimulator (D180A, Digitimer, UK). Stimuli were given through two Ag-AgCl cup electrodes (9 mm in diameter) fixed to the scalp; the cathode was placed at the vertex, the anode over the hand motor area (about 5-6 cm lateral to the vertex). Each of the conditioning and test magnetic stimuli was given with a figure-of-eight-shaped coil (outer diameter of each wing was 7 cm) connected to a Magstim 200 magnetic stimulator (The Magstim Company, UK).

Before the main series of experiments, we determined the current directions in which I1- or I3-waves were preferentially elicited in each subject. A figure-of-eight-shaped coil was placed over the hand motor area and held at eight different orientations, each separated by 45 deg. We chose two current directions: the direction that elicited responses with a latency some 1.5 ms longer (I1-waves) than those evoked by electrical stimulation (D-waves) and the direction required to evoke responses about 4.5 ms later (I3-waves) than the D-wave. In most subjects, anteriorly directed currents induced in the brain preferentially elicited I1-waves, and posteriorly directed induced currents evoked I3-waves. For descriptive purposes only, when presenting the average data obtained from all the subjects as one group, we refer to the induced current that preferentially activates I1-waves as an anteriorly directed induced current and that which activates I3-waves as a posteriorly directed induced current. In this paper, we use the direction of induced currents in the brain, not the direction of coil currents, when describing different directions of stimulation. That is, ‘the medially directed conditioning stimulus’ means the conditioning stimulus eliciting medially directed induced currents in the brain.

In the main set of experiments, we used two magnetic stimulators. Each stimulator was connected to a figure-of-eight-shaped coil. The test stimulus was given over the left-hand motor area (test motor cortex), and the conditioning stimulus over the right-hand motor area (conditioning motor cortex). In most of the experiments, the conditioning coil was oriented to induce medially directed currents in the brain. In a few experiments, posteriorly directed currents in the brain were used for the conditioning stimulus. To investigate the effect of the intensity of the conditioning stimulus, we employed intensities that were 5, 10, 20 and 30 % of the maximum stimulator output above the threshold at the motor point for active left FDI. In the present paper, we express all the stimulus intensities (for conditioning and test stimuli) relative to threshold at the motor point for active target muscles. We defined threshold as the lowest intensity that evoked a small response (about 50 μV) in active muscles in more than half of the trials. The test stimulus was given after the conditioning stimulus at interstimulus intervals (ISIs) of 3-15 ms. The intensity was adjusted to evoke a response with an amplitude of approximately 0.2-0.4 mV peak to peak in the active right FDI. To study the effect of the conditioning stimulus on D-waves, we applied electrical cortical stimuli over the left-hand motor area as a test stimulus. We used a randomized conditioning-test design similar to that reported previously (Hanajima et al. 1996, 1998). In short, various conditions (the test or conditioning stimulus given alone, or the test stimulus preceded by the conditioning stimulus at various ISIs) were intermixed randomly in one block. Several blocks of trials were performed to investigate the complete time course of the effect. Eight to ten responses were collected and averaged for each condition in which both stimuli were given, and 20 responses for a control condition in which the test stimulus was given alone. The amplitude of each single response in each condition was measured in order to compare amplitudes of the control and conditioned responses in the same block with Student's t test corrected for multiple comparisons (Fisher's protected least-significant difference) in each single subject. We calculated the ratio of the mean amplitude of the conditioned response to that of the control response for every ISI in each subject. The graphs plot the mean (±s.e.m.) time course of the effect of the conditioning stimulus averaged over all subjects. Since we considered that there were two effects at different ISIs judging from the time courses (see below), we defined two values representing these two effects. (1) The average size ratio (4–5 ms) for the early facilitation. Taking differences in the latency of responses into consideration, we also calculated the average size ratio (7–8 ms) for I1-waves and the average size ratio (9–10 ms) for D-waves and used these values in the analysis of the early facilitation. (2) The average size ratio (11–15 ms) for the late inhibition. Because the suspected ISIs for the early facilitation were 7–8 ms for I1-waves and 9–10 ms for D-waves, we used an average size ratio (11–15 ms) to compare the pure inhibitory influence (effect at late ISIs) among different conditions. For example, if we use an average size ratio (8–15 ms) facilitatory and inhibitory effects may overlap in the case of I1-waves or D-waves. Under several different combinations of the conditioning and test stimuli, using these two values, we determined whether or not a significant effect was evoked.

In seven subjects, we determined the best position for the conditioning stimulus for the early facilitatory effect (see below). The conditioning stimulus was moved in 2 cm steps around the right-hand motor area, while the coil for the test stimulus was fixed at the left-hand motor area. Medially directed currents in the brain were used for the conditioning stimulus, and posteriorly directed currents for the test stimulus. The intensity of the conditioning stimulus was set at 5 % above the threshold at the motor point for active muscles. The test stimulus was given 4 ms after the conditioning stimulus (ISI = 4 ms). The ratio of the mean amplitude of conditioned responses to that of control responses for each point was calculated. The best position for the conditioning stimulus was defined as the point having a maximal size ratio.

In order to study whether interhemispheric facilitation (see below) is produced by direct activation of output cells or interneurones in the right motor cortex (conditioning motor cortex), we compared the transcallosal effect evoked by an anodal electrical conditioning stimulus with that evoked by a magnetic conditioning stimulus. The intensity of the conditioning electrical stimulus was fixed at 2 % above the threshold for the active left FDI. In addition, we studied the effect of an anodal conditioning stimulus given at a point 2 cm anterior to the motor point in the right hemisphere. In this experiment, the cathode was fixed at a point 2 cm anterior to the mid-central position of the international 10–20 system (Cz).

RESULTS

Effects of magnetic conditioning stimuli

Figure 1 shows one example of the effects of a magnetic conditioning stimulus that induced medially directed currents in the brain. The test stimulus evoked control responses of about 0.3 mV in the right FDI (top traces in Fig. 1A-C). The onset latency of the test response elicited by posteriorly directed currents in the brain was 24.3 ms, corresponding to an I3-wave (top trace of Fig. 1A). The latency was 21.5 ms when the test stimulus induced anteriorly directed currents in the brain, corresponding to an I1-wave (top trace of Fig. 1C). Electrical stimulation evoked responses occurring at 19.7 ms, corresponding to a D-wave (top trace of Fig. 1D). The intensity of the conditioning stimulus was fixed at 5 % above the threshold for active left FDI (+5 %). At an ISI of 4 ms, responses to I3-waves were significantly larger than the control responses (P < 0.02,t test corrected for multiple comparisons). In contrast, neither responses to I1-waves nor those to D-waves were affected by the same conditioning stimulus at ISIs of 4 or 5 ms. The conditioning stimulus had no influence on them even when ISIs were adjusted to compensate for the differences in response latencies (at ISIs of 6, 7 and 8 ms for I1-waves and 8, 9 and 10 ms for D-waves, responses at ISIs of 7 and 9 ms are shown). Similar patterns of effects were observed in all the subjects.

Figure 1.

Comparisons among effects on responses to different kinds of test stimulus applied over the left-hand motor area (test motor cortex)

Conditioning stimuli were applied over the motor point in the right hemisphere (conditioning motor cortex) to induce medially directed currents in the brain. Responses evoked by left motor cortical stimulation were recorded from the right first dorsal interosseous (FDI) muscle with surface cup electrodes. Responses to antero-posteriorly (posteriorly) directed induced currents (test stimulus) (a), those to postero-anteriorly (anteriorly) directed induced currents (a) and those to anodal electrical stimuli (ES; C) are shown. The top trace shows a control response, and responses to the test stimulus preceded by the conditioning stimulus by 4, 5, 7 and 9 ms are shown in the subsequent four rows of traces in A-C. The intensity of a conditioning stimulus was fixed at 5 % above the threshold for active muscles at the motor point. The size of the control responses was about 0.3 mV. The latency of control responses to posteriorly directed induced currents was 24.3 ms, which corresponds to an I3-wave (a), that to anteriorly directed induced currents was 21.5 ms (I1-wave; B) and that to anodal electrical stimulation was 19.7 ms (D-wave; C). Responses to I3-waves were significantly enlarged by the conditioning stimulus at an ISI of 4 ms (a test corrected for multiple comparisons (Fisher's protected least-significant difference), P < 0.02). In contrast, amplitudes of responses to I1-waves or D-waves were unaffected by the conditioning stimulus at ISIs of 4 and 5 ms. Even at ISIs of 7 and 9 ms, compensated for latency differences among control responses (at an ISI of 7 ms for I1-waves and 9 ms for D-waves), the conditioning stimulus had no influence on the size of responses.

Figure 2 shows mean (±s.e.m.) time courses of the effect with different combinations of the conditioning and test stimuli. Average size ratios are summarized in Table 1. Responses to I3-waves (test stimulus: posteriorly directed induced currents in the left motor cortex) were enlarged by medially directed induced currents in the right motor cortex at ISIs of 4 and 5 ms (early facilitation), if the intensity of the conditioning stimulus was +5 or +10 % (Fig. 2A; see Table 1). At larger intervals, responses to I3-waves were significantly suppressed by the conditioning stimulus when its intensity was higher than +5 % (late inhibition; P < 0.05 for +10 %; P < 0.01 for +20 and +30 %; Table 1 and Fig. 2A). Higher intensities of the conditioning stimulus produced greater late inhibition. Neither early facilitation nor late inhibition was evoked by medially directed induced currents if the intensity was less than or equal to the threshold. When the conditioning stimulus was stronger than +5 %, the latency of responses of the left FDI elicited by medially directed currents in the right-hand motor area was the same as that of electrical cortical responses (D-waves).

Figure 2.

Mean (±s.e.m.) time courses of effects on three different control responses

In all combinations of the conditioning and test stimuli, time courses for different intensities of the conditioning stimulus (5, 10, 20 and 30 % above the active threshold, T) are shown. A, effects on responses to posteriorly directed induced currents in the left motor cortex (test motor cortex) (I3-waves). Facilitation was evoked at ISIs of 4 and 5 ms (early facilitation) only when the conditioning stimulus was 5 or 10 % above the threshold (+5 or +10 %). At later intervals, responses were significantly suppressed by the conditioning stimulus, if the conditioning stimulus was +10 % or more (late inhibition). The stronger the conditioning stimulus was, the deeper was the inhibition. B, effects on responses to anteriorly directed currents in the test motor cortex (I1-waves). The size of responses to anteriorly directed currents was not facilitated at early intervals by any conditioning stimuli. In contrast, responses were suppressed at late intervals (11-15 ms) when the conditioning stimulus was +10 % or more. C, effects on responses to anodal electrical stimulation of the left motor cortex (D-waves). Neither facilitation nor inhibition was elicited by any conditioning stimuli.

Table 1. Early facilitation and late inhibition under different combinations of the conditioning and test stimulus
  Average size ratio 4, 5 msAverage size ratio I1-wave, 7–8 ms D-wave, 9–10 msAverage size ratio 11–15 ms
 Intensity (%)Mean s.e.m. P Mean s.e.m. P Mean s.e.m. P
  1. M−P, medially directed conditioning and posteriorly directed test magnetic stimuli. M−A, medially directed conditioning and anteriorly directed test magnetic stimuli. M−E, medially directed conditioning magnetic and anodal electrical test stimuli. P−P, posteriorly directed conditioning and test magnetic stimuli. E(+)−P, anodal electrical conditioning and posteriorly directed magnetic test stimuli. E(ant)−P, anodal electrical conditioning stimulus given at a point 2 cm anterior to the motor point and posteriorly directed magnetic test stimulus. Intensity values are shown as a percentage above the threshold for active muscles. Asterisks indicate a significant difference from the size of control responses: *P < 0.05,**P < 0.01, Student's t test.

M−P+51.240.14*0.870.03
(I3-wave)+101.140.05**0.750.09*
 +201.100.080.500.11**
 +301.100.080.580.09**
M−A+50.980.090.940.061.050.10
(I1-wave)+101.140.110.970.090.830.08
 +200.990.060.111.100.680.08**
 +301.090.070.990.100.690.10**
M−E+50.960.060.930.070.980.05
(D-wave)+100.980.091.010.041.020.07
 +200.900.051.060.080.920.04
 +300.900.050.890.040.900.04
P−P+51.030.120.810.07*
 +101.030.080.520.07**
 +201.020.090.460.08**
 +301.110.150.450.10**
E(+)−P+01.210.08*0.960.07
E(ant)−P+01.020.020.850.04*

Responses to I1-waves never showed any significant early facilitation (Fig. 2B). At ISIs compensated for the latency difference between I3- and I1-waves (at an ISI of 7 or 8 ms), a slight but not significant facilitation was seen in responses to I1-waves. In contrast, late inhibition of responses to I1-waves occurred when the conditioning stimulus was stronger than +10 % (P < 0.01 for +20 and +30 %), even though the amount of inhibition was smaller than that seen in responses to I3-waves (Table 1).

Responses to electrical test stimuli showed neither early facilitation nor late inhibition (Fig. 2C), even if ISIs were adjusted for the latency difference between I3-waves and D-waves (at an ISI of 9 or 10 ms).

Best site of the conditioning stimulus for the early facilitation

Figure 3 shows how the response size ratios at an ISI of 4 ms varied according to the position of a medially directed conditioning stimulus over the right hemisphere. The mean size ratio obtained from seven subjects is shown as a bar at each point. The point (0, 0) was the motor point for the left FDI in each subject (the hand motor area in the right hemisphere (conditioning hemisphere)). The strongest facilitation occurred when the conditioning stimulus was given at the point (0, 0). No significant facilitation was evoked at any other positions.

Figure 3.

Mapping of the conditioning stimulus over the right (conditioning) hemisphere for facilitatory effect at an ISI of 4 ms

At various positions of the conditioning stimulus in the right hemisphere, size ratios at an ISI of 4 ms were obtained from each subject under the combination of medially directed induced currents at an intensity of +5 % for the conditioning stimulus and posteriorly directed induced currents for the test stimulus. The mean of the values from seven subjects was plotted against the position of the conditioning stimulus (shown as the distance from the motor point in cm). Point (0, 0) indicates a FDI motor point in the right hemisphere. At the point (0, 0), significant facilitation was elicited. At all other points, no facilitatory effect was seen.

Different kinds of conditioning stimuli

No early facilitation was evoked by posteriorly directed induced currents in the right motor cortex (conditioning motor cortex) even in responses to I3-waves (Fig. 4A and Table 1). In contrast, the late inhibition was elicited by this conditioning stimulus at all intensities (Table 1). The threshold for the posteriorly directed conditioning stimulus was about 10 % greater than that for the medially directed one. Therefore, the actual value of +10 % for posteriorly directed currents was almost the same as that of +20 % for medially directed currents. The degree of late inhibition evoked by a posteriorly directed conditioning stimulus with an intensity of +10 % (▪ in Fig. 4A) was almost the same as that evoked by a medially directed conditioning stimulus of +20 % (▵ in Fig. 2A; Table 1, 0.50 for +20 % of M-P and 0.52 for +10 % of P-P). The degree of late inhibition saturated for conditioning stimuli of these intensities.

Figure 4.

Effects of different conditioning stimuli on responses to posteriorly directed induced currents (I3-waves) in the left motor cortex

A, mean (±s.e.m.) time courses of effects elicited by posteriorly directed conditioning stimuli over the conditioning motor cortex. Early facilitation (4, 5 ms) was not evoked, whereas late inhibition (11-15 ms) was elicited by stimuli at all intensities. B, mean (±s.e.m.) time courses of effects evoked by anodal electrical stimulation (ES) over the FDI motor point of the conditioning hemisphere (•) and at a point 2 cm anterior to the motor point (▪). Conditioning stimuli over the motor point evoked facilitation at ISIs of 4 and 5 ms, but did not elicit late inhibition. In contrast, conditioning stimuli at a point 2 cm anterior to the motor point did not elicit early facilitation but evoked late inhibition. C, typical responses when the conditioning electrical stimulus was given over the motor point in the conditioning hemisphere. The intensity of the conditioning stimulus was fixed at 2 % above the threshold for an active FDI. The top trace shows a control response, and responses to the test stimulus preceded by the conditioning stimulus by 3, 4 and 5 ms are shown in the subsequent three rows of traces. An anodal electrical conditioning stimulus over the hand motor area evoked latero-medially directed currents in the conditioning motor cortex. The size of control responses was about 0.3 mV. The conditioned response was larger than the control response at an ISI of 4 ms (P < 0.02,t test corrected for multiple comparisons).

Figure 4 B and C shows the effect of an electrical conditioning stimulus. Electrical conditioning stimuli at an intensity of +2 % were applied either over the right-hand motor area or at a point 2 cm anterior to it. Posteriorly directed magnetic test stimuli were given to the left-hand motor area (test motor cortex). Figure 4C shows typical responses when the conditioning stimulus was applied over the motor point. Time courses of the effect when the conditioning stimulus was applied over the motor point (•) or at a point 2 cm anterior to it (▪) are shown in Fig. 4B. The test magnetic stimulus evoked responses of about 0.3 mV in amplitude in the right FDI (Fig. 4C). The conditioned response was larger than the control response at an ISI of 4 ms. In all subjects, there was significant facilitation at ISIs of 4 and 5 ms when the conditioning stimulus was applied over the motor point (Table 1). No late inhibition was elicited. If the conditioning electrical stimulus was applied 2 cm anterior to the motor point (Fig. 4B, ▪; Table 1), there was no early facilitation, although late inhibition was clear. The onset of this late inhibition (12 ms, Fig. 4B) was a few milliseconds later than that observed (9 or 10 ms) when using a medially or posteriorly directed magnetic conditioning stimulus over the motor point (Figs 2A and 4A).

DISCUSSION

Cortical stimulation over the motor cortex of one hemisphere (conditioning motor cortex) had two effects on responses to stimulation over the motor cortex of the other hemisphere (test motor cortex). (1) Responses to the test stimulus could be facilitated at short ISIs (early facilitation). This only occurred if (a) the test stimulus evoked I3-waves and (b) medially directed currents were used for the conditioning stimulus (induced by magnetic or electrical stimulation). Test responses produced by I1-waves or D-waves were never facilitated. No early facilitation was observed if posteriorly directed induced currents were used for the conditioning stimulus, or if anodal electrical stimulation was applied over a point 2 cm anterior to the motor point of the conditioning hemisphere. Facilitation was most prominent if the conditioning stimulus was over the hand motor area. (2) Responses to both I1- and I3-waves were suppressed by a conditioning stimulus at ISIs of 11-15 ms (late inhibition), but those to D-waves were not. This suppression was evoked by magnetic conditioning stimuli with medially or posteriorly directed currents in the conditioning motor cortex or by anodal electrical conditioning stimuli applied over a point 2 cm anterior to the motor point. There was no inhibition if the conditioning stimulus was an anodal electrical shock over the conditioning motor cortex at an intensity of +2 %. The amount of inhibition depended upon the absolute intensity of the magnetic conditioning stimuli irrespective of their current directions.

The lack of facilitation of I1-waves, even at ISIs of 7 or 8 ms, in the present study seems at first sight to be inconsistent with our previous report that facilitation of I1-waves could occur at these ISIs (Ugawa et al. 1993). However, this is probably due to the different size of the control responses in the two experiments. In our previous report, facilitation at these ISIs was only seen when the test responses were smaller than 0.1 mV (0.05 mV). There was no facilitation of responses larger than 0.2 mV (Ugawa et al. 1993). In the present experiments, we studied effects on responses of around 0.3 mV, and hence did not find any facilitation. The intensity of the conditioning stimulus (+5 %) was about the same in the two sets of experiments.

One explanation for these results is that the facilitatory effect on I1-waves is very weak and difficult to observe unless the I1-wave is small. Another possibility is that the early facilitation of I1-waves can be masked by late inhibition. However, responses to I1-waves were not enlarged at ISIs of 7 or 8 ms even when the intensity of the conditioning stimulus (+5 %) was adjusted to evoke no late inhibition. Thus inhibitory masking of facilitation seems unlikely even though we cannot completely exclude that possibility. Another possibility is that I1-waves are saturated when test responses have an amplitude > 0.2 mV. However, previous experiments with single motor unit and surface EMG recordings have shown that this is not the case, and that larger I1-waves can be evoked at higher stimulus intensities (Hanajima et al. 1998). Therefore saturation of I1-waves cannot explain the lack of facilitation. Our conclusion is that early facilitation of I1-waves occurs, but is too weak to have an influence on responses larger than 0.2 mV. In summary, we consider that early facilitation occurs moderately in I3-waves, weakly in I1-waves and does not occur in D-waves. We hereafter discuss the above two effects separately.

Early facilitation

The fact that the best site for the conditioning stimulus for early facilitation was over the hand motor area of the conditioning hemisphere indicates that the facilitation is not due to current spread of the conditioning stimulus to the opposite hemisphere. The fact that test responses evoked by D-waves were not facilitated suggests that the effect occurs within the test motor cortex, which is consistent with a previous report (Salerno & Georgesco, 1996). The ISIs for the facilitation (4-5 ms) are compatible with the transcallosal effect. Cracco et al. (1989) reported that motor cortical stimulation evoked a potential over the contralateral motor cortex with an onset latency of 8.8-12.2 ms. In patients with cortical myoclonus (Shibasaki et al. 1978; Wilkins et al. 1984; Brown et al. 1991), differences in the onset latency between jerks of homologous muscles on the two sides of the body were about 10 ms. A posteriorly directed test stimulus over the test motor cortex is thought to activate corticospinal neurones about 4.5 ms after the test stimulus is given (I3-waves). Therefore, the facilitation at ISIs of 4 or 5 ms means that facilitation occurs in corticospinal neurones of the test motor cortex about 10 ms after the conditioning stimulus (4-5 ms plus 4.5 ms). A slight facilitation of small I1-waves at ISIs of 7-8 ms is also consistent with a transcallosal effect because ISIs of 7 or 8 ms added to the latency difference between I1- and D-waves (1.5 ms) makes about 10 ms. The best position for the conditioning stimulus for facilitation is also compatible with the interhemispheric facilitatory connection between homotopic areas of the motor cortices in the two hemispheres in animals (Asanuma & Okamoto, 1959, 1962; Naitou, 1970). Based on these arguments, we suppose that early facilitation of the test motor cortex is produced by transcallosal inputs evoked by cortical stimulation of the conditioning motor cortex.

Which structure was activated by the conditioning stimulus? Facilitation was only evoked by medially directed currents in the conditioning motor cortex, whether these were induced by magnetic or electrical stimulation. Posteriorly directed induced currents never produced facilitation. Previous work (Werhahn et al. 1994) has shown that medially directed currents preferentially activate the pyramidal tract neurones directly and posteriorly directed currents tend to activate them via interneurones in the motor cortex. This was confirmed in the present study, since the latency of responses in the left FDI was compatible with D-waves if the conditioning stimulus used medially directed currents. Low intensity posteriorly directed currents evoked responses that had a latency compatible with I3-wave activation. Thus the type of stimulation that produced the best early facilitation was the same as that which activated the corticospinal neurones directly. Indeed, the intensity required to produce the two effects was also very similar. It is therefore possible that collateral fibres of corticospinal neurones activated by the conditioning stimulus facilitate the contralateral test motor cortex through the corpus callosum. However, anatomical evidence for the existence of such collaterals is under debate (Künzle, 1975; Castman-Berrevoets et al. 1980; Matsunami & Hamada, 1984; Ezrokhi et al. 1985). Another possibility is that the conditioning stimulus activates small or medium-sized pyramidal neurones in layer III or V projecting to the corpus callosum (Jacobson & Trojanowski, 1974; Wise, 1975; Wise & Jones, 1976; Beck & Kaas, 1994). This seems less likely than the former possibility because smaller pyramidal neurones should have a higher threshold than the corticospinal neurones.

Why was early facilitation absent if posteriorly directed currents were used as the conditioning stimulus? After all, the conditioning stimulus still activates the corticospinal tract, and therefore should activate the postulated collaterals to the opposite hemisphere. The explanation is probably that the latency of any facilitation evoked by posteriorly directed conditioning stimuli will overlap with the late inhibition. Thus, facilitation would begin at I3-wave latency (4-5 ms) plus 4-5 ms (the ISI for early facilitation elicited by a medially directed conditioning stimulus), i.e. 8-10 ms in total, which is the same as the onset latency of late inhibition. In addition, the absolute intensity of a posteriorly directed conditioning stimulus was 10-15 % higher than an anteriorly directed one and would have elicited strong late inhibition, which could easily mask any small facilitation.

It is unlikely that the facilitatory and inhibitory effects on the excitability of the test hemisphere are the result of direct projections to large pyramidal cells in layer V. Such transcallosal connections have not been described in animal experiments (Chang, 1953; Jacobson & Marcus, 1970; Jones et al. 1979). Instead it seems more probable that callosal fibres activated in the present experiments affect cortical interneurones in the test motor cortex. The experiments of Schnitzler et al. (1996) in man are also compatible with this.

Previous studies in humans emphasized the inhibitory interaction between the hand motor areas in the two hemispheres (Ferbert et al. 1992; Meyer et al. 1995, 1998; Boroojerdi et al. 1996; Gerloff et al. 1998). Ferbert et al. (1992) showed inconsistent early facilitation in some subjects, and we later showed that this facilitation was consistently evoked only when small test responses (< 0.1 mV) were conditioned by a weak stimulus (Ugawa et al. 1993). In animals, stimulation of the motor cortex in one hemisphere also evokes a minimal facilitation in the motor cortex of the contralateral hemisphere followed by inhibition at late intervals (Chang, 1953; Asanuma & Okamoto, 1959, 1962; Matsunami & Hamada, 1984). However, even in animal experiments, this facilitation was not always present. When high intensity conditioning stimuli were applied, the facilitation was masked by suppression (Chang, 1953; Asanuma & Okamoto, 1959, 1962). Asanuma & Okuda (1962) showed that the hand motor area had a facilitatory connection with the homotopic area in the contralateral motor cortex, which was surrounded by a larger area of more powerful inhibition (surround inhibition). Because this surround effect is so prominent, interhemispheric facilitation may be difficult to observe in humans using transcranial stimulation. However, in the present paper, we identified the conditions under which it can be consistently studied. There are two methodological reasons for our success. One is that we used test responses that result from I3-wave volleys to the spinal cord. In most previous papers, test responses were elicited by anteriorly directed currents in the brain, which preferentially evoke I1-waves. I1-waves are facilitated much less. The other reason is that we used medially directed induced currents as a conditioning stimulus. In many previous reports, anteriorly directed induced currents were used as a conditioning stimulus. One possible explanation for our result that I3-waves were more enhanced by the early facilitation than I1-waves is that I3-wave generation involves more synapses in the motor cortex than I1-wave generation, and synapses are very susceptible to excitability changes. Another possibility is that different sets of cortical interneurones are responsible for I1- and I3-waves, and those for I3-waves are susceptible to transcallosal inputs.

Late inhibition

Inhibition was provoked by the medially and posteriorly directed conditioning stimuli at ISIs later than 11 ms. These intervals are consistent with the ISIs for the interhemispheric inhibition reported by Ferbert et al. (1992). They proposed that the inhibition occurs at the cortical level because no inhibition was evoked in responses to electrical stimuli. Direct recordings of descending volleys (Di Lazzaro et al. 1999) also confirmed that the inhibition occurs at the cortex. Studies of patients with a lesion in the corpus callosum (Meyer et al. 1995, 1998; Boorojerdi et al. 1996) confirmed that this inhibition is transcallosal. Our results show that early facilitation was not associated with late inhibition when the conditioning stimulus was a magnetic stimulus of +5 % intensity or an electrical stimulus of +2 % intensity. This suggests that the inhibition is not a rebound phenomenon following the early facilitation. We consider that the late inhibition reported here is the same effect previously reported by Ferbert et al. (1992). We did not study this inhibition in detail for this reason.

At the motor point of the conditioning hemisphere, the threshold of the conditioning stimulus for late inhibition was higher than that for early facilitation. The amount of inhibition was positively correlated with the intensity of a conditioning stimulus, irrespective of its current direction. These results suggest that activation of large areas surrounding the motor point of the conditioning motor cortex or many callosal fibres is required for late inhibition. Many callosal fibres are activated by a strong conditioning stimulus and such large callosal inputs presumably evoke powerful surround inhibition in the test motor cortex. The fact that a conditioning electrical stimulus applied over a point 2 cm anterior to the motor point of the conditioning motor cortex evoked late inhibition without the early facilitation also supports the idea that the concept of surround inhibition is relevant to the mechanisms of late inhibition.

When late inhibition was elicited by electrical stimulation over a point 2 cm anterior to the motor point, it occurred at an ISI of 12 ms (Fig. 4B), which was a few milliseconds later than the onset of the late inhibition evoked by the medially or posteriorly directed magnetic conditioning stimulus over the motor point (Figs 2A and 4A). This may be due to the fact that activation of neurones surrounding the motor point at a point 2 cm anterior to it needs a few milliseconds more to affect the corticospinal neurones in either the conditioning or test motor cortex than that just over the motor point. The fact that even the shortest onset of late inhibition was longer than that of facilitation is also compatible with surround inhibition.

We have demonstrated a transcallosal facilitatory connection between the homotopic hand motor areas in the two hemispheres at short intervals in humans. At later intervals this was replaced by inhibition. This pattern of facilitation followed by inhibition is consistent with the effect of callosal stimulation on the motor cortex in animal experiments (Chang, 1953; Asanuma & Okamoto, 1959, 1962).

Acknowledgements

Part of this work was supported by a grant from the Life Science Foundation of Japan and Research Project Grant-in-aid for Scientific Research no. 12680768 from the Ministry of Education, Science, Sports and Culture of Japan. Two of the authors (R.H. and Y.T.) are supported by Research Fellowships of the Japan Society for the Promotion of Science for Young Scientists. We are very grateful to Professor J. C. Rothwell for his comments on this work and English editing. We wish to thank Dr Tantirige Ravi Parashara Ruberu for skillful editing.

Ancillary