- Email: [email protected]
Anticipation and execution of a simple reading task enhance corticospinal excitability
Clinical Neurophysiology 110 (1999) 424±429
Anticipation and execution of a simple reading task enhance corticospinal excitability M. Seyal*, B. Mull, N. Bhullar, T. Ahmad, B. Gage Department of Neurology, University of California, Davis, Davis Medical Center, 2315 Stockton Boulevard, Sacramento, CA 95817, USA Accepted 23 September 1998
Abstract Objective: Electromyographic responses (EMG) evoked in the right hand by transcranial magnetic stimulation (TMS) of the left motor cortex are enhanced during continuous reading. This enhancement is the result of increased excitability of the motor cortex. We proposed that anticipation and reading of single words would also enhance corticospinal excitability. We studied the temporal course of corticospinal excitability changes following left and right hemisphere TMS. Methods: Ten normal volunteers were studied. A warning stimulus (S1) was followed by an imperative stimulus (S2) whereupon a word was presented. Subjects responded by reading the word aloud or reading it silently. In other conditions, no word was displayed and the subjects responded to S2 by saying the word `Cat', pursing their lips, or doing nothing. EMG was recorded over the contralateral hand following a TMS pulse over the motor cortex during and after the S1±S2 period. Results: Enhancement of EMG amplitudes was signi®cantly greater following left hemisphere TMS. The enhancement in the S1±S2 period and that following S2 had a time course similar to several event-related brain potentials. Conclusions: There may be a common mechanism underlying both corticospinal excitability and the contingent negative variation, readiness potential and N400. q 1999 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Corticospinal excitability; Electromyographic responses; Reading; Transcranial magnetic stimulation
1. Introduction Cerebral processes involved with anticipation of a task and with processing of speech have electrophysiological concomitants with well-de®ned temporal characteristics. For example, the contingent negative variation (CNV), a slow negative potential, can be recorded over both hemispheres when a warning stimulus (S1) is followed at least 1 s later by an imperative stimulus (S2) to which a vocalization or other cognitive response is required. The CNV preceding speech is related to cerebral dominance (Tecce and Cattanach, 1993). Similarly, the RP precedes the onset of speaking of self-initiated single words by at least 1 s and is lateralized to the left hemisphere (Deecke et al., 1986). The N400 may be recorded about 400 ms after single words or sentences with semantically inappropriate endings are read. * Corresponding author. Tel.: 1 1-916-734-2636; fax: 1 1-916-4522739. E-mail address: [email protected] (M. Seyal)
Transcranial magnetic pulse stimulation (TMS) of motor cortex can be used to evoke electromyographic (EMG) responses in the contralateral hand. EMG activity evoked in the dominant hand is enhanced during a continuous reading task while subjects are reading aloud (Tokimura et al., 1996). There is no such enhancement of EMG responses when subjects make sounds without speaking (Tokimura et al., 1996). Furthermore, the speech-associated EMG enhancement is caused by an increase in cortical, rather than spinal, excitability (Tokimura et al., 1996). Thus, TMS-evoked changes in EMG amplitude may be used as a measure of cortical excitability during speech-related tasks. A relationship between the CNV/RP and excitability of the motor cortex would be suggested if one could demonstrate a temporal and spatial concomitance between the two. The RP is associated with depolarization of pyramidal tract neurons (Hashimoto et al., 1979; Brunia, 1993). The enhanced corticospinal excitability during a continuous reading task reported by Tokimura et al., (1996) may be related to repetitive increases in the excitability of pyrami-
1388-2457/99/$ - see front matter q 1999 Elsevier Science Ireland Ltd. All rights reserved. PII: S13 88-2457(98)0001 9-4
CLINPH 98063
M. Seyal et al. / Clinical Neurophysiology 110 (1999) 424±429
425
Fig. 1. (A) RWA condition with left-hemisphere TMS applied in the S1±S2 period and TMS applied after S2. EMG amplitude changes are depicted as % control amplitude with the control EMG amplitude at each latency set at 100%. The ®lled circles indicate the mean % change at each interval, the vertical bars ^ 1 SE. S1 was presented at 2 1500 ms. The vertical hatched bar depicts the 50 ms duration of S2. The arrow indicates onset of word presentation. Mean EMG amplitude changes that are signi®cantly different from control are annotated with a (*) (P , 0:05). The horizontal hatched bar depicts the mean duration of vocalization. The (#) indicates a result of marginal signi®cance. (B) RWA condition with right-hemisphere TMS applied in the S1±S2 period and TMS applied after S2. The inset box following S2 is a tracing of the change in luminance detected by a photodiode (Burr-Brown OPT202) placed in front of the LCD screen during word presentation. Increasing darkness gives a downward de¯ection. Peak change in luminance of the displayed letters is achieved approximately 47 ms after S2 (50% change in luminance in approximately 27 ms after S2).
dal cells as each word is read. We thus chose, as a ®rst step, to study the change in EMG enhancement during anticipation of single words. One would then expect that as isolated words are anticipated the time course of changes in cortical excitability would re¯ect that of the CNV/RP. We also looked at changes in TMS-induced EMG following S2 to see if the temporal course of any change in excitability was similar to any known electrophysiological events. We studied changes in TMS-evoked EMG, in the period
between S1 (warning light) and S2 (tone) and following S2 under the following 5 conditions. RWA condition: when the required response was to read a word aloud presented on a computer screen. RWS condition: when the required response was to read a word silently presented on a computer screen. CAT condition: when the required response was to say the word `Cat'. PL condition: when the required response was to purse the lips without speaking. NR condition: when no response was required.
426
M. Seyal et al. / Clinical Neurophysiology 110 (1999) 424±429
Table 1 Left hemisphere stimulation Testing % Change in Statistical condition a EMG amplitude signi®cance (mean of differences) Read word aloud 1000B 500B 100B 25B 50A 250A 400A 600A 1500A Read word silently 25B 250A 400A Cat 25B 50A 250A 400A Purse Lips 25B 250A 400A No Response 25B 250A 400A a b
Right hemisphere stimulation % Change in Statistical EMG amplitude signi®cance b (mean of differences)
1 22 1 28 1 15 1 40 1 18 1 57 1 69 1 42 1 33
NS P 0.021 NS P 0.004 NS P 0.004 P , 0.002 P 0.071 NS
1 13 18 1 28 1 45 17 1 33 1 32 1 29 1 22
NS NS NS NS NS P 0.008 NS P 0.005 NS
1 28 1 16 13
NS NS NS
1 15 1 49 14
NS P 0.074 NS
1 25 1 66 1 51 1 41
P 0.032 NS NS NS
16 17 1 43 1 35
NS NS NS NS
1 38 1 41 1 46
NS NS P 0.067
25 1 40 1 29
NS P 0.049 NS
2 18 11 2 13
P 0.028 NS NS
10 1 12 23
NS NS NS
TMS, timing in ms (`B' before S2, `A' after S2) NS, not signi®cant.
We wanted to determine the temporal characteristics of any changes in EMG enhancement, and whether there is hemispheric lateralization of EMG enhancement. 2. Methods Ten healthy human subjects (3 females) including the authors were studied after informed consent was obtained. The age range was 25±49 years. All subjects were righthanded as determined by the Edinburgh Handedness Inventory (Old®eld, 1971). TMS of the motor cortex was produced using a Cadwell MES-10 stimulator and a ®gure of 8 coil. The coil was placed on the scalp and the orientation was adjusted to optimize responses in the contralateral ®rst dorsal interosseous (FDI) muscle. EMG was recorded using a surface electrode placed over the FDI with a reference electrode located over the metacarpophalangeal joint. The output of the MES-
10 was adjusted to approximately 10% above the level required to cause a visible contraction of the relaxed hand. Subjects were instructed to sit with their hands open and relaxed. Baseline EMG was monitored on an oscilloscope at high gain ( £ 10 000) and any trials that occurred during voluntary hand movement were discarded. Vocalization was recorded simultaneously with the EMG using a microphone af®xed to the anterior neck. The time of onset and offset of vocalization were determined for each individual. Experimental trials began with a warning stimulus (S1), which was a 3 ms ¯ash from a light-emitting diode. This was followed 1500 ms later by an imperative stimulus (S2) which was a 50 ms tone at a frequency of 3.5 kHz. Subjects were required to make a response according to the condition being tested (RWA, RWS, CAT, PL). In the RWA and RWS conditions, a 4 letter word was presented on a liquid-crystal display (LCD) immediately following S2 (Fig. 1B, inset depicts LCD luminance change during word presentation). The word was selected randomly by the computer from a list of 16. TMS was administered at some point during the S1±S2 period, or following S2, in order to evoke an EMG response. For the RWA condition, these intervals were 25, 250, 500 and 1000 ms before S2, and 50, 250, 400, 600 and 1500 ms after S2. To shorten testing time, for all other conditions, the intervals were 25 ms before S2, and 250 and 400 ms after S2. The CAT condition was also studied at 50 ms after S2. Each subject was tested at all intervals, with the order chosen randomly. Left and right-hemisphere TMS were administered on separate days of testing. Trials were repeated every 5±10 s in blocks of 12 trials with the same interval and condition. For each block, 3 subsets of 4 recti®ed EMG responses each were averaged and saved on disk. Experimental blocks were alternated with control blocks in which there was no S1 and S2, and the subjects merely had to sit with their hands relaxed while TMS was administered. Twelve trials were similarly averaged for each control block. The baseline to peak EMG amplitudes (compound muscle action potential) were measured. The mean EMG amplitude for each experimental block was compared with the mean EMG amplitude of the two control blocks acquired before and after each experimental set. Statistical analysis was done with the non-parametric Wilcoxon Signed Rank test using a commercially available statistical package (SigmaStat). 3. Results This group of subjects was mostly right-handed (mean laterality quotient 94). The subjects had no dif®culty maintaining complete relaxation of the hand musculature. On rare occasions, inadvertent hand movements occurred which resulted in phasic EMG bursts and these trials were rejected.
M. Seyal et al. / Clinical Neurophysiology 110 (1999) 424±429
427
was maximal at 400 ms. The magnitude of the EMG enhancement diminished over the next 1100 ms. A similar, smaller enhancement occurred following right-hemisphere TMS (Table 1 and Fig. 1B). At 250 ms after S2, enhancement with left-hemisphere TMS was signi®cantly higher than with right-hemisphere TMS (P 0:028). At 400 ms after S2, the interhemispheric difference was marginally signi®cant (P 0:07). No signi®cant inter-hemispheric difference was present at other intervals. In the RWS condition, there was no signi®cant change in EMG amplitude with left- or right-hemisphere TMS either before or following S2 (Table 1). In the CAT condition, a signi®cant increase in EMG amplitude occurred only following left-hemisphere TMS 25 ms before S2 (Table 1). In the PL condition, EMG amplitude enhancement occurred, but only differed signi®cantly from control following right-hemisphere TMS applied 250 ms after S2 (Table 1). The magnitude of EMG enhancement in the RWA condition relative to the PL condition at 250 ms after S2 was signi®cantly higher for left-hemisphere TMS than for right-hemisphere TMS (P 0:04). No signi®cant inter-hemispheric difference was present at other intervals. In the NR condition, left-hemisphere TMS 25 ms before S2 resulted in a decrease in EMG amplitude (P 0:028). There was no signi®cant change in EMG amplitude with left-hemisphere TMS after S2, nor was there any change with right-hemisphere TMS (Table 1). Fig. 2. Grand averages of recti®ed EMG responses from one individual recorded in the RWA condition. Each continuous line tracing on the left is the average of 24 control responses (12 responses were recorded just prior to each RWA set and 12 responses following each RWA set). The tracings on the right are the averaged EMG response in the RWA condition at the latencies shown. At the 250A latency, the dotted tracings are subsets of the grand average superimposed to depict the variability in response amplitude from trial to trial. Calibration 10 ms, 50 mV.
The mean time to onset of vocalization in the RWA condition was 459 ms (SD 82) after S2. Mean offset latency was 825 ms (SD 91) after S2. In the CAT condition, mean onset of vocalization was 340 ms (SD 64) after S2, and mean offset was 664 ms (SD 112). The mean control EMG amplitude was 641 mV (SD 112). In the RWA condition, when left-hemisphere TMS was applied in the S1±S2 period (i.e. before display of a randomly-generated word), there was a ramp-like increase in the amplitude of the EMG relative to control (Table 1; Figs. 1A and 2). This EMG enhancement peaked when TMS occurred 25 ms before word presentation. At this interval the EMG amplitude was signi®cantly higher than control (P 0:004). A similar trend was present following righthemisphere TMS, though it did not reach statistical signi®cance (Table 1 and Fig. 1B). When left-hemisphere TMS was applied following S2 in the RWA condition, there was also a signi®cant increase in EMG amplitude relative to control (Table 1; Figs. 1A and 2). This increase was apparent at 50 ms following S2 and
4. Discussion We found that, when a subject is anticipating the presentation of a word to be read aloud, there is a larger TMSevoked EMG amplitude during the task, compared to TMSevoked EMG amplitude when the subject is at rest. Enhancement is more robust with left-hemisphere TMS but does occur with right hemispheric TMS. It is maximal just prior to presentation of words. It is possible that simple motor activation of the facial musculature associated with speech may have contributed to the signi®cant EMG enhancement occurring after S2 in the RWA condition. In the PL condition there were bilateral but non-signi®cant increases in EMG amplitude. We observed that when the EMG enhancement in the PL condition is subtracted from that in the RWA condition, the residual EMG enhancement is signi®cantly greater for left hemisphere stimulation at 250 ms after S2. This lateralized enhancement may be related to left hemispheric language processing. In the present study, bilateral increases in EMG enhancement in the RWS condition were smaller than in the RWA condition and did not reach statistical signi®cance. Functional magnetic resonance imaging (fMRI) has demonstrated increased activation of the frontal lobes bilaterally during silent word-generation tasks (Cuenod et al., 1995). This activation is most pronounced over the left hemisphere
428
M. Seyal et al. / Clinical Neurophysiology 110 (1999) 424±429
and includes primary motor cortex. The lack of signi®cance in the RWS condition may re¯ect a Type II statistical error given the relatively small enhancement and large variability. There are distinct similarities between the TMS-induced EMG enhancement we observed, and the temporal course of the CNV and RP. The CNV has a ramp-like shape that is maximal at the time of S2 (Coles and Rugg, 1995). The terminal segment of the CNV overlaps with the RP. For verbal tasks, the CNV amplitudes are higher over the left hemisphere, but no signi®cant left-right differences are present when the task involves jaw-opening or vocalization of meaningless sounds (Rebert and Lowe, 1980; Ohsawa et al., 1996). The RP is present bilaterally for about 2 s prior to onset of speech but becomes lateralized to the left hemisphere during the last 100±200 ms of the preparation period (Deecke et al., 1986). The temporal course of enhanced EMG in the RWA condition also exhibits a ramp-like shape that peaks just prior to S2. Furthermore, although the EMG enhancement occurred bilaterally, the change was signi®cant only for left-hemisphere TMS. No signi®cant enhancement occurs when subjects are required say the word `Cat', or purse the lips, tasks which arguably require less attentiveness and motor preparation than the RWA condition. Similarly, the CNV is enhanced when careful S2 discrimination requires greater attentiveness (Nakamura et al., 1979). In this context, it is interesting that in the NR condition there is actually a decrease in EMG amplitude prior to S2. Note that the NR condition was the same as the control condition except that S1 and S2 were present but had to be ignored. This decreased amplitude could represent suppression of cortical circuits involved in making a response to the imperative stimulus. The similarity of the TMS-induced EMG enhancement to the CNV and RP suggests that a common neurophysiological mechanism may underlie both. The RP is recorded from restricted cortical sites including the precentral gyrus, premotor cortex, primary somatosensory cortex and the supplementary motor area. The increasing surface negativity corresponding to the RP re¯ects increasing EPSPs at apical dendrites of pyramidal tract neurons (Brunia, 1993). These cellular changes could readily account for EMG enhancement demonstrated by TMS in the S1±S2 period. A pronounced enhancement of TMS-induced EMG amplitudes around 400 ms following the word stimulus was observed. The temporal course of this EMG enhancement resembles that of the N400. The N400 has a widespread, bilateral scalp distribution with maximum negativity at centro-parietal sites and with no evident focal topographic features (Curran et al., 1993). Large N400 responses are present in response to isolated words (Coles and Rugg, 1995). When a sentence is read, the N400 amplitude is modulated as a function of semantic relatedness between a word and its sentence context (Coles and
Rugg, 1995). If the EMG enhancement, presumably re¯ecting enhanced corticospinal excitability, and the N400 are both a function of the same underlying neurophysiological mechanism, then the level of corticospinal excitability should be sensitive to semantic processing. Why should a reading task result in enhanced responses in a muscle not directly relevant to that task? Tokimura et al. (1996) suggest that this increased excitability in the hand muscles may be an integral part of the program for speech or perhaps a non-speci®c spread of activation destined for the mouth and jaw. Preparation for speech appears to result in an increase in cortical excitability that extends beyond the cortical regions directly involved in speech. Brunia (1982) have demonstrated changes in the ankle jerk during preparation for a voluntary foot movement. This monosynaptic re¯ex was enhanced in the leg involved in the foot movement as well as in the uninvolved leg. The changes in spinal excitability may not be relevant to the results of the present study since during a reading task the H-re¯ex in the forearm is suppressed rather than enhanced (Tokimura et al., 1996). Our ®nding of hemispheric asymmetry in excitability changes, probably related to the lateralization of language processing, provides further support for a cortical origin for the observed EMG enhancement. We have established a correlation between a measure of corticospinal excitability (i.e. TMS-induced EMG change) and event-related potentials. Further research is needed to more ®rmly establish whether these phenomena share a common neurophysiologic basis.
References Brunia CHM, Scheirs JGM, Haagh SAVM. Changes of achilles tendon re¯exes during a ®xed fore period of four seconds. Psychopysiology 1982;19:63±70. Brunia CHM. Waiting in readiness: gating in attention and motor preparation. Psychophysiology 1993;30:327±339. Coles MGH, Rugg MD. Event-related brain potentials: an introduction. In: Rugg MD, Coles MGH, editors. Electrophysiology of the mind, London: Oxford University Press, 1995, pp. 1±26. Cuenod CA, Bookheimer SY, Hertz-Pannier L, Zef®ro TA, Theodore WH, LeBihan D. Functional MRI during word generation, using conventional equipment: a potential tool for language localization in the clinical environment. Neurology 1995;45:1821±1827. Curran T, Tucker DM, Kutas M, Posner M. Topography of the N400: brain electrical activity re¯ecting semantic expectancy. Electroenceph. clin. Neurophysiol 1993;88:188±209. Deecke L, Engel M, Lang W, Kornhuber HH. Bereitschaftspotential preceding speech after holding breath. Exp Brain Res 1986;65:219± 223. Hashimoto S, Gemba H, Sasaki K. Analysis of slow cortical potentials preceding self-paced hand movements in the monkey. Exp Neurol 1979;65:218±229. Nakamura M, Fukui Y, Kadobayashi I, Kato N. A comparison of the CNV in young and old subjects. Its relationship to memory and personality. Electroenceph clin Neurophysiol 1979;46:337±344. Ohsawa K, Yamaguchi T, Murata N, Kanazawa K, Uchida A, Nakajima I. Contingent negative variations associated with vocalization in humans. No to Shinkei. Brain and Nerve 1996;48:357±361.
M. Seyal et al. / Clinical Neurophysiology 110 (1999) 424±429 Old®eld R. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia 1971;9:97±113. Rebert C, Lowe R. Task-related hemispheric asymmetry of contingent negative variation. Prog Brain Res 1980;54:776±781. Tecce J, Cattanach L. Contingent negative variation. In: Niedermeyer E,
429
Lopes Da Silva F, editors. Electroencephalography, Baltimore: Williams and Wilkins, 1993, pp. 887±910. Tokimura H, Tokimura Y, Oliviero A, Asakura T, Rothwell JC. Speech-induced changes in corticospinal excitability. Ann Neurol 1996;40: 628±634.
Anticipation and execution of a simple reading task enhance corticospinal excitability M. Seyal*, B. Mull, N. Bhullar, T. Ahmad, B. Gage Department of Neurology, University of California, Davis, Davis Medical Center, 2315 Stockton Boulevard, Sacramento, CA 95817, USA Accepted 23 September 1998
Abstract Objective: Electromyographic responses (EMG) evoked in the right hand by transcranial magnetic stimulation (TMS) of the left motor cortex are enhanced during continuous reading. This enhancement is the result of increased excitability of the motor cortex. We proposed that anticipation and reading of single words would also enhance corticospinal excitability. We studied the temporal course of corticospinal excitability changes following left and right hemisphere TMS. Methods: Ten normal volunteers were studied. A warning stimulus (S1) was followed by an imperative stimulus (S2) whereupon a word was presented. Subjects responded by reading the word aloud or reading it silently. In other conditions, no word was displayed and the subjects responded to S2 by saying the word `Cat', pursing their lips, or doing nothing. EMG was recorded over the contralateral hand following a TMS pulse over the motor cortex during and after the S1±S2 period. Results: Enhancement of EMG amplitudes was signi®cantly greater following left hemisphere TMS. The enhancement in the S1±S2 period and that following S2 had a time course similar to several event-related brain potentials. Conclusions: There may be a common mechanism underlying both corticospinal excitability and the contingent negative variation, readiness potential and N400. q 1999 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Corticospinal excitability; Electromyographic responses; Reading; Transcranial magnetic stimulation
1. Introduction Cerebral processes involved with anticipation of a task and with processing of speech have electrophysiological concomitants with well-de®ned temporal characteristics. For example, the contingent negative variation (CNV), a slow negative potential, can be recorded over both hemispheres when a warning stimulus (S1) is followed at least 1 s later by an imperative stimulus (S2) to which a vocalization or other cognitive response is required. The CNV preceding speech is related to cerebral dominance (Tecce and Cattanach, 1993). Similarly, the RP precedes the onset of speaking of self-initiated single words by at least 1 s and is lateralized to the left hemisphere (Deecke et al., 1986). The N400 may be recorded about 400 ms after single words or sentences with semantically inappropriate endings are read. * Corresponding author. Tel.: 1 1-916-734-2636; fax: 1 1-916-4522739. E-mail address: [email protected] (M. Seyal)
Transcranial magnetic pulse stimulation (TMS) of motor cortex can be used to evoke electromyographic (EMG) responses in the contralateral hand. EMG activity evoked in the dominant hand is enhanced during a continuous reading task while subjects are reading aloud (Tokimura et al., 1996). There is no such enhancement of EMG responses when subjects make sounds without speaking (Tokimura et al., 1996). Furthermore, the speech-associated EMG enhancement is caused by an increase in cortical, rather than spinal, excitability (Tokimura et al., 1996). Thus, TMS-evoked changes in EMG amplitude may be used as a measure of cortical excitability during speech-related tasks. A relationship between the CNV/RP and excitability of the motor cortex would be suggested if one could demonstrate a temporal and spatial concomitance between the two. The RP is associated with depolarization of pyramidal tract neurons (Hashimoto et al., 1979; Brunia, 1993). The enhanced corticospinal excitability during a continuous reading task reported by Tokimura et al., (1996) may be related to repetitive increases in the excitability of pyrami-
1388-2457/99/$ - see front matter q 1999 Elsevier Science Ireland Ltd. All rights reserved. PII: S13 88-2457(98)0001 9-4
CLINPH 98063
M. Seyal et al. / Clinical Neurophysiology 110 (1999) 424±429
425
Fig. 1. (A) RWA condition with left-hemisphere TMS applied in the S1±S2 period and TMS applied after S2. EMG amplitude changes are depicted as % control amplitude with the control EMG amplitude at each latency set at 100%. The ®lled circles indicate the mean % change at each interval, the vertical bars ^ 1 SE. S1 was presented at 2 1500 ms. The vertical hatched bar depicts the 50 ms duration of S2. The arrow indicates onset of word presentation. Mean EMG amplitude changes that are signi®cantly different from control are annotated with a (*) (P , 0:05). The horizontal hatched bar depicts the mean duration of vocalization. The (#) indicates a result of marginal signi®cance. (B) RWA condition with right-hemisphere TMS applied in the S1±S2 period and TMS applied after S2. The inset box following S2 is a tracing of the change in luminance detected by a photodiode (Burr-Brown OPT202) placed in front of the LCD screen during word presentation. Increasing darkness gives a downward de¯ection. Peak change in luminance of the displayed letters is achieved approximately 47 ms after S2 (50% change in luminance in approximately 27 ms after S2).
dal cells as each word is read. We thus chose, as a ®rst step, to study the change in EMG enhancement during anticipation of single words. One would then expect that as isolated words are anticipated the time course of changes in cortical excitability would re¯ect that of the CNV/RP. We also looked at changes in TMS-induced EMG following S2 to see if the temporal course of any change in excitability was similar to any known electrophysiological events. We studied changes in TMS-evoked EMG, in the period
between S1 (warning light) and S2 (tone) and following S2 under the following 5 conditions. RWA condition: when the required response was to read a word aloud presented on a computer screen. RWS condition: when the required response was to read a word silently presented on a computer screen. CAT condition: when the required response was to say the word `Cat'. PL condition: when the required response was to purse the lips without speaking. NR condition: when no response was required.
426
M. Seyal et al. / Clinical Neurophysiology 110 (1999) 424±429
Table 1 Left hemisphere stimulation Testing % Change in Statistical condition a EMG amplitude signi®cance (mean of differences) Read word aloud 1000B 500B 100B 25B 50A 250A 400A 600A 1500A Read word silently 25B 250A 400A Cat 25B 50A 250A 400A Purse Lips 25B 250A 400A No Response 25B 250A 400A a b
Right hemisphere stimulation % Change in Statistical EMG amplitude signi®cance b (mean of differences)
1 22 1 28 1 15 1 40 1 18 1 57 1 69 1 42 1 33
NS P 0.021 NS P 0.004 NS P 0.004 P , 0.002 P 0.071 NS
1 13 18 1 28 1 45 17 1 33 1 32 1 29 1 22
NS NS NS NS NS P 0.008 NS P 0.005 NS
1 28 1 16 13
NS NS NS
1 15 1 49 14
NS P 0.074 NS
1 25 1 66 1 51 1 41
P 0.032 NS NS NS
16 17 1 43 1 35
NS NS NS NS
1 38 1 41 1 46
NS NS P 0.067
25 1 40 1 29
NS P 0.049 NS
2 18 11 2 13
P 0.028 NS NS
10 1 12 23
NS NS NS
TMS, timing in ms (`B' before S2, `A' after S2) NS, not signi®cant.
We wanted to determine the temporal characteristics of any changes in EMG enhancement, and whether there is hemispheric lateralization of EMG enhancement. 2. Methods Ten healthy human subjects (3 females) including the authors were studied after informed consent was obtained. The age range was 25±49 years. All subjects were righthanded as determined by the Edinburgh Handedness Inventory (Old®eld, 1971). TMS of the motor cortex was produced using a Cadwell MES-10 stimulator and a ®gure of 8 coil. The coil was placed on the scalp and the orientation was adjusted to optimize responses in the contralateral ®rst dorsal interosseous (FDI) muscle. EMG was recorded using a surface electrode placed over the FDI with a reference electrode located over the metacarpophalangeal joint. The output of the MES-
10 was adjusted to approximately 10% above the level required to cause a visible contraction of the relaxed hand. Subjects were instructed to sit with their hands open and relaxed. Baseline EMG was monitored on an oscilloscope at high gain ( £ 10 000) and any trials that occurred during voluntary hand movement were discarded. Vocalization was recorded simultaneously with the EMG using a microphone af®xed to the anterior neck. The time of onset and offset of vocalization were determined for each individual. Experimental trials began with a warning stimulus (S1), which was a 3 ms ¯ash from a light-emitting diode. This was followed 1500 ms later by an imperative stimulus (S2) which was a 50 ms tone at a frequency of 3.5 kHz. Subjects were required to make a response according to the condition being tested (RWA, RWS, CAT, PL). In the RWA and RWS conditions, a 4 letter word was presented on a liquid-crystal display (LCD) immediately following S2 (Fig. 1B, inset depicts LCD luminance change during word presentation). The word was selected randomly by the computer from a list of 16. TMS was administered at some point during the S1±S2 period, or following S2, in order to evoke an EMG response. For the RWA condition, these intervals were 25, 250, 500 and 1000 ms before S2, and 50, 250, 400, 600 and 1500 ms after S2. To shorten testing time, for all other conditions, the intervals were 25 ms before S2, and 250 and 400 ms after S2. The CAT condition was also studied at 50 ms after S2. Each subject was tested at all intervals, with the order chosen randomly. Left and right-hemisphere TMS were administered on separate days of testing. Trials were repeated every 5±10 s in blocks of 12 trials with the same interval and condition. For each block, 3 subsets of 4 recti®ed EMG responses each were averaged and saved on disk. Experimental blocks were alternated with control blocks in which there was no S1 and S2, and the subjects merely had to sit with their hands relaxed while TMS was administered. Twelve trials were similarly averaged for each control block. The baseline to peak EMG amplitudes (compound muscle action potential) were measured. The mean EMG amplitude for each experimental block was compared with the mean EMG amplitude of the two control blocks acquired before and after each experimental set. Statistical analysis was done with the non-parametric Wilcoxon Signed Rank test using a commercially available statistical package (SigmaStat). 3. Results This group of subjects was mostly right-handed (mean laterality quotient 94). The subjects had no dif®culty maintaining complete relaxation of the hand musculature. On rare occasions, inadvertent hand movements occurred which resulted in phasic EMG bursts and these trials were rejected.
M. Seyal et al. / Clinical Neurophysiology 110 (1999) 424±429
427
was maximal at 400 ms. The magnitude of the EMG enhancement diminished over the next 1100 ms. A similar, smaller enhancement occurred following right-hemisphere TMS (Table 1 and Fig. 1B). At 250 ms after S2, enhancement with left-hemisphere TMS was signi®cantly higher than with right-hemisphere TMS (P 0:028). At 400 ms after S2, the interhemispheric difference was marginally signi®cant (P 0:07). No signi®cant inter-hemispheric difference was present at other intervals. In the RWS condition, there was no signi®cant change in EMG amplitude with left- or right-hemisphere TMS either before or following S2 (Table 1). In the CAT condition, a signi®cant increase in EMG amplitude occurred only following left-hemisphere TMS 25 ms before S2 (Table 1). In the PL condition, EMG amplitude enhancement occurred, but only differed signi®cantly from control following right-hemisphere TMS applied 250 ms after S2 (Table 1). The magnitude of EMG enhancement in the RWA condition relative to the PL condition at 250 ms after S2 was signi®cantly higher for left-hemisphere TMS than for right-hemisphere TMS (P 0:04). No signi®cant inter-hemispheric difference was present at other intervals. In the NR condition, left-hemisphere TMS 25 ms before S2 resulted in a decrease in EMG amplitude (P 0:028). There was no signi®cant change in EMG amplitude with left-hemisphere TMS after S2, nor was there any change with right-hemisphere TMS (Table 1). Fig. 2. Grand averages of recti®ed EMG responses from one individual recorded in the RWA condition. Each continuous line tracing on the left is the average of 24 control responses (12 responses were recorded just prior to each RWA set and 12 responses following each RWA set). The tracings on the right are the averaged EMG response in the RWA condition at the latencies shown. At the 250A latency, the dotted tracings are subsets of the grand average superimposed to depict the variability in response amplitude from trial to trial. Calibration 10 ms, 50 mV.
The mean time to onset of vocalization in the RWA condition was 459 ms (SD 82) after S2. Mean offset latency was 825 ms (SD 91) after S2. In the CAT condition, mean onset of vocalization was 340 ms (SD 64) after S2, and mean offset was 664 ms (SD 112). The mean control EMG amplitude was 641 mV (SD 112). In the RWA condition, when left-hemisphere TMS was applied in the S1±S2 period (i.e. before display of a randomly-generated word), there was a ramp-like increase in the amplitude of the EMG relative to control (Table 1; Figs. 1A and 2). This EMG enhancement peaked when TMS occurred 25 ms before word presentation. At this interval the EMG amplitude was signi®cantly higher than control (P 0:004). A similar trend was present following righthemisphere TMS, though it did not reach statistical signi®cance (Table 1 and Fig. 1B). When left-hemisphere TMS was applied following S2 in the RWA condition, there was also a signi®cant increase in EMG amplitude relative to control (Table 1; Figs. 1A and 2). This increase was apparent at 50 ms following S2 and
4. Discussion We found that, when a subject is anticipating the presentation of a word to be read aloud, there is a larger TMSevoked EMG amplitude during the task, compared to TMSevoked EMG amplitude when the subject is at rest. Enhancement is more robust with left-hemisphere TMS but does occur with right hemispheric TMS. It is maximal just prior to presentation of words. It is possible that simple motor activation of the facial musculature associated with speech may have contributed to the signi®cant EMG enhancement occurring after S2 in the RWA condition. In the PL condition there were bilateral but non-signi®cant increases in EMG amplitude. We observed that when the EMG enhancement in the PL condition is subtracted from that in the RWA condition, the residual EMG enhancement is signi®cantly greater for left hemisphere stimulation at 250 ms after S2. This lateralized enhancement may be related to left hemispheric language processing. In the present study, bilateral increases in EMG enhancement in the RWS condition were smaller than in the RWA condition and did not reach statistical signi®cance. Functional magnetic resonance imaging (fMRI) has demonstrated increased activation of the frontal lobes bilaterally during silent word-generation tasks (Cuenod et al., 1995). This activation is most pronounced over the left hemisphere
428
M. Seyal et al. / Clinical Neurophysiology 110 (1999) 424±429
and includes primary motor cortex. The lack of signi®cance in the RWS condition may re¯ect a Type II statistical error given the relatively small enhancement and large variability. There are distinct similarities between the TMS-induced EMG enhancement we observed, and the temporal course of the CNV and RP. The CNV has a ramp-like shape that is maximal at the time of S2 (Coles and Rugg, 1995). The terminal segment of the CNV overlaps with the RP. For verbal tasks, the CNV amplitudes are higher over the left hemisphere, but no signi®cant left-right differences are present when the task involves jaw-opening or vocalization of meaningless sounds (Rebert and Lowe, 1980; Ohsawa et al., 1996). The RP is present bilaterally for about 2 s prior to onset of speech but becomes lateralized to the left hemisphere during the last 100±200 ms of the preparation period (Deecke et al., 1986). The temporal course of enhanced EMG in the RWA condition also exhibits a ramp-like shape that peaks just prior to S2. Furthermore, although the EMG enhancement occurred bilaterally, the change was signi®cant only for left-hemisphere TMS. No signi®cant enhancement occurs when subjects are required say the word `Cat', or purse the lips, tasks which arguably require less attentiveness and motor preparation than the RWA condition. Similarly, the CNV is enhanced when careful S2 discrimination requires greater attentiveness (Nakamura et al., 1979). In this context, it is interesting that in the NR condition there is actually a decrease in EMG amplitude prior to S2. Note that the NR condition was the same as the control condition except that S1 and S2 were present but had to be ignored. This decreased amplitude could represent suppression of cortical circuits involved in making a response to the imperative stimulus. The similarity of the TMS-induced EMG enhancement to the CNV and RP suggests that a common neurophysiological mechanism may underlie both. The RP is recorded from restricted cortical sites including the precentral gyrus, premotor cortex, primary somatosensory cortex and the supplementary motor area. The increasing surface negativity corresponding to the RP re¯ects increasing EPSPs at apical dendrites of pyramidal tract neurons (Brunia, 1993). These cellular changes could readily account for EMG enhancement demonstrated by TMS in the S1±S2 period. A pronounced enhancement of TMS-induced EMG amplitudes around 400 ms following the word stimulus was observed. The temporal course of this EMG enhancement resembles that of the N400. The N400 has a widespread, bilateral scalp distribution with maximum negativity at centro-parietal sites and with no evident focal topographic features (Curran et al., 1993). Large N400 responses are present in response to isolated words (Coles and Rugg, 1995). When a sentence is read, the N400 amplitude is modulated as a function of semantic relatedness between a word and its sentence context (Coles and
Rugg, 1995). If the EMG enhancement, presumably re¯ecting enhanced corticospinal excitability, and the N400 are both a function of the same underlying neurophysiological mechanism, then the level of corticospinal excitability should be sensitive to semantic processing. Why should a reading task result in enhanced responses in a muscle not directly relevant to that task? Tokimura et al. (1996) suggest that this increased excitability in the hand muscles may be an integral part of the program for speech or perhaps a non-speci®c spread of activation destined for the mouth and jaw. Preparation for speech appears to result in an increase in cortical excitability that extends beyond the cortical regions directly involved in speech. Brunia (1982) have demonstrated changes in the ankle jerk during preparation for a voluntary foot movement. This monosynaptic re¯ex was enhanced in the leg involved in the foot movement as well as in the uninvolved leg. The changes in spinal excitability may not be relevant to the results of the present study since during a reading task the H-re¯ex in the forearm is suppressed rather than enhanced (Tokimura et al., 1996). Our ®nding of hemispheric asymmetry in excitability changes, probably related to the lateralization of language processing, provides further support for a cortical origin for the observed EMG enhancement. We have established a correlation between a measure of corticospinal excitability (i.e. TMS-induced EMG change) and event-related potentials. Further research is needed to more ®rmly establish whether these phenomena share a common neurophysiologic basis.
References Brunia CHM, Scheirs JGM, Haagh SAVM. Changes of achilles tendon re¯exes during a ®xed fore period of four seconds. Psychopysiology 1982;19:63±70. Brunia CHM. Waiting in readiness: gating in attention and motor preparation. Psychophysiology 1993;30:327±339. Coles MGH, Rugg MD. Event-related brain potentials: an introduction. In: Rugg MD, Coles MGH, editors. Electrophysiology of the mind, London: Oxford University Press, 1995, pp. 1±26. Cuenod CA, Bookheimer SY, Hertz-Pannier L, Zef®ro TA, Theodore WH, LeBihan D. Functional MRI during word generation, using conventional equipment: a potential tool for language localization in the clinical environment. Neurology 1995;45:1821±1827. Curran T, Tucker DM, Kutas M, Posner M. Topography of the N400: brain electrical activity re¯ecting semantic expectancy. Electroenceph. clin. Neurophysiol 1993;88:188±209. Deecke L, Engel M, Lang W, Kornhuber HH. Bereitschaftspotential preceding speech after holding breath. Exp Brain Res 1986;65:219± 223. Hashimoto S, Gemba H, Sasaki K. Analysis of slow cortical potentials preceding self-paced hand movements in the monkey. Exp Neurol 1979;65:218±229. Nakamura M, Fukui Y, Kadobayashi I, Kato N. A comparison of the CNV in young and old subjects. Its relationship to memory and personality. Electroenceph clin Neurophysiol 1979;46:337±344. Ohsawa K, Yamaguchi T, Murata N, Kanazawa K, Uchida A, Nakajima I. Contingent negative variations associated with vocalization in humans. No to Shinkei. Brain and Nerve 1996;48:357±361.
M. Seyal et al. / Clinical Neurophysiology 110 (1999) 424±429 Old®eld R. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia 1971;9:97±113. Rebert C, Lowe R. Task-related hemispheric asymmetry of contingent negative variation. Prog Brain Res 1980;54:776±781. Tecce J, Cattanach L. Contingent negative variation. In: Niedermeyer E,
429
Lopes Da Silva F, editors. Electroencephalography, Baltimore: Williams and Wilkins, 1993, pp. 887±910. Tokimura H, Tokimura Y, Oliviero A, Asakura T, Rothwell JC. Speech-induced changes in corticospinal excitability. Ann Neurol 1996;40: 628±634.