JOURNALOF NEUROPHYSIOLOGY Vol. 68. No. 3, September 1992. Prlntcd

111 C’.S..d.

GABAergic Inhibition of Neuronal Activity in the Primate Motor and Premotor Cortex During Voluntary Movement MICHIKAZU Department

MATSUMURA, of Neurophysiology,

SUMMARY

AND

TOSHIYUKI

SAWAGUCHI,

Primate Research Institute,

AND

KISOU

Kyoto University,

KUBOTA Kanrin,

Inuyama,

Aichi 484, Japan

CONCLUSIONS

play a role in the generation of spatiotemporal patterns of electromyographic (EMG) activity of the target muscles required for the intended movement (Georgopoulos et al. 1988; Humphrey et al. 1970; Tanji and Evarts 1976). Recently, we showed at the behavioral level that GABAergic inhibition in the motor cortex is involved in generation of spatiotemporal patterns of muscle activity (Matsumura et al. 199 1). When an antagonist of y-aminobutyric acid (GABA), namely, bicuculline methiodide (BMI), was injected locally into the motor cortex, both agonist and antagonist muscles were cocontracted at the wrist movement, and muscular activity was elevated. When a GABA agonist, namely, muscimol (MUS), was injected, manipulative movement became weaker, and muscles were less active. From these results we suggested that intracortical GABAergic inhibitory mechanisms within the motor cortex may be involved in smooth reciprocal or directional movements. The roles of synaptic inhibition mediated by GABAergic neurons have been investigated at the single-neuron level in the cortex of anesthetized animals (Dykes et al. 1984; Sillito 1975 ). In the primary visual cortex of the cat, both orientational and directional specificities are disturbed by iontophoretically applied BMI, which renders the neurons responsive to other orientations and directions ( Sillito 1975 ). In the somatosensory cortex, receptive fields of tactile neurons became wider when BMI was injected locally (Dykes et al. 1984). A similar contribution of GABAergic inhibition might be expected at the single-neuron level in the motor cortex. Immunohistochemical studies of GABA or glutamic acid decarboxylase (GAD; an enzyme required for the synthesis of GABA) revealed abundant GABA-sensitive neurons in the motor cortex; these neurons have numerous terminals at the initial segments of pyramidal neurons in layers II, III, and V in the motor cortex (DeFelipe et al. 1985; Hendry and Jones 1981; Ribak 1985). Motor cortex neurons are already known to be sensitive to iontophoretically applied GABA and BMI in anesthetized animals (Curtis and Crawford 1969; Curtis and Felix 197 1; KrnjeINTRODUCTION vie 1974; Krnjevic and Schwartz 1967). Fine movements of distal muscles of the primate hand The present study was designed to assessthe importance are the results of programmed reciprocal activation of ago- of GABAergic inhibition by an analysis of single-unit activnists and antagonists in various joints ( Sherrington 1906 ) . ity in the forearm-hand area of the precentral motor and Under certain conditions, motor cortex neurons show direc- premotor cortex, and by observations of changes in activity tional specificities, known as unidirectional changes, or a when GABA, its agonist, or its antagonist were applied ionpreferred direction of firing activity before and/or during tophoretically while the monkey was performing a visual the reciprocal movement of the related joints (Georgopoureaction-time task. A preliminary report of this study was 10s et al. 1988; Kubota and Funahashi 1982). Such neuropublished previously as an abstract (Matsumura et al. nal activity or groups of activities in the motor cortex may 1985).

1. The functional role of GABAergic inhibition in neuronal activity in the forearm-hand area of the motor cortex and the postarcuate premotor cortex was studied while monkeys pressed and released a lever in response to a visual cue. y-Aminobutyric acid (GABA), its agonist muscimol (MUS), and its antagonist bicuculline methiodide ( BMI ) , as well as acetylcholine, noradrenaline, and sodium glutamate, were applied iontophoretically to isolated single neurons whose activity was recorded via glass micropipettes that contained carbon fibers. 2. The activity from single neurons recorded in the motor and premotor cortex showed changes during the press or release of the lever by movement of the contralateral wrist. Discharge of most of the movement-related neurons (~90%) was decreased or completely suppressed by iontophoretically applied GABA or MUS. 3. The activity of the movement-related neurons increased after application of BMI. In 70% of neurons tested, the activity during application of BMI was specifically enhanced at or near the phase of their peaks of activity, with or without a noticeable elevation in background activity. 4. About 10% of the neurons that had been unidirectional (i.e., neurons that showed a change in activity at either the lever-press or lever-release phase) became bidirectional (i.e., they showed changes in activity at both phases) when GABA transmission was blocked by the application of BMI. Bidirectional neurons also showed a reduction in the value of the directionality index. 5. One-half of the silent neurons, which had not shown any activity during either the lever-release or the lever-press phase, became active during the movement phases that followed application of BMI. 6. Most of the cortical neurons in layers II-VI in the motor area were found to be subject to GABAergic inhibition during voluntary movement. 7. We conclude that GABAergic inhibition plays a role in regulating the population of task-related neurons, and the levels of the task-related activity. GABAergic inhibition also improves directionality index in the motor cortex neurons to control the activity of target muscles.

692

0022-3077192$2.00 Copyright 0 1992The American Physiological Society

Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (163.015.154.053) on November 17, 2018. Copyright © 1992 the American Physiological Society. All rights reserved.

GABA

INHIBITION

IN

MOTOR

METHODS

Training

of‘ . monkeys

Four macaquemonkeys(3 Macaca fuscata and 1Macaca mulatta, weighing3.8-6.2 kg) weretrained to perform a simplevisual reaction-timetaskby pressinga lever with the left or the right hand (Fig. 1A). The task was the sameasthat in both the preceding study (Matsumura et al. 1991) and in earlierstudiesin our laboratory (Sawaguchi 1987; Sawaguchiet al. 1986b). The task was initiated when the monkey presseda lever, locatedin front of its hand at the thoracic level. After 3 s, a greenwarning light, 30 cm from the monkey’seyes,cameon. After a random period of 1.53.5 s, the warning light turned red (GO). If the monkey released the lever within 600 ms, it was rewardedwith a drop of water delivered via a solenoidvalve. The monkey usually started the next trial by pressingthe lever OS- 1s after delivery of the reward.

Surgical procedures Beforethe recording sessions, the monkeysunderwent aseptic surgeryfor head stabilization and implantation of electrodes.In brief, the animalswere anesthetizedwith sodium pentobarbital (Nembutal, 35 mg/ kg ip; Abott), and standardcraniotomy was performed.Numerous,small,stainlesssteelscrewsand two stainlesssteeltubes(7 mm ID) wereimplantedfor headfixation during recordings.In two monkeysa conventionalstimuluselectrodewas alsoimplanted in the medullary pyramid (at stereotaxiccoordinate AP = 0.0 mm) for identification of pyramidal tract neurons ( PTNs) via double electrical pulses(separatedby an interval of 3.0 ms) of 0.05-msduration, and of up to 1 mA, at 1 Hz (Phillips 1956). The skull over the ,presumedforearm-hand area of the motor cortex (from A = 8.0- 16.0, L = 10.0-l 8.0) wasmarked stereotaxicallyand covered with a thin layer ( - 1 mm thick) of dental cementto protect the areafrom infection and drying. After

A GREEN LIGHT

RED j---

I t

1

JUICE RELEASE

A20

Al7

Al4

Al2

A10

FIG. 1. Task and sites of recordings. A : schematic diagram of a visual reaction-time task. When a green warning signal changed to red (GO signal), the monkey had to release a lever within 600 ms. B: entry points of the electrodes for single neuronal recordings at the cortical surface from 6 hemispheres. C: examples of coronal sections obtained from 1 hemisphere. Representative recording tracks are shown by thick lines. Sections were obtained from the anterior 10.0-20.0 mm of a stereotaxic coordinate. CS, central sulcus: AS, arcuate sulcus.

AND

PREMOTOR

CORTEX

693

a I-wk recovery period, each monkey was retrained for an additional week to perform the task with its headrigidly attachedto a recording chair, which was custom designedto have the same stereotaxiccoordinatesasthe stereotaxicframe usedin acute experiments. On the days on which recordingswere made,the monkey was weakly anesthetizedwith Halothanegas( 1%with 1 l/ min nitrous oxide and 2 l/min oxygen, Takeda) after or without an induction dosageof Sernylan (phencyclidine hydrochloride, Bio-ceutic, St. Joseph,MO; 1 mg/ kg) asa preparationfor microsurgery.Then its headwasattachedto the recordingchair, and the stereotaxicposition wasadjusted.A small hole (3 mm diam) wasdrilled in the marked skull area through the dental cement. The dura wasincised,andthe cortical surfacewasexposedwith the aid of a binocular microscope. After the recordingsfrom one hemisphere,two of the monkeys were retrained to perform the task with the opposite hand for further recordingsfrom the other sideof the motor cortex. At all times the animalswere treated in accordancewith the Guide for Care and Use of Laboratory Animals (National Institutes of Health, 1978and 1985), and the Guide for Careand Useof Laboratory Primates(Primate Res.Inst., Kyoto Univ., 1986).

Recordings and iontophoresis A multibarreledglasselectrode(3-7 barrels), containing a carbon fiber (Armstrong-Jamesand Millar 1979) of 7 pm diam (Torayca-T300, Toray, Tokyo) in the center shaft of the pipette, was insertedinto the cortex by a pulsemotor-driven micromanipulator (MO-90; Narishige,Tokyo). The electrodewasfilled with 2 M NaCl in the centershaft, and solutionsof GABA (Wako, Tokyo), BMI (Sigma, St. Louis, MO), MUS (Sigma), acetylcholinechloride ( ACh; Nakarai, Kyoto), sodium glutamate (GLU; Wako) and noradrenaline (NA, arterenol hydrochloride; Sigma) were placedin the surroundingpipettes.All drugsweredissolvedat 10 mM in saline.The centershaft of the pipette wasledto a high-impedance amplifier through a Teflon-coated Pt-Ir wire. Other barrelswerealsoled to a four-channelcurrent-injection amplifier (DPI-30; Dia Medical, Tokyo). During recordings,weakbacking currents(usually < 10nA with oppositepolarities)wereappliedto the drug-containingbarrelsto prevent the solutionsfrom leaking. The drilled hole wascovered by a plug of 4% agardissolvedin physiologicalsaline.When all necessarypreparationsfor recording had been made, Halothane gaswas turned off to allow the monkey to recover from anesthesia. After an hour or so,the monkey startedto perform the task. Becausethe reactiontime wasnot significantlydifferent from that observedduring training, the aftereffectsof the anesthesiawerejudged to be negligible. Oncethe activity of a singleneuron wasisolatedwhile the monkey wasperforming the task, individual drugswereiontophoretitally applied in routine fashion. All drugsexcept for GLU were appliedby positivecurrent. To reducedirect effectsof the current, the amount of injected current waslimited to ~50 nA. Whenever changesoccurredin neuronalactivity immediatelyafter the injection of a drug, sodiumions were applied with the samepositive current. Only neuronsthat werenot driven by direct current were taken assourcesof data. In casesof applicationof BMI, in particular, the injected current was carefully adjustedto prevent any “traumatic” burst activity, which appearedwith frequenciesbetween0.5 and 2 Hz after the application of excessiveamountsof BMI. Usually 30-50 trials werenecessaryfor the routine testingof all drugs. The neuronalactivity, task status,and data about the injected current wererecordedon a seven-channelFM tape recorder(SR30; TEAC, Tokyo). Action potentialswere led to a time-amplitude window discriminator ( DIS- 1; BAK, Germantwon, MD) to generatea pulsetrain of transistortransistorlogic signallevel that

Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (163.015.154.053) on November 17, 2018. Copyright © 1992 the American Physiological Society. All rights reserved.

694

MATSUMURA,

SAWAGUCHI,

interfaced with a parallel input-output buffer. Only neurons with clear isolation (with amplitudes of more than twice the height of baseline noise) were included in the results. On-line task control, histogram displays, off-line histograms, and raster displays of neuronal activity were processed by personal computers (MZ-80 and MZ-2200; Sharp, Osaka). All processed data were stored on floppy disks for later calculations.

Averaging and statistics Recorded neuronal activity was averaged at onsets of various aspects of the task, such as the lever-press phase, green warning light, red GO light, and the lever-release phase. The present results were based mainly on averaging at the time of releasing and of pressing of the lever. Neurons were judged to be “task related” when the three continuous bins of activity exceeded -+2 SDS of the activity during the baseline period, which was recorded from 1,000 to 500 ms before the onset of the GO signal. The effect of each drug was judged by the statistical significance of differences in the averaged levels of activity before and during the application of a drug. When an increase or a decrease was observed at the peak of activity, the mean levels of the activity were compared in a zone where the amplitude exceeded baseline activity by t2 SDS. Significance was indicated by a probability level of 0.05.

Histology At the recording sites, both an anodal and a cathodal direct current of 10 PA were passed for 10 s through a carbon-fiber barrel to leave a carbon deposit as a mark (Sawaguchi et al. 1986a). After carbon had been deposited, the electrode was withdrawn from the cortex. The opening of the incised dura was so small that it closed by itself soon after the electrode was withdrawn. The hole in the skull was covered by dental cement. Then an antibiotic was given, and the monkey was taken back to its own individual cage where it rested for 1 or 2 days. At the end of the recording sessions, monkeys were deeply anesthetized with Nembutal and perfused with saline and then with 10% Formalin. The surface of the cortex was photographed to aid in histological reconstruction. Cortical blocks were sliced coronally or parasagittally in loo-pm sections. All serial sections were stained by the standard Nissl method. Sites of penetration and carbon deposits were confirmed histologically. The discrepancy between the depth of the site of the lesion recorded from the manipulator and actual depth revealed by histology was < 10% (Sawaguchi et al. 1989). Cortical layers and areas were determined in accordance with the descriptions of von Bonin and Bailey ( 1947 ). We also referred to the criteria adopted by Barbas and Pandya ( 1987). We classified the motor cortex as their area 4 (FC), and the premotor cortex as their areas 4C, 6DC, and 6Va (FB). However, no physiological attempts were made to delineate the motor from the premotor cortex.

KUBOTA

circles on the surface of the cortex in Fig. 1B. Neurons recorded on the cytoarchitectural borderline between FC and FB were classified as motor cortex neurons because they showed response-related behaviors similar to those of motor cortex neurons when BMI was injected. Examples of serial sections obtained from one hemisphere are shown, with penetration tracks indicated by thick lines, in Fig. 1C. The lengths of the lines correspond to the maximum depths of the recordings, which were determined from the locations of the carbon deposits and the readings on the micromanipulator. Among the recorded neurons, 156 neurons (recorded through layers II-VI) showed changes in activity at the lever-release or lever-press phases of the reaction-time task. Of these 156 neurons, a total of 119, including 19 identified as PTNs, were recorded from the motor cortex. About 60% of the motor and premotor cortex neurons showed changes in activity either at the lever-release phase (n = 59) or at the lever-press phase (n = 34); these neurons were designated unidirectional neurons (Kubota and Funahashi 1982). Neurons that showed changes in activity at both the leverrelease and the lever-press (n = 63) phase were designated bidirectional neurons (Kubota and Funahashi 1982; Schieber and Thach 1985). Table 1 summarizes the distribution of the recorded neurons of the motor and premotor cortex. The ratio of bidirectional to unidirectional neurons was slightly higher in the premotor cortex than in the motor cortex. In addition to the 156 neurons, 20 neurons that showed no firing activity during performance of the task became active after the injection of BMI. They were recorded along the same electrode tracks in the motor cortex as those along which the task-related neurons were recorded. Thus they were referred to as silent, task-unrelated neurons, and their activities were also examined routinely after iontophoretic application of the various drugs. Five other neurons in the motor cortex exhibited spontaneous activity but were not task modulated; their activities were also recorded, and the effects of drugs on them were also examined. Responses of motor cortex neurons to iontophoretically applied GABA, MUS, and BA4I Once a task-related neuronal activity had been isolated, drugs were applied iontophoretically in a routine manner to the recorded neuron through one of the barrels in the electrode. Each drug was applied l-2 min after the effects of the previous drug had completely disappeared. A typical examTABLE

RESULTS

Sites of. recording of’task-related . and ‘Went ” neurons

AND

1.

Directionality of movement-relatedneurons Motor Cortex

neurons

We recorded data from a total of 18 1 neurons: 144 neurons from 23 penetrations in the forearm-hand motor area and 37 neurons from four penetrations in the postarcuate premotor area (Fig. 1). Because we did not apply physiological criteria such as intracortical microstimulation, the classification of the recorded sites was based only on histological observations. Entry points of all penetrations, separately in each hemisphere of four monkeys, are shown as closed

Directionality Unidirectional Lever-press phase Lever-release phase Bidirectional Total

Premotor Cortex 19 (51)

6 (16) 13 (35) 18 (49) 37 (100)

Non-

Task-related 74 28 46 45 119

(62) (23) (39) (38) (100)

task-related

Silent

5

20

Values are numbers of neurons; numbers in parentheses are percentages.

Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (163.015.154.053) on November 17, 2018. Copyright © 1992 the American Physiological Society. All rights reserved.

GABA INHIBITION

IN MOTOR

ple of effects of applications of these drugs on a motor cortex neuron is shown in Fig. 2. This neuron showed a unidirectional change in activity before the lever-release phase and showed peak activity 120 ms before the onset of the lever-release phase (control). After six trials under the control conditions, GABA was applied to this neuron (20 nA). Within 5 s after the application of GABA, the neuron showed decreases in both spontaneous and task-related activity, and by the third post-GABA trial ( 15 s) it was almost silent. After the neuron recovered from the effects of GABA, MUS (30 nA) was applied. Similar suppression was obtained, although its onset was slower and the effect was milder than that induced by GABA. Similar results were obtained in neurons related to the lever-press phase: GABA and MUS decreased both the spontaneous and the peak activity at the lever-press phase. GABA and MUS produced similar depressions of activity in bidirectional neurons during both the lever-press and the lever-release phases. Likewise, they depressed the activity of premotor neurons that were related to movements of the lever. In contrast to the effects of GABA, when BMI (30 nA) was applied, both spontaneous activity and that related to release of the lever were greatly enhanced (Fig. 2). This effect occurred within 3 s and was more persistent than that induced by GABA and MUS, lasting for >60 s after application of BMI was terminated. The effects of the drugs on the motor cortex neurons (y1= 92) and premotor cortex neurons (n = 2 1) are summarized in Table 2. Neurons that were not activated by application of GLU are omitted from the table; silent neurons are also not included. Because motor cortex neurons and premotor cortex neurons showed similar changes in activity, data for both are combined in the table. GABA ( lo-30 nA) decreased activity in all neurons tested (n = 62), whereas MUS ( lo-40 nA) decreased activity in 12 of 13 neurons ( 92% ) tested. BMI, by contrast, increased activity in 103 of 107 neurons (96%). There were no significant differences in sensitivity by drugs between the unidirectional and bidirectional neurons and between the neurons related to the

. . . ... . . ..e..-m...w.. . .. . .. . .. . .

. ..-..--. . . . . . . . . ..m...m . .“. . .. .

.

.-.

........ ..... -

. . ..--w-w. -...-. . . . . Mm. . ..--I .. . . ..---. . m . . -w .e--. . . i .

.... . m .-

..

. ...... e.. .. .

... . ....... . .....

.

... ... ..... . -e-w-m-e--m -. . . . . . . ” ....... ........ .......... . ...... .

........

...s.....--. ..e .. ----e-w .......

. -.

.

. . . . . . . . . . . ... -. . . . .

1, *

.

. . . . .... . .--we .... . . .... . -.m. . . ..s. . . ... ..--,... . ... ... -1.a . ...... . --r --aa . . ... . I--. e ..m -. .. . m.... me...... . . . . ..w e -.--. - .... SM.. .... -w.-...-.. ...... .-.......... ... . . . ..m ..... ........... -.-me .......... . .-we . .

-

*oo

Control

IIAI .I. , I

1

1\

GABA

I \ MUS

i

I

CORTEX

695

TABLE 2. Eficts . ofiontophoretically . applied drugs on neurons in the motor cortex and premotor cortex Effects of Neuronal Activity Drug

Increase

Decrease

No effect

y-Aminobutyric acid Muscimol Bicuculline methiodide Acetylcholine Noradrenaline Sodium glutamate

0 (0) 0 (0)

62 (100) 12 (92)

0 (0) 1 (8)

62 (100) 13 (100)

103 (96) 64 (63) 1 (2) 40 (100)

0 (0) 6 (6) 44 (70) 0 (0)

4 (4) 31 (31) 18 (29) 0 (0)

107 (100) 101 (100) 63 (100) 40 (100)

Total

Values are numbers of neurons; numbers in parentheses are percentages.

lever-release phase and those related to the lever-press phase. There was no significant difference between motor cortex and premotor cortex neurons in terms of the proportion of BMI-sensitive neurons, although the latter needed significantly more current for demonstration of the modulatory effects (5-30 nA in the motor cortex and 20-40 nA in the premotor cortex, P < 0.05, t = 3.04, df = 82). Application of ACh mainly increased the activity, whereas NA mainly decreased it. Changes in neuronal activity induced by iontophoretically applied BMI As shown in Fig. 2 and Table 2, all of the tested neurons showed a decrease in or complete suppression of spike activity after iontophoretic application of GABA with a relatively small current (~30 nA). This effect could be weakened or completely nullified by simultaneous application of BMI, a GABA antagonist. Figure 3 shows a single-trace histogram of neuronal activity during performance of the task. Originally, this neuron showed increases in spike activity during the lever-release phase. When GABA ( 10 nA) was applied, the spike activity was almost completely suppressed. After the neuron had recovered from the suppression of the discharge, a small amount of BMI (30 nA) was applied and then, after a brief interval, GABA was applied again at 10 nA. The combined application of GABA and BMI produced no change from baseline neuronal activity. This result indicates that iontophoretically applied BMI can antagonize the suppression of spike activity that is induced by GABA. This effect was confirmed in all of eight neurons tested in this way. bicucullins

100 sp/s GABA I10 nAl

BMI

2. Typical example of iontophoretic application of drugs to a taskrelated neuron, showing changes in activity before release of the lever (control). Raster display (left) and averaged histogram (right) were aligned at the onset of release of the lever ( vertical lines). Neuronal activity was quickly depressed by the application of y-aminobutyric acid (GABA; 20 nA ) and became completely silent. Muscimol (MUS) had a similar effect. When bicuculline methiodide (BMI) was applied, the increase in activity before release of the lever was enhanced. FIG.

AND PREMOTOR

I30 nA 1

GABAl

20 set

3. Single-trace histogram of neuronal activity during performance of the task. This neuron showed increases in spike discharges during the lever-release phase (A ). When y-aminobutyric acid (GABA) (20 nA) was applied (horizontal bar on the /&), the spike discharges were completely suppressed. After recovery from this suppression of discharge, BMI (bicucullne, 20 nA) was applied simultaneously with GABA (20 nA, horizontal bars on the right). In this case GABA did not cause any noticeable suppression of activity. FIG.

Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (163.015.154.053) on November 17, 2018. Copyright © 1992 the American Physiological Society. All rights reserved.

MATSUMURA,

SAWAGUCHI,

Because BMI was able to antagonize the inhibitory action of iontophoretically applied GABA, we anticipated that BMI, if applied continuously, might expose excitatory activity during voluntary movement that had been hidden by GABAergic inhibition. Examples of changes in neuronal activity after microinjection of BMI are shown in Fig. 4. Neuron A showed a change in activity before the lever-release phase, with peak activity occurring at the onset of the lever-release phase. The activity of this neuron was decreased when ACh (50 nA) was continuously applied during the task, and it was almost completely suppressed when GABA (20 nA) was applied. However, the activity of this neuron was enhanced at the lever-release phase when BMI (20 nA) was applied continuously during the eight trials under these conditions. BMI enhances neuronal activity in a different manner from the way in which either ACh or GLU act. An example of the difference is well represented by y2euroy2B in Fig. 4. This neuron showed peak activity before the lever-release phase, with some background activity (5 spikes/s). ACh not only enhanced neuronal activity at the peak of the histogram before the lever-release phase, but it also elevated baseline activity throughout the trials ( 15 spikes/s). GLU produced similar enhancement (not shown). BMI also enhanced neuronal activity, but only when such activity was at its peak; background activity was not significantly affected by the injection of BMI (7 spikes/s). NA, by contrast, decreased neuronal activity. The application of BMI enhanced both the spontaneous and the movement-related neuronal activity or only one or the other. The ratio of the baseline-to-peak amplitude of the neuronal activity at the lever-movement phases recorded during injection of BMI was compared with that during the control activity. The enhancement ratio at baseline (&) and at the peak activity ( RP) was calculated as follows

A

ACH

B

ACH

I

l-4 GABA

500 I-m 50

/

,3,*

spls Tdb NA

100

sp/s

l---v-

FIG. 4. Examples of changes in neuronal activity after microinjection of bicuculline methiodide (BMI), shown as perievent time histograms of spike activities at the onset of the lever-release phase (vertical lines), averaged and normalized for 5-8 trials under each set of conditions. A : change in neuronal activity at the lever-release phase. Activity was decreased when acetylcholine ( ACh: 50 nA) was applied, and it was completely suppressed by r-aminobutyric acid (GABA; 20 nA). Activity was enhanced at the lever-release phase when BMI (20 nA) was applied. B: ACh enhanced activity not only at the peak of the histogram, but also at the baseline. Noradrenaline (NA) decreased activity. Application of BMI enhanced neuronal activity strongly at the peak of the activity, but not background activity.

AND KUBOTA

2 Control

PC

0c

I

I

6

I

a

10

Enhancement Ratio ( Rp 1 FIG. 5. Frequency distribution of the enhancement ratio (R,) for the activity peak after the application of bicuculline methiodide (BMI). The height of the activity peak, measured from the baseline after application of BMI, was divided by that measured in the control. Pc, peak activity at the control; Pb, peak activity during BMI application; R, = Pb/ Pc, enhancement ratio at the peak activity; Bc, baseline activity at the control; Bb, baseline activity during BMI. Ratio values for the motor cortex neurons ( CI) and the premotor cortex neurons ( q ) are shown separately. R, = BbIBc

R, = Pb/Pc

where Bc is the firing frequency of baseline at control, Bb is the frequency during BMI application, PC is the peak activity frequency at the control, and pb is the peak frequency during BMI application (see the inset in Fig. 5). The value of R, ranged from 1 ( no enhancement, 4 cases) to 10 after a relatively small application of BMI (~30 nA in the motor cortex, and ~40 nA in the premotor cortex). The mean of R, was 3.0 t 1.5 (mean t SD, n = 107), and that of R, was 1.4 $- 0.3 (n = 107). The difference was statistically significant (P < 0.005, t = 2.8 1, df = 2 13). Distributions of R, among the motor cortex neurons and among the premotor cortex neurons are shown separately in Fig. 5. The mean value of R, was 3.4 t 1.7 (n = 86) for motor cortex neurons and 1.7 t 0.8 (n = 2 1) for premotor cortex neurons. Thus the extent of enhancement was significantly greater in the motor cortex than in the premotor cortex (P < 0.005, t = 2.75, df = 106). Changes in activity pattern after application

of BMI

Of the 103 BMI-sensitive neurons tested, 11 neurons showed changes in their patterns of activity after application of BMI. Examples of the changes in patterns for unidirectionally activated neurons are shown in Fig. 6. The activity of neuron A showed a unidirectional change, i.e., an increase at the lever-press phase (see top right histograms), but no increase at the lever-release phase. This activity was completely suppressed by GABA ( 20 nA ) . BMI enhanced activity not only at the lever-press phase, but also at the lever-release phase, although the magnitude of the change was greater for the peak activity at the lever-press phase. A similar change in pattern was also observed in 5 PTNs and 18 unidentified and non-PTNs. The PTN in Fig. 6 B showed a change in unidirectional activity before the leverrelease phase after the injection of BMI (20 nA). In addition, new activity appeared before the lever-press phase

Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (163.015.154.053) on November 17, 2018. Copyright © 1992 the American Physiological Society. All rights reserved.

GABA INHIBITION RELEASE

.

--_.

Control

._

.e.--

.

PRESS

_ _-.

.-.

.

IN MOTOR

---. .

-

II...,

.

. .

GABA

BMI

1 200 sp/s 500

B

ms

(PTN) Control -L

/

FIG. 6. Examples of changes in pattern from unidirectional to bidirectional activity after application of bicuculline methiodide (BMI) in nonpyramidal tract neuron (non-PTN; A) and PTN (B). Dot display (left) was aligned at release of the lever (vertical line). Histograms were averaged at lever-release phase (middle) and at the lever-press phase ( right). A : activity of the neuron showed unidirectional change only at the lever-press phase (see top right histograms). This activity was completely suppressed by y-aminobutyric acid (GABA; 20 nA). BMI not only enhanced the lever of the peak activity at the lever-press phase, but it also caused an increase during the lever-release phase. B: a PTN with antidromic latency of 1.2 ms originally showed an increase in activity at the lever-release phase. After application of BMI, the activity appeared at the lever-press phase as well as at the lever-release phase.

after the injection. Thus the original unidirectional activity became bidirectional with the iontophoretic application of BMI. Bidirectional neurons, which exhibited activity at the press and the release phases of the task, were also influenced by the injection of BMI. To evaluate the change of directionality, we calculated an index for the directional specificity before and after the application of BMI in 30 bidirectional neurons, and after the application of GLU (~1 = 15), ACh (~2 = 9), and NA (~2 = 10) in the motor cortex. The index (D) was calculated as follows D = AWPW

- P(PH/MAwfYr),

P(P)1

where P(r) is the peak of activity above the background at the lever-release phase; P( p) is the peak of activity above the background at the lever-press phase; ABS indicates an absolute value, with the variables in the brackets; and MAX indicates the maximum value of peak activity at either the lever-press or the lever-release phase. When there were no differences in the peak values, this index was equal to 0. If a neuron showed completely unidirectional activity, the value of the index was 1. All values fell between 0 and 1. Figure 7 shows distribution of the values of this index in these neurons before and during application of BMI (top le/l), GLU (top right), ACh ( bottom left), and NA (bottom right), together with regression lines and their statistical values. With the exception of two neurons, the index became smaller after the application of BMI. Smaller values of the index, as seen in most of the cases, indicate lower degrees of directionality. The decrease of the index value was mainly brought by an increase of denominator (maximum of the peak activity). The numerator (absolute value

AND PREMOTOR

CORTEX

697

of the peak difference) rather stayed unchanged. This result suggested that the directionality itself was still preserved at the low level during BMI application, although the directionality index was decreased. The average value of the index for neurons was 0.72 t 0.3 1 (mean t SD) before and 0.43 t 0.20 after the application of BMI. These differences were statistically significant (P < 0.00 1, t test for paired samples, t = 3.19, df = 59). Because the background activity of the motor cortex neurons was usually very low, and because it remained at the low level after the application of BMI, these indexes were unchanged even when P(r) and P(p) were not corrected by subtraction of the baseline activity (0.74 t 0.30 before and 0.45 t 0.22 after BMI). Similar results were obtained from 12 bidirectional neurons in the premotor cortex (0.62 t 0.29 before and 0.41 t 0.22 after injection of BMI, P < 0.01, t = 2.53, df = 23). We also calculated the index value in different way by using the average activity during the lever press and release, instead of the peak activity. The results were similar to the present one. In contrast to BMI, GLU or ACh did not change these indexes; these drugs mainly enhanced the baseline activity. When enhancement of the peak activity was recognized in neurons tested, the enhancement was proportional for both peaks after the injection of GLU or ACh. The injection of NA, by contrast, caused a slight increase in the directionality index. Because NA had a nonspecific inhibitory effect on the neurons to which it was applied, the values of the denominator (the peak activity) became smaller, whereas the numerator (the difference between the 2 peak activities) tended to remain unchanged. A similar enhancement of the signal-to-noise ratio of neuronal activity, as a result of the application of NA, has been described elsewhere (Matsumura et al. 1990). Laminar distribution ofneuronal of GABAergic inhibit&

activity and the efect

The changes in patterns of neuronal activity induced by the application of BMI were observed in neurons in different layers. In most of the recording sessions, neuronal activity was obtained at different depths within the motor cortex along the track of the electrode. An example of such successive recordings of neuronal activity is shown in Fig. 8. The electrode was advanced into the motor cortex at an angle of loo from the vertical plane, as shown in Fig. 8. At the deepest point at which neuronal activity was obtained, a direct current was passed to leave a carbon deposit as a marker; this deposit, 150 pm diam, was found in the deeper part of layer VI. Along the track we recorded activity from six single neurons and estimated their locations within layers with an accuracy of 90% (Sawaguchi et al. 1989). One, located in layer V, was identified as a PTN (neuron 5), by pyramidal tract stimulation, with an antidromic latency of 1.2 ms. Neuron 6 was located at the site of the carbon deposit, at the bottom of layer VI. No neuronal activity was obtained at deeper sites. The averaged histograms for each neuron were aligned at the onset of the lever-release phase and are shown in the center (Control) and on the right (BMI) of Fig. 8. All neurons showed increases in activity after the application of

Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (163.015.154.053) on November 17, 2018. Copyright © 1992 the American Physiological Society. All rights reserved.

MATSUMURA,

698

1.0

SAWAGUCHI,

GLU

BMI

0.8

/

l

/

0

0 /0 l

0.4

/

0 . ./’ e 00 3’

//

.m + 0a

/ 0

0

0.6

s0

AND KUBOTA

/’

/

/

/0

0.2

l 04 0A’..

‘0 0. / u.

l T

y=o.71x+o.13

y=O.63x-0.02

:

0

A

r=o.79

t=6.72

r=0.77

t=4.38

c -i w

0*

1.0

ACh

NA

L

0.8

0.6

0.4

/

/

/

/

/

/

/ /

0.2 0

//

/

/

W”

y=o.97x-0.03 l

r=O.87

/

’

4

l

./ /

t=4.73

/

/

/

/

/

y=1.12x+o.o2 r=0.95

t=8.74

/

I

0.2

0.4

0.6

0.8

1.0

0.2

I

0.4

I

0.6

I

0.8

1

1.0

Pre-Injection FIG. 7. Distributions of directionality indexes ( seetext) in the bidirectional neurons before (x-axis) and after (y-axis) the application of BMI ( topk/i), sodium glutamate (GLU; top right), ACh ( bottomZ@), and NA (bottom right). Straight line (with 45” gradient) indicates index values when no drug effect occurred. Linear regression line (- - -) is shown in each figure, with its equation, correlation coefficient ( r>, and t value (t). For abbreviations, see Fig. 4 legend.

BMI. The histograms for y2euyoy1s4 and 5 showed activity ACh. These neurons were recorded via the same electrode peaks before the lever-release phase and subsequent second with which the activity of task-related neurons was repeaks, which corresponded to the lever-press phase. Neu- corded in the motor cortex. Sixty percent of these neurons rons showed greater increases in activity during the lever- showed changes in activity at one or both of the movement release phase after the application of BMI. The changes in phases after the application of BMI. Figure 9 shows typical activity at the lever-press phase also became apparent in examples of the changes in activity of silent neurons. Neumost neurons. Neuron 6 showed neither task-related activ- ron A, recorded in the lower part of layer III, and neuron B, ity under the control condition nor enhancement of the recorded in layer V, were located along the same electrode activity during any of the movement phases after the applitrack as the neuron whose activity is shown in Fig. 4A. Both cation of BMI, although baseline background activity was neurons A and B lacked spontaneous spike activity during slightly increased. the performance of the task. When GLU (neuron A) or Task-related activity both at the lever-press phase and at ACh (neuron B) was applied, the baseline activity was enthe release phase was, thus, observed in many layers, and hanced throughout the performance of the task. When BMI such activity was enhanced by BMI with or without specific was applied, these neurons showed enhanced activity at the activation at the task-related peaks. Similar enhancement lever-release phase and some changes in background levels. of neuronal activity by BMI was observed in all layers, with This observation clearly indicates that excitatory inputs exception of layer I, along almost all penetration tracks. were arriving at neurons that were silent during the task. Loss of directionality was observed in neurons located in Neuron C, recorded in layer VI, also showed an enhancealmost all layers. ment of baseline activity, but the application of BMI did not reveal any clear change in activity during the moveEfict of BiW on silent neurons ment phases. Although application of BMI did not reveal The effects of BMI were examined in 20 neurons that any task-related enhancement in the activity of this type of were silent during the task but were activated by GLU or silent neuron, the neuron also received tonic inhibition

Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (163.015.154.053) on November 17, 2018. Copyright © 1992 the American Physiological Society. All rights reserved.

GABA INHIBITION

IN MOTOR

BMI

CONTROL

VI

---\

6 \

5

I

6.1

b 100

500

zip/s

I

ms

8. Example of successive neuronal recordings along an electrode track. Lcifi : histological representation showing 6 single neurons recorded along the track; 1 of them was identified as a pyramidal tract neuron ( newon 5 ) . Middle and right: histograms of the activity of the 6 neurons on the /c/i, aligned at the release of the lever (vertical lines). All neurons were activated by bicuculline methiodide ( BMI ), and most became bidirectional after the application of BMI. FIG.

during the performance of the task (ACh and GLU were not tested in this neuron). Silent neurons were found in all layers except layer I (2 in layer II, 5 each in layers III and V, and 8 in layer VI). Among these neurons, there were two subclasses: task-related silent neurons, observed primarily in layers III (n = 5) and V (n = 4); and task-unrelated silent neurons, located mainly in layer VI ( n = 6 of 8). Thus all silent neurons showed enhanced activity after the injection of BMI, whether or not the activity was task related. The silent neurons remained silent when they were subject to GABAergic inhibition during the performance of the task. Miscellaneous

AND PREMOTOR

A

Control

Control

BMI

b

C

C Control

I 50spls

ACh

1

I 500 ms BMI

BMI

FIG. 9. Examples of the changes in activity of silent neurons. Neurons A and L3completely lacked spontaneous spike activity (control). Application of sodium glutamate (neuron A ) or acetylcholine (neuron B) only enhanced the baseline activity during the task. Bicuculline methiodide ( BMI ), however, enhanced activity specifically at the lever-release phase without any significant elevation of background levels. Although neuron C also showed an enhancement of baseline activity, the application of BMI did not reveal anv specific changes in activity during arm movement.

1 500

T

100

ms

spls

D

zY#iirx BMI

BMI

I3

ACH

Enhancement of the changes in activity, loss of directional specificity, and recruitment of task-related neurons were the most commonly observed effects of the application of BMI. In addition to these changes, BMI modulated B

699

neuronal activity in various ways. Neuron A in Fig. 10, for example, showed decreased activity before and after the onset of the lever-release phase. The application of ACh elevated the level of neuronal activity slightly, but neuronal activity around the lever-release phase was still decreased. However, the decrease was almost masked when BMI was applied. This masking effect can be explained simply by the blocking of GABAergic inhibition around the lever-release phase. Similar changes were observed in two other neurons. Neuron B in Fig. 10 also showed a similar, statistically nonsignificant decrease in activity before and after the lever-release phase. This depression could be more clearly observed when ACh was applied. However, the application of BMI clearly caused an increase in activity during the lever-release phase. Such changes in pattern were observed in two neurons. In two neurons the application of BMI revealed excitatory input that arrived at times different from the time of peak activity observed under the control conditions. For example, neuron C in Fig. 10 showed peak activity at the lever-release phase under the control conditions. The application of BMI caused the peak activity to occur 120 ms earlier than the original peak. In three other cases, BMI elevated baseline activity except around the lever-release phase (neuron D in Fig. 10). These neurons had been subject to tonic background activity. We have not, however, investigated systematically

changes in pattern induced by BMI

A

CORTEX

BMI

I

FIG. 10. Other types of change in patterns of activity induced by bicuculline methiodide (BMI) in 4 neurons (A-D). All histograms are aligned at the release of the lever. A: decrease in neuronal activity around the time of release of the lever under the control conditions (top), which continued after the application of acetylcholine ( ACh) (middk). This decrease was almost concealed by BMI (bottom). B: depression of activity could be more easily observed when ACh was applied (middk). This time, BMI did not simply mask the depression but also caused an increase in activity during the lever-release phase. C: this neuron showed peak activity at the lever-release phase under the control conditions (top). The application of BMI moved the location of the peak activity to 120 ms before the original peak. D: this neuron had a weak change in activity before release of the lever under the control conditions (top). Application of BMI specifically elevated the baseline activity except around the time of the release of the lever.

Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (163.015.154.053) on November 17, 2018. Copyright © 1992 the American Physiological Society. All rights reserved.

700

MATSUMURA,

SAWAGUCHI,

whether this decrease in activity was due to the absence of excitatory input or to other inhibitory mechanisms. There were task-unrelated neurons in the motor cortex that had some spontaneous background activity. These neurons showed enhanced background activity after the application of BMI, but they did not show any task-related modulations in activity. An example of this type is y2eu~oy26 in Fig. 8. Although these task-unrelated neurons were not routinely investigated in this study, all five neurons tested showed similar changes. DISCUSSION

General results In the present study we investigated the sensitivity of motor and premotor cortex neurons to GABA, as well as to an agonist (MUS) and an antagonist (BMI) of GABA, receptors, in behaving monkeys. In almost all neurons tested, which were located in cortical layers II-VI, activity was suppressed by the application of GABA or MUS. By contrast, the application of BMI enhanced almost all task-related activity with or without spontaneous activity. These results showed that almost all neurons from which recordings were made in the motor and premotor cortex were under influence of GABAergic transmission, and these inhibitory pathways appear to be involved when the monkey presses or releases a lever. The application of BMI enhanced task-related activity more strongly than the spontaneous activity (peak 3.0 t 1.5 vs. baseline 1.4 t 0.3). This phenomenon was markedly different from the effect of a general excitant such as GLU or ACh, which showed stronger enhancement of baseline activity (3.4 t 1.9 in the case of GLU and 2.7 t 1.5 in the case of ACh) than of the peak activity ( 1.3 t 0.8 in the case of GLU and 1.5 t 1.O in the case of ACh) (see also Matsumura et al. 1990). This specific enhancement of task-related activity by BMI suggests that the intensity or effect of GABAergic inhibition may not be tonically constant for the duration of all aspects of the task. Rather, GABAergic inhibition was stronger during the performance-related phases. The effects of BMI on premotor cortex neurons were essentially similar to, but less intense than those, on the motor cortex neurons. Inhihitorv . interneurons in the motor pathways Although, in the present study, we could not determine whether the GABAergic inhibitory neurons were located within or outside the motor cortex, several types of motor cortex neuron are known to be GABAergic. Histochemical studies of GABA and GAD have shown that aspiny basket cells or Chandelier cells are GABAergic (DeFelipe and Jones 1985; DeFelipe et al. 1985; Hendry and Jones 198 1; Hendry et al. 1987; Peters et al. 1982; Ribak et al. 1979; Ribak 1985) and are located mainly in layers II, III, and VI. Chandelier cells are known to have terminals on the initial segment of pyramidal neurons in layers II, III, and V (DeFelipe et al. 1985; Ribak 1985). Because the initial segment is known to be the most sensitive aspect of the initiation of action potentials (Eccles 1957), the inhibitory input terminal near the initial segment may control generation of firing

AND

KUBOTA

patterns in pyramidal cells more efficiently than a simple numerical summation of depolarization and hyperpolarization caused by somatic or dendritic postsynaptic potentials. It seems likely that these inhibitory mechanisms within the motor cortex could exert powerful gating control of the descending output signals from PTNs. Almost all tested neurons in the motor and premotor cortex were sensitive to GABA or MUS. In fact, the activity of most of these neurons was enhanced when GABA transmission was blocked by BMI. These results imply that the synaptic connections from inhibitory interneurons reach most of the motor and premotor neurons. The input sources of these inhibitory neurons are not yet known. Physiological studies have revealed polysynaptic inhibition from many pathways that involve, for example, thalamocortical (Deschenes et al. 1979) and callosal (Asanuma and Okamoto 1959) neurons. Recurrent inhibition has been observed from PTNs (Endo and Araki 1972; Stefanis and Jasper 1964) and from the intracortical circuit (Asanuma and Rosen 1973; Matsumura and Kubota 1983). Thus many inputs from a variety of pathways may participate in the inhibitory control of the target neuronal activity in the motor and premotor cortex. Functional roles of GABAergic inhibition in task-related neuronal activity The present results indicate that GABAergic inhibition plays a role in the generation of movement-specific patterns of discharge from motor cortex neurons. After the application of BMI, unidirectional activity was enhanced or changed to bidirectional activity, probably by revealing the subthreshold excitatory input. The bidirectional pattern became less marked, as seen from the change in the directionality index. Even if an equal amount of cutoff of the activity was caused by GABA transmission at both the lever-press and release phases, the relative influence might be greater for the smaller activity peak than the larger one. The smaller peak activity would be more quickly suppressed than the larger one. We do not anticipate that intracortical GABAergic inhibition has a special function for the gating of specific activities from nondifferential input sources. Because the directionality was not totally lost after the application of BMI, excitatory input already possesses, to some extent, an asymmetry of the activity. Intracortical GABAergic neural network should rather be regarded as to amplify the small differences in activity that accompany different movements. In contrast, general excitants such as GLU and ACh did not significantly reduce the value of the directionality index, because they simply modulate the background activity. We also showed that inhibitory neuromodulator such as NA application also caused slight increase of the directionality index. Tonic inhibition, caused by an application of NA, would influence more strongly for the smaller activity peak than the larger one. However, the effect of GABAergic inhibition might not be tonic throughout the task phases. The level of GABAergic inhibition was greater at the task-related phases than at the background phase. Thus GABAergic inhibition could include more complex neuronal mechanisms than the tonic inhibitory modulators.

Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (163.015.154.053) on November 17, 2018. Copyright © 1992 the American Physiological Society. All rights reserved.

GABA INHIBITION

IN MOTOR

These results match those of our Drevious study, which * showed that local pressure injection of BMI into the motor cortex induced behavioral deficits, which included increased activity of muscle and cocontraction during performance of a task (Matsumura et al. 199 1). The results also seem consistent with intracellular data that show changes in membrane potentials during voluntary movement (Matsumura 1979). Neurons with unidirectional activity exhibited subthreshold depolarization during flexion or extension of the wrist without action potentials, and they exhibited suprathreshold depolarization during the opposite movement (see Fig. 6 in Matsumura 1979). The application of BMI also revealed task-related activity that was hidden under the normal conditions. Excitatory synaptic inputs well beyond the threshold levels for firing were arriving at the majority of the neurons in the motor cortex. The actual generation of action potentials was under the powerful control of GABAergic inhibition. These silent neurons merely showed an elevated baseline activity after application of general excitant such as GLU or ACh. The existence of task-related silent neurons suggests that GABAergic inhibition might regulate the population of task-related neurons. In view of the variety of patterns of activity among the motor cortex neurons, it seems that they are generated not only by a variety of excitatory inputs but are also subject to a variety of GABAergic inhibitory mechanisms during the voluntary movement. Although many of corticospinal neurons originate from the hand area of premotor cortex as well as from the hand area of motor cortex (Dum and Strick 199 1 ), local pressure injection of BMI into premotor cortex does not cause severe deficits, compared with injections into motor cortex (Matsumura et al. 199 1). This might be partially because the effects of iontophoretic application of BMI on premotor cortex neurons were less intense than those on the motor cortex neurons, as shown in the present study. Inhibitory connectivities are also present in the spinal cord (Cheney et al. 1985; Hultborn et al. 197 1; Kasser and Cheney 1985). Corticospinal neurons are known to have extensive branching within the spinal cord, sending collaterals to different motoneuron pools (Shinoda et al. 198 1). To establish the finally specified activity of motoneurons from these diffuse projections, inhibitory mechanisms might exert an additional and auxiliary role, refocusing the final patterns of activity in the spinal cord. In summary, the descending motor system from the cortex to the target muscles may involve multiple inhibitory organizations, both in the motor cortex and in the spinal cord, which together form hierarchical motor structures. These collaborations of inhibitory systems in the corticomotoneuronal pathways, as well as other descending and ascending pathways, may subserve the performance of smooth movements. We gratefully acknowledge the assistance of T. Miwa and K. SawaguchiWatanabe in the histological analysis. This study was supported by Grants-in-Aid for Special Project Research (nos.61 131006,62121006,and63112005)for 1986-1988,andGrants-inAid on Priority Area (nos. 02255 103 and 0325 1103) for 1990 and 199 1 from the Ministry of Education, Science, and Culture of Japan. Preparation of the manuscript was supported in part by Grant RR-00 166 from the Division of Research Resources of the National Institutes of Health to the

AND PREMOTOR

CORTEX

701

Regional Primate Research Center at the University of Washington, where M. Matsumura was a visiting scientist. Present address of M. Matsumura: Dept. of Health and Exercise, College of Liberal Arts and Sciences, Kyoto University, Yoshida-Nihonmatsu-cho, Sakvo-ku, Kvoto 606-O 1, Japan. Address for reprint requests: K. Kubota, Dept. of Neurophysiology, Primate Research Institute, Kyoto University, Kanrin, Inuyama, Aichi 484, Japan. Received 20 August 199 1; accepted in final form 22 April 1992.

REFERENCES M. AND MILLAR, J. Carbon fibre microelectrodes. J. 1: 279-287, 1979. ASANUMA, H. AND OKAMOTO, K. Unitary study on evoked activity of callosal neurons and its effect on pyramidal tract cell activity on cats. Jpn. J. Physiol. 9: 473-483, 1959. ASANUMA, H. AND ROSEN, I. Spread of mono- and polysynaptic connections within cat motor cortex. Exp. Brain Res. 16: 507-520, 1973. BARBAS, H. AND PANDYA, D. N. Architecture and frontal cortical connections of the premotor cortex (area 6) in the rhesus monkey. J. Comp. Newel. 256: 2 1l-228, 1987. VON BONIN, G. AND BAILEY, P. The Neocortex o/‘Macaca Mulatta. Urbana, IL: Univ. of Illinois Press, 1947. CHENEY, P. D., FETZ, E. E., AND PALMER, S. S. Patterns of facilitation and suppression of antagonist forelimb muscles from motor cortex sites in the awake monkey. J. Ncurophysiol. 53: 805-820, 1985. CURTIS, D. R. AND CRAWFORD, J. M. Central synaptic transmission-microelectrophoretic studies. Annu. Rev. Pharmacol. 9: 209-240, 1969. CURTIS, D. R. AND FELIX, D. The effect of bicuculline upon synaptic inhibition in the cerebral and cerebellar cortices of the cat. Brain Res. 34: 301-321, 1971. DEFELIPE, J., HENDRY, S. H. C., JONES, E. G., AND SCHMECHEL, D. Variability in the terminations of GABAergic chandelier cell axons on initial segments of pyramidal cell axons in the monkey sensory-motor cortex. J. Cbmp. Newel. 23 1: 364-384, 1985. DEFELIPE, J. AND JONES, E. G. Vertical organization of y-amino-butyric acid-accumulating intrinsic neuronal systems in monkey cerebral cortex. ,J. Newosc’i. 5: 3246-3260, 1985. DESCHENES, M., LABELLE, A., AND LANDRY, P. A comparative study of ventrolateral and recurrent excitatory postsynaptic potentials in large pyramidal tract cells in the cat. Brain Rcs. 160: 37-46, 1979. DUM, R. P. AND STRICK, P. L. The origin of corticospinal projections from the premotor areas in the frontal lobe. J. Nerrrosci. 1 1: 667-689, 199 1. DYKES, R. W., LANDRY, P., METHERATE, R., AND HICKS, T. P. Functional role of GABA in cat primary somatosensory cortex: shaping receptive fields of cortical neurons. J. Newophysid. 52: 1066- 1093, 1984. ECCLES, J. C. The Physiology o/‘Nerve C’cj//.y. Baltimore, MD: Johns Hopkins Univ. Press, 1957. ENDO, K. AND ARAKI, T. Two components of the antidromic IPSPs in the pyramidal tract cells. Brain Rev. 39: 5 10-5 14, 1972. GEORGOPOULOS, A. P., KETTNER, R. E., AND SCHWARTZ, A. B. Primate motor cortex and free arm movements to visual targets in three-dimensional space. II. Coding of the direction of movement by a neuronal population. J. Newo.sci. 8: 2928-2937, 1988. HENDRY, S. H. C. AND JONES, E. G. Sizes and distributions of intrinsic neurons incorporating tritiated GABA in monkey sensory-motor cortex. J. Nczrrosci. 1: 390-408, 198 1. HENDRY, S. H. C., SCHWANK, H. D., JONES, E. G., AND YAN, J. Numbers and proportions of GABA-immunoreactive neurons in different areas of monkey cerebral cortex. J. Nerrrosci. 7: 1503- 15 19, 1987. HULTBORN, H., JANKOWSKA, E., AND LINDSTROM, S. Recurrent inhibition from motor axon collaterals of transmission in the Ia inhibitory pathway to motoneurones. J. Physiol. Land. 2 15: 59 l-6 12, 197 1. HUMPHREY, D. R., SCHMIDT, E. M., AND THOMPSON, W. D. Predicting measures of motor performance from multiple spike trains. Science Wash. DC’ 170: 758-762, 1970. KASSER, R. J. AND CHENEY, P. D. Characteristics of corticomotoneuronal postspike facilitation and reciprocal suppression of EMG activity in the monkey. J. Nczlr~)~~h.v.siol. 53: 959-978, 1985. KRNJEVIC, K. Chemical nature of synaptic transmission in vertebrates. Physiol. Rev. 54: 4 18-540, 1974. KRNJEVIC, K. AND SCHWARTZ, S. The action of gamma-aminobutyric acid on cortical neurones. Exp. Brain Res. 3: 320-336, 1967. ARMSTRONG-JAMES,

Newosci.

Methods

Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (163.015.154.053) on November 17, 2018. Copyright © 1992 the American Physiological Society. All rights reserved.

702

MATSUMURA,

SAWAGUCHI,

K. AND FUNAHASHI, S. Direction-specific activities of dorsolatera1 prefrontal and motor cortex pyramidal tract neurons during visual tracking. J. Neurophysiol. 47: 362-376, 1982. MATSUMURA, M. Intracellular synaptic potentials of primate motor cortex neurons during voluntary movement. Brain Res. 163: 33-48, 1979. MATSUMURA, M. AND KUBOTA, K. Spatial distribution of neurons with monosynaptic inputs to adjacent neurons in the primate motor cortex. Sot. Neurosci. Ahstr. 9: 492, 1983. MATSUMURA, M., SAWAGUCHI, T., AND KUBOTA, K. GABAergic inhibitions in the primate motor cortex and the control of voluntary movement. Sot. Neurosci. Abstr. 1 1: 1274, 1985. MATSUMURA, M., SAWAGUCHI, T., AND KUBOTA, K. Modulation of neuronal activities by iontophoretically applied catecholamines and acetylcholine in the primate motor cortex during a visual reaction time task. New-ok. Res. 8: 138- 145, 1990. MATSUMURA, M., SAWAGUCHI, T., OISHI, T., UEIU, K., AND KUBOTA, K. Behavioral deficits induced by local injection of bicuculline and muscimol into the primate motor cortex. J. Neurophysiol. 65: 1542-l 553, 1991. PETERS, A., PROSKAUER, C. C., AND RIBAK, C. E. Chandelier cells in rat visual cortex. J. C’omp. Neural. 206: 397-416, 1982. PHILLIPS, C. G. Intracellular records from Betz cells in the cat. Q. J. Exp. Physiol. 4 1: 58-69, 1956. RIBAK, C. E., HARRIS, A. B., VAUGHN, J. E., AND ROBERTS, E. Inhibitory, GABAergic nerve terminals decrease at many sites of focal epilepsy. Science Wash. DC 205: 2 1l-2 14, 1979. RIBAK, C. E. Axon terminals of GABAergic Chandelier cells are lost at epileptic foci. Brain Res. 326: 25 l-260, 1985. SAWAGUCHI, T. Properties of neuronal activity related to a visual reaction KUBOTA,

AND KUBOTA

time task in the monkey prefrontal cortex. J. Neurophysiol. 58: 10801099, 1987. SAWAGUCHI, T., MATSUMURA, M., AND KUBOTA, K. Long-lasting marks of extracellularly recorded sites by carbon fiber glass micropipettes in the frontal cortex of chronic monkeys. J. Neurosci. Methods 15: 34 l-348, 1986a. SAWAGUCHI, T., MATSUMURA, M., AND KUBOTA, K. Dopamine modulates neuronal activities related to motor performance in the monkey prefrontal cortex. Brain Res. 37 1: 404-408, 1986b. SAWAGUCHI, T., MATSUMURA, M., AND KUBOTA, K. Depth distribution of neuronal activity related to a visual reaction time task in the monkey prefrontal cortex. J. Neurophysiol. 6 1: 435-446, 1989. SCHIEBER, M. A. AND THACH, W. T. Trained slow tracking. II. Bidirectional discharge patterns of cerebellar nuclear, motor cortex, and spindle afferent neurons. J. Neurophysiol. 54: 1228-1270, 1985. SHERRINGTON, C. S. The Integrative Action of the Nervous System (2nd ed.). New York: Cambridge Univ. Press, 1906, p. 270-307. SHINODA, Y., YOKOTA, J., AND FUTAMI, T. Divergent projection ofindividual corticospinal axons to motoneurons of multiple muscles in the monkey. Neurosci. Lett. 23: 7-12, 198 1. SILLITO, A. M. The contribution of inhibitory mechanisms to the receptive field properties of neurones in the striate cortex of the cat. J. Physiol. Lond. 250: 305-329, 1975. STEFANIS, C. AND JASPER, H. H. Recurrent collateral inhibition in pyramidal tract neurons. J. Neurophysiol. 27: 855-877, 1964. TANJI, J. AND EVARTS, E. V. Anticipatory activity of motor cortex neurons in relation to direction of an intended movement. J. Neurophysiol. 39: 1062-1068, 1976.

Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (163.015.154.053) on November 17, 2018. Copyright © 1992 the American Physiological Society. All rights reserved.