INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A loop linking the cerebral cortex and the basal ganglia is important for the control of voluntary movement. In the current model of the basal ganglia, information derived from the cortex is processed in the basal ganglia and returns to the cortex via the thalamus (Alexander and Crutcher 1990a; Alexander et al. 1986). The striatum, composed of the putamen (Put) and the caudate nucleus, is considered as a main input station of the basal ganglia. It receives excitatory afferents from anatomically and functionally diverse areas of the entire cerebral cortex. In the last 10 years, there have been two opposing views of how the functional cortical map is transposed through the cortico-basal ganglia loop (Parent and Hazrati 1995). One is the "information-funneling" hypothesis that emphasizes the convergent nature of corticostriatal projections and subsequent striatopallidal and striatonigral projections (Percheron et al. 1994). Additionally, it has been proposed that cortical areas that are densely interconnected project to overlapping regions of the striatum (Yeterian and Van Hoesen 1978). In contrast, the "parallel-processing" hypothesis proposes that signals originating from functionally distinct cortical areas are processed in separate striatal territories and remain segregated in the striato-pallidal/nigral projection (Alexander and Crutcher 1990a; Alexander et al. 1986; Strick et al. 1995).

In our recent work, the identified forelimb regions of both the primary motor cortex (MI) and the supplementary motor area (SMA) of individual monkeys were each injected with a different anterograde tracer, and we analyzed the detailed organization of corticostriatal motor projections that represent the first step of the cortico-basal ganglia loop (Takada et al. 1998a,b). The forelimb region of the MI projects mainly to the lateral part of the Put, whereas that of the SMA projects predominantly to its medial counterpart. Such segregation of the corticostriatal input zones from the MI and SMA favors the parallel-processing hypothesis. These studies also revealed that the terminal zones from the MI and SMA partly overlap in the Put, especially in its mediolaterally located central zone. However, it remained unknown whether inputs from the MI and SMA converged onto individual neurons in this region of overlap. The primary objective of the present study is to investigate electrophysiologically the organization of the corticostriatal projections from the MI and SMA at the single-neuron level.

It has been reported that Put neurons change their activity in relation to movements of the limbs and other body parts. Neurons in the lateral and medial parts of the Put show activity changes in different aspects of motor behavior (Alexander and Crutcher 1990b; Liles 1983): laterally situated Put neurons are active in relation to simple movements, whereas medially situated ones are active in relation to complex movements. The differential activity changes of laterally versus medially situated Put neurons may be ascribed to cortical inputs of different origins (MI vs. SMA). The second objective of the present study is to examine whether the nature of sensorimotor responses of Put neurons receiving inputs from the MI and those receiving inputs from the SMA are different.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Surgery

Seven Japanese monkeys (Macaca fuscata) of either sex, weighing 5.0-6.0 kg, were used in this study. The use of animals in the present study was approved by the Animal Care and Use Committee of the Tokyo Metropolitan Institute for Neuroscience. Under general anesthesia with pentobarbital sodium (25 mg/kg body wt iv) after induction with ketamine hydrochloride (10 mg/kg im) and xylazine hydrochloride (1-2 mg/kg im), the monkeys received a surgical operation to fix their heads in a stereotaxic frame attached to a monkey chair. Briefly, each monkey was positioned in the stereotaxic apparatus. The skull was widely exposed, and small stainless screws were attached to the skull as anchors. The exposed skull and screws were completely covered with transparent acrylic resin. Two stainless steel pipes were mounted in parallel over the frontal and occipital areas for head fixation, and a small pin was fixed on the skull at A0, L0, H+40 as a stereotaxic reference point. All surgical procedures were performed under aseptic conditions.

Implantation of stimulating electrodes into cerebral cortex

A few days after surgery, stimulating electrodes were chronically implanted into the forelimb regions of the MI and SMA after electrophysiological mapping (Fig. 1). Each monkey was anesthetized with ketamine hydrochloride (10 mg/kg im) and xylazine hydrochloride (1-2 mg/kg im) and seated quietly in a monkey chair with the head fixed in the stereotaxic frame. After removal of a portion of the skull over the midline and the central sulcus, a glass-coated Elgiloy-alloy microelectrode (0.5-1.5 M at 1 kHz) was inserted perpendicular to the cortical surface at 1,000- to 1,500-µm intervals. Extracellular unit activity was recorded, and neuronal responses to somatosensory stimuli (skin touch and passive joint movement) were examined. Following extracellular unit recordings, intracortical microstimulation (ICMS) was performed. Currents of less than 50 µA were delivered with a train of 12 (for MI mapping) or 22 (for SMA mapping) cathodal pulses (200-µs duration at 333 Hz), and the evoked movements of various body parts were observed. According to this electrophysiological mapping, pairs of bipolar stimulating electrodes (made of enamel-coated stainless steel wire with a diameter of 200 µm; intertip distance, 2 mm; inserted length from the dural surface, 4 mm for the MI and 5 mm for the SMA) were implanted into the forelimb regions of the MI and SMA, and covered with transparent acrylic resin.

View larger version (36K): [in this window] [in a new window]   Fig. 1. Cortical mapping (monkey Put2) for implantation of bipolar stimulating electrodes. A: top view of the monkey brain showing the mapped area (). B, 1 and 2: somatotopic mapping of the supplementary motor area (SMA) and the primary motor cortex (MI). Each letter indicates the body part of which movement was evoked by intracortical microstimulation (ICMS) or the receptive field to somatosensory stimuli. D, digit; E, elbow; F, foot; H, hip; S, shoulder; V, visual response; W, wrist. Somatotopy in the mesial surface (B1) and in the rostral bank of the central sulcus (B2) are also shown along with depths from the cortical surface. Three pairs of bipolar stimulating electrodes were implanted into the loci (): one into the distal forelimb region of the MI (MId), one into the proximal forelimb region of the MI (MIp), and another into the forelimb region of the SMA. C, 1 and 2: rostrocaudally arranged frontal sections showing the traces of the stimulating electrodes (). Each number corresponds to the electrode number shown in B, 1 and 2.

Recording of Put neuronal activity

Each monkey was anesthetized with ketamine hydrochloride (10 mg/kg im) and xylazine hydrochloride (1-2 mg/kg im) and seated in a monkey chair with the head restrained. A hole (10-15 mm diam) was made in the skull over the orofacial area of the MI. A rectangular chamber enclosing the hole was fixed for the following experimental sessions. A glass-coated Elgiloy-alloy microelectrode (0.5-1.5 M at 1 kHz) was inserted obliquely (45° from vertical in the frontal plane) through the dura into the Put to record neuronal activity using a hydraulic microdrive (Narishige Scientific Instrument, Tokyo, Japan). The caudal part of the Put where corticostriatal projections terminate densely (Takada et al. 1998a,b) was the main area investigated. Signals from the electrode were amplified (8,000 times) and filtered (200-2 kHz). The responses of Put neurons to cortical electrical stimulation (300-µs duration single pulse, strength of less than 0.5 mA, sometimes up to 0.7 mA, at 0.4-0.8 Hz) were observed and stored on a computer. Several traces were superimposed, and the latency of the responses was determined by the onset of the earliest response. These responses were further analyzed by constructing peri-stimulus time histograms (PSTHs; binwidth, 0.5 ms) using a window discriminator and a computer. PSTHs were usually summed 100 times. Then the responses of Put neurons to somatosensory stimuli (by passive joint movement and muscle palpation) and/or active movements of the forelimb (elicited by offering food in various locations) were examined. Following the extracellular unit recordings, microstimulation through the microelectrode was performed. Currents of less than 50 µA (sometimes up to 60 µA) were delivered with a train of 40 cathodal pulses (200-µs duration at 333 Hz), and the evoked movements of various body parts were carefully observed. Stimulus parameters used in the present study were in the same range as those in previous studies (40 symmetric biphasic paired pulses, each pulse with 300-µs duration at 400 Hz, currents not exceeding 40 µA) (Alexander and DeLong 1985a,b)

Histology

After the electrophysiological mapping, several recording sites were marked by passing cathodal DC current (20 µA for 30 s) through the electrode. At the end of the final experiment, the monkeys were anesthetized deeply with pentobarbital sodium (50 mg/kg iv) and perfused transcardially with 2 l of phosphate-buffered saline, pH 7.3, followed by 5 l of 8% formalin in 0.1 M phosphate buffer (PB), pH 7.3, 3 l of 0.1 M PB containing 10% sucrose, and, finally, with 2 l of 0.1 M PB containing 30% sucrose. The monkeys were attached to the stereotaxic frame, and the skulls were removed. The brains were cut into blocks in the frontal plane, removed and kept in 0.1 M PB containing 30% sucrose at 4°C, and then cut serially into 60-µm-thick frontal sections on a freezing microtome. The sections were mounted onto gelatin-coated glass slides and Nissl-stained with 1% Neutral Red. The recording sites were reconstructed according to the lesions made by current injection and the traces of electrode tracks. The positions of the cortical stimulating electrodes were also confirmed histologically.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Implantation of stimulating electrodes into MI and SMA

The forelimb regions of the MI and SMA were identified successfully on the anterior bank of the central sulcus and on the mesial wall of the hemisphere, respectively, by electrophysiological mapping as exemplified in Fig. 1. The arrangements of body part representations in the MI and SMA obtained from our mapping of these areas were consistent with previous reports (Kwan et al. 1978; Luppino et al. 1991; Mitz and Wise 1987; Park et al. 2001). In the forelimb region of the MI, the distal part of the forelimb was represented more laterally than the proximal part. Two pairs of stimulating electrodes were implanted into the forelimb region of the MI: one pair into the distal part (MI_d in Fig. 1B2) and the other into the proximal part (MI_p in Fig. 1B2). Although the distal and proximal parts of the forelimb were somewhat intermingled in the mediolateral central zone, MI_d and MI_p electrodes were implanted at sites mainly representing the distal and proximal parts of the forelimb, respectively, in all of the seven monkeys (Table 1). In the SMA, the forelimb region was represented anteriorly to the hindlimb region. A pair of stimulating electrodes was implanted into the forelimb region of the SMA (Fig. 1B1). During daily experimental sessions, cortical stimulation (less than 0.5 mA) through the electrodes in the MI_d and MI_p evoked movements of the distal (digits and wrist) and proximal (elbow and shoulder) parts of the forelimb, respectively. Movements evoked by even stronger stimulation (up to 0.7 mA) through the MI electrodes were restricted to the forelimb, suggesting that current spread to neighboring regions, such as the orofacial and hindlimb regions of the MI, was negligible. Cortical stimulation (up to 0.7 mA) through the SMA electrodes evoked no obvious movements, suggesting no current spread to the MI. Histological analysis confirmed that the tips of the stimulating electrodes were properly placed in the gray matter of the MI and SMA (Fig. 1, C1 and C2, shaded areas). The actual depths of implanted electrodes from the cortical surface determined histologically in each monkey were dependent on the thickness of the dura mater and the subdural space, and therefore different among monkeys [the MI_d and MI_p electrodes were (in mm) P1, 2.3, 2.4; P2, 2.2, 1.7; Put2, 3.0, 2.9 (Fig. 1C2); Put3, 2.2, 2.6; Put4, 2.7, 2.8; K4, 2.5, 2.6; K6, 2.8, 3.5. The SMA electrodes were (in mm) P1, 4.0; P2, 3.5; Put2, 2.1 (Fig. 1C1); Put3, 2.8; Put4, 2.6; K4, 4.2; K6, 2.2].

Inputs from MI and SMA to Put neurons

Based on the patterns of spontaneous activity, Put neurons are classified into two groups: phasically active neurons (PANs) that are silent at rest but phasically active during voluntary movement and tonically active neurons (TANs) that exhibit tonic background discharges at about 2-10 Hz and have action potentials with longer duration (Alexander and DeLong 1985b; Aosaki et al. 1994). A total of 425 Put neurons (349 PANs and 76 TANs) displaying excitatory responses to electrical stimulation in the MI and/or SMA were analyzed in seven monkeys (Table 2).

Typical examples of the responses of PANs to electrical stimulation in the MI and/or SMA are shown in Fig. 2. The Put neuron shown in Fig. 2A received input exclusively from the MI (MI-recipient PAN). Stimulation in the distal forelimb region of the MI induced orthodromic responses (Fig. 2A) with a threshold of 0.28 mA. Such a single stimulation in the MI evoked a few spikes, whereas stimulation in the SMA (up to 0.7 mA) evoked no response in this neuron (not shown). The Put neuron shown in Fig. 2B received convergent inputs from both the MI and SMA (MI + SMA-recipient PAN). Stimulation in the distal forelimb region of the MI induced orthodromic responses (Fig. 2B1) with a threshold of 0.5 mA. Stimulation in the SMA also evoked orthodromic responses (Fig. 2B2) with a threshold of 0.16 mA. SMA stimulation was considered to activate the same neuron as MI stimulation did because the voltage traces of the action potentials of these responses were identical (Fig. 2B3). The Put neuron shown in Fig. 2C received input exclusively from the SMA (SMA-recipient PAN). Stimulation in the SMA induced orthodromic responses (Fig. 2C) with a threshold of 0.35 mA, but stimulation in the MI did not (not shown). The weak and long-latency effects of cortical stimulation can be observed by constructing PSTHs. Cortical stimulation usually induced a simple monophasic excitation in PANs (Fig. 2D). When low-frequency (within PAN level) spontaneous spikes were observed, an inhibition lasting for around 100 ms following the excitation was observed in PANs (Fig. 2E). The activity returned to the prestimulus level 100-200 ms after cortical stimulation.

View larger version (30K): [in this window] [in a new window]   Fig. 2. Spike discharges of phasically active neurons (PANs) in the putamen (Put) that were orthodromically evoked by cortical stimulation. A: a PAN responding to MId stimulation but not to SMA stimulation (MI-recipient PAN). Around 10 traces were superimposed. B: a PAN receiving convergent inputs from both the MI and SMA (MI + SMA-recipient PAN). B1: responses to MId stimulation. B2: responses to SMA stimulation. B3: voltage traces of action potentials evoked by MId stimulation () and SMA stimulation (- - -). C: a PAN responding to SMA stimulation but not to MI stimulation (SMA-recipient PAN). D: a peri-stimulus time histogram (PSTH; binwidth, 0.5 ms) of a PAN (the same neuron as in A) responding to MId stimulation. Cortex was stimulated at time 0 (). E: a PSTH of a PAN responding to SMA stimulation (another SMA-recipient PAN). Voltage calibration in A is applicable to all traces in this figure. Time calibration in A is applicable to A, B, 1 and 2 and C. All PSTHs were summed 100 times.

TANs also responded to stimulation in the MI and/or SMA (Fig. 3). A TAN in Fig. 3A received input exclusively from the SMA (SMA-recipient TAN). This neuron was spontaneously active at 4 Hz and had action potentials with a long duration (Fig. 3A2). Stimulation in the SMA induced orthodromic responses (Fig. 3A1) with a threshold of 0.5 mA, but stimulation in the MI did not (not shown). Early excitation was followed by a period of inhibition and a long-latency excitation (Fig. 3A3). Another example of an SMA-recipient TAN is shown in Fig. 3B. In this neuron, SMA stimulation induced an early excitation, a subsequent inhibition, and a late excitation (Fig. 3B1). However, MI stimulation induced an inhibition and a subsequent excitation without an early excitation (Fig. 3B2). As the inhibitory response may be due to synaptic events within the cortex (for details, see DISCUSSION), only early excitations were taken into account, and this neuron was classified as an SMA-recipient TAN. Put neurons lacking early excitations to all cortical stimulation were not sampled for further analysis.

View larger version (23K): [in this window] [in a new window]   Fig. 3. Spike discharges of tonically active neurons (TANs) in the Put that were orthodromically evoked by cortical stimulation. A: a TAN responding to SMA stimulation but not to MI stimulation (SMA-recipient TAN). A1: responses to SMA stimulation. A2: voltage traces of action potentials. Voltage calibration in A1 is applicable to A2. A3: a PSTH showing responses to SMA stimulation. B: another SMA-recipient TAN. B1: responses to SMA stimulation. B2: responses to MIp stimulation. This neuron was classified as an SMA-recipient TAN as it lacked an early excitation to MIp stimulation. All PSTHs were summed 100 times.

Both PANs and TANs could be classified into MI-, MI + SMA-, or SMA-recipient neurons (Table 2). Although the percentages of each group varied somewhat among individual monkeys, around 20% of all the Put neurons examined (19% of PANs and 27% of TANs) were found to receive convergent inputs from both the MI and SMA.

Among 174 MI-recipient PANs, 80 neurons responded exclusively to stimulation in the distal forelimb region of the MI (MI_d-recipient neurons), 54 neurons to stimulation in the proximal forelimb region (MI_p-recipient neurons), and 40 neurons to stimulation in both the distal and proximal regions (MI_d+p-recipient neurons), as shown in Table 3. This suggests that cortical stimulation (up to 0.5 mA) excites the distal and proximal forelimb regions of the MI separately (for details, see DISCUSSION). A PAN shown in Fig. 2A, for example, responded to MI_d stimulation but not to MI_p stimulation; this neuron should be an MI_d -recipient neuron. According to the same criteria, MI + SMA-recipient Put neurons could be classified into MI_d + SMA-, MI_d+p + SMA-, or MI_p + SMA-recipient neurons (Table 3).

The latencies of excitations in Put neurons evoked by each cortical stimulation are compared in Fig. 4. The latency of MI-induced excitation [10.2 ± 2.5 (SD) ms] was significantly shorter than that of SMA-induced excitation (13.8 ± 3.6 ms; ANOVA with Bonferroni/Dunn post hoc tests, P < 0.01). The latencies of the excitations evoked by stimulation in the same cortical area were comparable between PANs and TANs and between neurons with converging inputs and neurons with a single cortical input (Fig. 4) except that the latency of the excitation evoked by MI stimulation in MI-recipient PANs (9.9 ± 2.3 ms) was shorter than that in MI + SMA-recipient PANs (11.2 ± 2.9 ms; ANOVA with Bonferroni/Dunn post hoc tests, P < 0.01).

View larger version (26K): [in this window] [in a new window]   Fig. 4. Distribution of the latencies of cortically evoked excitatory responses in Put neurons. Ordinates indicate the number of neurons. A: latencies of MI ()- and MI + SMA ()-recipient PANs to MI stimulation. B: latencies of MI ()- and MI + SMA ()-recipient TANs to MI stimulation. C: latencies of SMA ()- and MI + SMA ()-recipient PANs to SMA stimulation. D: latencies of SMA ()- and MI + SMA ()-recipient TANs to SMA stimulation. Values represent the means ± SD expressed in milliseconds. *,  significantly different from each other (ANOVA with Bonferroni/Dunn post hoc tests, P < 0.01)

Sensorimotor response properties of Put neurons and movements evoked by Put microstimulation

To clarify the nature of MI-, MI + SMA-, and SMA-recipient Put neurons, sensorimotor responses of each neuron and the effects of microstimulation at each recording site were investigated. Around 60% of PANs and TANs were activated during passive manipulation or active movements of the forelimb, as shown in Table 4. The percentage of responding neurons tended to be higher in MI- and MI + SMA-recipient neurons and lower in SMA-recipient neurons. The distribution patterns of Put neurons according to the origins of cortical inputs and the receptive fields in the forelimb to sensorimotor stimuli are presented in Fig. 5. Among PANs activated during passive manipulation or active movements of the forelimb, most of the MI_d-recipient PANs responded to somatosensory stimuli of the digits, and most of the MI_p-recipient PANs responded to those of the shoulder. Thus a close somatotopic correspondence was detected between the response properties of MI-recipient PANs to somatosensory stimuli and the cortical inputs. On the other hand, MI + SMA-recipient PANs were dominated by proximal somatosensory inputs, although the distal/proximal ratio was higher in MI_d + SMA- and MI_d+p + SMA-recipient PANs than in MI_p + SMA-recipient PANs. In SMA-recipient PANs, the proximal parts of the forelimb were more strongly represented than the distal parts. The somatosensory inputs to TANs seemed to parallel those found in PANs except for the absence of distal somatosensory inputs to MI_d-recipient TANs.

View larger version (56K): [in this window] [in a new window]   Fig. 5. Distribution patterns of PANs (left) and TANs (right) according to the origins of cortical inputs and the receptive fields in the forelimb to sensorimotor stimuli. Abscissas indicate the percentage of neurons per category. Each figure above each bar indicates the number of responsive neurons. Other, neurons that responded to stimulation of body parts other than the forelimb; active movement, neurons that became active only during active forelimb movements; negative, neurons that did not respond to any sensorimotor stimuli.

The effects of microstimulation varied significantly between MI- and MI + SMA-, and SMA-recipient Put zones (Table 4). Microstimulation at more than 80% of the sites around MI-recipient neurons and at 70% of the sites around MI + SMA-recipient neurons evoked conspicuous movements of the forelimb, while such movements were rarely (around 10%) evoked by microstimulation in the neighborhood of SMA-recipient neurons (² test, P < 0.01). The distribution patterns of Put sites according to the origins of cortical inputs and the body parts displaying movements evoked by Put microstimulation are shown in Fig. 6. Most of the sites within the MI_d-recipient zone were associated with movements of the digits, and more than half of the sites within the MI_p-recipient zone were related to movements of the shoulder or elbow. This indicates a somatotopic correspondence between microstimulation-evoked movements and cortical inputs. The MI + SMA-recipient zone also showed a similar somatotopic correspondence. In rare instances, microstimulation in the SMA-recipient zone evoked movements of the distal as well as the proximal part of the forelimb. The distribution of threshold currents to evoke movements by Put microstimulation according to the origins of cortical inputs and the body parts displaying evoked movements is shown in Fig. 7. In this figure, the difference in threshold currents indicates the differences in microexcitability among the MI-, MI + SMA-, and SMA-recipient zones. Threshold currents in the MI-recipient zone of the Put (26 ± 12 µA) were significantly lower than those in the MI + SMA-recipient zone (33 ± 11 µA) and in the SMA-recipient zone (38 ± 12 µA; ANOVA with Bonferroni/Dunn post hoc tests, P < 0.01). The MI-recipient zone representing the digits was found to have the lowest threshold (23 ± 11 µA).

View larger version (38K): [in this window] [in a new window]   Fig. 6. Distribution patterns of Put sites according to the origins of cortical inputs and the body parts displaying movements evoked by Put microstimulation. Abscissas indicate the percentage of sites per category. Each figure above each bar indicates the number of effective microstimulation sites. Other, sites where movements of body parts other than the forelimb were evoked; negative, sites where no movements were evoked.

View larger version (30K): [in this window] [in a new window]   Fig. 7. Distribution of threshold currents to evoke movements by Put microstimulation separately in the MI (top)-, MI + SMA (middle)-, and SMA (bottom)-recipient zones, and the body parts displaying movements evoked by microstimulation. Ordinates indicate the number of sites where movements were evoked.

Locations of recorded Put neurons

The locations of recorded neurons in the Put are plotted in Figs. 8 and 9. The neurons that responded to stimulation in the forelimb regions of the MI and/or SMA were distributed in a band extending from the ventrolateral to dorsomedial part of the Put. Within this band, MI-recipient neurons (Fig. 8A, circles) were located mainly in the lateral part, whereas SMA-recipient neurons (squares) were located predominantly in the medial part. MI + SMA-recipient neurons (triangles) were located in between. Put neurons situated dorsally to the MI-recipient zone often responded to manipulation of the hip joint, and microstimulation in this area evoked movements of the hip joint (not shown, but see Fig. 10). In contrast, Put neurons situated ventrally to the MI-recipient zone responded to manipulation of the orofacial region, and microstimulation in this area evoked orofacial movements. TANs were observed to be sparse (shaded symbols in Fig. 8A), and distributed to share cortical inputs with neighboring PANs.

View larger version (24K): [in this window] [in a new window]   Fig. 8. Locations of Put neurons recorded in monkey Put2 in frontal sections according to cortical inputs. In each row, 3 sections are arranged rostrocaudally from left to right (A16, A14, A12). A: MI-, MI + SMA-, and SMA-recipient PANs and TANs. B: Put neurons with input from the MId, MId+p, and MIp.

View larger version (29K): [in this window] [in a new window]   Fig. 9. Locations of Put neurons recorded in monkey Put2 in frontal sections displaying sensorimotor responses and movements evoked by microstimulation. In each row, 3 sections are arranged rostrocaudally from left to right (A16, A14, A12). A: receptive fields of Put neurons to sensorimotor stimuli. B: the body parts displaying movements evoked by Put microstimulation. C: threshold currents for Put microstimulation to evoke movements.

View larger version (35K): [in this window] [in a new window]   Fig. 10. Summary diagram showing the somatotopic organization of the Put according to cortical motor inputs from the MI and SMA. The results of the present study (Figs. 8 and 9) and the previous reports (Takada et al. 1998a,b) are combined. Drawings in black lines represent somatotopically arranged cortical inputs from the MI, whereas those in gray lines roughly represent somatotopic inputs from the SMA. Inputs from both the MI and SMA converge onto the mediolateral central part of the Put. There is a distal-proximal somatotopic organization in the MI- and MI + SMA-recipient forelimb regions.

In the MI- and MI + SMA-recipient zones of the Put, MI_d- and MI_d + SMA-recipient neurons (Fig. 8B, closed circles) were located mainly in their ventral parts, whereas MI_p- (open circles), MI_d+p-, and MI_d+p + SMA- (shaded circles) recipient neurons were located more dorsally.

Put neurons displaying sensorimotor responses are plotted in Fig. 9A. In the MI-recipient zone, the neurons responding to somatosensory stimuli of the digits (closed circles) or wrist (closed triangles) were located in the ventral aspect, whereas neurons responding to stimuli of the shoulder (open circles) or elbow (open triangles) were distributed more dorsally. In the SMA-recipient zone, most neurons responded to somatosensory stimuli of the shoulder or elbow, with neurons related to the shoulder tending to be located more dorsolaterally.

Put sites where microstimulation evoked forelimb movements are plotted in Fig. 9B. In the MI- and MI + SMA-recipient zones of the Put, microstimulation in the ventral part evoked movements of the digits (closed circles) or wrist (closed triangles), whereas microstimulation in the dorsal part elicited movements of the shoulder (open circles) or elbow (open triangles). In the SMA-recipient zone of the Put, microstimulation rarely evoked any movement. Threshold currents for microstimulation to evoke movements at each Put site are shown in Fig. 9C. More excitable sites (i.e., those with lower thresholds, represented by larger circles) were greatly accumulated in the MI-recipient zone of the Put, especially in its ventral part representing the digits.

In the MI- and MI + SMA-recipient Put zones, the cortical input map (Fig. 8B), sensorimotor response map (Fig. 9A), and microstimulation map (Fig. 9B) corresponded well to one another. Such a somatotopic organization revealed in these maps suggests a ventrodorsal topography of representations from the distal to proximal part of the forelimb (see Fig. 10). In the SMA-recipient zone, on the other hand, the proximal part of the forelimb was represented almost exclusively, so no clear somatotopic arrangement was observed.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study has revealed that corticostriatal inputs from the forelimb regions of the MI and SMA are largely segregated. On the other hand, convergent inputs from the MI and SMA were noted on single neurons located at the junction between the two input zones. This suggests that information derived from the MI and SMA is at least partly integrated in the Put. The present study has also demonstrated a distal-proximal somatotopic organization in the Put according to the corticostriatal inputs from the MI (see Fig. 10).

Largely segregated but partly convergent inputs from MI and SMA

The extent of the current spread from stimulating electrodes implanted in the cortex was estimated to be less than 2.5 mm because electrophysiological recording at the distance of 2.5 mm from the electrode implantation sites showed no excitation of cortical neurons after stimulation (up to 0.7 mA) by the electrode (not shown). MI_d and MI_p electrode implantation sites were 2.5-6.2 (mean 3.6) mm apart, and stimulation in the MI_d and MI_p is considered to have distinctly stimulated the distal and proximal forelimb regions of the MI, respectively. If MI_d stimulation spread to MI_p, or vice versa, stimulation in either the MI_d or the MI_p would have excited most MI-recipient neurons. However, the actual number of MI_d+p-recipient Put neurons was comparable to that of MI_d- or MI_p-recipient Put neurons (Table 3). Thus stimulation in the MI_d, MI_p, and SMA is considered to have excited each cortical area specifically. MI_d+p-recipient neurons may receive convergent inputs from both the distal and proximal forelimb regions of the MI, and/or inputs from corticostriatal neurons located in the intermediate region between the distal and proximal forelimb regions of the MI.

The orthodromic responses evoked by cortical stimulation in the present study are considered to be mediated by direct corticostriatal projections based on comparisons with previous findings. The distribution of the orthodromically activated Put neurons corresponds well to that of MI- and SMA-derived corticostriatal terminals reported previously (Künzle 1975; Liles and Updyke 1985; Strick et al. 1995; Takada et al. 1998a,b). The latency of MI-evoked orthodromic responses of Put neurons in this study (Fig. 4) was within the same range as that of corticostriatal evoked field potentials in the monkey (Liles 1975), orthodromic activation of striatal neurons in the cat (Liles 1974), and rat (Kawaguchi et al. 1989; Kitai et al. 1976; Wilson 1986), Put-evoked antidromic activation of MI neurons (Bauswein et al. 1989; Turner and DeLong 2000), and the latency difference between cortically evoked inhibitory responses and striatally evoked inhibitory responses in the globus pallidus (Yoshida et al. 1993).

It might be argued that excitatory responses in MI + SMA-recipient Put neurons evoked by SMA stimulation could be mediated by the SMA-MI projection (Tokuno and Nambu 2000) and then the MI-Put projection but not by the direct SMA-Put projection. If this was the case, the neurons in the center of the MI-recipient Put zone might also be expected to respond to SMA stimulation. However, the MI + SMA-recipient Put neurons were located only in the intermediate zone between the laterally situated MI-recipient and medially situated SMA-recipient zones (Fig. 8), corresponding well to the distribution patterns of corticostriatal terminals from these cortical areas (Takada et al. 1998a,b). The fact that the latency of MI + SMA-recipient Put neurons to SMA stimulation was similar to that of SMA-recipient Put neurons to SMA stimulation (Fig. 4) also supports the argument that SMA stimulation does not activate MI + SMA-recipient Put neurons indirectly. Thus MI + SMA-recipient neurons in the present study must receive convergent inputs from both the MI and SMA directly.

There has been some controversy about whether corticostriatal inputs from somatotopically corresponding regions of the MI and SMA converge in the Put. An early report on this subject showed that striatal areas receiving afferents from forelimb representations of the MI and SMA were segregated from each other (Alexander et al. 1988). Another study indicated a similar segregated pattern in which efferents from the MI and SMA terminated most densely in separate regions of the Put, although some terminals from the MI were found in the dorsal region of the Put that was the site of dense termination from the SMA (Strick et al. 1995). In contrast, a recent study reported that the forelimb regions of the MI and SMA sent overlapping projections to the Put (Inase et al. 1996). Our anatomical data have shown that corticostriatal input zones from somatotopic regions of the MI are located predominantly in the lateral part of the Put, whereas those from SMA regions are located more medially (Takada et al. 1998a,b). Thus somatotopically corresponding regions of the MI and SMA tend to have separate input zones within the Put, as previously reported (Alexander et al. 1988; Strick et al. 1995). However, a partial overlap of input zones from these cortical areas takes place in the mediolateral central part of the Put (Takada et al. 1998a,b). In the present electrophysiological study, MI-recipient neurons (about 45% of the total Put neurons examined) were located in the lateral part, while SMA-recipient neurons (about 35% of the Put neurons) were located in the medial part (Table 2, Fig. 8). Moreover, about 20% of the Put neurons were found to receive convergent inputs from both the MI and SMA and to be distributed in the mediolateral central part. These results confirm our previous anatomical data at the single-neuron level (Takada et al. 1998a,b) (see also Fig. 10).

Our systematic studies (Inase et al. 1999; Takada et al. 1998a,b, 2001) indicate that corticostriatal input zones from the motor-related areas of the frontal lobe are distributed in an orderly topographical fashion and display complex patterns of segregation versus overlap of one another, presumably based on some rules. Projections from anatomically different cortical areas basically terminate in segregated regions of the striatum, but projections from areas with similar hodological and functional aspects might converge at least partly in the striatum. For example, projections from the MI and higher-order motor areas, such as the presupplementary motor area and the rostral cingulate motor area, terminate in segregated regions of the striatum. Conversely, projections from the MI, SMA, and premotor cortex partly converge in a common region of the striatum. These observations lead us to the idea that the convergence of corticostriatal projections from different cortical areas may have some functional significance. In this study, the latency to MI stimulation of MI + SMA-recipient PANs was longer than that of MI-recipient PANs (Fig. 4), suggesting that the convergence is not merely a result of a common border between the MI and SMA input zones. Further studies, such as recording of neuronal activity in convergent zones and blocking of these zones during task performance, are necessary to clarify the functional role of corticostriatal convergence.

Previous anatomical studies have repeatedly demonstrated that striatal zones receiving corticostriatal afferents from the MI and SMA are composed of multiple dispersed patches or modules called matrisomes (Flaherty and Graybiel 1993; Parthasarathy and Graybiel 1997; Takada et al. 1998a,b). It has also been reported that the dendritic fields of striatal spiny neurons observe compartmental boundaries (Kawaguchi et al. 1989). It follows that PANs recorded in the present study should be located mainly in the matrisomes. However, responsive neurons and microexcitable sites seem to be distributed in a continuous fashion rather than in patches (Figs. 8 and 9). Electrophysiological methods used in the present study may not be able to detect the anatomical compartmental boundaries precisely: extracellular recordings may detect the activity of neurons within a certain distance, and microstimulation currents may spread over a certain distance. Similar discrepancies between the anatomical and electrophysiological results have been pointed out previously (Kawaguchi et al. 1989), and the authors discussed other explanations, including minor cross-compartmental corticostriatal projections and polysynaptic intrastriatal excitatory connections between the compartments.

In the present study, cortical stimulation evoked early excitation and subsequent inhibition in both PANs and TANs and, in addition, a late excitation in TANs. Intracellular recording studies of striatal spiny neurons, which are considered to be PANs, showed similar response patterns to cortical stimulation, and the origin of each response was analyzed (Kawaguchi et al. 1989; Kita 1993, 1996; Kita et al. 1985; Kitai et al. 1976; Wilson 1986; Wilson et al. 1982, 1983; see also Wilson 1998). The initial excitatory postsynaptic potential (EPSP) was found to be followed by a long-lasting (200-500 ms) hyperpolarization and, on its termination, by a period of depolarization and increased synaptic noise. The initial EPSP was considered to be caused by monosynaptic corticostriatal projections because the latency remained constant despite changes in stimulus intensity or frequency (Kitai et al. 1976). A similar EPSP evoked by intrastriatal (Kita et al. 1985) or cortical (Kawaguchi et al. 1989) stimulation in the striatal slice also supports this idea. The possible involvement of polysynaptic pathways, such as intrastriatal excitatory connections (although no excitatory interneuron has been identified in the striatum) and the cortico-thalamo-striatal pathway (Wilson et al. 1982), in the later component of the initial EPSP cannot be excluded. The hyperpolarization observed was composed of a small, short-lived inhibitory postsynaptic potential (IPSP) and a long-lasting hyperpolarization. The short IPSP was probably produced by GABAergic striatal interneurons (Kita 1993, 1996), while it is likely that the long-lasting hyperpolarization was generated by disfacilitation, i.e., the removal of excitatory cortical inputs (Wilson et al. 1983). The late depolarization following the long-lasting hyperpolarization was probably due to resumption of a tonic excitatory influence from the cortex (Wilson 1986). Because the long-lasting hyperpolarization and the following depolarization may have reflected synaptic events within the cortex, only the early excitatory response of Put neurons was taken into account in the present study.

Somatotopic organization of Put

The somatotopic organization in the Put has been previously reported as having a dorsolateral to ventromedial topography of representations from hindlimb to face, with the forelimb being represented in an intermediate region, based on the somatosensory responses, the evoked movements by microstimulation (Alexander and DeLong 1985a,b), the neuronal activity related to movements (Crutcher and DeLong 1984), and the corticostriatal projections (Künzle 1975; Liles 1975; Takada et al. 1998a,b). The results of the present study concurred well with such a somatotopic organization in the Put (see Fig. 10). However, the proximal versus distal representations of the forelimb region were not reported in previous electrophysiological studies. The present study showed that the distal and proximal forelimb parts were represented along the ventrodorsal axis of the MI- and MI + SMA-recipient Put zones (Figs. 8 and 9; see Fig. 10), corresponding well to the organization of corticostriatal terminal zones from the distal and proximal forelimb regions of the MI (Tokuno et al. 1999).

Liles and Updyke (1985) demonstrated a close relationship between corticostriatal input and physiological activity of individual Put neurons. Alexander and DeLong (1985b) also showed a coincidence between the sensorimotor response map and the microstimulation map. The present study essentially supports these ideas (Figs. 8 and 9). The Put neurons receiving inputs from the distal forelimb region of the MI had somatosensory responses to the distal forelimb, and microstimulation adjacent to these neurons induced movements of the distal forelimb. Analogous relationships were true for Put neurons receiving inputs from the proximal forelimb region of the MI. Thus the activity of Put neurons appears to be dominated by their corticostriatal projections. However, there was some discrepancy between the response properties to somatosensory stimuli (Fig. 5) and the movements evoked by microstimulation (Fig. 6). Among the PANs activated during passive manipulation or active movements of the forelimb, only MI_d-recipient PANs were dominated by distal somatosensory inputs. Other categories of PANs, including MI_d+p-, MI_d + SMA-, MI_d+p + SMA-, and SMA-recipient PANs, were dominated by proximal somatosensory inputs. This may reflect the nature of cortical area of input origin (see Table 1). Conversely, the Put zones subjected to microstimulation, with the exceptions of the MI_p- and MI_p + SMA-recipient zones, were dominated by distal movements. This may be ascribed primarily to the greater excitability of Put zones for distal movements compared with those for proximal movements (see Fig. 7). Another possibility is that the input information from the proximal forelimb might be linked to the output information to the distal forelimb in these Put areas.

The source of the somatosensory responses of Put neurons is likely to be the MI and the primary somatosensory cortex (SI) because projection fibers from the SI terminate in MI-recipient striatal zones (Flaherty and Graybiel 1993). Another possible source is the thalamus (Matsumoto et al. 2001; McFarland and Haber 2000). Matsumoto et al. (2001) have shown that neurons in the centromedian-parafascicular nuclear complex supply striatal neurons with information about behaviorally significant sensory events that can activate conditional responses of striatal neurons.

Activity of TANs

An anatomical study (Lapper and Bolam 1992) failed to show synaptic termination from the cortex on cholinergic striatal interneurons, which are considered to be TANs. However, an intracellular recording study showed that cholinergic interneurons were directly excited by cortical stimulation (Wilson et al. 1990). Recently, Thomas et al. (2000) have described cortical inputs to the distal dendrites of cholinergic interneurons in rats and monkeys. In the present study, TANs responded to cortical stimulation with a similar latency to that of PANs. It is reasonable to assume that TANs are likely to share cortical input with neighboring PANs. Previous studies reported that TANs did not fire in relation to body part movements per se but did respond specifically to conditioned sensory stimuli (Alexander and DeLong 1985b; Aosaki et al. 1994). In the present study, a considerable number of TANs responded to somatosensory stimulation (Table 4). This might be explained by the method used in the present study, whereby only TANs with cortical inputs were sampled. Further studies are necessary to clarify the properties of the inputs to TANs.

Functional considerations

The changes in activity of Put neurons are correlated with body part movements during the performance of motor tasks. Neurons in the medial and lateral parts of the Put seem to show activity changes in relation to different aspects of motor tasks. Put neurons in the lateral part had firing patterns that closely resembled the activity in agonist muscles, whereas those in the medial part did not (Liles 1983). When monkeys were trained to perform a visuomotor step-tracking task in which elbow movements were made with or without prior instruction concerning the direction of a forthcoming movement, Put neurons manifesting preparatory activity, i.e., those displaying task-related changes in the discharge rate during the postinstruction (preparatory) interval, were located more rostrally and medially than those showing movement-related activity only (Alexander and Crutcher 1990b). These authors further reported that the proportion of neurons showing preparatory activity in the SMA was larger than that in the MI. These differences in the activity of Put neurons may be ascribed to distinct cortical motor inputs arising from the MI versus SMA.

Alexander and DeLong (1985a) showed that movements evoked by microstimulation in the Put resulted from the activation of Put projection neurons but not from the antidromic activation of the cortex nor from the current spread to the internal capsule. This conclusion was based on three findings. First, the microstimulation effects were abolished by fiber-sparing lesions produced by microinjection of the neurotoxin ibotenic acid into the Put. Second, the effective radius of microstimulation (40 µA) was estimated to be approximately 150 µm. Finally, the chronaxie value in the Put was significantly longer than that in the internal capsule. In the present study, both the MI- and MI + SMA-recipient Put zones were microexcitable, whereas the SMA-recipient Put zone was far less microexcitable. This suggests that both the MI- and MI + SMA-recipient Put neurons may predominantly project to the area of the globus pallidus that is directly related to motor execution, whereas the SMA-recipient Put neurons may project to the pallidal area that is involved in higher-order motor control. These observations favor the notion that motor information processing in the striato-pallido-thalamo-cortical projections is essentially governed by a parallel rule. The effects of signals derived from the MI-, MI + SMA-, and SMA-recipient Put neurons on motor control merit further investigation to advance our understanding of the functional significance of information processing in the Put.