| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Department of Physiology, Hirosaki University School of Medicine, Hirosaki, Japan
Submitted 9 February 2007; accepted in final form 4 August 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
; Hoshi and Tanji 2006; Rizzolatti and Luppino 2001; Schwartz et al. 2004; Tanji et al. 1987, 1988). One important integration is the transformation of coordinates from visual to motor space for reaching (Alexander and Crutcher 1990; Cisek et al. 2003; Kakei et al. 2001; Kurata and Hoshi 2002; Pesaran et al. 2006). We perform reaching movements toward a target under visual guidance on a daily basis. During reaching, our CNS apparently first maps the location of the target in a retinocentric reference frame and then transforms the map into final motor commands via several steps (Andersen et al. 1993; Atkeson 1989; Kawato et al. 1988). Recently it has been suggested that the ventral premotor cortex (PMv) and the primary motor cortex (MI) in monkeys (Fig. 1A) play important roles in this transformation by providing distinct movement-related neuronal activities, one reflecting a visual (extrinsic) frame of reference and the other reflecting a motor (intrinsic or muscle-like) frame of reference (Kakei et al. 1999, 2001; Kurata and Hoshi 2002; Rizzolatti and Luppino 2001; Schwartz et al. 2004). Among those studies, we have focused on movement-related activities in the PMv and MI at the time that the visuomotor transformation for reaching was expected to occur. We have trained monkeys to perform a reaching task in which the visual and motor spaces were dissociated using shift prisms, and we have shown that movement-related activity reflecting the visual frame of reference (termed V-type activity) was more frequently recorded in the PMv than in the MI, whereas activity reflecting the motor frame of reference (termed M-type activity) was recorded in both the MI and PMv (Kurata and Hoshi 2002). Other than the V- and M-types, we also recorded a number of movement-related activities that were not closely related to visual or motor space (termed B- and N-types), and they exhibited properties intermediate between the V- and M-types. Based on the observation, we suggested that the activities in the PMv and MI could contribute to coordinate transformations from visual to motor space (Kurata and Hoshi 2002). However, it remains to specify what role the activity types play in the transformation of coordinates.
|
; see Porter and Lemon 1993 for review) and in monkeys (Tanji et al. 1988), laterality of movement-related activity is a good indicator to examine how movement-related activity reflects final motor commands for forelimb muscles. Although it should be remembered that the MI and nonprimary motor areas of humans (Hanakawa et al. 2005; Kansaku et al. 2005; Kim et al. 1993; Kollias et al. 2001; Kurata et al. 2000) and of monkeys (Cisek et al. 2003; Evarts 1966; Tanji et al. 1987, 1988) are involved in the bilateral and/or ipsilateral motor control of distal limbs as well as proximal limbs and the body axis, those findings also imply higher brain mechanisms in the cortical motor area such as sensorimotor transformation especially from visual inputs rather than direct motor control. Supporting this view, it has been reported that neuronal activity in the PMv is influenced by visual inputs from contra- and ipsilateral hemifields (Boussaoud and Wise 1993a,b; Schwartz et al. 2004). Furthermore, Hoshi and Tanji (2002, 2006) examined how PMv neurons responded to two successively presented visuospatial cues with intervening delays that instructed either future target locations or arms to be used for reaching. Throughout the two instructed delay periods, a majority of the PMv neurons responded to the cues and exhibited a tight relation to their spatial positions, regardless of whether the cues were presented in the contra- or ipsilateral visual hemifield. Based on the findings, it is possible that, if the movement-related activity reflects visual space, it could exhibit less laterality than that reflecting motor space. By contrast, if activity reflecting motor space represents final motor commands, it could be selectively active during contralateral forelimb movements.
Thus in the present study, we examined laterality as well as relationship to visual and motor space of the movement-related activity we have previously found (Kurata and Hoshi 2002) to address important questions as to what role the activity plays in the transformation of coordinates and how the transformations are processed in the PMv and MI. Using laterality and correlations to visual and motor frames of reference, we are able to characterize activities not only the V- and M-types, but also the B- and N-types, and discuss their roles in the transformation of coordinates. Accordingly, we trained monkeys to reach with either the left or right arm toward targets with visual and motor coordinates that had been dissociated by shift prisms as in our previous study and specifically examined laterality and relation to visual and motor space of movement-related activity in the PMv and MI. Throughout the experiment, we focused on movement-related neurons with activity that changed during the reaction time (RT) period, when the visuomotor transformation for goal-directed reaching was expected to occur, and compared the activity of these neurons during the movement time (MT) period.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
All experiments were approved by the Animal Research Committee of Hirosaki University, Japan. Experiments were conducted in compliance with the Guide for the Care and Use of Mammals in Neuroscience and Behavioral Research by the Committee on Guidelines for the Use of Animals in Neuroscience and Behavioral Research (National Research Council, Washington, DC).
We used three male Japanese monkeys (Macaca fuscata, 5.5–8.2 kg). The monkeys sat comfortably in a primate chair, facing a 21-in CRT screen covered with a transparent touch panel that monitored the position of the monkey's hand on the screen by detecting local pressure. The position of the monkey's hand on the touch screen was sampled at 500 Hz through an eight-channel, 12-bit A/D converter, and the data were stored in a laboratory computer. The screen was placed 30 cm away from the monkey's eyes, and the vertical center of the screen was aligned with the monkey's head and body. An apparatus with two pairs of 4 x 4-cm wedge prisms (10° to the left or right) was placed 6 cm away from the monkey's eyes. The same prism apparatus has been used in previously published experiments (Kurata and Hoshi 1999, 2002). Two switches, made of 5 x 10-cm acrylic plates, were symmetrically placed at the ends of the right and left armrests to hold the keys. The keys and armrests were positioned 20 cm below the animal's eyes. Eye movements were not monitored in monkey 1 but were monitored in monkeys 2 and 3 using an infrared oculometer system (R21CA, RMS, Hirosaki, Japan). The infrared light was presented by the oculometer and reflected by a half mirror placed in a 6-cm space between the monkey's eyes and the prism holder (Fig. 1B). All monkeys were free to move their eyes at any time during a trial. An opaque barrier immediately below a spout for juice rewards (Fig. 1B) blocked a view to their arms until a reaching arm was seen through the prisms and prevented the mirror from being displaced or broken by the monkeys.
Behavioral task
The three monkeys were trained to make quick reaching movements with the right or left hand, toward a target. A trial began when the monkey used both hands to depress the keys (Fig. 1C). Then, 500 ms after trial initiation, either a 300-Hz or 1-kHz tone was randomly selected and presented to instruct the monkey to prepare for a left- or right-arm movement, respectively. Within 1.5–3.0 s after the instruction signal, a blue rectangular target appeared on the 21-in monitor with a touch sensor (Fig. 1C). If the monkey released the hold key by the instructed arm (movement onset) within 500 ms after target presentation, reached the touch screen within 700 ms of the movement onset, and hit the correct target, then a drop of juice (0.1 ml) was delivered as a reward. The RT was defined as the period between the appearance of the target and the onset of movement by the instructed arm. The MT was defined as the period between the onset of movement and contact with the screen. In any prism conditions, the monkeys were allowed to hit the screen first (not necessarily the target) and then move their hand to the target. However, the first contact point on the screen was generally close to the target irrespective of the prism condition after prism adaptation (see RESULTS). The monkeys were outfitted with prisms that shifted the image of the target 10° to either the left or right or were allowed to view the target normally. The prisms dissociated the visual space from the motor space, enabling us to examine which reference frame, visual or motor, was reflected in the movement-related neuronal activity. During each trial, the instruction signal for either a right or left arm movement was pseudorandomly selected, and a target was also pseudorandomly selected from among nine visual locations that were common under the three prism conditions (Fig. 1D). The task was performed in a block of 200 trials under each prism condition. During the trial block, data were evenly sampled for the nine target locations and the two arms.
Neuronal and EMG recordings
After completion of the behavioral training, the monkeys were surgically prepared under aseptic conditions using nitrous oxide and isofluorothane anesthesia afterng induction with ketamine hydrochloride (8 mg/kg im) and atropine sulfate. Four head-restraining bolts and one rectangular stainless steel recording chamber (27 x 27 mm) were implanted in the skull of each monkey. The chamber was centered at anterior 12.0 mm and lateral 18.0 mm, according to the Horsley-Clarke stereotaxic frame. Antibiotics and analgesics were used to prevent postsurgical infection and pain.
After complete recovery from the surgery (>7 day), we recorded the neuronal activity from the PMv and the MI (Fig. 1A) in the four hemispheres of each monkey during performance of the behavioral task. The monitored areas were selected based on the central and arcuate sulci and the arcuate spur observed during surgery. We confirmed by histological reconstruction that the areas covered the proximal forelimb representations of the PMv and MI (Gentilucci et al. 1988; Kurata and Hoshi 2002; Kurata and Tanji 1986). For the single-unit recordings, we used glass-insulated Elgiloy microelectrodes (1.0–1.5 M
at 333 Hz) inserted through the dura mater using a hydraulic microdrive (MO95; Narishige, Tokyo, Japan). Electrode signals were amplified, filtered, and sorted (MCP and MSD; Alpha-Omega, Nazareth, Israel). A multi-spike detector (MSD) was used to sort spikes, allowing up to three isolated neurons to be recorded simultaneously.
We also bilaterally sampled electromyographic (EMG) activity with wire electrodes from the anterior deltoid, trapezius, supraspinatus, infraspinatus, pectoralis major, rhomboid, thoracic paravertebral, biceps, and triceps brachii muscles. The EMG data were band-pass filtered between 20 Hz and 5 kHz and were sampled at 100 Hz through the A/D converter in a laboratory computer.
Data analysis
Our database included only those neurons for which activity was stably recorded during >100 trials under each of the three prism conditions. For quantitative analysis, the data from the first 10 trials during adaptation to the prisms were not used because it was essential to obtain data after readaptation to the prisms in each condition. The significance level of all statistical tests was P < 0.01.
For each of the nine targets under the three prism conditions, a raster display of recorded neuronal activity was aligned with onsets of contra- and ipsilateral arm movements, and a peri-event histogram with a 20-ms bin width was created to show neuronal activity for each target under each prism condition (Figs. 3 and 5). The mean discharge rate and its SD during the interval of 0.5–1.5 s before target presentation (premovement control period) were calculated first. If the neuron increased its activity before movement onset after a target was presented and its activity exceeded 2.56 SD (P < 0.01) during the RT in at least two consecutive bins (40 ms) of at least three of the nine histograms in the no-prism condition, using either the contra- or ipsilateral arm, then the activity was tentatively considered movement related. During recording, we also examined visual response of the neurons. For this purpose, a visual stimulus identical to the target was presented in optionally selected trials under different prism conditions. The test visual stimulus was presented 200 ms after trial initiation for 100 ms at selected target positions where the neuron exhibited activity change during the RT (Fig. 4A). During the period immediately after trial initiation, the monkeys were required to withhold an arm movement. When their activity changed before the onset of movement in response to the target presentation but did not change in response to the test visual stimulus, we regarded the activity as movement-related. We did not apply the test visual stimulus during the preparation period after instruction signal presentation for two reasons. First, if the stimulus is presented during the preparation period, the monkeys would initiate a movement in response to the stimulus because there was no difference between the stimulus and the target. Second, we could use a stimulus with a different property (e.g., color) for it. In that case, however, the monkeys are required to withhold the movement, and we could obtain neuronal responses to "NoGo" signals as we have previously found in the supplementary motor area (Kurata and Tanji 1985) and in the ventral and dorsal premotor cortex of the same monkeys (unpublished observations). In the two monkeys the eye movements of which were monitored, we confirmed that the visual stimulus was presented while the eyes were open. When the eyes were open, recorded eye position traces showed continuity (Fig. 4A). When the eyes were closed, the traces were discontinuous and located at a default extreme position.
|
|
|
|
 | (1) |
An LI value of 1.0 indicates that the neuronal activity is associated with contralateral arm movements only. An LI value of –1.0 shows movement-related neuronal activity immediately before and during ipsilateral arm movements only. Neuron activity with an LI value of 0 is equally active immediately before and during contra- and ipsilateral arm movements. When the neuronal activity associated with contralateral movements is 50% greater than that associated with ipsilateral movements, the LI would be 0.20. Using ANOVA (SYSTAT for Windows, version 8.0.2, Chicago, IL), we statistically compared laterality by examining the discharge rates during contra- versus ipsilateral movements for each neuron during the RT and MT, after matching sampling numbers for the targets and the arm used (see Figs. 3 and 5). Bilateral activity was defined as the situation in which there was no statistically significant difference in the discharge rate between contra- and ipsilateral forelimb movements (ANOVA, P > 0.01). For statistically significant differences (P < 0.01), the contra- or ipsilateral type of activity was defined, depending on the positive or negative LI value, respectively. For each movement-related activity, the contra- or ipsilateral arm that exhibited a higher absolute LI value during the RT was termed the preferred arm, and the other was called the nonpreferred arm.
After defining the preferred and nonpreferred arms, we analyzed the neuronal activity during each of the trials with the preferred and nonpreferred arms to determine whether it reflected visual or motor space. We classified the neuronal activity using a previously published statistical method (Kurata and Hoshi 2002). The mean discharge rates during sampling times (either RT or MT) of each trial were compared statistically, using the general linear model (GLM) of SYSTAT for Windows. First we selected neurons that showed activity dependent on target location in either the visual or motor coordinates because, without variation, the activity cannot be judged as reflecting visual or motor space. Second, to determine whether the activity reflected visual or motor space and/or prism effects, we statistically compared the neuronal activity during the RT or MT at visual and motor target locations (Fig. 1D) in the three prism conditions, using the following linear regression models
 | (2) |
 | (3) |
Throughout the analyses using Eqs. 2 and 3, we classified only activity with statistically significant variation dependent on target location (targetv or targetm) to judge whether the activity reflected visual or motor space (Kurata and Hoshi 2002). Otherwise, neuronal activity was termed unclassified (e.g., activity with high laterality during the nonpreferred arm trials, as shown in Fig. 5, D–G). If the factor analysis for prism was not statistically significant in Eq. 2 but was statistically significant in Eq. 3 and if the factor analysis for targetv was statistically significant in Eq. 2, then the activity was judged to be consistently dependent on target location for the visual coordinate only regardless of the presence or absence of prisms (termed V-type). Similarly, if the factor analysis for prism was not statistically significant in Eq. 3 but was statistically significant in Eq. 2 and if that for targetm was statistically significant in Eq. 2, then the activity was judged to be consistently dependent on target location for the motor coordinate only (termed M-type). If the factor analysis was statistically significant in both equations, then the activity was considered to be differentially active for both coordinates (termed B-type). Finally, if the factor analysis was not statistically significant in either equation, then the activity was deemed to be not differentially active for the two coordinates (termed N-type).
We classified neuronal activities associated with the preferred or nonpreferred arm into the four activity types described in the preceding text and computed the correlation coefficients of each activity to the visual and motor coordinates (rvisual and rmotor, respectively) (Kurata and Hoshi 2002), using the following formula
 | (4) |
In the calculation for rvisual, xi is the average firing rate while an animal reached for a visual target i under the no-prism condition, yi is the rate while reaching for the same target under either the left- or right-prism condition, 
is the average of xi, 
is the average of yi, and n is the number of targets (9) that overlapped under the two prism conditions. For the similar calculation of rmotor, n is the number of motor targets (6) that overlapped under the two prism conditions (Fig. 1D). A value of 1 for the coefficient rvisual indicates that the neuronal activity perfectly reflected visual space. In this case, the value for rmotor in the neuron should be close to 0. On the other hand, if the value of rmotor is 1, then the activity reflects the motor space. We sought to determine which visual frame of reference, head-centered or retinocentric, the V- and other types of neuronal activity reflected. For this purpose, a 10 x 10° grid on the retinocentric frame was used to define target areas (Fig. 4C) to compute coefficients (rretina) using Eq. 4, and we compared them with rvisual of the neurons. We rectified the digitized EMG activity, and analyzed it in the same manner as the neuronal activity.
Histology
When all the experiments were completed, electrolytic marking lesions were produced by passing 20 µA of cathodal DC through the microelectrodes for 15 s. Nine to 10 day later, the monkeys were deeply anesthetized with pentobarbital (50 mg/kg im) and were perfused through the heart with saline, followed by a fixative containing 3.7% formaldehyde in 0.1 M phosphate buffer, pH 7.4, which was followed by 10 and 20% sucrose solutions in 0.1 M phosphate buffer, pH 7.4.
After marking the location of the recording chamber with five pins at known electrode coordinates, the brain was removed from the skull and photographed. Later it was sectioned serially at 50 µm in the frontal plane using a freezing microtome. The PMv was defined as the area within the dysgranular frontal cortex rostral to the MI and lateral to the arcuate spur, where intracortical microstimulation (pulse width: 0.2 ms, frequency: 333 pulse/s, and train pulses: 12) in layer III or V did not evoke muscle activity at an intensity <50 µA (Barbas and Pandya 1987; Hoshi and Tanji 2002, 2006; Kurata 1993, 1994; Kurata and Hoffman 1994; Kurata and Hoshi 1999, 2002; Matelli et al. 1985).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
Each monkey adapted to the prisms, typically within 10–15 trials (Kurata and Hoshi 1999), and hit the targets correctly with relatively constant RTs in every prism condition regardless of the arm moved, prism condition, or target location. The average horizontal errors from the center of the targets were between 3.6 and 6.6 mm. The RTs ranged from 275 ± 55 to 365 ± 56 ms. The MT to a given motor target did not vary with the prism condition, although the MT did vary depending on the target location, ranging from 269 ± 52 to 509 ± 71 ms.
EMG analyses showed that shoulder muscles were movement-related, but no systematic changes in activity were detected during the other task periods. Of the muscles recorded, the anterior deltoid was a prime mover for the reaching movements in all three monkeys. As shown in Fig. 2, A–C, the right anterior deltoid muscle of monkey 2 was active immediately before and during reaching movements of the right limb but did not show even slight activity changes when the limb was required to not move during instructed delay periods. The muscle did not show activity change throughout the left arm trials (Fig. 2, D–F). The movement-related EMG activity in association with the right forelimb was similarly active at the same motor target location, regardless of the prism condition and thus depended on motor, but not visual, target locations (Fig. 2G). The EMG activity was classified as M-type (see METHODS). The LIs (see METHODS) during the RT and MT of the muscle were 0.76 and 0.69, respectively (Table 1, monkey 2). Other shoulder muscles, such as the rhomboid, pectralis major, and teres major, were active during the task, and the majority of their activities were classified as M-type. They were selectively active before and during the movements of the limb to which they belonged, and the LIs of the EMG activities were 0.23–0.92. Other proximal and distal forelimb muscles changed activity after the onset of reaching movements. The muscle activities during the RT and MT were classified as M- and other types (B- and N-types, see METHODS), but no activity reflecting visual coordinates (V-type) was found in any of these muscles. The classified EMG activities, along with their LIs, are summarized in Table 1. When an arm contralateral to the recorded side was used to reach for a target, the muscles showed irregular activities, which were unclassified.
|
Neuronal activity was recorded from 545 and 374 neurons in the PMv and MI, respectively, of the three monkeys. Of these, 202 PMv and 141 MI neurons with movement-related activity were well isolated long enough to analyze their relation to a visual or motor frame of reference under different prism conditions. The neuronal activities during RT and MT were classified as V-, M-, B-, or N-type after their laterality was analyzed (see METHODS).
Movement-related neuronal activity dependent on location of visual targets
Figure 3 shows an example of movement-related activity recorded in the left PMv that was dependent on the location of visual, but not motor, targets. The LI (see Eq. 1 in METHODS) of the activity during the RT was 0.02, indicating nearly the same activity during right and left arm movements. As the LI had a positive value, the right (contralateral) and left (ipsilateral) arms were defined as the preferred and nonpreferred arms, respectively (see METHODS for definitions). The mean discharge rates during the RT of the preferred and nonpreferred arm trials are indicated in Fig. 3G. When the right arm was used, the neuronal activities (Fig. 3, A–C and G, top left) toward the same visual targets were similar under the three prism conditions, whereas the activities toward the motor targets differed (Fig. 3G, top right). Using statistical criteria (see METHODS), the activity during the RT in the preferred arm trials was classified as V-type, reflecting the visual frame of reference. When the left arm was used for reaching (Fig. 3, D–F and G, bottom), this neuron showed a pattern of activity that depended on the location of the visual target, similar to the finding for the right arm. The neuronal activity during the RT in the nonpreferred arm trials was also classified as V-type.
When the neuronal activity shown in Fig. 3 was recorded, eye positions were also recorded. Figure 4A shows four eye position traces (2 traces each under no and left prism conditions) along with behavioral events when the monkey initiated reaching movements toward the same visuospatial target (V08). In two of the four trials (1 trial each under no and left prism conditions), the test visual stimulus was applied at V08 immediately after initiation of the trials. We confirmed that the neuron changed its activity when reaching movements were initiated in response to the target presentation (V08) but did not respond to the same stimulus was applied 200 ms after trial initiation for 100 ms under no prism and left prism conditions (Fig. 4A). We assume that the monkey detected the test stimulus because the eye position was nearly at the center when the test stimulus was applied, and the eyes were open, judging from the eye position traces. Thus we regarded the activity as movement-related as well as other neurons with V-type activity in the PMv and MI (METHODS). Figure 4A also shows that the activity was not related to saccadic eye movements and that the eyes were located at variable positions when the same target (V08) was presented, although eyes were frequently located at the central position as indicated by the peaks on the histograms in Fig. 4B. Figure 4C shows target locations on a retinocentric coordinate, calculated from the data shown in Fig. 4B. The trends shown in Fig. 4, B and C, were similar in monkeys 2 and 3 throughout recording sessions. Based on the target locations on the head-centered and retinocentric coordinates, contour plots of the neuronal activity during the RT shown in Fig. 3 were created (Fig. 4, D and E). The contour plot of the head-centered coordinate seems smoother than that of the retinocentric coordinates. This observation was confirmed in that the correlation coefficient of the head-centered coordinate (rvisual) of the activity was 0.82, whereas that on the retinocentric coordinate (rretina, see METHODS) was 0.08. For 15 V-type neurons in the PMv of the two monkeys the eye movements of which were monitored, rvisual was always higher than rretina. Mean rvisual and rretina values of the V-type activity during the RT were 0.81 and 0.10, respectively; the difference was statistically significant (Student's t-test, P < 0.01). The results indicated that the V-type activity reflected head-centered visual space better than retinocentric space.
Movement-related neuronal activity dependent on location of motor targets
Figure 5 shows another example of a movement-related neuron recorded in the left PMv. This neuron changed its activity during the RT when the monkey used his right arm contralateral to the recorded hemisphere. Thus the right arm was defined as the preferred arm. The activity patterns during the RT in the preferred arm trials were similar toward the same motor target in the presence and absence of the prism, whereas the same neuron was differently active toward the same visual target under the three prism conditions. We determined statistically that the activity was closely related to the motor frame of reference and was thus classified as M-type (see METHODS). This neuron was preferentially active when the contralateral right arm (preferred arm) was moved and was nearly silent during left (nonpreferred) arm movements (Fig. 5, D–F and G, bottom). The LI of the activity during the RT was 0.72. When the nonpreferred arm was used, the activity was not classified as any type (V-, M-, B-, or N-type) and was thus termed unclassified. The activity pattern was similar to that of the prime mover muscle (anterior deltoid). Other than the V- and M-types, we recorded movement-related activities that reflecting neither visual nor motor coordinates, which were classified as B- and N-types, according to Eqs. 2 and 3 in METHODS (cf. Figs. 9 and 10 of Kurata and Hoshi 2002).
Classified movement-related activities during the RT and MT
Tables 2 and 3 summarize the numbers of neurons classified as V-, M-, B-, and N-types in the PMv and MI of the three monkeys during the RT and MT. Throughout the RT and MT in the preferred arm trials, the most remarkable distinction between the PMv and MI was the proportion of neurons with V-type activity. During the RT, the neurons with V-type activity constituted 11.4% in the PMv but only 0.7% in the MI (1 neuron), whereas M-type neurons were more numerous in the MI (23.4%) than in the PMv (17.3%). In the two areas, B- and N-type neurons were more numerous than V- and M-type neurons. The trends were similar in the two areas during the MT (Table 2).
|
|
). Of the 202 PMv and 141 MI neurons with activities categorized into the four types during the RT of preferred arm trials, 152 (75.2%) and 122 (86.5%) neurons, respectively, were also active during the MT of the trials, demonstrating the presence of neurons that were specifically active during the RT but inactive during the MT (Fig. 5). The proportion of each classification type changed greatly between the RT and the MT (Table 3). Of the 23 V-type neurons active during the RT, 4 neurons remained as V-type and 19 changed to other types during the MT. On the other hand, about half of the neurons that were M-, B-, and N-types during the RT were classified as that same type during the MT, and the rest were classified as other types. Notably, no neuron classified as M-type during the RT was classified as V-type during the MT.
Laterality of the classified movement-related activities during the RT and MT
Figure 6 shows the histograms of the LI values for each type of movement-related activity in the PMv and MI during the RT and MT of the preferred arm trials. We found statistically defined contra-, ipsi-, and bilateral activities (see METHODS) for each of the V-, M-, B-, and N-types; the frequencies are shown in Fig. 6. Each of the V-, M-, B-, and N-type activities exhibited a wide variety of LIs, between –1.0 and 1.0. In the PMv, however, M-type activities during the RT exhibited the highest mean LI (0.55, see Table 4), and the differences in LI among the four classified types were statistically significant (Bonferroni, P < 0.01). During the RT, 39% of the V-type neurons (9/23) showed bilaterality, whereas 61% of them exhibited laterality (Fig. 6). By contrast, a vast majority of the M-type neurons in the PMv exhibited laterality (33/35). The bilateral type of activity was seen in 21.7 and 44.2% of the B- and N-type neurons, respectively. Multiple statistical comparisons among the four types of activity showed that the M-type activities during the RT had significantly larger LIs than the V-type (Bonferroni, P < 0.01), although the LIs of the V-type activities were similar to those of the B- and N-type activities (Bonferroni, P > 0.01). The trend is shown in cumulative sum plots of the absolute LI values of the V-, M-, B-, and N- types during the RT (Fig. 7A). During the MT, V-type activities exhibited the lowest mean LI (0.11, Table 4) in the PMv, and the differences in LI among the four classified types were statistically significant (Bonferroni, P < 0.01). In the MI, on the other hand, M-type activities exhibited relatively higher LIs than the B- and N-types during the RT and MT. During the RT, the V-type activity was classified as bilateral type, whereas 66.7, 9.1, and 24.2% of the M-type neurons were classified as contra-, ipsi-, and bilateral types, respectively. The contra-, ipsi-, and bilateral types of activity were seen in 74.1, 11.1, and 14.8% of the B neurons and in 56.6, 30.2, and 13.2% of the N neurons, respectively. The trends were similar during the MT (Fig. 6).
|
|
|
|
We performed a statistical discrimination analysis (Systat) using the three parameters, rvisual, rmotor, and LI to examine whether the four classified types (V-, M-, B-, and N-types) could be statistically discriminated. The four classified types were significantly discriminated during each period of RT and MT in both the PMv and MI (P < 0.01). The means and SDs of the values, rvisual, rmotor, and LI are summarized in Table 4. Other than the statistically significant difference in LI between the classified types described in the previous section, the PMv V-type activities during the RT and MT had significantly higher rvisual values than the other activity types (Bonferroni, P < 0.01), whereas the M-type activities of the PMv and MI during the RT and MT had significantly higher rmotor values than the other activity types (Bonferroni, P < 0.01). For the V-, M-, B-, and N-types during the RT recorded in the PMv and MI of monkeys 2 and 3, the overall mean of the rretina values (METHODS) was 0.06, and the rretina values were significantly lower than the rvisual and rmotor values throughout the activity types (ANOVA, P < 0.01). Accordingly, we did not analyze the rretina values further.
To characterize the V-, M-, B-, and N- types of activity, we focused on activity during the RT of the preferred arm trials and created their cumulative sum plots using LI, rvisual, and rmotor. In addition to the absolute values of the LIs (Fig. 7A), Fig. 7, B and C, shows the cumulative sum plot of the rvisual and rmotor values of the four types. For rvisual (Fig. 7B), the B-type activities exhibited values between the V- and M-types in the PMv and MI, whereas the curves of the N- and M-types largely overlapped in the PMv, although the curve of the N-type was located between the B- and M-types. For rmotor (Fig. 7C), the B- and N-types exhibited values between the V- and M-types in the PMv and MI.
Using the three values (rvisual, rmotor, and |LI|), we introduced another index to characterize the types, defined as rvisual x (1 – rmotor) x (1 – |LI|). The index becomes closer to 1 if the activity reflects visual space and bilaterality, whereas it is closer to 0 if the activity reflects motor space and laterality. The cumulative sum plots of the index (Fig. 7D) show that the V- and M- types were distinguished from the B- and N-types, although curves of the B- and N-types largely overlapped.
We analyzed the lead time from the onset of the V-, M-, B-, and N-type activities to movement onset of the preferred arm (see METHODS). Table 6 shows the mean lead times (ms) of each classified type during RT in the PMv and MI of the three monkeys. The lead times of the movement-related activities in the PMv were significantly longer than those in the MI (ANOVA, P < 0.05). In both the PMv and MI, the difference in lead time among the V-, M-, B-, and N-types was not statistically significant (ANOVA, P > 0.05), except for the M-type in the MI of monkey 1, the lead times of which were significantly shorter than the B- and N-types in the region (Bonferroni, P < 0.05).
|
Figure 8 shows the distribution of movement-related neurons, classified by activity during the RT of the preferred arm trials, in the PMv and MI of monkey 2. In the MI, the neurons were located primarily in the proximal forelimb and trunk representation areas identified by intracortical microstimulation (see METHODS). In the PMv, the majority of these neurons were located in the caudal part of the PMv close to the border between the PMv and MI. The V-, M-, B-, and N-type neurons were similarly distributed in the PMv and MI (Fig. 8, B–E). Figure 8, F–H, shows the distribution of the movement-related neurons with laterality preference (contra-, ipsi-, bilateral), as determined by ANOVA (see METHODS). In the distribution maps, the neurons with each classified activity were distributed in parallel with the density of the total classified neurons (Fig. 8A); we could not discern any tendency toward clustering from surface views. The classified neurons were recorded mainly in layers III and V and were intermingled without any particular tendency for the locations. The classified neurons were located similarly in all three monkeys during the RT and MT of the preferred arm trials, and the distributions were consistent with those reported previously (Kurata and Hoshi 2002). We also recorded a number of movement-related neurons with activity that changed after the movement onset but did not precede it. The majority of these were located in the distal representations of the MI and PMv and were not included in the present study.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
) that movement-related neuronal activities can be statistically classified into several types: activity closely reflecting either visual space (V-type) or motor space (M-type) and activity reflecting both (B-type) or neither (N-type) frame(s) of reference. Our new findings in this study are that the V-type activity is likely to reflect head-centered visual space and that the V-type neurons exhibit significantly less laterality than the M-type. Furthermore, the B- and N-types exhibit intermediate properties between the V- and M-types, judging from laterality and correlations to visual and motor space. In the following sections, we will discuss functional significance of the different activity types and our view how transformation of coordinates from visual to motor space is processed in the PMv and MI. Functional significance of the V- and M-types of movement-related activity
We found that the V-type activities most frequently recorded in the PMv were likely to reflect a head-centered, rather than retinocentric, visual frame of reference. They were regarded as "movement-related" but not "signal-related" because they were active when the monkeys initiated movements in response to target presentation as "go" signal; however, they did not respond to the same visuospatial stimuli when they were presented immediately after trial initiation. Most importantly, the V-type activity showed significantly less laterality than the M-type. Thus we suggest that V-type movement-related activity could be emerged from visuospatial inputs (Boussaoud and Wise 1993a; Fogassi et al. 1996) and represents a stage between the visuospatial inputs and final motor commands. However, it should be noted that, although 39% of the V-type neurons showed similar activity in association with contra- as well as ipsilateral arm movements, 61% of them exhibited laterality. We suggest that the former V-type activity may reflect a stage in the transformation process that occurs before final motor commands are generated by specifying the dynamics of either right or left forelimb movements, whereas the latter may be more processed for final motor commands, although both of them convey visuospatial information on target locations.
M-type activity likely reflects a motor frame of reference very similar to those of EMGs in prime movers. The correspondence strongly suggests that the M-type neuronal activities in the cortical motor areas represent final motor commands for controlling muscle activities in reaching. We also observed that the entire population of the classified movement-related activities showed a preference for contralateral reaching movements, although a number of neurons were also active in association with ipsilateral arm movements (Cisek et al. 2003; Evarts 1966). This suggests that M-type activity not only conveys final motor commands but also plays a role in sensorimotor integration for coordinate transformation. Such activity may contribute to activation of the ipsilateral cortical motor areas, including the MI, in human subjects (Hanakawa et al. 2005; Kansaku et al. 2005; Kim et al. 1993; Kollias et al. 2001; Kurata et al. 2000).
Functional significance of the B- and N-types of movement-related activity
We found that among the whole population of movement-related activities in the PMv and MI, the majority was classified as either the B- or N-types. Our analysis of the B- and N-type activities revealed that the classified types exhibited intermediate properties between the V- and M-types when laterality and correlations to visual or motor space were jointly analyzed. Because the movement-related activities were classified based on statistical criteria (METHODS), it is possible that the number of the classified activities could change if we used different level of statistical significance. However, it is clear that the PMv and MI contain a number of neurons without a close relationship to visual or motor coordinates. The presence of the neurons can be interpreted as follows.
First, the B- and N-types could convey information on motor space or represent final motor commands because similar activities were found in EMGs or the upper arm and shoulder. In reaching, vectors of muscle activities other than prime movers may not necessarily be directed toward motor target locations. We found that some classified activities during the RT changed their type during the MT and that shifts in laterality of the classified neuronal activities occurred during the sampling periods. However, this does not necessarily mean drastic changes in the properties of each movement-related neuron. These changes may be partially explained by the nature of the reaching movement, which requires a spatially and temporally complex response and orchestrated muscle activities at multiple joints.
Second, the intermediate properties of the B- and N-type activities between the V- and M-types may reflect an intermediate stage of processing that is necessary for the coordinate transformation, and they could dynamically modify relationship between visual and motor space, namely in case of prism adaptation. Supporting this idea, we previously found that transient inactivation of the PMv resulted in reacquisition deficits in prism adaptation (Kurata and Hoshi 1999).
Finally, the B- and N-types may represent noisy, but functionally efficient, elements in the centrally generated motor commands necessary for achievement of overall accuracy in reaching (Harris and Wolpert 1998; Todorov 2004). If the B- and N-types contributed to motor performance in this way, they could also play a role in the prism adaptation by optimizing their efficiency in the control system.
Information flows for visuospatial transformation in the PMv and MI
Our main findings give support to the view that a series of transformations from a visuospatial (extrinsic) frame of reference to representations of muscle activation patterns in an intrinsic coordinate space occurs in the primate brain to generate goal-directed movements (Alexander and Crutcher 1990; Andersen et al. 1993; Atkeson 1989; Kawato et al. 1988; Soechting and Flanders 1992) and that the PMv and MI contribute to these transformations (Kakei et al. 1999, 2001; Kurata and Hoshi 2002; Schwartz et al. 2004). Our data show that during the RT, V-type activity, reflecting a head-centered visual frame of reference, was preferentially recorded in the PMv. By contrast, the M-, B-, and N-types were distributed in the proximal representation of the PMv and MI (Gentilucci et al. 1988; Kurata and Tanji 1986) but were not clustered in selective regions within the two cortical motor areas. Furthermore, our analysis on lead times of the classified types before movement onset showed that there was no statistically significant difference in lead times between the classified types when they were compared within each cortical area; however, there was a statistically significant difference in lead times between the PMv and MI. Accordingly we assume that information processing for coordinate transformation may not take place solely within local circuits in the PMv or MI in a serial and hierarchical manner. Instead information at various stages is transferred and processed in a direction from the PMv to the MI in a parallel distributed manner, based on our observations on location of the activity types and difference in neuronal lead time between the areas. Consistent with this, it has recently been reported (Rubino et al. 2006) that high-frequency oscillations in the beta range were propagated from the rostral to caudal parts of the MI in the direction of anatomical connections from the PMv to the MI (Kurata 1991; Muakkassa and Strick 1979). Although it is unclear whether the propagating waves mediate information transfer across the different cortical motor areas, the anatomical linkage between the PMv and MI provides a potential pathway for information flow. This idea is supported by a report that electrical stimulation of the PMv exerted powerful facilitation of MI outputs to upper limb motor neurons in the spinal cord (Shimazu et al. 2004).
|
GRANTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
|
FOOTNOTES |
|---|
Address for reprint requests and other correspondence: Dept. of Physiology, Hirosaki University School of Medicine, Hirosaki 036-8562, Japan (E-mail: kkurata-ns{at}umin.net)
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
Andersen RA, Snyder LH, Li CS, Stricanne B. Coordinate transformations in the representation of spatial information. Curr Opinion Neurobiol 3: 171–176, 1993.[CrossRef][Medline]
Atkeson CG. Learning arm kinematics and dynamics. Annu Rev Neurosci 12: 157–183, 1989.[CrossRef][ISI][Medline]
Barbas H, Pandya DN. Architecture and frontal cortical connections of the premotor cortex (area 6) in the rhesus monkey. J Comp Neurol 256: 211–228, 1987.[CrossRef][ISI][Medline]
Boussaoud D, Wise SP. Primate frontal cortex: effects of stimulus and movement. Exp Brain Res 95: 28–40, 1993a.[ISI][Medline]
Boussaoud D, Wise SP. Primate frontal cortex: neuronal activity following attentional versus intentional cues. Exp Brain Res 95: 15–27, 1993b.[ISI][Medline]
Cisek P, Crammond DJ, Kalaska JF. Neural activity in primary motor and dorsal premotor cortex in reaching tasks with the contralateral versus ipsilateral arm. J Neurophysiol 89: 922–942, 2003.
Evarts EV. Pyramidal tract activity associated with a conditioned hand movement in the monkey. J Neurophysiol 29: 1011–1027, 1966.
Fogassi L, Gallese V, Fadiga L, Luppino G, Matelli M, Rizzolatti G. Coding of peripersonal space in inferior premotor cortex (area F4). J Neurophysiol 76: 141–157, 1996.
Gentilucci M, Fogassi L, Luppino G, Matelli M, Camarda R, Rizzolatti G. Functional organization of inferior area 6 in the macaque monkey. I. Somatotopy and the control of proximal movements. Exp Brain Res 71: 475–490, 1988.[CrossRef][ISI][Medline]
Hanakawa T, Parikh S, Bruno MK, Hallett M. Finger and face representations in the ipsilateral precentral motor areas in humans. J Neurophysiol 93: 2950–2958, 2005.
Harris CM, Wolpert DM. Signal-dependent noise determines motor planning. Nature 394: 780–784, 1998.[CrossRef][Medline]
Hoshi E, Tanji J. Contrasting neuronal activity in the dorsal and ventral premotor areas during preparation to reach. J Neurophysiol 87: 1123–1128, 2002.
Hoshi E, Tanji J. Differential involvement of neurons in the dorsal and ventral premotor cortex during processing of visual signals for action planning. J Neurophysiol 95: 3596–3616, 2006.
Kakei S, Hoffman DS, Strick PL. Muscle and movement representations in the primary motor cortex. Science 285: 2136–2139, 1999.
Kakei S, Hoffman DS, Strick PL. Direction of action is represented in the ventral premotor cortex. Nat Neurosci 4: 1020–1025, 2001.[CrossRef][ISI][Medline]
Kansaku K, Muraki S, Umeyama S, Nishimori Y, Kochiyama T, Yamane S, Kitazawa
, mean LI (see 
