| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1Unité de physiologie et biomécanique de la locomotion, Département d'éducation physique et de réadaptation, Université Catholique de Louvain, 1348 Louvain-la-Neuve, Belgium; and 2Department of Physiology, Université de Montréal, Montreal, Quebec H3C 3J7, Canada
Submitted 10 December 2003; accepted in final form 26 May 2004
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
; Horak and Macpherson 1996; Massion 1992). Such APAs are particularly evident for large movements of the arm or trunk or when that limb supports the body (Belenkii et al. 1967; Béraud and Gahéry 1995; Bouisset and Zattara 1981; Brown and Frank 1987; Burleigh et al. 1994; Cordo and Nashner 1982; Crenna and Frigo 1991; de Wolf et al. 1998; Horak et al. 1984; Lee et al. 1990; Mouchnino et al. 1992; Nashner and Forssberg 1986). The postural adjustments that precede the movement, often by more than 100 ms, have been referred to as preparatory APAs (pAPAs; Gahéry 1987). There are also postural responses that occur during the movement and that serve to stabilize the body or body segment during the execution of the movement itself. These postural responses are also anticipatory in nature because they occur before there is any possibility of feedback from the movement itself influencing the response (Massion 1992). They are referred to by Gahéry (1987) as accompanying APAs (aAPAs).
The control mechanisms responsible for the production of these different types of postural response and for their integration with the movement are only poorly understood and the extent to which posture and movement are expressed either as a unity or as independent signals is a controversial issue (Massion 1992). There is certainly evidence to support some level of independence in the signals responsible for the production of the pAPAs and the movement (Brown and Frank 1987; Cordo and Nashner 1982; de Wolf et al. 1998; Horak et al. 1984; Massion 1992), including our own recent study (Schepens and Drew 2003a). However, it is difficult to determine whether the dissociation of timing in the pAPAs and the movement that we observed reflects independence of planning or independence at a later stage in the execution of the movement. In the case of the postural responses that accompany the movement, the data, at least in the cat, suggest that they are tightly coordinated with the movement (Alstermark and Wessberg 1985; Schepens and Drew 2003a) and might be initiated by a single motor command (see Gahéry and Nieoullon 1978; Kably and Drew 1998b).
The neural signals responsible for the production of the postural responses and for their integration with the cortical commands for movement are even less well understood. Nevertheless, there is abundant evidence from lesion studies to suggest that the brain stem reticular formation makes an important contribution to this function. For example, Kuypers (1963; Lawrence and Kuypers 1968) demonstrated that damage to the reticulo- and vestibulospinal tracts, in both cats and primates, leads to gross ataxia and a wide variety of balance problems while leaving fine control of the distal arm mostly intact. Similarly, during locomotion, damage to these same tracts also leads to problems in balance control as well as in weight support (Afelt 1974; Brustein and Rossignol 1998; Eidelberg 1981; Gorska et al. 1990). In addition, studies by Luccarini et al. (1990) have shown that injection of a cholinergic agonist into the pontine reticular formation reduces the postural responses elicited by cortical stimulation. This supports a role for the reticular formation in the integration of posture and movement, although it leaves open the nature of the contribution. Indeed, there have been very few studies that have examined the activity of neurons in the pontomedullary reticular formation (PMRF) during discrete voluntary movements (Gibson et al. 1998; although, for information in primates on the contribution of the mesencephalic reticular formation, see Stuphorn et al. 1999; Werner et al. 1997) and none that was directly related to the question of how the reticulospinal system contributes to the regulation of posture.
The present study was designed to address this issue by recording the activity of neurons in the PMRF, including reticulospinal neurons (RSNs), during a reaching task in the cat. Our recent kinetic and electromyographic (EMG) study of the strategy adopted by cats during a reaching task (Schepens and Drew 2003a) showed that the pAPAs followed the GO signal instructing the cat to make the movement at a relatively fixed latency and were temporally decoupled from the movement itself. Consequently, the latency of the onset of the movement itself showed no fixed relationship to the stimulus. This temporal decoupling provides an ideal theoretical framework in which to determine the extent to which neurons in the PMRF are related to different components of the overall reaching strategy. For example, changes in activity in neurons that contribute to the initiation of the pAPAs would be expected to be tightly linked to the stimulus for the movement (the GO signal) and would be temporally decoupled from the movement. Conversely, changes in activity in neurons related to the initiation of the movement, or to the postural responses that accompany the movement, would bear no temporal relationship to the stimulus and would covary with the onset of the movement. The results suggest that both types of neurons may be found within the same areas of the PMRF.
Preliminary results of these findings were previously published in abstract form (Schepens and Drew 2000, 2001, 2003b).
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Experiments were performed on the same 2 animals as used in a previous publication (Schepens and Drew 2003a). All training procedures and most surgical interventions were documented in that publication. In brief, the animals were trained to reach to a tube placed medially and at shoulder height and to retrieve a morsel of food. Between trials the tube was closed by an opaque shutter that was opened, under computer control, to provide access to the food (Fig. 1). After a period of training, the cats were anesthetized with isoflurane (2–3% with oxygen) and prepared for surgery. A stainless steel, rectangular plate (internal dimensions 8 x 6 mm) was attached to the cranium to form a recording chamber that gave access to the PMRF. Three microwires (50 µm diameter) were implanted in the lumbar spinal cord (L2) to allow antidromic identification of RSNs (Drew et al. 1986; Prentice and Drew 2001). Pairs of stainless steel wires were inserted into the bellies of selected muscles in all 4 limbs to record EMG activity. Note that the recording chamber was implanted over the left PMRF: EMGs from the left limbs are thus ipsilateral to the recording site. All procedures were approved by the institutional Ethics Committee and followed national guidelines.
|
Experiments were started 1 wk after the surgery. In each experimental session, a conventional, glass-insulated, tungsten microelectrode (impedance 0.5–1.5 M
), held in a custom-made microdrive, was advanced to a position just above the brain stem while the cat was held on the experimenter's lap. The cat was then transferred to a treadmill where it rested quietly while the electrode was advanced slowly through the brain stem. As the electrode was advanced, stimuli were applied to the most effective spinal electrode to help in isolating RSNs with low or no spontaneous activity. When a neuron was isolated, stimulation was applied to each of the spinal electrodes in turn to determine whether the cell could be activated antidromically from the spinal cord. Cells that were activated at constant latency from the spinal stimulation and that exhibited collisions with spontaneous spikes at appropriate latencies (Lipski 1981; see Fig. 1) were classed as RSNs. All other cells were classified as Unidentified neurons. Both classes of neurons were recorded during the task, providing that the recording was stable.
All isolated neurons, with the exception of those that showed only saccade-related activity (found in and close to the abducens nucleus), were recorded during a period of locomotion. The cat was then transferred to the apparatus used for the reaching task and neuronal activity was recorded while the cat stood with each foot on a force platform (Fig. 1; see also Schepens and Drew 2003a). When the cat was standing quietly, data collection was initiated while the cat reached with the left or the right forelimb. Sets of data in which the cat reached with the left or right limb were normally interspersed in blocks of 5 or 10 trials. During experimental sessions in which neuronal recordings remained stable, further data trials were recorded either with the reaching target placed at different heights or distances. The electrode was then advanced and another cell isolated. Experimental sessions lasted 2–4 h and ended when the cats were no longer willing to work for a food reward. In this paper, we report only on the neuronal discharge characteristics of the cells during reaching to a standard target position with the left, ipsilateral limb. In cat RS22 the target was normally either 29 (30% of recordings) or 27 (69%) cm from the center of the front force-platforms and at a height of 22 cm (81% of recordings), wereas in cat RS23 it was 22 cm (96% of recordings) from the center of the front force-platforms and at a height of 16 cm (100% of recordings).
All neuronal and EMG data were recorded on-line during the experiments together with the forces (F) and moments exerted by the cat against the platform in all 3 planes (subscripts: V, vertical; AP, anteroposterior; and ML, mediolateral). Force and EMG data were sampled at 1 kHz. Neuronal data were sampled at 100 kHz. Video recordings were made of all experiments and a digital time code recorded simultaneously on the computer and on the video recordings allowed these recordings to be synchronized to the EMG and neuronal recordings.
Data analysis
Because of the inherent incertitude in ensuring adequate on-line cell discrimination during experiments in unrestrained animals, all neuronal recordings were discriminated off-line. The recordings of neuronal activity were displayed on a computer monitor and an interactive, time-amplitude, software-discrimination routine was used to identify individual neurons. Because only neurons that appeared well isolated during the experiments were recorded, no special spike-isolation routines were necessary. Normally, only one cell was isolated from each record; in exceptional cases 2 cells were sometimes clearly identifiable using time-amplitude discrimination. Given that the goal of the analyses in these experiments was to determine the relationships between cell activity and behavior, we rejected from the analysis only those trials in which the cat began the movement before the GO signal (end of tone and opening of shutter). Variability in the reaction time and in the force levels was advantageous for performing regression analyses between different aspects of the neuronal discharge activity and the behavioral measures (see following text). Force data were filtered and the bias removed as described in our previous publication (Schepens and Drew 2003a). Only neurons for which a minimum of 5 reaches with the limb ipsilateral to the recording site were included in the database.
Initially, each trial was examined individually and the onset and offset of the period of activity in selected flexors and extensors was marked using a cursor and interactive software; we also marked the onset and offset of the periods of cell activity, when they were sufficiently discrete. The onset of activity in the ipsilateral cleidobrachialis (iClB) and in the ipsilateral brachialis (iBr) muscles was always measured because one or the other of these simultaneously activated muscles was used as the standard indicator of movement onset (Prentice and Drew 2001; Schepens and Drew 2003a). Raster displays and postevent histograms (PEHs) of instantaneous frequency (Drew and Doucet 1991; Udo et al. 1982) were always generated with respect to the opening of the shutter (GO signal) and to the onset of the iClB. Rasters and PEHs were sometimes generated for other events and muscles depending on the nature of the discharge activity. Scatterplots and linear regressions were automatically generated for the time of onset and offset for the cell discharge as a function of the time and offset of each recorded muscle.
We always calculated linear regressions for both the onset of cell discharge and lead time as a function of the onset of movement. In an analogous manner to our analysis of the relationship between the onset of EMG activity and the onset of the APA in our previous paper (Schepens and Drew 2003a), lead time was defined as the onset of movement minus the onset of the variable under study (e.g., onset of cell discharge or onset of initial force change) (see Fig. 2). This latter relationship was used to determine whether the onset of cell discharge, or of the respective variable, was better related to the GO signal or to the onset of the movement itself. A slope of 1.0 for lead time versus movement onset indicates that the variable is better related to the GO signal, whereas a slope of 0.0 indicates a better relationship to the movement (Chapman et al. 1986; Vicario et al. 1983) (Fig. 2). Only cells for which the regression was significant (P < 0.05) were classified as being unambiguously related to either the GO signal or the movement.
|
). For cat RS22, force onset had to begin within 210 ms of the GO signal and for cat RS23 within 112 ms. To ensure that average values were not unduly biased by slow movements we also calculated the mode and interquartile ranges for the same data set for the onset of activity in the iClB. On this basis, to be included in the averages, the onset of iClB had to be <563 ms in cat RS22 and <400 ms in cat RS23. No limits were used when compiling any of the linear regressions.
|
50 ms (if we examined the relationship for the whole trial). Within each bin, we integrated the EMG activity or the force data and we calculated the average instantaneous cell-discharge frequency within this bin. Linear regressions were then automatically made between the intensity of the cell discharge and the magnitude of all of the EMG and force traces that were recorded (
64 measures in all). These were then rank-ordered according to the coefficient of determination. For the purposes of classifying cells, neuronal discharge activity was synchronized with respect to the onset of activity in the iClB of the reaching limb and averaged. This averaged trace was then filtered using the same digital filter as that for the EMGs (see Fig. 4). The initial classification of neurons into those that showed or did not show a significant change in discharge activity during the task was based on whether cell discharge after the GO signal significantly deviated from the control level of discharge (increase or decrease of ±2SD). Cells that showed no significant deviation from control levels were classified as unresponsive; cells that showed significant changes in activity were classified either as cells showing an increase or a decrease in discharge according to the sign of the initial significant change.
Cells exhibiting a significant increase in discharge frequency were further objectively classified as discharging phasically, tonically, or with a phasic/tonic pattern (see Fig. 4). As a basis for this classification, we automatically measured cell discharge at 3 defined times during the trial: 1) The level of control activity was measured for a 1-s period beginning 2 s before iClB onset (control period). 2) Dynamic activity during the reach was defined as the overall maximum cell discharge that occurred from the onset of the period of activity in iClB to the time that the paw entered the tube (dynamic period); in situations in which information on the time of entry into the tube was not available we used a fixed movement time of 300 ms. 3) The level of cell discharge after the reaching movement, indicating the level of maintained, or tonic, discharge was measured for a period of 500 ms beginning 1 s after iClB onset (static period). Phasic cells (Fig. 4A) generally exhibited a ratio of tonic level/control level of <150%. Because many of these cells had very low levels of activity during the control period, we also included in this category cells in which the discharge frequency during the static period did not exceed that during control by more than 10 spikes/s. Tonic cells (Fig. 4B) exhibited a ratio of tonic/control level of >150% and showed a maintained discharge after the time that the paw reached the tube in which the ratio of the maximum discharge during the movement/tonic period was 125%. Phasic/tonic cells (Fig. 4C) equally exhibited a ratio of tonic/control level of >150% and exhibited a ratio of movement/static period of >125%.
Histology
At the end of the recording sessions, electrolytic lesions (20–50 µA) were made at selected regions within the recording area. The cats were deeply anesthetized and perfused through the aorta. The brain stem was removed and processed before being stained with cresyl violet. The location of each penetration was traced onto standard sections of Berman's atlas (1968) using the electrolytic lesions as a guide. This allowed us to determine the approximate stereotaxic location of each of the recorded neurons (see Fig. 3).
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
The reaching task that we used in this study required that the cat adjust its posture before initiating the movement. This can be seen from inspection of Fig. 1, which shows selected, averaged force and EMG data during a left (ipsilateral to the recording chamber) reach. As detailed in a previous publication (Schepens and Drew 2003a), the reach is preceded by a loading in the ipsilateral reaching limb (iFLV) and an unloading in the contralateral, supporting, limb (coFLV) that define the preparatory anticipatory postural responses (pAPAs) (Fig. 2). Their presumed function is to produce a transfer of the center of mass over the supporting limbs, ensuring stability and contributing to the initiation of the movement (Ioffé et al. 1982; Schepens and Drew 2003a). There is also a loading of the ipsilateral hindlimb (iHLV) and an unloading of the contralateral hindlimb (coHLV). All of these changes are actively produced as they are preceded and accompanied by EMG changes in the appropriate muscles. Figure 4D, for example, illustrates the change in activity in the iTriL that precedes the first detectable change in force in the reaching limb (iFLV). Similarly, the decrease in activity in the coTriL precedes the decrease in force observed in the respective force recording (coFLV).
In our task, we define the onset of movement as coinciding with the onset of activity in the iClB or iBr. As demonstrated in Schepens and Drew (2003a), onset of activity in these muscles is tightly correlated to the lift of the limb (force under the paw drops to zero) and precedes lift by an average of 83 ms in cat RS22 and by 64 ms in cat RS23. The onset of activity in the iClB is accompanied by an increase in activity in the extensor muscles of the supporting forelimbs (see coTriL). The change in activity in the iClB and the coTriL is coincident and strongly correlated (Alstermark and Wessberg 1985; Schepens and Drew 2003a). These postural changes in the supporting forelimb thus occur as feed forward signals that anticipate the destabilization produced during the movement. Following Gahéry (1987) and Massion (1992), we refer to these changes as anticipatory postural responses that accompany the movement (aAPAs). Note that some muscles (e.g., coTriL) may contribute to both the preparatory responses before the movement as well as to the postural responses that accompany the movement (Schepens and Drew 2003a).
Database and general characteristics
A total of 142 neurons (89 in cat RS22 and 53 in cat RS23), histologically determined to have been recorded within the PMRF, and which fulfilled all of the criteria detailed in METHODS, were recorded during ipsilateral reach. Of these 142 neurons, 60 (42%) were identified as RSNs.
The location of these neurons is illustrated in Fig. 3, A and B on a standard section taken from the atlas produced by Berman (1968). The majority of the neurons was recorded within the nucleus reticularis gigantocellularis (NRGc), primarily in the more rostral regions. A smaller proportion of cells was recorded from the caudal region of the nucleus reticularis pontis caudalis (NRPc) and from within the nucleus reticularis magnocellularis (NRMc). Both RSNs and Unidentified neurons were recorded throughout the explored area and all neurons were located within 1.6 mm of the midline. Within the population of RSNs, the majority of the neurons that we recorded had axons that conducted at velocities >90 m s–1 (mean 98 m s–1); similar values were obtained in each cat (Fig. 3C). As such, in terms of both location and conduction velocity of axon, the characteristics of this database are generally similar to those used in our previous studies of PMRF function during locomotion (Drew et al. 1986; Prentice and Drew 2001), albeit with a smaller proportion of cells from the NRPc. Differences in the populations of RSNs and Unidentified neurons, when present, are addressed later.
The majority of the neurons that we recorded (98/142, 69%) exhibited a significant increase in their cell-discharge activity and for the majority of this subpopulation (82/98, 84%) the increase in activity began before the onset of activity in the iClB. This latter population of cells consisted of the 3 main groups (Fig. 4), defined according to the criteria in the METHODS, that will form the major focus of the paper: that is, cells discharging phasically (22/82, 27%), tonically (10/82, 12%), and with a phasic/tonic discharge pattern (41/82, 50%). Other neurons that showed an initial, significant increase in cell discharge showed an irregular pattern of activity and did not fall into any of these classifications (9/82, 11%). A further 29/142 (20%) neurons exhibited a decrease in cell-discharge frequency after the GO signal that continued after the onset of activity in the ClB of the reaching limb; 11/29 of these showed a subsequent increase in activity. Only a few neurons (15/142, 11%) showed no significant change in activity after the GO signal and of these about one half (9/15) were completely silent throughout the task.
Peak discharge rates for neurons showing increased activity during the task were relatively high (Table 1). Neurons with a tonic or phasic/tonic pattern of activity discharged at notably higher rates of discharge during the dynamic period than did those with a purely phasic pattern of activity. Control rates in the 3 groups were relatively similar. As expected, on the basis of the definitions used for classifying these cells, the discharge rate in the static period was greatest in neurons discharging tonically and least in phasic neurons.
|
Overall, the discharge patterns of the neurons in the 3 major groups that we defined formed a continuum, from cells that discharged in a purely tonic pattern to those that discharged in a purely phasic pattern (Fig. 5). Thus although objective criteria were used to define 3 groups of cells that accord with the classifications most commonly used by neurophysiologists, the data do not support a clearly demarcated division between them. For this reason, we have chosen to treat the database as a whole, emphasizing the major differences between tonic, phasic, and phasic/tonic neurons where appropriate. Moreover, as explained in the INTRODUCTION, one of the major goals of these experiments was to determine whether the temporal relationships of the reticular cells to the GO signal and to the onset of movement were similar to those that were previously detailed in our behavioral study (Schepens and Drew 2003a) for the kinetic and EMG measures.
|
Figure 6 illustrates one example of a neuron in which the initial change in discharge activity was temporally related to the GO signal. Activity in this neuron increased shortly, and abruptly, after the GO signal. It then diminished before again increasing and remaining elevated throughout the dynamic stage of the reach (Fig. 6, A and B). Examination of the discharge activity on a trial-by-trial basis showed that, in most trials, the onset of the period of increased discharge activity began at a relatively constant latency after the onset of the GO signal (Fig. 6B) and preceded the initial change in force (
iFLV). The later bursts of activity were more variable and occurred later in trials in which movement was delayed (Fig. 6, A and B, right). As a result, the initial increase in activity is most obvious in the PEH triggered on the Shutter (Fig. 6A, left) and the later burst is clearer in the PEH triggered on the movement (Fig. 6A, right).
|
iFLV) and the latency of the onset of activity in the iClB shows that the slope (m = 0.06: open symbols) of the relationship is closer to zero than to 1, confirming the temporal dissociation of these 2 components of the reach. Even for reach movements that were initiated relatively slowly, the initial postural change was elicited at an almost constant latency. As we previously demonstrated, this relationship is maintained over a wide range of reaction times (see Fig. 12A; Schepens and Drew 2003a). In contrast, plotting the lead time of the force onset as a function of the onset of activity in iClB reveals a much stronger relationship between these 2 values (m = 0.94, R2 = 0.99: filled symbols), confirming that iFLV is strongly linked to the onset of the GO signal (see METHODS and Schepens and Drew 2003a). The change in activity in the iBr is different from that observed for iFLV. Activity in this muscle is very tightly linked to the change in activity in the iClB and shows no temporal relationship to the GO signal, as indicated by the plot of iBr lead time as a function of iClB onset. As we previously discussed in detail (Schepens and Drew 2003a), this simple analysis provides an excellent method of determining whether a component of the reaching strategy is better related to the GO signal, and by extension to the initial changes in force that define the pAPA, or to the onset of activity in iClB, and by extension the reaching movement and the postural responses that accompany that movement.
|
iFLV and iBr. For this analysis, we measured the onset of cell activity only in those trials in which there was a clear change in frequency that corresponded to the initial peak observed in the PEH. This scatterplot clearly shows that the temporal relationships of this reticular neuron are very similar to those observed for iFLV. That is, the onset of activity in this neuron is temporally decoupled from the onset of movement and is much more strongly related to the GO signal, as illustrated by the plot of cell lead time as a function of iClB (m = 1.04, R2 = 0.92: filled symbols). This neuron could thus contribute to the initiation of the preparatory APA that follows the GO signal with a similar time course (Fig. 6B, see also Fig. 12). This suggestion would be strengthened by directly showing a relationship between the onset of cell activity and either iFLV or the onset of one of the shoulder or extensor muscles that show a similar temporal link to the GO signal. However, in this and in most of the other cells, all such regressions were nonsignificant (not illustrated). This is not surprising, given the small variation in latency between these 2 values. Similar nonsignificant relationships, for a similar reason, were observed between the onset of activity in the extensor muscles and iFLV (see Fig. 12G; Schepens and Drew 2003a). Similar relationships between the onset of the initial period of increased activity in cell discharge and the GO signal were observed in 6/11 of the phasically discharging neurons for which we were able to measure the latency of the change in discharge activity in individual trials. Similar relationships were also quantitatively verified in 14/29 of the neurons discharging with a phasic/tonic pattern of activity that showed an initial increase in discharge activity (see, e.g., Fig. 10).
|
These data suggest that, whereas the initial period of activity in these phasic neurons might be better related to the initial postural response (pAPA; see DISCUSSION), some of them probably also contribute to the dynamic period of the reach. Similarly, in the tonic and phasic/tonic cells that showed an initial period of discharge related to the GO signal, the more pronounced and maintained discharge activity that followed this almost certainly contributed to the movement and/or to the postural responses occurring during the movement (aAPAs).
In a number of the tonic and phasic/tonic neurons there was a clear decrease in the level of cell activity. One example of such a RSN is illustrated in Fig. 7. Quantitative analysis of this decrease of activity (Burst_1) showed that there was a strong relationship between lead time and the onset of activity in the iClB (Fig. 7B, left) confirming that it was better related to the GO signal than to the movement. Note that the intercept of the regression (66 ms) is similar to that observed for the increase in cell activity illustrated in Fig. 6 (80 ms). After the decrease in cell activity there was a large increase in cell discharge (Burst_2). Analysis of this increase showed that, although the coefficient of determination for the relationship between lead time and the onset of activity in the iClB was only slightly >0.5 (Fig. 7B, right), it was significant, suggesting that this period of activity was also better related to the GO signal.
|
Cells temporally related to the initiation of the movement
A large proportion of the cells showed an initial change in cell discharge that was better related to the movement than to the GO signal. Figure 8 shows one such phasically discharging neuron. In this example, there was a discrete burst of activity that occurred progressively later in trials in which the onset of movement was delayed. This can be readily seen in both the PEH and raster displays of Fig. 8A as well as in the individual trials illustrated in Fig. 8B. The scatterplots of Fig. 8C indicate that there was a strong linear relationship between the onset of the cell discharge and the onset of the iClB (m = 0.77, R2 = 0.81) but only a weak relationship between the lead time and the onset of activity in the iClB (m = 0.23, R2 = 0.27). This is the opposite of the relationship demonstrated by the cell illustrated in Fig. 6. A similar relationship to the movement was observed in 2 other phasically discharging neurons. The cessation of the period of discharge of this neuron occurred coincidentally with the end of the period of activity of the iBr (Fig. 8A, right). This relationship was significant, as it was in the other 2 phasic movement–related neurons (not illustrated).
|
|
Summary of temporal relationships
Analysis of the temporal relationships between neuronal activity and the 2 major components of the reaching strategy (the GO signal and the onset of the iClB) was performed for a total of 71 periods of modified cell-discharge activity measured in 51 cells. A majority of these events (58/71, 82%) were significantly related to either the GO signal or to the movement (Fig. 11A). The remaining 13 periods of activity either showed no significant relationship to either event (n = 7) or showed a significant relationship to each (n = 6). In the latter case, this could be explained by a lack of variation in the reaction times in these experiments. The slopes of those neurons showing a significant relationship to one or the other event formed 2 distinct groups (Fig. 11B). This bimodal distribution is similar to that obtained, in the same experiments, from plotting the slopes calculated from regressions of the onset in the change of force and the time of lift as a function of the onset of activity in iClB (Fig. 11C).
|
|
A large proportion of the neurons exhibited changes in activity that were better related to the movement than to the GO signal (Table 2 and Fig. 12D). The overall average latency of the onset of activity in these movement-related cells was 239 ms (Table 2). Although the average latency in the phasic cells was less than that in the other 2 groups, there was complete overlap with the shorter latency responses of the neurons in the tonic group. In most cases, the onset of the change in activity in these movement-related cells followed the initial change in force and preceded the onset of the change in activity in the iClB (Fig. 12D).
As demonstrated by the scatterplots of Fig. 12, the relationship between the latency (or lead time) of each population of neurons and the onset of the iClB was similar to that illustrated for each of the individual cells of Figs. 6–10. This supports the view that those measured events that showed significant relationships either to the GO signal or to the onset of the iClB form a part of 2 separate signals related to different events in the behavioral strategy.
Quantitative relationships with the level of EMG and force
A quantitative analysis of the intensity of cell discharge as a function of EMG and force magnitude was performed for all cells. For phasic neurons we used a binwidth of 20 ms and analyzed only the data from the GO signal to the time that the paw entered the tube, whereas for the other cells we used a binwidth of 50 ms and examined the whole trial, from the initial tone to 3 s after the shutter opened.
For the phasic neurons, only relatively weak correlations (R2 < 0.25) were seen, despite the fact that the discharge frequency in each cell was regressed against the total of 64 EMG or force traces. For the neurons showing a tonic component there was some evidence of stronger correlations. Figure 13 illustrates some of the results from 4 cells: 2 tonic cells (Fig. 13A, 2 leftmost cells) and 2 phasic/tonic cells (Fig. 13A, 2 rightmost cells). In all 4 examples, there was a fair linear correlation between cell discharge and the level of activity in the contralateral triceps (coTriL), which is increased throughout the reaching (see Figs. 1 and 4). Examination of the value of R2 for this relationship, for the whole population of tonically discharging cells, showed that 26/51 (51%) cells showed a value of R2 > 0.3 and 12/51 (24%) had a value >0.4 (see Fig. 13B). However, as illustrated in Fig. 13, A and B, it should be emphasized that the correlation with the coTriL was not noticeably better than that obtained with some of the other variables. For example, the correlation with the FV in the left hindlimb (iHLV) was as good as that with the coTriL in cell RS22T07G and the correlation with iClB was as good as that with the coTriL in cell RS22T23B. Nevertheless, it is noteworthy that of the 64 variables with which cell intensity was correlated for each cell, the relationship with coTriL was found in the top 5 correlations in 21/51 (41%) of the tonically discharging cells, and iClB was in the top 5 in 25/51 (49%) of the cells. In comparison, iHLV was in the top 5 correlations in only 6/41 (15%) of the cells, although either iHLV or iVL was in the top 5 in 14/51 (28%) cells.
|
Three other types of cell discharge were observed in more than 10 cells.
In one of these subpopulations the neurons (11/142) showed a decrease in activity that continued throughout the trial. Figure 14A illustrates an example of such a neuron. Inspection of this figure shows that this decrease in activity began relatively late and followed the onset of activity in the iClB. Indeed, in this example, the decrease in activity began at about the same time as the paw entered the tube to obtain the food reward (not illustrated). This pattern of activity was typical for those cells showing a sustained decrease in activity. From 29 individual trials, taken from 4 cells, in which it was possible to accurately measure latencies, the onset of activity in the iClB began 233 ± 84 ms after the GO signal, whereas the initial decrease in cell activity began at 425 ± 224 ms after the GO signal (the large SD was caused by one cell in which the decrease began noticeably earlier).
|
The other pattern of discharge activity that we observed (16/142 cells) included all of those neurons that showed an initial increase in activity that followed the onset of activity in the iClB (Fig. 14C). The onset of activity in many of these neurons was frequently well related to events occurring at the end of the movement. In 6/8 cells for which latencies could be measured, the onset of cell discharge was significantly related to the end of the period of activity in the iBr or to the time that the paw entered the food tube (P < 0.05, R2: 0.51–0.95). The average latency of the onset of activity in these cells was 426 ± 148 ms (71 trials, 8 cells).
Discharge activity during the pretrigger period
For most neurons, the initial change in activity occurred after the GO signal. However, in a number of neurons there were ramp increases or decreases of activity that occurred before the GO signal. Of the 73/142 phasic, tonic, and phasic/tonic neurons, 26/73 (36%) showed significant changes in activity during this pretrigger period. We refer to this activity in the current manuscript as pretrigger activity to avoid confusion with the terminology used to describe the different types of anticipatory postural activity that are subsequent to the GO signal but before the movement.
An example of one such cell, showing one of the largest changes that we observed, is illustrated in Fig. 15. This neuron showed a phasic/tonic pattern of discharge after the GO signal but also showed a significant increase in activity before the onset of the GO signal that occurred at approximately the same time as the onset of the Cue tone (Fig. 15B). At the same time, there was a significant modification of the anteroposterior component of the center of vertical pressure (CVP) during this same period, indicating that the cat was leaning forward in anticipation of the opening of the shutter. There was also a slight increase in the FV exerted by the 2 forelimbs (e.g., iFLV in Fig. 15A) and there was an increase in activity in the ipsilateral acromiotrapezius (iAcT) muscle at about the same time. Although these changes were clear in the illustrated cell, they were not so obvious in some of the other cells. Moreover, even in the illustrated example (Fig. 15), a clear change in pretrigger activity was evident only in the iAcT, although at high gain, increases could also be seen in the coAcT and in the iTrM. The changes in CVP thus likely reflect the aggregate sum of subtle changes in activity in a number of muscles, rather than large changes in a few muscles.
|
Figure 15B shows that the change in cell activity occurred at approximately the time that the cue tone began. However, a regression analysis of the change in cell-discharge activity as a function of the time of onset of the cue, as measured from individual trials from 5 cells showing clear pretrigger activity, revealed only a poor relationship (Fig. 15C). In contrast, there was a strong relationship between the onset of the cell discharge and the first detectable change in the anteroposterior component of the CVP (Fig. 15D). This suggests that the anticipatory changes in posture and the associated changes in cell activity were not directly triggered by the Cue signal but reflect more general changes in the overall posture dictated by the motivation of the cat in anticipation of the GO stimulus.
Localization of the different groups
The database was examined for localization along several lines using a discriminant analysis that examined the anteroposterior and dorsolateral location of the cells. As in Fig. 3, no account was taken of the mediolateral location of the cells because of the limited range of recording sites along this axis. We examined 3 aspects of the localization: whether there was any difference in the location of cells classified according to differences in discharge pattern; whether there was any difference in the localization of cells divided into the 4 classes illustrated in Fig. 12; and whether there was any difference in the properties of cells localized within the NRGc, NRPc, or NRMc. In all comparisons the probability that any one group or class was differentially localized within the recording area was >0.01 and, in a majority of cases, >0.05. We emphasize, however, that these recordings were made in a quite restricted region of the brain stem and that the NRPc and NRMc were only poorly represented in the overall database.
RSNs versus unidentified neurons
The populations of cells showing temporal relationships to the initial changes in posture and to the movement included both identified RSNs and Unidentified neurons. Overall, a comparison of the characteristics of the RSNs and the Unidentified neurons showed no differences, although it should be noted that phasically discharging neurons were predominantly Unidentified (18/22: 82%). Nevertheless, even among the phasic group the Unidentified neurons included cells related both to the GO signal and to the movement. Similarly, within the phasic/tonic group there was no one pattern of discharge that was found only in RSNs or Unidentified neurons. For example, the complicated discharge pattern observed in the Unidentified neuron illustrated in Fig. 10 was also observed in RSNs. The only other distinction that was observed was a tendency for cells showing decreased activity to include more Unidentified neurons than RSNs (23/29: 79%).
We also examined whether the cells organized into the different classes with respect to the relationship of the discharge to the GO signal or to the movement (see, e.g., Fig. 12) showed any differences with respect to the number of RSNs or Unidentified neurons. However, RSNs and Unidentified neurons were almost equally represented in each class.
|
DISCUSSION |
|---|
) forming the database in the manuscript. Neurons from cats RS22 (A) and RS23 (B) have been plotted onto separate standard sections of laterality (L) 1.2 taken from the atlas of Berman (1968). C: conduction velocity of the axons of the RSNs; values from cat RS23 are indicated by the cross-hatched bars. Conduction velocities were calculated on the basis of the distance from the microwires inserted into the spinal cord to stereotaxic coordinate P-7. Note that the latency of 2 cells was not recorded.
), Tonic (), and Phasic/Tonic (
) groups. Inset: histograms illustrate example neurons.
