| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1Department of Integrative Physiology, National Institute for Physiological Sciences and 2Department of Physiological Sciences, Graduate University for Advanced Studies, Myodaiji, Okazaki, Aichi, Japan; and 3Graduate School of Frontier Biosciences, Osaka University, Toyonaka, Osaka, Japan
Submitted 28 May 2004; accepted in final form 23 August 2004
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
). It has been suggested that the saccade performance is well explained by the "dynamic interactions model" in which different motor commands compete for dominance in a winner-takes-all fashion with excitatory interactions occurring between motor commands coding movements to closely clustered targets (Clark 1999; Findlay and Walker 1999; Godijn and Theeuwes 2002; Munoz and Fecteau 2002; Trappenberg et al. 2001). Several lines of evidence have suggested that the intermediate layer of superior colliculus (SC) is the field on which these interactions occur. The SC integrates input from several saccade-related areas and sends its output to the brain stem saccade generator circuits (Sparks and Hartwich-Young 1989). SC neurons are organized into a retinotopically coded motor map, which specifies saccades toward the contralateral visual field. (Robinson 1972; Schiller and Stryker 1972). The SC has been suggested to have intrinsic circuits, which are known to have excitatory interactions between sites of close proximity (McIlwain 1982; Moschovakis et al. 1988; Pettit et al. 1999; Saito and Isa 2003) and inhibitory interactions between remote sites (Meredith and Ramoa 1998; Munoz and Istvan 1998). Furthermore, it has been shown that the activities of SC neurons correlate with the psychophysical phenomena (Dorris and Munoz 1995; Edelman and Keller 1998; Glimcher and Sparks 1993b; McPeek et al. 2003; Olivier et al. 1999; Port and Wurtz 2003; van Opstal and van Gisbergen 1990). Accordingly, the competition and cooperation within the SC motor map have been suggested to implement the dynamic interactions model. However, recent findings have cast doubts on the role of the SC in the competitive selection of different motor plans (McPeek and Keller 2002; McPeek et al. 2003; Ozen et al. 2004; Pare and Hanes 2003; Port and Wurtz 2003).
The manipulation of visual stimuli not only modifies the activities of SC neurons but also affects the activities of neurons in widely distributed regions in the brain. Therefore to examine the direct relationship between SC neural activity and behavior, artificial manipulation of SC neuronal activity is required (Stanford 2004). Electrical stimulation has been frequently used to activate neurons close to the stimulation site, and the results are consistent with the predictions from the dynamic interactions model (Carello and Krauzlis 2003; Glimcher and Sparks 1993a; McPeek et al. 2003; Munoz and Wurtz 1993b). However, electrical currents stimulate not only the cell bodies of neurons directly but also antidromically and orthodromically activate fibers of passage as well. These artifacts could create an abnormal spatial distribution of activation states across various brain regions. Consequently, the resultant behavior could be caused by a mechanism different from that which operates naturally. The injection of bicuculline, a GABAA receptor antagonist, is another method that has been used to selectively activate SC neurons without affecting axons (Hikosaka and Wurtz 1983, 1985a; Munoz and Wurtz 1993b). However, it is difficult to quantitatively analyze the effects on purposive saccades because bicuculline injection in the caudal SC prevented monkeys from performing saccade tasks byfrequently evoking fixation-breaking saccades toward the movement field of neurons at the injection site (affected area) in addition to inducing nystagmus (Hikosaka and Wurtz 1983, 1985a). Moreover, in the case of bicuculline injection, the resultant saccades would not reflect the change in neural activity in GABAergic systems.
It has been shown that the SC receives cholinergic projections from the pedunculopontine tegmental nucleus (Beninato and Spencer 1986; Graybiel 1978; Hall et al. 1989; Illing and Graybiel 1985) the neurons of which exhibit activities related to performance of saccades (Kobayashi et al. 2002). We have clarified that cholinergic input excite the majority of neurons in the SC via activation of 
4
2-type nicotinic receptors and M3-type muscarinic receptors and inhibit GABAergic synaptic transmission via activation of M1 and M3 muscarinic receptors on presynaptic terminals in rodent SC slices (Isa et al. 1998; Li et al. 2004; Sooksawate and Isa, unpublished observations). We previously reported that local microinjection of the cholinergic agonist nicotine into the monkey SC decreased the reaction time of saccades toward targets presented at the affected area (Aizawa et al. 1999). In addition, we confirmed recovery from the nicotinic drug effect within the experimental session (Aizawa et al. 1999). Furthermore, it seems likely nicotinic agonism may directly elevate the activation of SC neurons while preserving the GABAergic inhibitory influence. For these reasons, it is likely that the microinjection of nicotine can be a powerful tool for modestly elevating the basal firing rate of SC neurons without stimulating fibers of passage. In this study, we used this method to alter the spatial distribution of neural activity across the SC and examined its effects on saccades. A preliminary report has been published elsewhere (Watanabe et al. 2003).
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
All experimental procedures were performed in accordance with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals and approved by the Committee for Animal Experimentation at Okazaki National Research Institutes. The details of the surgical and data-acquisition methods have been published previously (Aizawa et al. 1999; Kobayashi et al. 2002). Briefly, two male (monkeys M and D) and one female (monkey H) Japanese monkeys (Macaca fuscata) weighing 8.5–12.5 kg were trained to perform a button-press task for a liquid reward while sitting in a primate chair. After the training was completed, the monkeys were anesthetized with isoflurane and implanted with scleral search coils (Fuchs and Robinson 1966) and head holders. The monkeys were allowed to recover for >3 wk and trained to perform visually guided saccade tasks with their heads in a fixed position. Recording chambers tilted 38° posterior to the vertical axis were mounted on the skull by a separate surgical procedure, and experiments were subsequently performed.
Behavioral procedures for visually guided saccades
Visual stimuli, behavioral tasks, and data acquisition were administered via the Tempo/Win computing system (Reflective Computing, St. Louis, MO). Horizontal and vertical eye positions were sampled at 1 kHz using the search-coil technique (Fuchs and Robinson 1966). Visual stimuli were back-projected on a tangential screen at a distance of 28 cm from the eye. The onset and offset of the visual stimuli were synchronized with the projector's vertical refresh (noninterlaced refresh rate of 60 Hz).
Each trial was preceded by an inter-trial interval (varied randomly between 500 and 1,500 ms) during which the screen was illuminated with diffuse white light to prevent dark adaptation. After the removal of the background light, a fixation point appeared at the center of the screen, and the monkeys were required to direct their eyes toward the target and maintain fixation for 400–800 ms within a fixation window (3–5°, square). The monkeys were trained to perform three different visually guided saccade tasks: step, overlap, and gap tasks (the overlap and step tasks are sometimes referred to as the no-gap task in the following text). In the step task, a visual target appeared simultaneously with the offset of the fixation point. In the overlap task, the target appeared while the fixation point remained visible until the end of a trial. In the gap task, the fixation point was extinguished 200 ms prior to the appearance of the target. During the gap period, the monkeys were required to maintain fixation in total darkness. The target presentation was the go signal for visually guided saccades in all of the three tasks. The monkeys were required to immediately make a saccade to the target within 500 ms from the target onset and then to maintain fixation for 250–300 ms within a target window (3–10° depending on target eccentricity and the performance of the monkeys). The monkeys were given a liquid reward when they correctly performed each trial. Two different tasks were usually included in a block of trials with equal probabilities (the gap and overlap tasks or the gap and step tasks). Visual targets were arranged to be symmetrically (for example, Fig. 3, inset) or asymmetrically (for example, Fig. 4, inset) distributed around the fixation point. These arrangements of visual targets were successively implemented when multiple microinjections were made within single canula tract (see Injection procedures). For experiments in which the target arrangement was symmetric (7 and 5 experiments in monkeys M and D, respectively), targets were presented at equally spaced eight different directions in all but one experiment. When targets were asymmetrically arranged (15 and 7 experiments in monkeys M and D, respectively), average numbers of targets in an experimental block were 6.7 and 4.7 in monkeys M and D, respectively. Stimulus conditions were randomly presented with equal probability.
|
|
|
The location of the SC was identified by single-unit recordings and magnetic resonance images. The extracellular activities of single neurons were recorded using tungsten microelectrodes (Frederick Haer, Bowdoinham, ME) with an impedance of 1–6 M
. Electrodes were positioned through stainless steel guide tubes (23 gauge) using a micromanipulator (MO-95, Narishige, Tokyo, Japan). The guide tubes were held in position with a delrin grid that was fixed to the recording cylinder (Crist et al. 1988). Activities of saccade-related neurons were recorded and their movement fields were qualitatively determined. The depth of the location of the neurons was confirmed in relation to the guide tubes. Functional identification of the SC was performed through electrical stimulation (<20 µA, 500 Hz, 50 biphasic pulses of 200-µs width) delivered via the electrodes, and evoked saccades were confirmed. The relationship between the evoked saccades and the position of electrodes on the grid was in agreement with the SC movement map reported previously (Robinson 1972). Stimulation sites where the amplitudes of evoked saccades were larger or smaller than 3° were referred to as being caudal or rostral regions, respectively.
Injection procedures
Nicotine (Sigma-RBI, St. Louis, MO) was dissolved in phosphate-buffered saline (PBS; pH 7.4, Gibco-BRL, Gaithersburg, MD) at concentrations of 10, 50, or 100 mM. To inject nicotine, a 30-gauge syringe needle with a microwire extending from the tip (Crist et al. 1988) was lowered into the SC to the same depth at which saccade-related neurons had been previously identified. Electrical stimulations were delivered via the microwire to confirm the location of the tip of the syringe needle. After control saccades were recorded, pressure injection of nicotine was made into the SC at a rate of 0.2 µl/min during a period of 1–5 min. The injected volumes ranged from 0.2 to 1.0 µl. In some experiments, after on-line qualitative verification of the recovery from the nicotine microinjection, a second microinjection of nicotine was performed at different depth within the same canula tract (difference: >200 µm). As a control, an equivalent volume of vehicle (PBS) was injected into the same location where functional effects of nicotine microinjection on saccades had previously been observed on different days. A summary of the injections is shown in Table 1.
|
The onset and offset of saccades were identified by radial eye velocity criteria (threshold: 30°/s) during off-line analysis. The center of the affected area was determined for each canula tract from saccades evoked by electrical stimulations delivered through the microwire of the syringe needle. In monkey H, only spontaneous saccades were analyzed. In monkeys M and D, spontaneous saccades, fixation-breaking saccades, and visually guided saccades were analyzed.
Reaction times and endpoints
The reaction time of visually guided saccades was defined as the time to initiation of the saccade after target onset. Saccades whose reaction times were <80 ms were regarded as fixation-breaking saccades and separately analyzed from visually guided saccades. Setting this criterion to 70 ms did not affect the following results (data not shown). The saccade endpoint was quantified as the distance from the center of the affected area while its initial point was aligned to the origin of the coordinate axes. A misdirected saccade was defined as a saccade, which did not terminate within an off-line window (a radius of 2.5°) placed at the center of a target. Misdirected saccades were included in the analysis of reaction times and endpoints of visually guided saccades if corrective saccades toward the target followed the misdirected saccades and then the eye position was confined within the off-line window during 80-ms period just after the offset of the target.
Curved trajectories
To quantify the curvature of visually guided saccades, the curvature index was defined as follows
 |
S denotes an area enclosed by the trajectory and the straight line passing the initial point and endpoint of the saccade. The sign of the S was positive if the segment of the trajectory in S was curved toward the affected area and negative if the curvature was opposite. To exclude an erroneous effect of misdirected saccades on the estimation of the median of the indices, this analysis was restricted to saccades ending within the off-line target window. It is still possible, however, that the distribution of saccade endpoints obtained prior to microinjection was different from that after microinjection even if they were confined within the off-line target window. To eliminate the difference, an additional spatial window was defined based on the saccade endpoints before microinjection. The saccade endpoints before microinjection were applied to a principle component analysis and their SDs were calculated for the first and second components. The additional window was set to cover 2 SDs from the mean of the endpoints. We did not analyze the data with saccade endpoints outside of this additional window. Furthermore, we excluded the most deviated saccade from data after microinjection when the median of the saccade endpoints before and after microinjection was significantly different (Mann-Whitney U test P < 0.05). The statistical tests and the exclusion of a saccade were then repeated until the tests did not reach statistical significance (P > 0.05). The curvature indices of the remaining saccades were normalized to the curvature indices before microinjection. Statistical tests
Data obtained during the 15-min period after microinjection (injection data) were statistically compared with data for the 15-min period prior to the start of the microinjection (baseline data). The time window was extended by 1 min if the number of trials included in this window was <10. The time periods during which nicotine affected saccades depended on several experimental conditions, such as the concentration and volume of nicotine and the location of the injection site (data not shown). Thus in the summaries of results (Figs. 2, 5, 8, and 10), the time window for the injection data was shifted by 1-min steps to maximize the absolute difference between the baseline and injection data (the extent of the differences was quantified by the probabilities derived from statistical tests). Saccade performance was rather variable in the behaving monkeys. To maintain the behavioral states of the monkeys, the shift of the time window for the injection data was shifted by 1 min and determined within 20 min after the start of microinjection so that the effect of microinjection became maximum. Mann-Whitney U tests with Bonferroni correction were used to perform statistical comparisons between the baseline and injection data (P = 0.05 was divided by the number of the statistical tests repeated). Observation of recovery from the effects of nicotine began 20 min after the start of the microinjection. Recovery was confirmed by comparing 15 min of recovery data with the injection data (data not shown).
|
|
|
|
For the endpoints and trajectories of visually guided saccades, task-dependent effects (gap vs. no-gap tasks) were examined for the injection data in which significant nicotinic effects (endpoint: a decrease in distances from the center of the affected area, trajectory: an increase in curvature indices, U test P < 0.05 with Bonferroni correction) were observed in at least one of the tasks and differences between the tasks were not observed in the corresponding baseline data (U test P > 0.05). Because saccade reaction times are shorter in the gap task than those in the no-gap task (Saslow 1967), a direct comparison between saccade reaction times after microinjection in the gap and no-gap task is not adequate. The effects of nicotine on saccade reaction times for each task were quantified using a receiver operating characteristic (ROC) analysis (Figs. 11 and 12) (Green and Swets 1966). In the ROC analysis, histograms of reaction times before and after microinjection were compared. Each point in the ROC curve was the proportion of reaction times after microinjection exceeding an arbitrary criterion value as a function of that of reaction times before microinjection exceeding the same criterion. Entire curves are obtained by sweeping the criterion value through the range of the data. The integrated area beneath each ROC curve (ROC area) indicates the difference between the two distributions of the reaction times before and after microinjection. If the two distributions were not significantly different, the ROC area would be nearly equal to 0.5. On the other hand, if the distribution of reaction times after microinjection had smaller or larger values than that before microinjection, the ROC area would approach zero or one, respectively. In this analysis, the windows of time for both the baseline and injection data were extended to 30 min to obtain a sufficient number of saccades for each task (
20 trials). The analysis was restricted to the data in which a decrease in the reaction times of saccades was observed after microinjection in at least one of the tasks (Permutation test P < 0.05 with Bonferroni correction) (Uka and DeAngelis 2004).
|
|
Analyses of task-dependent effects on fixation-breaking saccades were restricted to the saccades triggered within 80 ms from the target onset. The effects on the frequency of fixation-breaking saccades were quantified for each task (gap or no-gap task) by subtracting the proportion of trials in which fixation-breaking saccades were generated before microinjection from that of trials after microinjection (Fig. 14A, error rate). The endpoints of fixation-breaking saccades after microinjection were also compared between the tasks (U test P < 0.05 with Bonferroni correction).
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Spontaneous saccades
Figure 1 shows the distribution of the endpoints of spontaneous saccades the initial points of which were aligned to the origin of the coordinate axes. Prior to microinjection of nicotine (Fig. 1A), the endpoints of spontaneous saccades were evenly distributed. In contrast, saccades were frequently directed toward the affected area after nicotine microinjection (Fig. 1B, 100 mM, 0.6 µl, experiment N15-1 compared with Fig. 1A, U test P < 0.0005). Saccades directed opposite to that of the affected area were also increased after nicotine microinjection (Fig. 1B). These are mostly return saccades to the center of the screen after the spontaneous saccades toward the affected area. The endpoint bias toward the affected area diminished with time (Fig. 1, C compared with B, U test P < 0.0005).
|
Figure 2 summarizes the effects on saccade endpoints from experiments in which the baseline and injection data could be compared. The distances of saccade endpoints from the center of the affected area were consistently decreased by nicotine microinjection (Fig. 2, monkey M: 11 of 20 experiments, monkey D: 8 of 12 experiments, monkey H: 2 of 3 experiments, U test P < 0.05 with Bonferroni correction, pooling all data, Wilcoxon test P < 0.0005), indicating that nicotine microinjection facilitates generation of spontaneous saccades toward the affected area. Remarkable effects were observed for the data points plotted at the bottom left portion of the summary figure. These plots were derived from experiments in which nicotine microinjection was made at the rostral region; this caused repeated generation of small contraversive staircase saccades toward the affected area (data not shown). Thus marked effects were observed when the distance between the affected area and intended endpoints of spontaneous saccades (mostly <20°, see Fig. 1A) was short.
Reaction times of visually guided saccades
INJECTIONS INTO CAUDAL REGIONS.
If there is mutual facilitation between proximal sites and mutual inhibition between remote sites on the SC map as suggested by the dynamic interactions model (Clark 1999; Findlay and Walker 1999; Godijn and Theeuwes 2002; Munoz and Fecteau 2002; Trappenberg et al. 2001), saccadic reaction times were expected to be decreased when targets were presented close to the affected area, whereas reaction times would be increased when targets were presented at a distance remote to the affected area after nicotine microinjection. Figure 3 shows the results from the same experiment shown in Fig. 1 in which targets were presented in eight directions (N15-1). Consistent with the prediction, the reaction times of saccades toward a visual target close to the center of the affected area were markedly decreased after nicotine microinjection in both the gap and overlap tasks (Fig. 3A, U test P < 0.0005). However, the reaction times of saccades toward a remote visual target were not increased (Fig. 3B, gap and overlap, U test P > 0.05). The same results were observed for other remote targets (Fig. 3C). It might be possible that the effect of nicotine at the injection site was not strong enough to drive the inhibitory interactions in the SC. However, fixation-breaking saccades were frequently evoked toward the affected area after microinjection of nicotine (Fig. 13), suggesting that neuronal activity at the injection site was sufficiently elevated via nicotine exposure.
|
; Krauzlis et al. 1997, 2000; Munoz and Wurtz 1993a). Accordingly, nicotinic activation of neurons in this region was hypothesized to increase the reaction times of saccades toward visual targets with large eccentricities and expected to decrease those of saccades toward visual targets with small eccentricities. Saccades toward targets with the same eccentricity were pooled because they showed similar results. In agreement with the hypothesis, the reaction times of saccades toward the targets with small eccentricity were decreased after microinjection (Fig. 4A, U test P < 0.0005). However, the reaction times of saccades toward the targets with large eccentricity were not increased (Fig. 4B, U test P > 0.05). The effects on the saccadic reaction times clearly depended on the target eccentricities (Fig. 4C). Fixation-breaking saccades were frequently evoked toward the affected area after nicotine microinjection in both tasks (data not shown), suggesting that neurons close to the injection site were highly activated. The data for the gap task were excluded from the current analysis because a sufficient number of visually guided saccades was not obtained due to frequent generation of fixation-breaking saccades immediately after the fixation point offset by microinjection at the rostral region (Fig. 15B).
|
Figure 5 summarizes the effects of nicotine microinjection on the reaction times of visually guided saccades (monkey M: 121 targets in 18 experiments, monkey D: 73 targets in 12 experiments). These results are shown in the coordinates of the SC motor map (Ottes et al. 1986). The amplitude and direction of each target were first normalized to 1 and 0 on the Cartesian coordinate, and the position of the corresponding affected area was also converted to maintain its relative position with the target. The normalized target and converted affected area were then plotted on the SC motor map. The scatter plots in Fig. 5 were created by applying this method to all targets. Although data for the gap task and those for the no-gap (step or overlap) task were pooled, the data points were almost evenly distributed when the analysis was applied to the data from different tasks separately (data not shown). A decrease in saccadic reaction times was restricted to the data points (injection sites) near the normalized target point (Fig. 5A, monkey M: 32 targets in 14 experiments, monkey D: 17 targets in 7 experiments, U test P < 0.05 with Bonferroni correction). Only two data points from one experiment showed an increase in saccadic reaction times (Fig. 5B, monkey D, U test P < 0.05 with Bonferroni correction). No changes in saccadic reaction times were observed for the data points remote from the normalized target point (Fig. 5C, monkey M: 89 targets in 17 experiments, monkey D: 54 targets in 12 experiments, U test P > 0.05 with Bonferroni correction).
Endpoints of visually guided saccades
INJECTIONS INTO CAUDAL REGIONS. In addition to the effects on reaction times, the endpoints of visually guided saccades were affected by nicotine exposure. Figure 6 shows endpoint effects from the same experiment shown in Figs. 1 and 3 (N15-1). The trajectories of saccades before and after microinjection toward three different targets are shown in Fig. 6, A and B, respectively. The endpoints of saccades for the target proximal to the affected area were biased toward the affected area (red lines in Fig. 6, A and B). The biased saccades were followed by the corrective saccades (Fig. 6B). This effect gradually diminished with time (Fig. 6C). In contrast, the endpoints of saccade toward the target remote from the affected area were not changed by the microinjection (black lines in Fig. 6, A and B). For the target presented between the above two targets, the endpoints of saccades for the target were also biased toward the affected area (blue lines in Fig. 6, A and B). The duration of this effect, however, was much shorter than that for the target close to the affected area (Fig. 6, D compared with C). The effects on the saccade endpoints were restricted to the targets close to the affected area (Fig. 6E).
INJECTIONS INTO ROSTRAL REGIONS. Figure 7 shows the effects of nicotine microinjection into the rostral region of the SC on the amplitude of saccades from the same experiment shown in Fig. 4 (N5-1). The amplitude of saccades toward the targets of intermediate eccentricity (15°) became shorter followed by gradually diminished effects with time (Fig. 7A). The amplitudes of saccades toward visual targets with large eccentricity (25°) also became shorter (Fig. 7B). However, the effect was short lasting (Fig. 7B) in contrast to the effect on the saccades to the targets with intermediate eccentricity (Fig. 7A). A significant effect on amplitude was found for saccades toward the above targets (Fig. 7C).
|
The effects of nicotine microinjection on the endpoints of visually guided saccades are summarized in Fig. 8, using the same format as Fig. 5 (monkey M: 121 targets in 18 experiments, monkey D: 73 targets in 12 experiments). A bias of saccade endpoints toward the affected area was mainly observed for the data points near the normalized target point (Fig. 8A, monkey M: 22 targets in 13 experiments, monkey D: 17 targets in 7 experiments, U test P < 0.05 with Bonferroni correction). Although a bias of saccade endpoints away from the affected area was also observed, the number of data points was small (Fig. 8B, monkey M: 5 targets in 4 experiments, monkey D: 2 targets in 2 experiments, U test P < 0.05 with Bonferroni correction). The endpoints of saccades toward targets remote from the affected area were not changed (Fig. 8C, monkey M: 94 targets in 18 experiments, monkey D: 54 targets in 12 experiments, U test P > 0.05 with Bonferroni correction).
Relationship between effects on reaction times and endpoints
As shown in Figs. 3 and 6 and 4 and 7, the effects on the reaction times of visually guided saccades were more localized than those on the endpoints of saccades. To compare the extent of nicotinic effects on the reaction times and endpoints of saccades, we calculated average distances between the affected area and targets toward which saccades were significantly modified (decreased reaction times or biased endpoints toward the affected area) for each experiment in which both reaction time and endpoint effects were observed. The average distances for reaction time data were significantly smaller than those for endpoint data (12 and 5 experiments in monkey M and D, respectively, pooling all data, Wilcoxon test P < 0.0005), indicating that nicotinic effects on the reaction times of saccades are more localized than those on the endpoints of saccades.
Curved trajectories of visually guided saccades
Recent studies using recordings of neural activity and electrical stimulation have suggested that the activities of SC neurons can affect saccade trajectories (McPeek et al. 2003; Port and Wurtz 2003). However, as previously mentioned in the introduction, these results should be revisited using microinjection of pharmacological agents, such as nicotine. Figure 9 shows an example of the effect of nicotine on saccade trajectories (N22-1). In this experiment, the endpoints of some saccades were biased toward the affected area (data not shown). Such biased saccades were excluded from analysis (see METHODS) and the records shown in Fig. 9 were restricted to saccades correctly directed toward the target. The postmicroinjection trajectories (injection, red line) were modestly curved toward the affected area compared with those recorded prior to microinjection (baseline, blue line; Fig. 9A). To quantify these trajectories, the curvature index was calculated for each saccade (see METHODS). A positive value of the index indicates a curved trajectory toward the affected area, whereas a negative value indicates curvature away from the affected area. The time course of changes in the normalized indices (Fig. 9B, see METHODS) shows that an increase in the indices emerged immediately following nicotine microinjection and persisted for 
15 min (U test P < 0.0005).
|
Summary of effects on trajectories
The effects of nicotine microinjection on the trajectories of visually guided saccades are summarized in Fig. 10 (monkey M: 117 targets in 18 experiments, monkey D: 72 targets in 12 experiments). Curved trajectories toward the affected area were more frequently observed for the data points near the normalized target point (Fig. 10A, monkey M: 14 targets in 9 experiments, monkey D: 13 targets in 8 experiments, U test P < 0.05 with Bonferroni correction). Although we observed trajectories curved away from the affected areas, the number of data points was small (Fig. 10B, 3 targets in 3 experiments in each monkey, U test P < 0.05 with Bonferroni correction).
Task-dependent effects on visually guided saccades
In most experiments, the gap and no-gap tasks were randomly mixed in an experimental block. The effects of nicotine microinjection on visually guided saccades were clearly observed in both tasks across several saccade parameters (Figs. 3, 6, and 9). It has been shown that the low-frequency preparatory activities of SC neurons are different between the tasks (Dorris and Munoz 1995; Dorris et al. 1997). Accordingly, it is possible that the effects of nicotine microinjection on visually guided saccades are task dependent. We did find task-dependent effects on saccade reaction times but not on the saccade endpoints (monkey M, all of 4 targets in 4 experiments; monkey D, 9 targets in 5 experiments of 11 targets in 6 experiments, U test P > 0.05 with Bonferroni correction) and trajectories (monkey M, all of 2 targets in 2 experiments; monkey D, 3 targets in 3 experiments of 5 targets in 3 experiments, U test P > 0.05 with Bonferroni correction). Figure 11 shows another example of the effects of nicotine microinjection on the reaction times of visually guided saccades (N19-1). In this experiment, the effect on saccadic reaction times lasted >1 h in the gap task, whereas saccadic reaction times for the step task (Fig. 11A) and the effects on spontaneous saccades during the task block (data not shown) were already recovered. ROC analysis was applied to quantify the difference between the pre- and postmicroinjection reaction times for each task (Fig. 11B). When the time windows for the data ranged from 0 to 30 min after the start of the microinjection, the ROC areas for the gap (black dotted line) and step (no-gap) tasks (black continuous line) were 0.12 and 0.17, respectively (Fig. 11B). The difference in the effects between the tasks was more evident when the time windows for the analysis were set to range from 30 to 60 min after the start of the microinjection: the ROC areas were 0.25 and 0.40 for the gap (red dotted line) and step (no-gap) tasks (red continuous line), respectively (Fig. 11B). Figure 12 summarizes the results from the analysis in which the time windows for injection data were set to range from 0 to 30 min after the start of microinjection. Most data points were above the equality line (monkey M: 12 targets in 7 experiments, monkey D: 10 targets in 4 experiments, pooling all data, Wilcoxon test P < 0.005), indicating that the effects for the gap task were more evident than those for the no-gap tasks. The same result was observed regardless of the position of the time window for the injection data (data not shown).
Task-dependent effects on fixation-breaking saccades
The frequency of fixation-breaking saccades was increased after microinjection of nicotine. Figure 13 shows the results from the same experiment shown in Figs. 1, 3, and 6 (N15-1). Compared with the baseline data (Fig. 13A), the number of fixation-breaking saccades during the gap task was markedly increased after microinjection, and these saccades were directed toward the affected area (Fig. 13, B and C). However, in the case of the overlap task, the fixation-breaking saccades were not so frequently generated even after microinjection, and their endpoints were not always associated with the affected area (Fig. 13, B and C). Figure 14 summarizes the effects of nicotine microinjection on fixation-breaking saccades (19 and 10 experiments in monkeys M and D, respectively). The error rate indicates the differences of the proportions of pre- and postmicroinjection trials in which fixation-breaking saccades were generated. A positive value of the error rate indicates more frequent generation of fixation-breaking saccades after nicotine microinjection. The distribution of the error rate was biased toward positive values in the gap task (sign test P = 0.06 in monkey M, P < 0.05 in monkey D), indicating that nicotine microinjection facilitates generation of fixation-breaking saccades in the gap task. Although the frequency of fixation-breaking saccades was also modestly increased in the no-gap task, the changes did not reach statistical significance (P > 0.3 in both monkeys). For experiments in which the frequency of fixation-breaking saccades was increased after microinjection in at least one of the tasks (15 and 9 experiments in monkeys M and D, respectively), the rates of increases were higher in the gap task than those in the no-gap task (Wilcoxon test P < 0.05 in both monkeys), indicating that facilitation of fixation-breaking saccades was more evident in the gap task than those in the no-gap task. The endpoints of fixation-breaking saccades after microinjection in the gap task were compared with those in the no-gap task (Fig. 14B, 7 experiments in both monkeys). The distance of saccade endpoints from the center of the affected area in the gap task was smaller than those in the no-gap task (2 and 5 experiments in monkeys M and D, respectively, U test P < 0.05 with Bonferroni correction, pooling all data, Wilcoxon test P < 0.005), indicating that fixation-breaking saccades in the gap task were more closely directed toward the affected area than those in the no-gap task.
Reaction times of fixation-breaking saccades from fixation point offset
Fixation-breaking saccades were frequently generated during the gap period of the gap task in addition to the 80-ms period after the target onset. Figure 15 shows the reaction times of fixation-breaking saccades from the fixation point offset in the gap task when microinjection was made at the caudal (Fig. 15A, N15-1, the same experiment shown in Figs. 1, 3, 6, and 13) or rostral (Fig. 15B, N1-1, amplitude of affected area <3°) region. The frequencies of fixation-breaking saccades were markedly increased after nicotine microinjection in both experiments (Fig. 15, A and B). However, a transient decrease in the reaction times of fixation-breaking saccades from the fixation point offset was observed only when microinjection was made at the rostral region (Fig. 15B, U test P < 0.0005). Consistent results were observed in other experiments in which the baseline and injection data could be compared (microinjection at rostral region: 3 experiments in monkey M, U test P < 0.05 with Bonferroni correction, microinjection at caudal region: 6 and 11 experiments in monkeys M and D, respectively, U test P > 0.05 with Bonferroni correction).
Vehicle injections
Vehicle microinjections of PBS were performed into the caudal SC at the same location where significant effects of nicotine on saccades had been observed on previous days. For spontaneous saccades, the distances of saccade endpoints from the center of the affected area were not consistently changed, although the changes often reached statistical significance (increase: 1 experiment in monkey M, decrease: 2 experiments in monkey M, U test P < 0.05 with Bonferroni correction). Figure 16 shows the reaction times of visually guided saccades from one of the experiments in which a significant decrease in the distances of the endpoints of spontaneous saccades from the center of the affected area was observed (P2-1). The reaction times of saccades toward a target close to the center of the affected area were unchanged after PBS microinjection (Fig. 16A, U test P > 0.05). The saccade reaction times were not changed by PBS microinjection regardless of the location of the targets (Fig. 16B, U test P > 0.05). Therefore the significant change in the endpoints of spontaneous saccades was not supposed to reflect the nonspecific actions of vehicle microinjection.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Limitation of nicotine microinjection
The excitatory effects of nicotine microinjection observed in this study are consistent with the previous report that application of acetylcholine agonists directly depolarized the postsynaptic membrane potential of SC neurons mainly via 42 nicotinic receptors in a rat slice preparation (Isa et al. 1998; Sooksawate and Isa, unpublished observations). As to the action of nicotine on the SC local circuits, we have found that not only excitatory neurons but also GABAergic inhibitory neurons are excited by nicotinic receptor activation (Sooksawate and T. Isa, unpublished observations). However, we did not observe any phenomena that are explained by enhanced inhibitory effects on saccade-related SC neurons. Therefore it seems likely that the net effect of nicotine on the activities of neurons at the injection site is excitatory. Second, it has previously been shown that the regions where immunoreactivity of choline acetyltransferase is high are patchy and distributed more frequently in the caudal region than the rostral region (Graybiel 1978; Ma et al. 1991). Accordingly, the effects of nicotine microinjection on saccades might depend on the location of the injection site relative to the "patch" of cholinergic fibers as well as the dose of nicotine. The answer to this question should wait for the detailed analysis of distribution of nicotinic acetylcholine receptors and patches of cholinergic fibers in the SC. However, we observed fairly pronounced excitatory effects of nicotine even when microinjection was made at the rostral region (Figs. 4, 7, and 15B). Although the detailed mechanism of nicotinic activation of SC neurons should be clarified in the future, we suggest that nicotine microinjection is a useful tool for locally activating SC neurons.
Modulation of neuronal activity at injection sites
In previous studies in behaving monkeys (Hikosaka and Wurtz 1983, 1985a), bicuculline was used to elevate neuronal excitation within the SC. Bicuculline blocks GABAA receptors and disinhibits neurons at the injection site, leading to fixation-breaking saccades toward the affected area and finally inducing nystagmus. These monkeys were unable to inhibit the generation of these eye movements in any way because the fixation system could not affect neural activity at the injection site. Consequently, bicuculline injection prevented the monkeys from performing saccade tasks. Moreover, in the case of bicuculline injection, it may not be possible to observe the effects directly caused by the alteration of the GABAergic systems. In contrast, GABAergic inhibition seems to be intact after nicotine exposure, allowing the monkeys to perform saccade tasks. The results shown in Figs. 13 and 14 are consistent with this notion due to the observation that after nicotine microinjection, fixation-breaking saccades directed toward the affected area were more frequently generated in the gap task than the no-gap task. During the gap period of the gap task, low-frequency preparatory activities of SC neurons are increased (Dorris and Munoz 1995; Dorris et al. 1997), which has been explained by decreased inhibition from the fixation system (Dorris and Munoz 1995; Dorris et al. 1997; Gore et al. 2002). If the activities of neurons at the injection site were under control of the fixation system as described in the preceding text, such disinhibition would facilitate the effects of nicotine on the neural activation and thus cause more frequent generation of the fixation-breaking saccades toward the affected area during the gap task. Task-dependent effects of nicotine on the reaction times of visually guided saccades (Figs. 11 and 12) also support the above discussion. Tonic elevation of the basal firing rate of SC neurons may contribute to a decrease in the saccadic reaction times in the gap and no-gap tasks. Moreover, judging from the stronger effects in the gap task (Figs. 11 and 12), it is likely that the rate of elevation of the preparatory activities after the offset of the fixation point is increased after nicotine microinjection. Based on these considerations, we suggest that nicotine microinjection increased the rate of change in the activities of neurons at the injection site in addition to elevating the baseline activity.
Spatial distribution of activities for generation of saccades
The reaction times of saccades toward the targets close to the affected area were decreased immediately after nicotine exposure (Figs. 3–5). The preparatory activities of SC neurons have been shown to correlate with saccadic reaction times (Dorris and Munoz 1998; Dorris et al. 1997; Everling et al. 1999; Sparks et al. 2000). Thus it is likely that nicotine enhanced the preparatory activation of neurons at the injection site and facilitated initiation of their presaccadic bursts. The endpoints of saccades toward the targets close to the affected area were biased to the affected area (Figs. 6–8), suggesting that the center of gravity of activities would be shifted toward the injection site (Badler and Keller 2002; Edelman and Keller 1998; Glimcher and Sparks 1993b; van Opstal and van Gisbergen 1990). Compared with the effects on the endpoints, the effects on the reaction times were more localized to the targets close to the affected area (Figs. 3 and 6, 4 and 7, and 5 and 8). This is in agreement with the assumptions of the SC model (Clark 1999; Findlay and Walker 1999; Godijn and Theeuwes 2002; Munoz and Fecteau 2002; Trappenberg et al. 2001) that saccades were triggered when the activities of neurons within a restricted region reach a constant threshold (Pare 2003), whereas the endpoints of the saccades were determined by the widespread activities across the SC (Lee et al. 1988; Robinson 1972). Our results do not exclude other possible mechanisms. For instance, the threshold for triggering saccades might be changed after nicotinic activation if the firing rates of neurons within the local region of the SC are not so strictly related to the initiation of saccades (Liston and Krauzlis 2003). Simultaneous recordings of neural activity during nicotine exposure are required to establish whether the assumptions of the SC model hold true even after excitatory perturbation is directly applied to the SC.
Curved trajectories of saccades
It has recently been shown that the trajectories of target-directed saccades can be curved by presentation of distractors (Arai et al. 2004; Doyle and Walker 2001, 2002; McPeek and Keller 2001; McPeek et al. 2003; Port and Wurtz 2003). When curved saccades are generated, the spatial distribution of activities in the SC motor map is dynamically
) and B (
) in the inset, respectively. C: mean reaction times against target eccentricities. The time windows for the baseline and injection data were included the 15-min period immediately before and after the start of the nicotine microinjection, respectively. · · ·, the amplitude of the center of the affected area.
and 
and 
and 
, data from monkeys M, D, and H, respectively. 
