INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In daily life, when an interesting object moves slowly across the visual field, primates use smooth pursuit eye movements to keep the retinal image of the object from slipping so that the image is not blurred. During movement of the head and/or whole body, the smooth pursuit system does not work independently but interacts with the vestibular system to maintain precise of eye movements in space (i.e., gaze movement). This interaction requires calculation of gaze-movement commands in order to match the eye-velocity in space to the actual target velocity (see Robinson 1981 for review). To drive ocular motoneurons during smooth gaze tracking, further signal conversions are necessary: 1) omnidirectional gaze velocity commands must be sorted into roughly horizontal and vertical components, and 2) such gaze velocity commands must be converted into oculomotor commands for the control of eye movements in the orbit.

Clinical and experimental observations have shown that the cerebellum (Burde et al. 1975; Straube et al. 1997; Westheimer and Blair 1973), particularly the flocculus and ventral paraflocculus, play a role in generating smooth pursuit (Lisberger and Fuchs 1978; Miles et al. 1980; Noda and Suzuki 1979; Stone and Lisberger 1990; Zee et al. 1981; see Leigh and Zee 1999 for review). These structures, together called the floccular lobe (Krauzlis and Lisberger 1996), are also thought to play a role in generating gaze-velocity signals during horizontal smooth gaze tracking. Although the existence of eye-velocity cells have been reported, gaze-velocity Purkinje (P-) cells constitute the majority of horizontal P-cells in the floccular lobe of macaque monkeys that respond during smooth pursuit and smooth gaze tracking during head movement but little, if at all, during the vestibuloocular reflex (VOR), which requires no gaze movement (Lisberger and Fuchs 1978; Miles and Fuller 1975; Miles et al. 1980; Stone and Lisberger 1990). Removal of the floccular lobe results in an impairment of pursuit and cancellation of the VOR, but the horizontal VOR itself is rarely impaired (Zee et al. 1981).

Whether or not these findings can be extended to gaze-velocity control in all directions remains controversial since the following observations taken together seem to suggest the importance of the floccular lobe in the control of eye movements (in the orbit) rather than gaze movements in macaques (also Krauzlis and Lisberger 1994). First, preferred activation directions of gaze-velocity signals carried by floccular lobe P-cells are either roughly horizontal or vertical, consistent with the pulling directions of extraocular muscles (Krauzlis and Lisberger 1996; Miles et al. 1980; Shidara and Kawano 1993; Stone and Lisberger 1990). Second, during pitch rotation there are relatively few vertical gaze-velocity P-cells in macaques (Fukushima et al. 1999), and the activity of many vertical P-cells that respond to smooth pursuit and vertical vestibular stimulation is not related to gaze-movement. Third, even for horizontal P-cells, using squirrel monkeys, Belton and McCrea (2000) reported that half of floccular P-cells are eye-velocity P-cells whose firing rates were related to eye movements during smooth pursuit and passive whole body rotation. Moreover, although the other half of the P-cells resembled gaze-velocity P-cells, their head velocity signals tended to be less than half of the eye-velocity-related signals (Belton and McCrea 2000).

The dorsal vermis (lobules VI-VII) also contains many P-cells that respond to smooth pursuit (Kase et al. 1979; Sato and Noda 1992; Suzuki and Keller 1982, 1988a,b; Suzuki et al. 1981; see Leigh and Zee 1999 for review). Activity of many of them is modulated during horizontal head rotation, and horizontal gaze-velocity signals are also found there (Suzuki and Keller 1982, 1988a,b; Suzuki et al. 1981). Smooth pursuit eye movements are impaired by lesions of the dorsal vermis (Takagi et al. 2000; Vahedi et al. 1995) and by chemical inactivation of the bilateral caudal fastigial nuclei (Robinson et al. 1997), the main output area of the vermal lobules VI and VII (Voogd and Glickstein 1998; Yamada and Noda 1987). Pursuit-like slow eye movements are induced by electrical microstimulation within the dorsal vermis (Krauzlis and Miles 1998). These results suggest that the dorsal vermis is also involved in the control of horizontal gaze-velocity during pursuit and smooth gaze tracking. However, the interaction between pursuit and vestibular signals for vermal P-cells has been tested only in the horizontal plane (Sato and Noda 1992; Suzuki and Keller 1982, 1988a), so it is unknown whether the dorsal vermis is involved in omnidirectional gaze-velocity control. Moreover, although it seems that both the floccular lobe and the dorsal vermis participate in the control of smooth pursuit, potential differences in their involvement are still incompletely understood (cf., Fuchs et al. 1994; Stone and Lisberger 1990; Suh et al. 2000).

To address these points and also understand the role of the dorsal vermis in signal conversion during smooth gaze tracking in pursuit-vestibular interactions, we recorded simple-spike activity of vermal P-cells in trained Japanese macaques and compared their discharge characteristics with those of pursuit-responding cells in the floccular lobe from previous studies using identical task conditions (Fukushima et al. 1999). We will show that signals carried by vermal P-cells are qualitatively different from those of floccular vertical P-cells. These differences suggest that the two cerebellar regions are involved in different kinds of processing of pursuit-vestibular interactions during smooth gaze tracking. Some of these results have been presented in preliminary form (Fukushima et al. 2000b; Shinmei et al. 2000).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Three Japanese monkeys (Macaca fuscata, 4.0-6.0 kg) were used. All experiments were performed in strict compliance with the Guide for the Care and Use of Laboratory Animals (DHEW Publication NIH85-23, 1985). Specific protocols were approved by the Animal Care and Use Committee of Hokkaido University School of Medicine (protocol 9290). Our methods for animal preparation and training are described in detail elsewhere (Fukushima et al. 1996b, 1999). Briefly, each monkey was sedated with ketamine hydrochloride (5 mg/kg im) and then anesthetized with pentobarbital sodium (Nembutal; 25 mg/kg ip) and additional anesthesia (0.5-1.0% halothane mixed with 50% nitrous oxide and 50% oxygen) was administered as necessary. Under aseptic conditions, two head-holders were installed over the skull, and a scleral search coil was implanted on the right eye to record vertical and horizontal components of eye movement (Fuchs and Robinson 1966; Judge et al. 1980). Analgesics (pentazocine, 0.2 mg/kg) and antibiotics (penicillin G sodium, 20,000 U) were administered postsurgically to reduce pain and prevent infection.

Monkeys' heads were held firmly in the primate chair in the stereotaxic plane. The chair was fixed to a turntable and was rotated sinusoidally in the vertical, horizontal, or oblique planes. A tangent screen was positioned 75 cm in front of the animals' eyes and subtended 60 by 80° of visual angle. The monkeys were trained for apple juice reward to track a laser spot (0.2° diam) that was back-projected onto the tangent screen in an otherwise completely dark room. The target moved sinusoidally in vertical, horizontal, or two oblique directions at 45 and 135° polar angle. Target position signals were first calibrated before a recording session by placing the target at known horizontal and vertical locations (0, ±10, and ±20°). Eye position signals were then calibrated to the target by requiring the animal to fixate the stationary target or pursue a slowly moving one.

Recording procedures and behavioral paradigms

Recordings were made in the dorsal vermis at Post. 13-16 mm and Lat. 2-4 mm in stereotaxic coordinates. All stimuli were applied sinusoidally using five task conditions as in our previous studies (Fukushima et al. 2000a). For our search stimulus, the target was moved obliquely (at 0.5 Hz, ±10°) in association with chair rotation at the same frequency either in the yaw or pitch plane. P-cells were identified by the existence of complex spikes as described previously (Fukushima et al. 1993, 1996a, 1999). Once responsive single cells were encountered as judged visually and on the audio monitor, smooth pursuit responses were tested in four planes (vertical, horizontal, and 2 diagonal planes at 45° angles) at 0.5 Hz (±10°) to determine the preferred direction for pursuit activation without chair rotation (Fig. 1, 1, smooth pursuit).

View larger version (31K): [in this window] [in a new window]   Fig. 1. Representative discharge of 3 types of vermal Purkinje- (P-) cells. Spike rasters and histograms with fit sinusoid (thin line) of gaze-velocity (A), eye/head-velocity group I (B), and eye/head-velocity group II (C) P-cells during smooth pursuit (A1, B1, and C1), vestibuloocular reflex (VOR) cancellation (A2, B2, and C2), VOR ×1 (A3, B3, and C3), and VOR ×2 (A4, B4, and C4). Traces from top to bottom in the 1st column are de-saccaded averaged eye velocity, target velocity, and chair velocity in the preferred direction for the cell shown in A. Stimulus frequency is 0.5 Hz (±10°) in all cases.

To dissociate eye movement in the orbit from that in space (i.e., gaze), and to compare the results with previous observations (Fukushima et al. 1999, 2000a), we employed two standard tracking conditions. During the VOR cancellation task (Fig. 1, 2), the monkeys tracked a target that moved in space with the same amplitude, direction, and phase as the chair. This condition required the monkeys to cancel the VOR so that the eyes remained virtually motionless in the orbit, and gaze therefore moved with the chair. (Although we called this condition "VOR suppression" previously, the term "VOR cancellation" is now commonly used.) During VOR ×1 (Fig. 1, 3), the target stayed stationary in space during chair rotation, which required perfect VOR and no gaze movement. In addition, we tested VOR enhancement (×2) condition (Fig. 1, 4). In this task, the target moved in space with the same amplitude and in the same plane as the chair but in antiphase. This condition required the monkeys to increase eye movement in the orbit to twice the chair movement but with the opposite direction and thereby caused a gaze movement equal to that produced during VOR cancellation but in the direction opposite to the chair movement. For some cells, responses to a variety of frequencies were examined (0.1-1.0 Hz, at ±10°) for each of the tracking conditions.

To examine how visual target inputs affected vestibular responses, chair rotation was also applied in complete darkness without a target. The monkeys were not required to perform any particular task during this condition but were kept alert by occasional drops of apple juice.

To examine whether pursuit-related neurons receive retinal image motion information about target movement, the stationary monkeys were rewarded for fixating a stationary laser spot (1st fixation target, 0.2° diam) while a second laser spot (0.6° diam) moved sinusoidally along one of the four directions. The fixation spot was extinguished periodically while the test laser spot was presented continuously. Extinction of the fixation spot cued the monkey to track the second spot. This procedure was used to reward the monkeys for pursuing the second spot so that it would not become behaviorally meaningless, requiring the monkeys attended to it. Responses to saccades were also examined. The monkeys first fixated a stationary target at the center of the screen. After 1-2 s, the target jumped to a new position 5 or 10° away from the center, and the monkeys made a saccade to the visible target as described previously (Fukushima et al. 2000a).

Data analysis

The data were analyzed off-line as previously described (Fukushima et al. 1999, 2000a). Simple spike discharge was discriminated with a dual time-amplitude-window discriminator and digitized together with eye position, chair position, and target position signals at 500 Hz using a 16-bit A/D board. All position signals were differentiated by software to obtain velocity. For this, we calculated the slope using a "sliding box car" method (Rabner and Gould 1975) using a least-squares fit for each of 9-11 consecutive points (Fukushima et al. 1996c). Those occasional bursts or pauses in cell discharge associated with saccades were marked manually and removed from the analysis. Rasters and histograms were constructed by averaging between 10 and 30 cycles. Each cycle was divided into 64 equal bins together with averaged velocity. Marked bursts or pauses in discharge did not contribute to the histograms, although they are shown in the cycle rasters.

To quantify responses, a sine function was fit to the cycle histograms of the cell's discharge, exclusive of the bins with a zero spike rate, by means of a least-squared error algorithm. Responses that had a harmonic distortion (HD) of more than 50% or a signal to noise ratio (S/N) of less than 1.0 were discarded; S/N was defined as the ratio of the amplitude of the fitted fundamental frequency to the root mean square amplitude of the 3rd through 8th harmonics and HD as the ratio of the amplitude of the 2nd harmonic to that of the fundamental (Wilson et al. 1984). The phase shift of the peak of the fitted-function relative to upward or rightward stimulus velocity was calculated as a difference in degrees. Gain of cell response was calculated as the peak amplitude of the fundamental component fitted to the cycle histogram divided by the peak amplitude of the fitted stimulus velocity (i.e., target velocity for pursuit or chair velocity for tasks that included chair rotation, Fig. 1). For each cell response, gain 0.10 spikes/s per deg/s was taken as significant modulation.

For responses to oblique stimulus directions, radial stimulus velocity was first calculated as the square root of the sum of the squares of the vertical and horizontal components, and gain was calculated by dividing amplitude of modulation of cell activity by radial stimulus velocity. The phase shift of responses for diagonal preferred directions was calculated relative to the rightward component of eye-, gaze-, or stimulus-velocity. Eye-velocity sensitivity was calculated during smooth pursuit or VOR ×1 as the peak amplitude of the fundamental component of the cell's response divided by the peak amplitude of the fitted eye-velocity.

Preferred direction of a cell's response was estimated by the method of Krauzlis and Lisberger (1996) using a Gaussian function. Responses to eight polar directions along the four stimulus planes were examined. We estimated the Gaussian fit by plotting gain (re: stimulus velocity) as in previous studies (e.g., Fukushima et al. 2000a). Gains were plotted as positive for the increasing discharge, and as negative for the direction for which discharge rate decreased.

To analyze retinal image-motion responses, all traces were aligned with the second target cycle. Traces that contained saccades or slow eye movements were removed since they were indicative of the monkeys' failure to fixate the stationary target, and only those traces with eye position changes of less than 1° during each cycle were analyzed (see Fig. 8) as in previous studies (Fukushima et al. 2000a). Mann-Whitney's U test and ² test were used to examine statistical significance.

Histological procedures

Near the conclusion of the recording period in each monkey, the sites of pursuit-related cell activity were marked by iron deposits or by electrolytic lesions made by passing negative current through the microelectrode (20 µA for 30 s). After recording was completed, each monkey was deeply anesthetized with Nembutal (50 mg/kg ip) and perfused with physiological saline followed by 3.5% Formalin for reconstruction of electrolytic lesions and with both Formalin and 2% ferrocyanide (Suzuki and Azuma 1976) for iron marking. After histological fixation, sagittal sections were cut at 100-µm thickness on a freezing microtome. The sections were then stained for cell bodies and fibers using the Klüver-Barrera method (1953), and the locations of recording sites were verified microscopically.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In the cerebellar dorsal vermis (lobules VI, VII; see Recording location) of three monkeys, we examined discharge characteristics of 95 cells that responded to smooth pursuit and/or chair rotation. Of those, 58 cells were identified as P-cells by their complex spikes (Fukushima et al. 1993, 1996a, 1999). Of the 58, 53 cells were tested during both smooth pursuit and chair rotation, and the remaining 5 were tested only during smooth pursuit. Five of the 53 P-cells responded only during chair rotation. Of the 48 P-cells that responded to both pursuit and chair rotation, 47 cells were recorded long enough to allow detailed analysis during the three (Fig. 1, A1-A3), and 45 cells were tested for all the 4 conditions (Fig. 1, A1-A4).

Overall mean (±SD) eye gains (re: stimulus velocity) were 0.75 ± 0.15 during smooth pursuit, 0.18 ± 0.12 during VOR cancellation, 0.99 ± 0.17 during VOR ×1, 1.86 ± 0.28 during VOR ×2, and 0.94 ± 0.14 during chair rotation in complete darkness.

Comparison of preferred activation directions of P-cells in the dorsal vermis during smooth pursuit and VOR cancellation

Gaze movement can be performed either by eye movement alone without head movement or during whole body rotation without appreciable eye movement by canceling the VOR. As illustrated in Fig. 1, A-C, for three representative cells, all pursuit-responding P-cells in the dorsal vermis also responded during VOR cancellation (Fig. 1, A1-C1, cf. A2-C2, respectively), and the majority of them (37/53 = 70%) showed similar preferred directions during these two tasks (e.g., Fig. 1, A1 and A2, B1 and B2). A representative P-cell response is shown in Fig. 2 during eye (smooth pursuit) and gaze (VOR cancellation) movement in different directions (Fig. 2, A2-A5, B2-B5). A clear modulation in firing rate is noticeable in this cell when downward pursuit and VOR cancellation were performed (Fig. 2, A3-A5, B3-B5). The peak discharge was near peak stimulus velocity. Directional tuning of this cell is plotted in Fig. 2, A6 and B6 (see METHODS). Preferred directions estimated by the Gaussian fit (Krauzlis and Lisberger 1996) were downward during both task conditions.

View larger version (38K): [in this window] [in a new window]   Fig. 2. Directional selectivity of a representative vermal P-cell during smooth pursuit (A2-A5) and VOR cancellation (B2-B5) in different stimulus directions (middle insets). A1 and B1: the idealized movement of gaze, target, and chair. Stimulus frequency is 0.5 Hz (±10°) in all cases. Directional tuning curves for smooth pursuit (A6) and VOR cancellation (B6) of this cell estimated using a Gaussian function by plotting gain (re: stimulus velocity). Gain values are plotted as positive for increasing discharge and as negative for decreasing. L, R, U, D indicate left, right, up, and down, respectively.

Figure 3 summarizes preferred activation directions of individual vermal P-cells during smooth pursuit (Fig. 3A1) and VOR cancellation (Fig. 3B1) in a polar format. Although we encountered P-cells with downward preferred directions more frequently than upward, preferred directions for the population of vermal P-cells are distributed in all directions for both task conditions. Their distribution seems to be different from that of floccular lobe P-cells, which was sorted into roughly horizontal and vertical components (Fukushima et al. 1999; Krauzlis and Lisberger 1996; Shidara and Kawano 1993). This suggests that signals carried by vermal P-cells are not sorted yet into horizontal and vertical components (see DISCUSSION).

View larger version (23K): [in this window] [in a new window]   Fig. 3. Preferred activation directions and gains of individual vermal P-cells. A1 and B1 summarize preferred directions of all vermal P-cells examined during smooth pursuit (A1) and VOR cancellation (B1) with respect to the recording side (ipsi). A2-A4 and B2-B4 show preferred directions and gains of 3 groups of vermal P-cells during pursuit (A2-A4) and VOR cancellation (B2-B4). Response of each cell is displayed in polar coordinates with the preferred direction plotted as polar angle and gain (re: stimulus velocity) as a radius (see inset). , gains exceeded plotting scale.

Response phase of the majority of vermal P-cells was mostly in phase with eye velocity during smooth pursuit and with chair velocity during VOR cancellation as illustrated in Figs. 1 (A1 and A2, C1 and C2) and 2 (A4 and B4). Mean (±SD) phase shifts for all cells relative to target velocity were 1.6 ± 39.1° lead, but eye velocity lagged target velocity with the mean of 4.3 ± 5.2°. As a result, the mean phase shift between cell response and eye velocity was 5.8 ± 39.7° lead. Figure 4A1 summarizes phase shifts (re: eye velocity) of individual P-cells along the preferred axis (METHODS) during smooth pursuit. Figure 4B1 summarizes phase shifts (re: chair velocity) of individual P-cells during VOR cancellation. The mean phase shift was 13.0 ± 41.8° lag with the median of 2° lead during VOR cancellation (Fig. 4B1).

View larger version (35K): [in this window] [in a new window]   Fig. 4. Histograms of phase shift of individual vermal P-cell modulation during smooth pursuit (A; re: eye velocity) and VOR cancellation (B; re: chair velocity). A1 and B1 summarize phase shift of all vermal P-cells examined during smooth pursuit (A1) and VOR cancellation (B1). A2-A4 and B2-B4 show phase shifts of the three groups of vermal P-cells as indicated.

To compare preferred activation directions of individual P-cells in these two task conditions, in Fig. 5A we plotted the preferred direction during VOR cancellation against the preferred direction during smooth pursuit for each cell. The points for the majority of P-cells (37/48 = 77%, open and filled circles) are distributed around the unity slope line confirming that the preferred directions are similar in the two tasks (Fig. 5A, dashed line). The preferred directions of the minority of P-cells (11/48 = 23%) are different by more than 90°, and some showed oppositely directed preferred directions (Fig. 5A, squares on 2 straight lines). Response magnitudes of these cells in these two task conditions are compared in Fig. 5B by plotting head velocity sensitivity during VOR cancellation against eye velocity sensitivity during pursuit. It is apparent that pursuit-responding P-cells in the dorsal vermis consist of different groups in terms of preferred directions and response magnitudes in the task conditions that require pursuit-vestibular interactions.

View larger version (23K): [in this window] [in a new window]   Fig. 5. Comparison of preferred activation directions during smooth pursuit with that of VOR cancellation for individual vermal P-cells (A) and sensitivity to eye velocity during smooth pursuit versus the sensitivity to head velocity during VOR cancellation for the 3 groups of P-cells (B). Dashed and straight line slope in A = 1. Filled circle, open circle, and squares indicate gaze-velocity, eye/head velocity group I, and group II P-cells, respectively, as indicated in B. Linear regressions are shown for each cell group with correlation coefficients in B.

Classification of P-cells in the dorsal vermis during pursuit-vestibular interactions

Since all vermal P-cells responding to pursuit (eye) in this study also responded during VOR cancellation (head) with similar preferred directions for the majority (77%) of them (Fig. 5A), their activity may be related to gaze-velocity. To test this possibility, we examined the responses of 47 vermal P-cells during the 3 tasks that dissociate eye movement in the orbit from eye movement in space (Fig. 1, 1 and 3; see METHODS). The majority of them (36/47 = 77%) showed similar preferred directions during pursuit and VOR cancellation. We classified P-cells as "gaze-velocity" if they met the standard criteria that characterized the horizontal gaze-velocity P-cells of Lisberger and Fuchs (1978; also Fukushima et al. 1999, 2000a): 1) modulation occurred for movements of the eye (pursuit) and the head (VOR cancellation) in the same direction, 2) modulation during one of these two tasks was less than twice that during the other, and 3) modulation during the VOR ×1 was less than that during VOR cancellation. Based on these criteria, 19 of the 36 were classified as gaze-velocity (Fig. 5, A and B, filled circles); the other 17 did not satisfy these criteria and were classified as eye/head-velocity group I P-cells (Fig. 5, A and B, open circles). The remaining 11 of the 47 were classified as eye/head-velocity group II P-cells (Fig. 5, A and B, squares; see EYE/HEAD-VELOCITY GROUP II P-CELLS). To test cell responses during pursuit-vestibular interaction conditions further, we examined responses of 45 of the 47 P-cells during VOR enhancement (×2) in which the target moved with equal amplitude as, but in the opposite direction to, the chair (Fig. 1, 4). All gaze-velocity neurons responded maximally for opposite directions (i.e., phase reversal) during VOR ×2 and cancellation.

GAZE-VELOCITY P-CELLS. Figure 1A shows the discharge of a representative gaze-velocity P-cell whose preferred direction during smooth pursuit was 315° in polar coordinates (Fig. 1A1). During VOR cancellation, it responded well for right-down (315°) head rotation (Fig. 1A2). During the VOR ×1 (target-stationary-in-space) condition, in which gaze was nearly stable (not shown), the modulation was minimal (Fig. 1A3). During VOR ×2, the response reversed its phase (Fig. 1A4) compared with that during VOR cancellation (Fig. 1A2). Thus the discharge pattern of this cell during these conditions defines a gaze-velocity P-cell with an oblique preferred direction.

View larger version (20K): [in this window] [in a new window]   Fig. 6. Amplitude of modulation plotted against peak gaze-velocity for 7 representative gaze-velocity P-cells (A-G) during different pursuit-vestibular interaction conditions at different stimulus frequencies as per Lisberger and Fuchs (1978). Linear regressions are shown for each cell with correlation coefficients.

EYE/HEAD-VELOCITY GROUP I P-CELLS. As described above, 16 of the 34 P-cells that had similar preferred directions during pursuit and VOR cancellation did not meet the gaze-velocity criteria and were classified as eye/head-velocity group I P-cells. A representative response is shown in Fig. 1B for a P-cell whose preferred direction during pursuit was 90° in the polar coordinates (Fig. 1B1). It clearly responded during VOR cancellation (Fig. 1B2) somewhat similar to gaze-velocity P-cells (Fig. 1, A1 vs. A2) and also during VOR ×1 when gaze remained stationary with the magnitude comparable with that during pursuit but with phase reversal (Fig. 1, B3 vs. B1). Moreover, its response during VOR ×2 (Fig. 1B4) was stronger compared with that during VOR cancellation. Thus the responses of this P-cell were not related to gaze-movement. This observation was confirmed by the loss of correlation between eye velocity sensitivity during smooth pursuit and head velocity sensitivity during VOR cancellation (Fig. 5B, open circles, r = 0.09). Also plotted in this figure are eye/head-velocity group II P-cells that did not show any significant correlation (squares; see EYE/HEAD-VELOCITY GROUP II P-CELLS).

View larger version (20K): [in this window] [in a new window]   Fig. 7. Comparison of smooth pursuit and vestibular-related responses of the 3 groups of vermal P-cells (keys) in their preferred directions. A: eye velocity sensitivity during VOR ×1 are plotted against eye velocity sensitivity during smooth pursuit. B: gains during VOR in complete darkness (re: chair velocity) are plotted against gains during VOR ×1. Head-velocity P-cell in B indicates a cell that responded to chair rotation but not during pursuit. Linear regression is shown for all 15 P-cells in B.

EYE/HEAD-VELOCITY GROUP II P-CELLS. The remaining 11 of the 47 vermal P-cells had oppositely directed (n = 9) and orthogonal (n = 2) eye and head movement sensitivity during pursuit and VOR cancellation. We call these cells eye/head-velocity group II P-cells in this study. An example is shown in Fig. 1C for a P-cell whose preferred direction during pursuit was 315° in the polar coordinates (Fig. 1C1). Although activity clearly increased toward 135° during VOR cancellation (Fig. 1C2), the responses during VOR ×1 (Fig. 1C3) and VOR ×2 (Fig. 1C4) were similar. Typically, modulation of this cell group was stronger during vestibular stimulation (Fig. 1, C2-C4) than during pursuit (Fig. 1C1).

VESTIBULAR RESPONSES IN COMPLETE DARKNESS. To investigate how visual inputs affected vestibular responses of vermal P-cells during VOR ×1, we examined responses of 15 eye/head-velocity P-cells (13 group I and 2 group II) during VOR in complete darkness. Figure 7B compares gains of these 15 cells (circles and squares, respectively) with their gains during VOR ×1 during chair rotation in complete darkness. This figure also includes five gaze-velocity P-cells (filled circles) and one P-cell (open triangle) that did not respond to pursuit. A significant correlation with a slope close to one (0.96) was observed between gains during chair rotation in complete darkness and those during VOR ×1 for eye/head-velocity group I P-cells (r = 0.71, P < 0.01). These results indicate that their responses were induced primarily by vestibular (but not visual) inputs.

Retinal image-motion response of vermal P-cells

For accurate gaze tracking of a moving target during whole body rotation, target-velocity-in-space information is necessary, and it can be computed from gaze-velocity by adding retinal image-slip-velocity of the target (Robinson 1981). It has been shown that P-cells in the dorsal vermis respond to retinal image-slip-velocity (Kase et al. 1979; Suzuki and Keller 1988a,b; Suzuki et al. 1981). To examine whether our pursuit-related vermal P-cells also respond to retinal image-motion, we tested the response of 13 P-cells (4 gaze-velocity, 5 eye/head-velocity group I, 3 eye/head-velocity group II, and 1 pursuit cell that was not tested for chair rotation) to a second laser spot moving sinusoidally while the monkeys fixated the stationary target. Of the 13, the majority (n = 8; 3/4 gaze-velocity, 2/5 eye/head-velocity group I, 2/3 eye/head-velocity, and 1/1 pursuit) showed a clear modulation for the movement of the second spot, while the remaining 5 showed no consistent response. Representative activity is shown in Fig. 8 for a gaze-velocity P-cell that had a preferred direction (inset) toward 315° in polar coordinates during pursuit (Fig. 8A) and VOR cancellation (not shown). During fixation with the second target task, this cell increased activity when the second spot moved toward 315° (Fig. 8B). The monkey fixated the first stationary spot well in this condition (eye position trace in Fig. 8B), and eye gain (re: velocity of 2nd spot) was only 0.056 during this task compared to an eye gain of 0.78 during smooth pursuit (Fig. 8A). Gains of this cell during pursuit and fixation with the second target tasks were 0.67 and 0.27 spikes/s per deg/s, respectively. If the visual response during the fixation with the second target task had been induced by small residual pursuit, the modulation should have been 0.67 × (0.056/0.782) = 0.048 spikes/s per deg/s. The fact that we actually observed modulation at 0.27 spikes/s per deg/s suggests that the modulation of cell activity during this task cannot reflect eye velocity. Rather, this cell responded to motion of the second target spot. Similarly, the average eye gains for the eight responding P-cells during pursuit and fixation with the second target tasks were 0.80 ± 0.12 and 0.04 ± 0.03, respectively, but the mean cell gain during the fixation with the second target task was one-third (0.34) of the gain during smooth pursuit. Figure 8C compares gains of the P-cells during these two tasks.

View larger version (25K): [in this window] [in a new window]   Fig. 8. Retinal motion response of vermal P-cells during the fixation with 2nd target. A and B: representative response of a gaze-velocity P-cell during pursuit (A) and fixation with 2nd target (B). C: gain during fixation with 2nd target plotted against gain during smooth pursuit for each P-cell with response groups indicated in the key. Asterisk indicates response of the P-cell shown in A and B. Pursuit P-cell in C indicates a pursuit-responding cell that was not tested for vestibular stimulation. Solid line in C slope = 1. LU and RD, left up and right down. + and , increasing and decreasing discharge, respectively.

Addition of eye- and head-velocity sensitivity of vermal P-cells during pursuit-vestibular interactions

Lisberger and Fuchs (1978) suggested that smooth pursuit eye velocity signals cancel vestibular head velocity signals during VOR ×1, and that the activity of floccular horizontal gaze-velocity P-cells could be well predicted by linear vector addition of separate eye- and head-velocity components. We looked for this linear interaction in the responses of vermal P-cells to understand whether discharge properties of our vermal P-cells are similar to those of floccular horizontal gaze-velocity P-cells. A representative analysis is shown in Fig. 9 (A-C) for three cells (gaze-velocity, eye/head-velocity group I, and eye/head-velocity group II P-cells, respectively). The solid curves are the sine functions fit to the cycle histograms of these P-cells. Dashed sine functions in Fig. 9 (A3-C3, VOR ×1) are the responses predicted by adding the response during VOR cancellation (vestibular input) to the oppositely directed smooth pursuit response for each cell. Similarly, dashed sine functions in Fig. 9 (A4-C4, VOR ×2) are predicted responses by adding the VOR ×1 response and the oppositely directed smooth pursuit response for each cell. Predicted responses in these three cells had correct phases, although the response magnitudes varied.

View larger version (31K): [in this window] [in a new window]   Fig. 9. Predicted and actual responses of 3 groups of pursuit-responding vermal P-cells (A-C) presented in Fig. 1 during pursuit and vestibular stimulation. Solid lines are sine functions fitted to the cell discharge during different task conditions. Dashed sine functions in A3, B3, and C3 (VOR ×1) are responses predicted by adding VOR cancellation response and oppositely directed smooth pursuit response for each cell. Dashed sine functions in A4, B4, and C4 (VOR ×2) are responses predicted by adding VOR ×1 response and oppositely directed smooth pursuit response for each cell. D: frequency histograms of phase difference between predicted response and actual response during VOR ×1 for the 3 groups of P-cells indicated in the key (D1). D2: frequency histograms of phase difference between predicted response and actual response during VOR ×2 for the 3 groups of P-cells.

We performed similar analyses for 43 (18 gaze-velocity, 16 eye/head-velocity group I, and 9 eye/head-velocity group II) of the 45 P-cells. Since preferred directions of the remaining two eye/head-velocity group II P-cells during pursuit and VOR cancellation were orthogonal (see above), they were excluded from the analysis. Figure 9D compares phase difference between predicted responses and actual responses for the three groups of P-cells during VOR ×1 and ×2 tasks. The majority of them showed phase differences of less than 90° with the overall means of 12.1 ± 86° lag during VOR ×1 and 4.3 ± 74° lead during VOR ×2, suggesting that actual responses were fairly well predicted by the linear addition of eye- and head-velocity sensitivities for these P-cells. Figure 10 compares predicted and actual gains of the three groups of P-cells during these two tasks. Correlations between predicted and observed gains during these two task conditions were slightly higher for eye/head-velocity group I and II P-cells (Fig. 10, A2 and A3, B2 and B3) than gaze-velocity P-cells (Fig. 10, A1 and B1) with higher correlation coefficients and with their linear regressions closer to the slope of one. Nevertheless, significant correlation was observed in all three cell groups with the slopes ranging from 0.56 to 1.60 with the mean of 0.91, indicating that eye and head velocity signals add linearly in vermal P-cells including the majority of gaze-velocity P-cells, similar to horizontal gaze-velocity P-cells in the floccular lobe (Lisberger and Fuchs 1978).

View larger version (22K): [in this window] [in a new window]   Fig. 10. A: comparison of predicted and actual gains for the 3 groups of vermal P-cells (A1-A3) during VOR ×1. Predicted gains during VOR ×1 (response during VOR cancellation  response during smooth pursuit) are plotted against actual gains during VOR ×1. B: comparison of predicted and actual gains during VOR ×2 for 3 groups of P-cells (B1-B3). Predicted gains during VOR ×2 (response during VOR ×1  response during smooth pursuit) are plotted against actual gains during VOR ×2. Linear regressions are shown for each cell group with correlation coefficients.

Activity during saccades

Of the 58 P-cells that responded during smooth pursuit, 30 were also tested during saccades (METHODS). Nine of the 30 showed burst discharge associated with saccades. Of the nine, the bursts of six P-cells were followed by tonic discharge. Three of these six P-cells had response phase shifted to eye position during pursuit (±58-86°; re: eye velocity), whereas the phase of the remaining 3 and 3 other P-cells that showed only bursts were closer to velocity (less than ±26°; re: eye velocity).

Incompletely identified cells

Among the 37 cells tested for smooth pursuit and/or chair rotation sensitivity in which complex spikes were not discernable, 13 were classified as gaze-velocity, 10 as eye/head-velocity group I, and 5 as eye/head-velocity group II. Of the remaining nine cells, three responded only during vestibular stimulation but not during smooth pursuit, so we classified them as head-velocity cells. The other six cells were tested only during smooth pursuit (pursuit cell). Of the cells with responses during smooth pursuit and VOR cancellation, the preferred activation directions were distributed in all directions. Thus discharge characteristics of our incompletely tested and unidentified vermal neurons seemed to be similar to those of the identified P-cell population.

Recording location

Figure 11 summarizes the recording locations in the three monkeys. Recording sites from neighboring sections (L2.0-L4.0) in each monkey (G, S, and C) were collapsed on the three representative sections to indicate the rostrocaudal extent of the responding sites. All responding cells were found in the lobules VI and VII of the dorsal vermis (Fig. 11). We often recorded cells with different properties in the same tracks, suggesting that they are intermingled in these areas.

View larger version (38K): [in this window] [in a new window]   Fig. 11. Reconstructed recording sites of vermal P-cells in a sagittal plane. Response groups are indicated in the key. Vestibular P-cell is a cell that responded to vestibular stimulation but that did not respond to pursuit. Pursuit P-cell is a cell that responded to pursuit but whose vestibular sensitivity was not tested. R, right; L, left.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study has extended previous results testing vermal lobe responses in the horizontal plane (Sato and Noda 1992; Suzuki and Keller 1982, 1988a; Suzuki et al. 1981) to include testing for vertical and oblique motion. The major conclusion is that pursuit-responding P-cells in the vermal lobules VI and VII consists of different groups that have preferred directions not sorted into either horizontal or vertical directions and a substantial percentage of gaze-velocity P-cells.

Different groups of pursuit-related P-cells in the cerebellar dorsal vermis during pursuit-vestibular interactions

Using behavioral paradigms that dissociate eye movement in the orbit from that in space (i.e., gaze) in various directions, we have shown that pursuit-responding individual vermal P-cells could be classified as either gaze-velocity (19/47 = 40%), eye/head-velocity group I (17/47 = 36%), or eye/head-velocity group II (11/47 = 24%) P-cells. We confirm gaze-velocity signals in the horizontal plane reported earlier in other macaque species in the dorsal vermis (Sato and Noda 1992 for Macaca menestrina; Suzuki and Keller 1982, 1988a; Suzuki et al. 1981 for Macaca fascicularis) and extend their observations by showing that gaze-velocity signals are also found in other, particularly oblique directions.

Previous studies reported some asymmetry in the percentage of preferred activation directions of individual vermal P-cells. For example, Suzuki and Keller (1988b) reported that 60% of vermal P-cells discharge for ipsilateral pursuit and 40% for contralateral pursuit in Macaca fascicularis. In our studies, P-cells with ipsilateral preferred directions outnumbered those with contralateral preferred directions in Macaca fuscata (Fig. 3A1). Fuchs et al. (1994) reported similar asymmetry in rhesus macaques; preferred activation directions of pursuit-related neurons in the caudal fastigial nucleus are either in the contralateral and/or downward (50/69 = 72%) or in the ipsilateral and/or upward directions (13/69 = 19%). In our results also, preferred directions of vermal P-cells with downward preferred directions outnumbered those with upward preferred directions (Fig. 3A1). Thus our results obtained in Japanese macaques seem to be consistent with these previous observations obtained in other macaque species.

Fuchs et al. (1994) reported the existence of many pursuit-related neurons with oblique preferred directions in the caudal fastigial nucleus. Since this nucleus receives direct inhibition from lobules VI and VII of the dorsal vermis (Yamada and Noda 1987), it seems that signals carried by output neurons in these two cerebellar regions are not sorted yet into horizontal or vertical components. It should be pointed out that since the deep cerebellar nuclei receive direct mossy fiber inputs by axon collaterals of pontine fibers projecting to the cerebellar cortex (e.g., Shinoda et al. 1992), signals in the fastigial nucleus probably are not determined solely by the overlying vermal P-cell responses.

Our results indicate that preferred activation directions of vermal P-cells during VOR cancellation were also distributed across all directions similar to their pursuit directions, although there was a paucity of upward preferred directions (Fig. 3, A1 and B1).

Comparison of response properties of pursuit-related P-cells in the dorsal vermis and the floccular lobe

Although P-cells in these two regions respond during VOR cancellation as well as smooth pursuit, there are several distinct differences in their response properties. First, in contrast to our results showing that all pursuit-related P-cells in the dorsal vermis also responded during VOR cancellation, the floccular lobe contains a substantial number of eye movement-related P-cells that do not respond during VOR cancellation in the yaw or pitch plane (Fukushima et al. 1999; Lisberger and Fuchs 1978; Miles et al. 1980). Second, gaze-velocity cells are a minority of the vertical P-cells in the floccular lobe of the same macaque species using identical task conditions (19%) (Fukushima et al. 1999), whereas gaze-velocity P-cells constituted 40% of pursuit-responding vermal P-cells, and the majority of them (31/47) had vertical preferred directions (Fig. 3, A2 and B2). Third, although vermal P-cells showed mostly velocity-related responses (Fig. 4A2), many floccular vertical P-cells in previous studies had eye position-related activity (Fukushima et al. 1999; Suh et al. 2000). Fourth, frank visual responses are reported to be minimal in most floccular P-cells (Belton and McCrea 2000; Stone and Lisberger 1990; Suh et al. 2000), whereas the activity of the majority of vermal P-cells was modulated by retinal image motion (Fig. 8B), consistent with previous reports (Kase et al. 1979; Suzuki and Keller 1988a,b; Suzuki et al. 1981). Fifth, the distribution of preferred directions of vermal P-cells (Fig. 3, A1 and B1) seem to contrast with that of floccular lobe P-cells that are either roughly horizontal or vertical (Fukushima et al. 1999; Krauzlis and Lisberger 1996; Miles et al. 1980; Shidara and Kawano 1993; Stone and Lisberger 1990). Preferred directions for the population of vermal P-cells were distributed in all directions for both smooth pursuit and VOR cancellation (Fig. 3, A and B). We therefore think that the directional organization of the vermis is qualitatively different from that of the floccular lobe. Finally, although the activity of floccular vertical P-cells was not well predicted by the linear addition of vertical pursuit and vestibular sensitivities (Fukushima et al. 1999), the activity of the majority of vermal P-cells is fairly well predicted (Figs. 9 and 10).

We do not have a compelling rationale for these differences. However, in contrast to horizontal eye movements that occur around a single (vertical) axis, vertical eye movements in the orbit occur along two (horizontal and torsional) axes using two pairs of muscles (i.e., vertical recti and oblique) and are more complex (Baker and Peterson 1991; Crawford et al. 1991). Similarly, the vertical VOR, which relies on the normal activity of four vertical semicircular canals and otolith inputs, differs in a fundamental way from the horizontal VOR, which relies solely on signals from two horizontal canals (see Leigh and Zee 1999 for review). Therefore control signals for orbital eye movements may be expected to be more complex for vertical than horizontal eye movement. The floccular lobe presumably is more important in eye movement control than gaze control (see INTRODUCTION). We think that this presumed complexity in the control of vertical eye movements may make simplified linear addition more difficult for floccular vertical P-cells. Consistent with this interpretation, many floccular vertical P-cells other than gaze-velocity [i.e., vertical eye/head-velocity and off-pitch vertical eye/head-velocity P-cells of Fukushima et al. (1999)] had vestibular direction tuning near the torsional plane and near the planes of vertical canals in contrast to the minority of vertical gaze-velocity P-cells that had vestibular direction tuning near the pitch plane (also Fukushima et al. 1993, 1996a for cats; Fukushima and Kaneko 1995; Powell et al. 1996). Vestibular direction tuning was not examined in vermal P-cells in this study because of mechanical limitations in our testing apparatus. We expect from previous results in monkeys and cats that vestibular direction tuning of vertical gaze-velocity P-cells in the dorsal vermis would also be near the pitch (not torsional or eye-muscle) plane (cf., Krauzlis and Lisberger 1996). It should be pointed out that vermal P-cell discharge is similar to that of periarcuate pursuit cells except for the percentages of gaze-velocity and eye/head-velocity cells using identical task conditions (Fukushima et al. 2000a).

Neural pathways for smooth gaze tracking and the cerebellar dorsal vermis

To track a slowly moving object accurately during whole body rotation, target-velocity-in-space signals must first be reconstructed in the brain by retinal image-slip-velocity of the target, eye velocity and head velocity (Robinson 1981). The latter two signals can provide gaze-velocity used to match the velocity of the eyes-in-space to target velocity. Gaze-velocity signals must eventually be converted into oculomotor signals to drive the eyes in the orbit. The medial superior temporal (MST) area is known to be essential for initiation and maintenance of smooth pursuit (see Keller and Heinen 1991; Lisberger et al. 1987 for reviews). The MST contains all the signal components needed to reconstruct target motion in space including retinal image-slip-velocity, eye velocity, and even gaze-velocity (Dicke and Thier 1999; Kawano et al. 1984; Komatsu and Wurtz 1988; Newsome et al. 1988; Sakata et al. 1983; Thier and Erickson 1992). The MST sends signals to the floccular lobe through the dorsolateral pontine nucleus (Kawano et al. 1992; see Leigh and Zee 1999 for review) and the frontal eye fields (FEF), and these two cortical areas have reciprocal connections (Stanton et al. 1993, 1995; Tian and Lynch 1996a,b; Tusa and Ungerleider 1988). Retinal image-slip-velocity of a target, eye-velocity during pursuit, and even gaze-velocity signals are also found in the periarcuate cortex in and around the FEF (Bruce and Goldberg 1985; Fukushima et al. 2000a; Gottlieb et al. 1994; MacAvoy et al. 1991; Tanaka and Fukushima 1998; Tian and Lynch 1996a). Moreover, gaze-movement- and/or retinal image-velocity-related signals are also found in the central thalamus (e.g., Schlag and Schlag-Rey 1986). It has been suggested that the MST forms an internal positive feedback circuit in the pursuit system that provides signals for the maintenance of pursuit (Newsome et al. 1988). This circuit may include both the periarcuate cortex and the central thalamus in addition to the MST.

Major outputs of the FEF regions are sent to the nucleus reticularis tegmenti pontis (NRTP) (Stanton et al. 1988), which projects primarily to the cerebellar dorsal vermis (also Gottlieb et al. 1994; MacAvoy et al. 1991; Suzuki et al. 1999; see Leigh and Zee 1999 for review). Gaze-velocity signals are also reported in the NRTP (Suzuki et al. 1996). As this study shows, a substantial percentage of gaze-velocity P-cells in the dorsal vermis carry both gaze-velocity and retinal image-slip-velocity signals, suggesting the importance of this structure in the control of gaze-movements. The gradual but significant decrease in the percentage of gaze-velocity-related neurons from the periarcuate cortex to the dorsal vermis (66% = 66/100 vs. 40% = 19/47, P < 0.02, ² test) and the subsequent increase in the percentage of eye/head-velocity neurons (34% in the periarcuate cortex vs. 60% in the vermis including group I and II) may suggest that the vermis is involved, at least in part, in signal conversion from gaze-velocity to eye-velocity for pursuit-vestibular interactions. Although we classified vermal P-cells into three groups in this study, we feel that they may not belong to three different cell types. Rather, they may constitute a single population with varying degrees of eye and head velocity sensitivities. Consistent with this view are the similar preferred directions (Fig. 3), similar response phases (Fig. 4), and similar retinal motion responses (Fig. 8) at least for gaze-velocity and eye/head-velocity group I P-cells.

One-third of our pursuit-related vermal P-cells (9/30) showed activity changes associated with visually induced saccades. Although we did not examine further the nature of saccade-related activity of vermal P-cells, neurons that respond both to saccades and pursuit have been reported previously in the cerebellar vermis (e.g., Suzuki and Keller 1988b), fastigial nucleus (e.g., Fuchs et al. 1994), and the periarcuate cortex in and around the FEF (Fukushima et al. 2000a). Surgical ablation of the dorsal vermis and chemical deactivation of the fastigial nucleus impair both smooth pursuit and visually induced saccades (Robinson et al. 1997; Takagi et al. 2000), suggesting that the same vermal pathways may control both smooth pursuit and visually induced saccades (cf., Krauzlis and Miles 1998; Krauzlis and Stone 1999). Although the details are still unknown, these observations substantiate the importance of gaze-related activity in the cerebellar dorsal vermis as well as the periarcuate cortical areas.

One may ask why there are gaze signals in so many places in the brain. We do not know the exact answer, but smooth gaze tracking of a slowly moving object is a complex function that requires multistage processing. For example, target movement is not limited in the frontal plane but occurs in three-dimensional (3-D) space. Accordingly, reconstruction of the target velocity in the brain and tracking gaze movement must be done in 3-D space by combining smooth pursuit, vergence eye movements, head velocity, and retinal signals (cf., Fukushima et al. 2001). Moreover, to maintain the target image on the fovea during tracking and to compensate for the long delays involved in processing visual motion information and/or eye/gaze velocity commands, prediction is required not only on the motor side as, for example, perseverance of ongoing movements, but also on the sensory and/or perception side about the direction and speed of the target movement. Such prediction signals must also use target-velocity-in-space and gaze-velocity signals, and they may also be needed to coordinate eye and hand movements to track and manipulate objects in 3-D space (cf., Mushiake et al. 1997). Therefore gaze signals are necessary in virtually every aspect of our daily life for our adequate behavior in 3-D space. Although gaze-related signals are represented in multiple areas in the brain, it is possible that different areas may be involved in processing of gaze signals in different aspects of the above functions (Dicke and Thier 1999; Fukushima 1997). Further studies are needed to test this speculation.