| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1Department of Physiology and Neuroscience, New York University School of Medicine, New York; 2The Center for Biomathematical Sciences, Mount Sinai School of Medicine, New York, New York; 3Laboratoire de Neurobiologie des Réseaux Sensorimoteurs, Centre National de la Recherche Scientifique Unité Mixte de Recherche 7060, Université Paris 5, Paris, France; and 4Fishberg Department of Neuroscience, Mount Sinai School of Medicine, New York, New York
Submitted 30 March 2006; accepted in final form 10 June 2006
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
90 ms) and matched both the build-up and the time course of the decay (19 s) in eye velocity storage. Latency of neuronal discharge to table steps (40 ms) was significantly longer than for eye movement (17 ms), but firing rate rose faster than eye velocity to steady-state levels. The velocity sensitivity of Area II neurons was shown to equal (±10%) the sum of eye- and head-velocity firing rates as has been observed in cerebellar Purkinje cells. These results demonstrate that Area II neuronal firing closely emulates oculomotor performance. Conjoint signaling of head and eye velocity together with the termination pattern of each Area II neuron in the vestibulolateral lobe presents a unique eye-velocity brain stem-cerebellar pathway, eliminating the conceptual requirement of motor error signaling. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
). Visuo-vestibular mismatches (toward high or low gain) reflexively produce compensatory changes to maximize oculomotor plant performance. Detection of oculomotor mistuning occurs within the vestibulocerebellum where feedback signals conveying head and eye velocity converge on and are integrated by Purkinje cells. Mismatched signals alter Purkinje cell firing activity, which, in turn, alters the activity of premotor nuclei in the brain stem. With prolonged exposure, the alterations in oculomotor output result in plastic changes within the brain stem that match the new level of performance (Lisberger 1988; Miles and Lisberger 1981).
Because feedback signals that convey both head and eye movements are required for the cerebellum to regulate and maintain oculomotor performance, identifying and analyzing the sources of cerebellar input is essential to understanding cerebellar function. The mammalian vestibulocerebellum largely receives mossy fiber input from three brain stem nuclei: vestibular, prepositus hypoglossi (NPH), and reticularis tegmenti pontis. The role of the vestibular nucleus in oculomotor behavior has a long history, and its function and signaling has been extensively characterized (Fuchs and Kimm 1975; Keller and Kamath 1975; Lisberger and Fuchs 1978b; Scudder and Fuchs 1992). For example, second-order vestibular neurons targeting the flocculus effectively encode a signal that is proportional to head velocity over a wide frequency range (0.1–1.0 Hz) (Blanks and Precht 1976; Fernandez and Goldberg 1971; Melvill Jones and Milsum 1971; see Precht 1979 for review).
In contrast to the vestibular system, there are numerous sources and types of mossy fiber input signaling eye movements to the vestibulocerebellum. In mammals, evidence of neural signaling of eye movements originating outside of motor nuclei was found first in recordings of medullary neurons in the cat NPH (Baker et al. 1975). These neurons were modulated in proportion to changes in eye position. Later work found that the NPH was an assemblage of neurons exhibiting a mixture of sensitivities to eye position and velocity, ranging from mostly position sensitive to mostly velocity sensitive (Escudero et al. 1992, 1996; Lopez-Barneo et al. 1982). Similarly, neuronal responses recorded from precerebellar cells in the NRTP were categorized as smooth-pursuit eye velocity and included separate populations encoding static eye position (Suzuki et al. 2003) and gaze velocity (Ono et al. 2004), suggesting numerous parallel eye, head, and visual inputs from the brain stem to the vestibulocerebellum (Takeichi et al. 2005).
Anatomical studies of the NPH have linked its role exclusively to oculomotor behavior. Detailed structural examination of the NPH connections have revealed major, reciprocal projections from the vestibular nuclei as well as the nodulus and flocculus of the cerebellar cortex. Other efferent targets of NPH also include the contralateral NPH, inferior olive, and extraocular motor nuclei (Brodal and Brodal 1983; McCrea and Baker 1985; also see McCrea and Horn 2006 for review). Together, the physiological and anatomical studies have demonstrated that the neurons in the NPH not only carry suitable eye-position and velocity-related signals but also are structurally situated to play an essential role in the control and plasticity of eye movements.
Whereas the mammalian NPH is heterogeneously organized, complicating further detailed study, an analogous region in the goldfish appears to be highly compartmentalized. Pastor et al. (1994b) demonstrated that goldfish have discrete, segmented areas in the hindbrain comprising neurons responding primarily to changes in either eye position or velocity. Combined, these regions are functionally analogous to the NPH in mammals. Individually, they are termed Area I and Area II and contain position- and velocity-sensitive neurons, respectively.
Recent investigations of goldfish Area I neuron morphology and physiology (Aksay et al. 2000, 2001) found that Area I predominantly targets neurons in the abducens nucleus. In contrast, the preceding report demonstrated that Area II neurons were exclusively precerebellar with 85% of Area II neurons exhibiting bilateral projections directly to the goldfish cerebellar vestibulolateral lobe (Straka et al. 2006). Area II neurons also received prominent disynaptic contralateral excitatory and ipsilateral inhibitory vestibular inputs as well as a multisynaptic excitatory drive from moving visual stimuli. Here we examine and quantify the response of Area II neurons under static conditions as well as their response to dynamic stimuli. These results represent an in-depth analysis of a precerebellar nucleus that fires with changes in eye and head velocity. We show that Area II neurons exhibit negligible position and saccadic sensitivity but respond to all changes in eye and head velocity with a proportional change in firing rate. The robust emulation of oculomotor performance by Area II suggests this nucleus to be the source of efferent head- and eye-velocity signaling found within the cerebellar vestibulolateral lobe. (Note: Supplemental movies referred to throughout the text depict the dynamic activity of Area II neurons in relation to eye movements presented in the figures.)
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Animal preparation and recording has been described previously (Aksay et al. 2000; Marsh and Baker 1997; Pastor et al. 1994a) and will be summarized briefly. Goldfish (Carassius auratus, n = 17) 10–13 cm in length from tip to peduncle were obtained from Hunting Creek Fisheries (Thurmont, MD) and maintained at 18°C in aquaria exposed to a 12:12 h light/dark cycle. Experiments were performed in compliance with the National Academy Press's Guide for the Care and Use of Laboratory Animals (1996) and the New York University School of Medicine Institutional Animal Care and Use Committee.
At least 24 h prior to an experiment, goldfish were deeply anesthetized with MS-222 (Sigma, St. Louis, MO) and the occipital bone prepared for an opening and two stabilizing bolts anchored with dental acrylic and self-tapping screws. In most circumstances, the spinal cord was compressed between C4 and C5 to minimize body motion during recordings. On the day of experiments, goldfish were wrapped in protective cotton gauze, placed in a white acrylic aquarium (27 cm diam) filled to a level below the cranial window and gently restrained in body-conforming acrylic holders that left the head and opercula uncovered. The recirculating aquarium water was maintained at 18°C with a thermoelectric device, filtered, and aerated. Stabilization was achieved by connecting the head bolts to a small aluminum plate anchored to the aquarium.
Eye position and electrophysiology recordings
Changes in eye position were measured using the scleral search coil technique (Fuchs and Robinson 1966) at a bandwidth of 320 Hz. Single-unit extracellular recordings were made using conventional physiological techniques. A previously made 5 x 5-mm window in the skull, in between the semi-circular canals and roughly centered on the occipital crest, was used to expose the brain for recording. Glass microelectrodes filled with 2 M NaCl were inserted through the facial lobe and into brain stem Area II with a three-axis micromanipulator that was positioned caudally to the animal, minimally interfering with the visual scene. Using the terminology of Duensing and Schaefer (1958), Area II neurons were recognized by their characteristic type II (contralateral) response to head velocity (HII) and type I (ipsilateral) response to eye velocity (EI) (Pastor et al. 1997; Straka et al. 2006). Behavior and neuronal activity were recorded simultaneously and remained paired during data analysis and subsequent presentation (also see supplementary movies 1 ).
Vestibular and optokinetic stimulation
The aquarium was secured to a vestibular turn-table with the goldfish head-centered in the vertical axis. A planetarium, illuminated with a blue light-emitting diode (LED) that cast a random pattern of spots on the inside walls of the aquarium, was centered above the tank and attached to the table to produce optokinetic stimulation. Both the table and planetarium were servo-controlled and stimuli were produced by a digital frequency generator (Exact Model 337). For Bode plots of the optokinetic reflex (OKR) and vestibuloocular reflex (VOR), sinusoidal stimulation was 0.065–4.0 Hz with a peak velocity of 16°/s. Bidirectional step stimuli varied in amplitude (4–32°/s) at 0.125 Hz. Constant velocity optokinetic steps were 16°/s for periods of 60–120 s. Because the planetarium was attached to the turn-table, VOR cancellation (gain 0) was achieved by eliminating the command signal to the illuminated planetarium. All other vestibular stimulation was performed in the dark to remove visual feedback.
Data storage and analysis
Voltages from the search coils and extracellular amplifier were acquired with hardware (Digidata 1200B or 1320A) and software made by Axon Instruments (Union City, CA) at 15 kHz. Spike sorting and data processing was accomplished using custom written routines in MATLAB (The Mathworks; Natick, MA) and have been previously described (Aksay et al. 2000). Additional routines are outlined below. In addition, unless otherwise stated, results are reported as means ± SE.
Prior to analysis, fast phases in eye velocity were identified by inspection of the eye-acceleration profile, a threshold was set, and an area 0.2 s before and 0.2–0.4 s after the peak eye acceleration was removed by computer. Blinks (Easter 1971; Pastor et al. 1991) and irregular eye movements were manually removed and excluded from analysis.
The amplitude, phase, and offset of eye velocity and neuronal firing rate in response to visual and vestibular sinusoidal stimuli were determined by least-squares fitting of a sinusoidal function over multiple stimulus cycles. Eye-velocity gain and neuronal firing rate responses, with respect to the stimuli, were calculated as the ratio of the peak-to-peak amplitude of the fitted sinusoid to the peak-to-peak amplitude of the stimulus velocity.
Eye velocity and neuronal firing rate responses to visual and vestibular bidirectional step velocity stimuli were averaged over 7–10 stimulus cycles. Latencies of eye velocity and neuronal firing rate were defined as the interval between stimulus onset and the point when the acceleration profile (i.e., derivative of eye velocity or firing rate) crossed the first SD of the acceleration mean and continued to increase. Mean acceleration ± SD were calculated for 1 s of the recording preceding the step. Eye-velocity gain for bidirectional steps was simply the ratio of average peak eye velocity to peak stimulus velocity. Neuronal activity sometimes ceased in response to steps in the nonpreferred direction; thus neuronal step response was defined as the ratio of the change in mean firing rate to zero-to-peak stimulus velocity.
Position sensitivity (K) was measured during 5-min periods of spontaneous eye movements in the light or dark and was the slope of the firing rate versus eye position. Saccadic eye-velocity sensitivity (R) was determined by comparing the difference in burst amplitude between the peak firing rate during the saccade and the mean firing rate during the previous fixation. Optokinetic and vestibular sensitivity measured during sinusoidal stimuli or step stimuli was calculated as slope of firing rate versus eye or head velocity in the neuron's preferred direction (Lisberger et al. 1994). The slope was calculated by least squares regression of the firing rate on eye velocity beginning 2.5°/s from the other side of zero velocity, e.g., –2.5°/s to >20°/s for a leftward preferred direction.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Single-unit extracellular recordings of goldfish Area II neurons were made bilaterally in hindbrain Area II. This nucleus extends rostrally and caudally 450 µm, centered on the caudal margin of the facial lobe, and is positioned 800 µm laterally from either side of the medial longitudinal fasciculus (Pastor et al. 1994b; Straka et al. 2006).
Because Area II neurons respond to visual stimuli, changes in firing activity during the testing paradigms might be influenced by whether the light was on or off. To account for this, several minutes of spontaneous saccadic activity were recorded while in the dark and light. Most Area II neurons exhibited a low, tonic firing rate that remained unchanged (2-tailed, paired t-test: P = 0.86) whether the animal was in the light (10.4 Hz ±1.4) or in the dark (10.5 Hz ±1.8; n = 17 neurons, 15 animals), also see movie 1.
Initial observations of Area II neurons noted a lack of position and saccadic sensitivity compared with neighboring Area I neurons (Pastor et al. 1994b). In general, the response of Area II neurons to changes in eye position and saccadic velocity were negligible (Fig. 1, movie 1). In the dark, Area II neurons exhibited a small eye-position sensitivity with an average K value (slope of firing rate vs. eye position) of 0.1 ± 0.02 spike/s/° (n = 19, P < 0.01; Fig. 1A, compare Area I and Area II curves in Fig. 1B). Saccadic sensitivity in the dark of most Area II neurons was also small for either the on-direction (Ripsi = –0.02 ± 0.01 spike/s per °/s) or off-direction of the neuron (Rcontra = –0.02 ± 0.02 spike/s per °/s; n = 19, P < 0.01; Fig. 1C). Position and saccadic sensitivity in the light was similar (not illustrated).
|
; Pastor et al. 1994b): they did not pause during oppositely directed saccades, they displayed marginal position sensitivity (approximately –0.18 spike/s/° for both), and after saccades, their firing rate quickly returned to presaccade rate (not shown). Neural response to sinusoidal visual and vestibular stimuli
Quantitative observations demonstrated that Area II neurons exhibited a firing rate that closely mirrored eye velocity over a broad frequency range of visual stimuli (0.065–1.0 Hz, ±16°/s; Fig. 2A, movies 2 and 3). Eye velocity approximated planetarium velocity and phase at lower frequencies but decreased with increasing frequency (Marsh and Baker 1997). Likewise, Area II neuronal activity demonstrated a strong response at lower frequencies that decreased in amplitude and shifted in phase with increasing planetarium frequency (Fig. 2A, inset, movies 2 and 3). A Bode plot generated from the pooled results demonstrate that Area II firing rate and phase varied consistently with eye velocity (0.0625–4.0 Hz, ±16°/s; Fig. 2B).
|
|
While sinusoidal analysis of Area II activity vis-à-vis eye velocity demonstrated a correlation over a wide range of frequencies for both visual and vestibular stimuli, the transient and dynamic components of visuomotor and Area II activity were examined with velocity steps (±16–20°/s, 0.125 Hz, movie 6). The response of the eye to visual velocity step stimuli comprise several well-defined features (Fig. 4A) : a small delay (latency) then a sharp increase in velocity (early component, OKRe) followed by a slower increase in velocity (delayed component/build-up, OKRd) until a constant velocity response (steady state) is reached (Cohen et al. 1977). Sometimes, a clearly identifiable plateau in eye velocity was not observed, and the maximum velocity obtained during OKRd was used to estimate the steady-state component.
|
) and may be the result of using bidirectional rather than unidirectional velocity steps. Having calculated the eye-velocity step response for each animal, the behavior was then used to help define onset latency in Area II neurons. The latency was 97 ± 5 ms (n = 6) and did not differ significantly from the latency of eye-velocity response (paired t-test, P = 0.93).
The mean accelerations of both the eye velocity and neuronal firing rate responses during the early and delayed components were also calculated (Lisberger et al. 1994; Marsh and Baker 1997). During OKRe, the mean eye acceleration was 37.5 ± 7.0°/s2 (n = 7), whereas the mean change in neuronal firing rate was 26.8 ± 3.2 spike/s2 (n = 7). By the end of the early component, eye velocity and the neuronal firing rate had risen to 59 ± 8 and 51 ± 12% (n = 7), respectively, of their steady-state responses. For OKRd, the mean eye acceleration was 3.4 ± 1.1°/s2 (n = 7), and the mean change in neuronal firing rate was 6.3 ± 2.3 spike/s2 (n = 7). The average eye-velocity gain during the steady-state component (OKRd) was 0.61 ± 0.08 (n = 7) and the average increase in Area II firing rate over spontaneous levels was 0.92 ± 0.18 spike/s per °/s (n = 7). Occasionally, eye velocity in the off-direction led to cessation of Area II activity. In sum, the profiles of eye velocity and firing rate responses to visual step stimuli were quite similar (Fig. 4A).
Because of the variability in the firing rate between individual neurons, eye velocity and neuronal firing rate were normalized as a fraction of their respective maximum values and examined at 0.25-s intervals during OKRd and steady state for step velocities 16 ± 2°/s. The results illustrate how closely Area II firing activity reflected eye velocity (Fig. 4B, n = 7). The ratio of normalized firing rate to normalized eye velocity was calculated (Fig. 4C) to create a measure of neuronal gain and to delineate the neuronal contribution during the step stimulus. Initially, neuronal gain was >1.0, indicating Area II activity was proportionally larger than eye velocity to the planetarium step. However, from 1.0 to 2.5 s, the neural gain began a gradual decline from its peak; and from 2.5 s to the end of the steady state, the gain was just below 1.0 as firing rate and eye velocity remained proportional to each other until the end of the step.
If Area II was truly signaling eye velocity to the cerebellum, then Area II neurons should exhibit activity during the discharge of stored eye velocity, i.e., in the absence of sensory input. To measure this directly, animals were exposed to long duration (90 s, 16°/s) optokinetic velocity steps (movie 7). After eye velocity matched stimulus velocity for 30 s, the light was extinguished, and the exponential decays in eye velocity and firing rate were calculated from the period beginning immediately after the initial drop in eye velocity (Fig. 5A, 1 and 2, asterisk). In a few cases, the decay of some pairs of eye-velocity storage and firing rate were not adequately described by an exponential function (not shown), so a simple linear fit was employed (Collewijn et al. 1980). Accordingly, decay time constants were calculated starting from the initial eye-velocity drop as described and ending when the response had decayed to 37% of its starting amplitude. There was no significant difference (paired t-test, P = 0.42) between the mean time constant for eye-velocity storage (15.7 ± 2.4 s, n = 8) and neuronal firing rate (18.9 ± 4.0 s, n = 8; Fig. 5B, 1 and 2).
|
Eye velocity and Area II neuronal firing rate responses to vestibular step stimuli in the dark were investigated using bidirectional head velocity trapezoids at 0.125 Hz with rise times of 100 ms and amplitudes of ±16°/s (movie 8). Table steps induced an eye velocity comprising a short latency and sharp rise (direct component, VORd) that often overshot the constant velocity or sustained component (VORs) and persisted until the end of the step (Fig. 6A). Response latency for eye velocity averaged 17 ± 3 ms (n = 5), VORd was 131 ± 12 ms (n = 5), and VORs was 1.5 ± 0.4 in length (n = 5, Pastor et al. 1992). The mean latency of neuronal firing rate response to stimulus onset was 41 ± 4 ms (n = 4) and differed significantly from eye-velocity latency (paired t-test, P < 0.002). Area II neuronal activity sometimes ceased in response to head velocity in the off direction. Average eye acceleration during VORd was 164.3 ± 37.3°/s2 (n = 5), whereas the change in neuronal firing rate was 246.5 ± 50.7 spike/s2 (n = 5). By the end of VORd, the eye velocity achieved 63 ± 14% of its maximum gain value, whereas the neuronal response reached 60 ± 15% of its maximum gain value.
|
Neural sensitivity to eye and head velocity
During either optokinetic or vestibular stimulation, Area II neurons may be conveying at least two velocity signals to the cerebellum: eye velocity and/or stimulus velocity. Firing rate-velocity plots for low-frequency stimuli (Fig. 7, A and D) demonstrated that both eye velocity (black, filled circles) and stimulus velocity (red, open squares) were collinear and well fit by linear regression. By examining rate-velocity plots at higher stimulus frequency, it was possible to separate the response of Area II neurons to eye and stimulus velocity during OKR but not VOR. The correlation exhibited between firing rate and slip velocity at low frequency (0.065 Hz; Fig. 7A, red, open squares) was not maintained with increasing frequency but persisted with eye velocity (Fig. 7, B and C, black, filled circles). In contrast to optokinetic stimulation, increasing table frequency during vestibular stimulation was not able to separate the collinear relationship between Area II firing rate and head and eye velocity (0.125–2.0 Hz, Fig. 7, D–F).
|
): an adjusted VOR sensitivity calculated by subtracting from the firing rate generated during VOR the Area II optokinetic sensitivity as a product of eye velocity evoked during VOR, i.e., FRhead(t) = FRVOR(t) – ROKR x 
VOR(t), (Fig. 8, A–C) (Lisberger et al. 1994) and VOR cancellation that permitted sensitivity computation directly from the evoked neuronal firing rate (Fig. 8D) (Miles et al. 1980). Thus if the contribution of eye and head velocity to Area II summed arithmetically, then either approach should yield approximately the same value for head velocity sensitivity. For most Area II neurons analyzed (9 of 12, Table 1; Fig. 8, C and D), there was only a small (13 ± 2.6%) difference between the firing rate-head velocity slopes calculated by the two methods. When examined as a group, there was no significant difference between the two measurements (2-tailed, paired t-test; P = 0.29, n = 12). For five additional neurons, head cancellation data were not available for comparison (Table 1); but, in two cases, the adjusted head sensitivity was almost zero (Fig. 8, E and F), suggesting that a small subset of Area II neurons may exclusively signal eye velocity.
|
|
The velocity sensitivity of Area II neurons to optokinetic and vestibular input was examined under conditions of changing sinusoidal stimulus frequency at a constant velocity amplitude (±16°/s) or changing bidirectional step velocity at a single frequency (0.125 Hz) to determine whether responses remained linear. The optokinetic sensitivity of Area II neurons was largely linear for varying frequencies and step velocities (Fig. 9, A and B). With increasing planetarium frequency, average optokinetic sensitivity decreased. However, a few (3 of 8) neurons demonstrated nonlinear behavior with inflections occurring at frequencies of 0.125–0.25 Hz (Fig. 9A). Two neurons with large sensitivities (>1.5 spike/s per °/s) dropped sharply between 1.0 and 2.0 Hz. During planetarium velocity steps, three of five Area II neurons tested demonstrated a roughly linear sensitivity with increasing eye velocity, whereas two dropped sharply as eye velocity increased (Fig. 9B). The dynamic vestibular response of Area II neurons was also largely linear for varying table frequencies and velocity trapezoids (Fig. 9, C and D). The average vestibular response remained flat at lower frequencies. However, as with the dynamic optokinetic sensitivity, there was a sharp drop at 2.0 Hz for five of eight neurons tested (Fig. 9C). Of the group, only one neuron demonstrated a strong nonlinearity at lower frequencies. For table steps (Fig. 9D), there was a general trend of decreasing sensitivity with increasing velocity. There was no correlation among neurons with nonlinear vestibular versus optokinetic sensitivities.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
; Straka et al. 2006) and neurophysiology of Area II is noteworthy for it affords one of the few documented cases where visual and vestibular motor commands are emulated and fed back to the cerebellum, thereby closing the oculomotor loop.
Characterizing Area II neurons using a linear systems approach has shown that the changes in eye- and table-velocity phase and gain are well matched by similar changes in Area II firing rate over a broad range of frequencies, extending previous but more limited observations (Pastor et al. 1994b). Two details are worth noting. First, the overall modulation of the neural response was greater for vestibular stimulation compared with optokinetic by almost a factor of 2 (cf. Figs. 2 and 3), and this difference was not fully explained by the eye-velocity gain between the two paradigms. Greater spike modulation during OKR might be possible using a more robust optokinetic stimulus, i.e., striped drum. Not only have stripes been shown to evoke higher OKR gains in goldfish (Keng and Anastasio 1997) than observed here and previously (Marsh and Baker 1997; Schairer and Bennett 1986), but the higher-contrast motion stimuli might also increase visual system activity as a whole. A more parsimonious explanation may be that there is a difference in the vestibular and visual sensitivity of Area II, such that the vestibular connections provide stronger and more deeply modulated input. Unlike the firing rates observed during OKR, Area II neuronal firing rate declined at high table frequencies (>1 Hz), whereas VOR eye velocity actually slightly increased. This high-frequency drop-off might also reflect the vestibular connectivity to Area II. It is possible that second-order vestibular pathways innervating Area II are low-pass filtered like eye-velocity signals, thereby producing a decrement in Area II modulation with increased frequency (Dickman and Angelaki 2004). More likely, the rise in eye velocity reflects the increased contribution of the direct pathway from second-order vestibular neurons to extraocular motor nuclei (Lorente de No 1933; Szentagothai 1950). These direct connections bypass Area II and suggest that Area II may not be involved in motor generation. Indeed, differences in both sensitivity and connectivity are complementary and, in combination, are likely to provide a suitable explanation.
When presented with dynamically changing stimuli in the form of planetarium and table velocity steps, the modulation in Area II firing rates also mirrored changes measured in eye velocity. However, as with sinusoidal stimuli, a difference between visual and vestibular steps was also observed. Onset latency of Area II firing rates and eye velocity were both 95 ms for optokinetic steps of 16°/s; and the average time course for the firing rate and eye velocity were remarkably similar during the course of the 0.125-Hz step (Fig. 4C). In contrast, there was a significant onset delay in Area II firing latency for vestibular steps. The eye responded within 14 ms after step onset while the neuron did not respond until 41 ms. The latency difference could also be the result of similar processes affecting modulation of Area II neurons during high-frequency table motion—low-pass filtering of second-order neurons combined with a large contribution from the direct pathway from vestibular to extraocular nuclei. Area II neuron response dynamics may instead be intrinsic to Area II. For example, a high firing threshold would insert a delay in signaling and likely preclude a shorter latency to stimulus input. Despite the long latency of Area II responses to vestibular steps, the neuronal firing rate rose faster than eye velocity and reached its steady state within the first second of the step (Fig. 6C). A comparison of the firing rate response to eye-velocity gain (neuronal gain) demonstrated that the response of Area II neurons in relation to eye velocity was steady over a range of planetarium velocities (Fig. 9B). Increasing table step velocity, however, yielded a general decrease in Area II firing rate response relative to eye velocity (Fig. 9D).
Constant velocity visual stimuli demonstrated that both eye velocity and firing rate were well matched during all phases of the paradigm. When the light was extinguished, both the eye-velocity and neuron firing rate persisted in the absence of a sensory input, decaying at approximately the same rate (Fig. 5, movie 7). It is unknown whether the sustained firing rate was the product of an intrinsic property of Area II neurons, generated by a local Area II neural network, or was being driven by another neuronal site such as the vestibular nucleus. Nevertheless, the stored eye-velocity signal originates in the brain stem (Area II) and is signaled directly to the cerebellum.
Relationship between Area II and prepositus neurons
The brain stem regions Areas I and II in the goldfish have been suggested to be functionally analogous to the mammalian NPH (Pastor et al. 1994b) as these regions comprise anatomically distinct clusters of neurons responding to eye position and eye velocity much like NPH neurons. Neurons comprising the mammalian NPH exhibit a range of sensitivities to eye position and velocity and have been identified as "position," "position-velocity," or "velocity-position" neurons (Baker 1977; Delgado-Garcia et al. 1989; Lopez-Barneo et al. 1982). Although the Area I neurons described by Aksay et al. (2000) are similar to the position and position-velocity neurons, Area II neurons are unlike any mammalian NPH neuron heretofore identified (Straka et al. 2006).
Even though Area II neurons may resemble mammalian velocity-position neurons in that they both possess increased eye-velocity sensitivity with reduced eye-position sensitivity, there are differences between the two neuron types. Area II neurons demonstrated type II vestibular sensitivity unlike the type I sensitivity found in some mammalian precerebellar velocity-position neurons (Delgado-Garcia et al. 1989; Lopez-Barneo et al. 1982). However, the majority of velocity-position neurons recorded by Lopez-Barneo et al. (1982) were type II neurons, although cerebellar connectivity was not tested. The average vestibular sensitivity of Area II neurons was much less than the 3.3 spike/s per °/s reported by Escudero et al. (1996) or the single example of 5.7 spike/s per °/s reported by Lopez-Barneo et al. (1982). In addition, unlike the mammalian velocity-position neurons (Delgado-Garcia et al. 1989; Escudero et al. 1996; Lopez-Barneo et al. 1982) or the adjacent goldfish Area I neurons (Aksay et al. 2000), Area II neurons had an average position sensitivity that was an order of magnitude less and relatively flat over the entire 40° range of eye movements (Fig. 1B, Movie 1). Mammalian velocity-position and goldfish Area I neurons, in contrast, were strongly rectified in the off direction and more strongly modulated with increasingly eccentric eye positions.
Similar observations were made when comparing saccadic sensitivity between Area II neurons and velocity-position/Area I neurons. The sensitivity of Area II neurons to saccades was an order of magnitude less than velocity-position/Area I neurons and, on average, close to zero. Furthermore, unlike velocity-position and Area I neurons, the two Area II neurons demonstrating saccadic sensitivity had no lasting postsaccadic change in firing rate and no appreciable position sensitivity (Fig. 1C). In addition, while velocity-position neurons are reported to make connections with oculomotor nuclei (Delgado-Garcia et al. 1989), there is no connectivity between Area II neurons and oculomotor nuclei (Straka et al. 2006). This clearly suggests that Area II does not have the requisite structure and physiology to play a significant role in maintaining eye position.
The physiological differences between mammalian velocity-position neurons and goldfish Area II neurons may simply reflect the fact that the mammalian NPH is a heterogeneous cluster of cell types; consequently, it is not tractable to find an analogous population of pure eye-velocity neurons in NPH as is found in Area II of the goldfish. Alternatively, Area-II-equivalent neurons in tetrapods may require concurrent velocity and position sensitivity for proper oculomotor function. These differences may explain why floccular Purkinje cells in mammals possess both eye position and velocity sensitivity (Miles et al. 1980; Nagao 1991; Noda and Warabi 1982), whereas goldfish Purkinje cells exhibit only velocity sensitivity (Pastor et al. 1997). Last, as hypothesized in the preceding report (Straka et al. 2006), the role of Area II neurons may have been superceded by the analogous and more derived pontine nuclei. If found in mammals, neurons responding like Area II would represent a new category of NPH neuron and should be termed "velocity" neurons based on the neurophysiology presented in this study. In sum, it is clear that whether an homologous (NPH) or functionally analogous group (pontine nuclei) of Area II neurons exists in mammals the identification of a distinct precerebellar population of eye-velocity neurons will permit the selective recording and manipulation of oculomotor signals destined to the cerebellum.
Area II neuron signaling
Analysis of the discharge properties of horizontal gaze-velocity Purkinje neurons (HGVP) within the vestibulocerebellum has suggested that both head and eye velocity independently contribute to the modulation of Purkinje cell firing rate (Lisberger and Fuchs 1978a; Miles et al. 1980; Pastor et al. 1997) and has been modeled, in a simple form, as Pfr(t) = a x 
(t) + b x (t), where Pfr is Purkinje cell firing rate, is head and is eye velocity, and a and b are the sensitivities to head and eye velocity, respectively. Because the majority of the input to the goldfish vestibulolateral lobe is derived from Area II (Straka et al. 2006), the logical extension regarding HGVP discharge properties, as well as those of eye-velocity Purkinje cells (Belton and McCrea 2000; Pastor et al. 1997), would imply that the eye-velocity sensitivity observed on goldfish Purkinje cells is derived exclusively from the firing of Area II neurons, although less at high frequencies (Fig. 2). This conclusion is supported by the extensive projections made by Area II neurons throughout the bilateral granule cell layer (Straka et al. 2006) as well as the highly correlated activity of Area II neurons in response to eye velocity.
The inability for Area II to adequately convey high-frequency information is not surprising for optokinetic stimuli. The visual system has been noted for its low-pass filtering characteristics (Collewijn 1971; Evinger and Fuchs 1978; Fuchs 1967; Hess et al. 1985; Marsh and Baker 1997). The limited sensitivity of Area II to high-frequency vestibular input, as evidenced by both the disproportionate drop in Area II discharge relative to eye velocity at high table frequencies (Fig. 3) as well as the 40-ms latency of Area II discharge during table velocity steps (Fig. 6), suggests that motor generation is not a function of Area II. Nevertheless, mammalian HGVP cells also exhibit similar long latencies to vestibular table steps (Lisberger et al. 1994; Miles et al. 1980) and reduced sensitivity in response to 2-Hz table sinusoids (Raymond and Lisberger 1998). Thus it is likely that the firing activity of goldfish HGVP cells, as in primates, is highly influenced by the activity of Area II neurons.
Even though head- and eye-velocity sensitivity can be separately determined, the parsimonious conclusion based on the collinear nature of the two signals (see Fig. 7) is that Area II neurons largely signal eye velocity under most behavioral conditions observed in goldfish. However, it is worthwhile to note that during VOR cancellation paradigms most Area II neurons exhibit a large head-velocity sensitivity (Table 1). Future investigations of the neural basis of oculomotor learning should analyze the role of Area II neurons during plasticity to ascertain differences in head- and eye-velocity sensitivity in Area II firing rates before and after oculomotor training (Blazquez et al. 2004; Lisberger et al. 1994; Miles and Lisberger 1981; Miles et al. 1980).
Role of Area II connections to cerebellum
The extensive bilateral termination of Area II neurons in the vestibulolateral lobe (Straka et al. 2006) and the signals they carry to the cerebellum explain the basis for the diverse HI, EI, EII, HIEII, and HIEI Purkinje cell sensitivities in goldfish (Pastor et al. 1997). The emulation of oculomotor activity by Area II is analogous to the efferent copy signaling attributed to NPH (McCrea and Baker 1985). Based on these neurophysiological results as well as the morphology and connectivity of Area II (Straka et al. 2006), previous circuit based descriptions of brain stem and cerebellar pathways can now be extended (Fig. 10) to accurately reflect direct structural connections that address conceptual inaccuracies such as the envisioned role of a motor error signal (Lisberger et al. 1994; Miles et al. 1980).
|
) and are likely to target different vestibular neuron subtypes (Fig. 10). The EII and HIEII pathway (Fig. 10, right Purkinje cell) modulates the type II inhibitory DO neurons hypothesized to suppress both type I inhibitory and excitatory DO neurons. The EI and HIEI (HGVP) Purkinje cells (Fig. 10, left) innervate the ipsilateral excitatory DT neurons to medial rectus motoneurons and ipsilateral inhibitory DO neuron. The HIEII Purkinje cell pathway thus inhibits the type II inhibitory DO neuron thereby disinhibiting the contralateral excitatory and ipsilateral inhibitory DO neurons. The DO neurons also receive excitatory vestibular nerve afferent input and, in turn, excite contralateral and inhibit ipsilateral Area II neurons. The contralateral Area II (Fig. 10, right) projects an eye-velocity signal to the cerebellum that reaches Purkinje cells and generates an HIEII signal. Inhibition of the ipsilateral Area II (Fig. 10, left) generates a reduction in EI eye-velocity signaling that is proportional to the rise in head velocity. Thus the activity of HIEI (HGVP) and HIEII Purkinje cells are modulated during VOR plasticity (Lisberger and Fuchs 1978a; Miles et al. 1980; Pastor et al. 1997). Indeed, other oculomotor adaptation paradigms, e.g., OKR, VOR light, or VOR cancellation, can be achieved based on the structure and signaling pathways shown in the summary figure (Fig. 10).
The bilateral representation of Area II provides the basis on which to produce monocular vestibulocerebellar control of eye velocity for each separate eye (Fig. 10). If precerebellar Area II activity (HIIEI) converging on Purkinje cells emulates the head- and eye-velocity commands for one eye, then the same coding within each nucleus would be sufficient to drive two disparate sets of Purkinje cells (EI, HIEI) and (EII, HIEII) for disjunctive rather than conjugate eye motion. This would also imply that each Purkinje cell type must target different sets of vestibular neurons (Fig. 10). Presumably, the use of monocular and vergence stimuli (Zhou and King 1998) will identify parallel, but separate, pathways in Area II that mirror the comparable signaling circuitry in the vestibulolateral lobe of the cerebellum.
Conclusion
Area II represents the major input provided by the oculomotor system to the cerebellum. Because Area II signals head/eye velocity that directly and dynamically mirror both the input and output of the motor system, this goldfish nucleus represents the ideal place to search for modifications in brain stem discharge rates associated with oculomotor plasticity.
|
GRANTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
FOOTNOTES |
|---|
1 The online version of this article contains supplemental data. 
Address for reprint requests and other correspondence: J. C. Beck, Dept. of Physiology and Neuroscience, NYU School of Medicine, 550 First Ave., New York, NY 10016 (E-mail: james.beck{at}med.nyu.edu)
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Aksay E, Gamkrelidze G, Seung HS, Baker R, and Tank DW. In vivo intracellular recording and perturbation of persistent activity in a neural integrator. Nat Neurosci 4: 184–193, 2001.[CrossRef][ISI][Medline]
Baker R. The nucleus prepositus hypoglossi. In: Eye Movements, edited by Brook B and Bajandes FJ. New York: Plenum, 1977, p. 145–178.
Baker R, Gresty M, and Berthoz A. Neuronal activity in the prepositus hypoglossi nucleus correlated with vertical and horizontal eye movement in the cat. Brain Res 101: 366–371, 1975.[CrossRef]
Belton T and McCrea RA. Role of the cerebellar flocculus region in cancellation of the VOR during passive whole body rotation. J Neurophysiol 84: 1599–1613, 2000.
Blanks RH and Precht W. Functional characterization of primary vestibular afferents in the frog. Exp Brain Res 25: 369–390, 1976.[ISI][Medline]
Blazquez PM, Hirata Y, and Highstein SM. The vestibulo-ocular reflex as a model system for motor learning: what is the role of the cerebellum? Cerebellum 3: 188–192, 2004.[CrossRef][ISI][Medline]
Brodal A and Brodal P. Observations on the projection from the perihypoglossal nuclei onto the cerebellum in the macaque monkey. Arch Ital Biol 121: 151–166, 1983.[ISI][Medline]
Cohen B, Matsuo V, and Raphan T. Quantitative analysis of the velocity characteristics of optokinetic nystagmus and optokinetic after-nystagmus. J Physiol 270: 321–344, 1977.
Collewijn H. The optokinetic system of the rabbit. Doc Ophthalmol 30: 205–226, 1971.[CrossRef][ISI][Medline]
Collewijn H, Winterson BJ, and van der Steen J. Post-rotary nystagmus and optokinetic after-nystagmus in the rabbit linear rather than exponential decay. Exp Brain Res 40: 330–338, 1980.[ISI][Medline]
Delgado-Garcia JM, Vidal PP Gomez C, and Berthoz A. A neurophysiological study of prepositus hypoglossi neurons projecting to oculomotor and preoculomotor nuclei in the alert cat. Neuroscience 29: 291–307, 1989.[CrossRef][ISI][Medline]
Dickman JD and Angelaki DE. Dynamics of vestibular neurons during rotational motion in alert rhesus monkeys. Exp Brain Res 155: 91–101, 2004.[CrossRef][ISI][Medline]
Duensing F and Schaefer KP. Die Aktivität einzelner Neurone im Bereich der Vestibulariskerne bei Horizontalbeschleunigungen unter besonderer Berücksichtigung des vestibulären Nystagmus. Arch Psychiatry Nervenkr 198: 225–252, 1958.
= 7.4 and 3.5°, respectively) and lag stimulus velocity at high frequency (
, where y is eye velocity or firing rate, y0 is initial y value, A is amplitude, x is time, and 
