Control of Spatial Orientation of the Angular Vestibulo-Ocular Reflex by the Nodulus and Uvula of the Vestibulocerebellum

Abstract

Eye velocity produced by the angular vestibulo-ocular reflex (aVOR) tends to align with the summed vector of gravity and other linear accelerations [gravito-inertial acceleration (GIA)]. Defined as “spatial orientation of the aVOR,” we propose that it is controlled by the nodulus and uvula of the vestibulocerebellum. Here, electrical stimulation, injections of the GABA_A agonist, muscimol, and single-cell recordings were utilized to investigate this spatial orientation. Stimulation, injection, and recording sites in the nodulus were determined in vivo by MRI and verified in histological sections. MRI proved to be a sensitive, reliable way to localize electrode placements. Electrical stimulation at sites in the nodulus and sublobule d of the uvula produced nystagmus whose slow-phase eye-velocity vectors were either head centric or spatially invariant. When head centric, the eye velocity vector remained within ± 45° of the vector obtained with the animal upright, regardless of head position with respect to gravity. When spatially oriented, the vector remained relatively constant in space in one on-side position, with respect to the vector determined with the animal upright. A majority of induced movements from the nodulus were spatially oriented. Spatially oriented movements were generally followed by after-nystagmus, which had the characteristics of optokinetic after-nystagmus (OKAN), including orientation to the GIA. After muscimol injections, horizontal-to-vertical cross-coupling was lost or reduced during OKAN in tilted positions. This supports the hypothesis that the nodulus mediates yaw-to-vertical or roll cross-coupling. The injections also shortened the yaw-axis time constant and produced contralateral horizontal spontaneous nystagmus, whose velocity varied as a function of head position with regard to gravity. Nodulus units were tested with static head tilt, sinusoidal oscillation around a spatial horizontal axis with the head in different orientations relative to the pitching plane, and off-vertical axis rotation (OVAR). The direction of the response vectors of the otolith-recipient units in the nodulus, determined from static and/or dynamic head tilts, were confirmed by OVAR. These vector directions lay close to the planes of the vertical canals in 7/10 units; many units also had convergent input from the vertical canals. It is postulated that the orientation properties of the aVOR result from a transfer of otolith input regarding head tilt along canal planes to canal-related zones of the nodulus. In turn, Purkinje cells in these zones project to vestibular nuclei neurons to control eye velocity around axes normal to these same canal planes.

INTRODUCTION

The angular VOR (aVOR) exhibits the property of spatial orientation. During nystagmus induced by rotation of the subject or surround, the yaw-axis component of slow phase velocity tends to align with gravity or with tilts of the gravito-inertial acceleration vector (GIA) with regard to the head.^e,1-12 We have shown that the central velocity storage mechanism is responsible for spatial orientation of the aVOR, and that the direct visual and vestibular pathways do not orient eye velocity to the GIA.¹² Thus, the study of the spatial orientation of the aVOR is essentially a study of the three-dimensional characteristics of velocity storage. Alignment of the eye-velocity vector to the GIA has been modeled by three processes: a reduction in the time constant of the dominant horizontal component; an increase in the torsional and/or vertical time constants; and the appearance of orthogonal vertical or torsional “cross-coupled” eye velocity.^{1,2,6-8,11,12} From the frequency characteristics of velocity storage, and from the finding that velocity storage is activated by visual, vestibular, and somatosensory input in both monkeys¹³ and humans,¹⁴ it has been inferred that this type of spatial orientation is likely to be important for postural and gaze stabilization during the centripetal accelerations experienced in circular locomotion¹⁵ and when passively transported around curves on bicycles, motorcycles, cars, and so on.

Behavioral, stimulation and lesion studies have shown that the nodulus and uvula control the horizontal time constant of velocity storage, one element in determining its spatial orientation. The horizontal aVOR time constant is reduced when the head is reoriented with respect to gravity from the upright position during postrotatory nystagmus (tilt dumping), both in humans^16,17 and in monkeys.^9,18-23 The horizontal time constant is also reduced when subjects view a relative stationary visual surround during vestibular nystagmus or OKAN (light dumping).^21,24-26 Electrical stimulation of the nodulus and lobule 9d of the ventral uvula reproduces this reduction.²³ Both tilt dumping and light dumping are lost after ablation of the nodulus and uvula.²¹ Electrical stimulation of the nodulus also produces after-nystagmus that has the characteristics of optokinetic after-nystagmus (OKAN).²⁷ This suggests that it is not only possible to discharge velocity storage, but also to induce it by nodulus stimulation.

The nodulus and ventral uvula are also involved in producing the changes in vertical and/or torsional aVOR components that tilt the eye-velocity vector during spatial orientation. After complete nodulo-uvulectomy, there is a loss of spatial orientation of the aVOR in the monkey.^1,2,9,10 Removal of only medial portions of the nodulus and uvula causes a loss of yaw-to-vertical and yaw-to-roll cross-coupling, and control of the horizontal aVOR time constant is maintained.² From this, it is likely that different portions of the vermal cortex of the nodulus and uvula participate in different parts of the process of spatial orientation.²

Much of the information about the representation of spatial orientation in the nodulus and uvula has come from studies utilizing visual motion around different axes. A striking feature of the cortex of the nodulus and uvula is its division into parasagittal zones that are innervated by different subnuclei of the inferior olive.^28-31 In the rabbit, activity induced by optokinetic stimulation about a vertical axis during yaw or horizontal optokinetic nystagmus (OKN) is processed in the caudal dorsal cap (cdc), while activity related to optokinetic stimulation about a horizontal axis parallel to the axis of the ipsilateral anterior canal (135°) reaches the cerebellar cortex through the rostral dorsal cap (rdc) and ventrolateral outgrowth (vlo).^32-39 Vestibular signals aligned approximately with the anterior- and posterior-canal axes project from the β nucleus of the inferior olive.^{32,33,36,40-43} Utricular signals project from the dorsomedial cell column (DMCC)⁴⁴ and the β nucleus.^41,45 As yet, no signals related to lateral canal activation have been found in the inferior olive or in climbing fiber responses.^33,42 From these data, it has been proposed that the three reciprocal planes of the semicircular canals are represented across the vermis of the nodulus in zones defined by inferior olive input.^33,46 From midline to the left, the right-anterior-left-posterior-canal (RALP) plane is represented in the medial 1.5 mm, yaw-axis visual movement (horizontal OKN) in the next 0.5–1.0 mm, and the left-anterior-right-posterior (LARP) plane 1.0–2.0 mm more laterally.^{42 47} These zones project to appropriate regions of the superior and medial vestibular nuclei where appropriate canal plane responses could be generated.⁴⁶ This could provide a cellular mechanism for realizing spatial orientation of the aVOR.² Since the major transmitter of Purkinje cells is GABA,⁴⁸ presumably the eye velocity changes produced by the direct projections of the nodular cortex to the vestibular nuclei occur through inhibition and disinhibition.

Four sagittal zones have been distinguished in the monkey, based on four white-matter compartments innervated by discrete subnuclei of the contralateral inferior olive.³¹ Olivocerebellar projections to the different zones in monkey have been inferred by analogy with the organization in rabbit. The most medial zone (Zone 1) extends from the midline to 0.8-mm lateral in the nodulus. It receives climbing fiber input from the β nucleus medially and the cdc laterally. The second zone (Zone 2) extends from 0.8 to 2.4 mm in the nodulus and uvula, and forms the lateral limits of the vermis over the posterior surface of the cerebellum. Zone 2 was labeled from injections including the vlo, rostral β, and rostral medial accessory olive (rMAO). Similar to the rabbit, the vlo projection is probably restricted to Zone 2 in the nodulus, while axons from the rostral β nucleus may extend into the uvula. Zone 3, innervated by the cdc, extends from 2.4 to 3.2 mm and is restricted to the nodulus. An auxiliary projection to this zone from the β nucleus cannot be excluded, but is unlikely in the nodulus. Zone 4 extends from 3.2 to 4mm lateral. Its olivocerebellar afferents are currently unidentified in the rhesus monkey, but by analogy with the lateral zone in rabbit, probably include the DMCC and rMAO.^32,36,49 If it is assumed that the same canal-, otolith- and OKN-related information is present in the inferior olive of the monkey as in the rabbit, Zone 1 has horizontal axis anterior canal (135°) VOR and otolith (β), as well as yaw-axis representation from the cdc (yaw OKN). Input to Zone 2 would be related to horizontal axis (135°) OKN and VOR and the utricle. Zone 3, extending laterally beyond the vermis, would receive information about yaw-axis OKN. This canal-related organization has been postulated as the basis for spatial orientation in the rabbit,³³ and we have proposed that it is also the basis for the spatial orientation of the aVOR in the monkey.^1,2 Recently, a canal-based orientation system was also found in the pigeon nodulus in response to visual flow along canal planes.⁸²

As noted in the preceding paragraphs, electrical stimulation of the nodulus and rostro-ventral uvula (lobule d), caused a reduction in the horizontal aVOR time constant in the monkey,²³ one component of spatial orientation of the aVOR.^1,2,6,7 When prolonged, such stimulation also produced after-nystagmus, which had the characteristics of OKAN.²⁷ Only horizontal eye movements were recorded, however, either for short²³ or longer stimuli,²⁷ and it is not known whether vertical and torsional components would also be produced by electrical stimulation of the nodulus or uvula or whether the OKAN-like after-nystagmus also has the property of alignment to the spatial vertical as does OKAN.⁷ There is also a paucity of information about whether the nodulus and uvula receive otolith input in the monkey. Otolith information, which codes tilts of the GIA with respect to the head, and which would be critical in orienting the aVOR, has been found in the cat.⁵⁰ Such cells have not been studied in the monkey, and the directions of their response vectors have not been determined. The purpose of this study was to determine how the nodulus organizes spatial orientation of the aVOR in the monkey. We wished to establish whether trajectories of eye velocity in response to electrical stimulation of the nodulus had spatial components, and how spatial orientation of the aVOR was altered after inactivation of the nodulus with the GABA_A agonist, muscimol. We also wished to determine the spatial organization of otolith-related units in the nodulus.

METHODS

Two rhesus and one cynomolgus monkey were used in this study. The experiments conformed to the Guide for the Care and Use of Laboratory Animals (National Research Council, 1996), and were approved by the Institutional Animal Care and Use Committee. Under sterile surgical conditions and anesthesia, an acrylic ring was implanted that held the animals’ heads painlessly during experiments.⁵¹ A Delrin bar, attached to this ring, permitted introduction of microelectrodes into deep brain structures. Two three-turn coils were implanted on one eye to measure horizontal, vertical, and torsional eye position.^22,52 To calibrate eye movements, the animals were rotated in light at 30°/s about the pitch, roll, and yaw axes. It was assumed that horizontal and vertical gains were close to unity in this condition. Roll gains were assumed to be 0.6 when rotation was around a naso–occipital axis aligned with the spatial vertical.⁵³ Eye velocities to the left, down, and counterclockwise from the animal’s point of view, are represented by downward deflections in the velocity traces in the figures.

During testing animals sat in a primate chair in a three-axis vestibular stimulator surrounded by an optokinetic drum. Each axis went through the center of rotation of the head. The stimulator has been described in detail in previous publications.^20,54,55 With the monkey upright, the yaw axis was aligned with gravity, and the horizontal stereotaxic plane was aligned with the spatial horizontal. Thus, the lateral semicircular canals were tilted up approximately 30° from the earth horizontal plane during the experiments.^53,56 Amphetamine sulfate (0.3 mg/kg) was given 30 min before testing to maintain alertness.

Electrical stimuli were 0.5-ms constant current, cathodal pulses of 40–60 μA given at a frequency of 200 Hz for 10–30 s. The pulses were monopolar, with the electrode tip referenced to the bolts implanted on the skull. For injection, a stainless-steel guide tube was inserted into the brain, aimed at the region of interest. Injections were made using a Hamilton pressure syringe inserted through the guide tube toward the designated injection site. Muscimol ( 1.2–1.4 μL of a 1-μg/μL solution) was injected over a period of 2 min. The syringe was left in place for an additional 3 min. Testing began 10 min after the start of injection and lasted for 2–3 h. In a control experiment, conducted on a separate day, 1.2–1.4 μL of 0.9% NaCl was injected into one of the effective locations.

Units were recorded extracellularly using 80-μ tungsten electrodes, varnished to the tip. Electrodes had a resistance of ≈ 1 mΩ tested at 1 kHz. Unit amplifiers had a bandwidth from 200 Hz-10 kHz. The location of the recording electrodes in the nodulus was based on stereotaxic coordinates and the location of characteristic activity related to eye movements in the abducens and rostral fastigial nuclei. The accuracy of electrode placement was confirmed in the MRI. Because the bodies of the animals were not restrained during recording, sinusoidal oscillation in unit activity during yaw oscillation of the body could reflect neck proprioceptive as well as semicircular-canal input. Vertical-canal input, on the other hand, could be separated from neck input on the basis of temporal phase of response. Therefore, testing during unit recording was limited to determining otolith sensitivity and vertical canal input, and we did not study potential lateral canal input.

Unit sensitivity to tilts of the GIA was determined in three conditions. The animal was statically tilted at 30° and/or 60° for 30 s with the head in different orientations to the plane of tilt. After each tilt, the animal was returned to the upright position for about 20 s. The animal was also sinusoidally oscillated around a spatial horizontal axis at ± 18° and with the head in different orientations to the plane of oscillation. Third, some of the units were tested during off-vertical axis rotation (OVAR). In our coordinate system, 0° corresponded to nose-down, 180° to nose-up, 90° to right side down (RSD), and 270° to left side down (LSD) tilts. Unit sensitivity is presented as imp*s⁻¹*g⁻¹. This was calculated by assuming that the resting discharge was the mean value of all temporal gains (imp*s⁻¹) obtained from a variety of head orientations to the plane of tilt or the plane of oscillation around a spatial horizontal axis. Unit sensitivity during static or dynamic tilts was plotted as a function of head orientation relative to the tilt plane. Sensitivity was positive when it was within ± 90° relative to head position, and as negative in other cases.

With the exception of optokinetic nystagmus (OKN), all testing was done in darkness. Per- and postrotatory nystagmus were induced by rotation in yaw around a vertical axis at a constant velocity at speeds of 30°/s and 60°/s (270°/s² acceleration and deceleration). OKN was induced by rotating the visual surround about the animal’s yaw axis at 60°/s for 30 s with the animal upright, left side down (LSD), or right side down (RSD). Optokinetic after-nystagmus (OKAN) was then recorded in darkness. The animals were also tilted statically toward the LSD or RSD positions at angles of 0° to 90° in 15° steps. Each angle of tilt was maintained for 60 s. Finally, animals were rotated in yaw at velocities of 30–120°/s about axes tilted from the vertical (off-vertical axis rotation, OVAR) at angles of 0° to 90° in 15° steps. Each axis of tilt was maintained for at least 40 s.

Eye-position voltages and voltages related to the velocity or position of the axes of rotation were recorded by amplifiers with a bandpass of DC to 40 Hz. Voltages were digitized at 600 Hz/channel with 12-bit resolution, and stored on disk. Eye-position voltages were digitally differentiated and saccades were removed. Slow phase velocities were analyzed from the onset of OKAN during optokinetic stimulation or from the initial jump in velocity during per-rotatory and postrotatory nystagmus to the point where the yaw-axis eye velocity decayed to zero. In all cases, the direction of nystagmus is indicated by the direction of the slow phase eye velocity. Time constants were determined by dividing the area under the slow phase velocity envelope by the initial jump in velocity.⁵⁷ Vectors of slow phase velocity of nystagmus induced by stimulation were determined in two dimensions, yaw and pitch, from responses that were greater than 37s in each of the three tested positions (upright, left, and right side down; Figs. 3 and 4). A mean vector was calculated for the average horizontal and vertical components of eye velocity during stimulation after subtracting any spontaneous velocities in the period just before stimulation.

FIGURE 3.

Stimulation at site a in nodulus, 2.1 mm to the right of the midline (see Fig. 2) with the animal (A) upright (B) left side down (LSD), or (C) right side down (RSD) that produced a head-centric response. There was a rapid rise in eye velocity at the onset of stimulation. The induced movement was predominantly horizontal to the right, but there were also vertical and roll components. Eye velocity fell rapidly at the end of stimulation and there was no after-nystagmus. (HEV, horizontal eye velocity; VEV, vertical eye velocity, REV, roll eye velocity.) Eye position traces are not shown. Eye velocities to the right, up and clockwise from the animal’s point of view produce upward deflections in the velocity traces. On the right are horizontal/vertical phase-plane plots. The plots start from positions close to the origin at the onset of stimulation and proceed eccentrically. They terminate at the end of stimulation. The arrows give the mean direction and amplitude of the per-stimulus eye-velocity vector. Note that the direction of the eye-velocity vector in upright and tilted positions remained relatively fixed with regard to the head, despite the change in head position with regard to the GIA.

FIGURE 4.

Simulation at site c in nodulus, 2.2 mm to the left of the midline in nodulus produced a spatially oriented response with the animal RSD (C). (A) Stimulation with the animal upright produced a slow rise in eye velocity to the left and down, followed by after-nystagmus that declined along the time constant of OKAN. The vector of the induced per-stimulus response was up and to the right both spatially and with respect to the head (phase-plane graph on right). The relationship of the induced vector to the monkey’s head is shown in the inset above. (B) With the animal LSD, the per-stimulus nystagmus was to the left, but the vertical component disappeared. The direction of the per-stimulus vector was upward with regard to the head and tilted with regard to gravity (phase-plane plot and inset on right). There was a cross-coupled, upward after-response that tended to swing the vector of the induced eye velocity toward the GIA. (C) When RSD, the horizontal component disappeared and down/roll eye velocities appeared. This produced a per-stimulus vector that was oriented spatially in the upward direction, similar to that induced with the animal upright (A). No after-nystagmus was induced in the RSD position.

Electrode locations were identified in vivo with MRI. Under ketamine/xylazine anesthesia, sharpened 120-μ tungsten wires were introduced as markers through the Delrin plate, and moved to positions 3–5 mm above the positions where the stimulating/recording electrodes or pipettes were located during experiments. Scans were performed on a whole-body 1.5-T scanner (GE Medical Systems, Milwaukee, WI). The sedated animal with the electrode array in place was put in the scanner in a supine position. A quadrature knee coil was placed around the animal’s head. The knee coil was used to maximize signal to noise. A T1-weighted, 3-D SPGR volume-acquisition scan was performed (TR = 24, TE = 5) using a 256 × 256 matrix and 12-cm × 12-cm field of view. The acquired images were 1.2 mm thick, with 60 images in total. Scan time for this sequence was approximately 6 min. The acquired images were transferred to an independent workstation (GE Medical Systems). Using the interactive image-analysis program, the acquired axial images were reformatted into sagittal and coronal images, and interactively adjusted to be tangential to the plane of the electrodes (Fig. 1).

FIGURE 1.

MRI of brainstem and cerebellum of monkey used for identification of electrode tracks. (A) Sagittal section 2.5 mm from midline on right showing two electrodes (arrows) flanking region of interest in cerebellum. The electrode tracks were separated by 8 mm. (B) Coronal section through posterior electrode pair. The electrode tracks (arrows), which were 5 mm apart, lay on either side of the vermis. (C) Coronal section about 2-mm caudal to abducens nucleus. Arrow points to electrode track 1.5-mm lateral on the right, at a location similar to the tracks shown in Figure 2. See text for details.

Upon completion of the experiments, seven electrolytic lesions (cathodal DC, 60 μA for 30 s) were placed at the same locations where electrical stimulation and muscimol injections were delivered. Several days later, the animals were deeply anesthetized and perfused intracardially with saline and 10% formalin. The brains were removed, embedded in gelatin, serially sectioned in sagittal stereotaxic planes, stained with cresyl violet, and the location of the recording and stimulation sites was determined.

RESULTS

Location of Sites of Stimulus and Injection

The MRI, taken to identify the location of the microelectrodes in vivo, was of good quality (Fig. 1A-1C), except in the immediate vicinity of the stainless-steel bolts implanted in the skull. The implanted tungsten electrodes, used as markers, did not cause significant distortion of the surrounding brain tissue. From the implanted electrodes, we were able to determine the rostral/caudal and medial/lateral limits of the regions that were explored, as well as the depth of the relevant structures. The most anterior electrodes were just rostral to lobules 1–3 of the anterior cerebellum, and the most posterior electrodes were close to the interface between sublobules c and d of the uvula (arrows, Fig. 1A). At the level of the cerebellum, they were separated by 8 mm. Electrodes straddling the area of interest in the posterior vermis were 5 mm apart (arrows, Fig. 1B). An electrode that penetrated the nodulus 1.8 mm to the right of the midline is shown in Figure 1C.

Microelectrode tracks similar to the track shown in Figure 1C were identified in sagittal histological sections about 1.5 mm to the right of the midline (Fig. 2A). This section also included the abducens nucleus (not shown) and the medial vestibular nucleus (MVN; Fig 2B). Dorsally, the electrode tracks traversed the fastigial nucleus (FN), and terminated at the sites marked by the lesions in the first and second folia of the posterior bank of the nodulus (Fig. 2A) or at a, d, and f in the diagram of Figure 2B. Additional sites, which were electrically stimulated and where injections of muscimol were placed, were located more laterally in the nodulus (c, e, and g) and more medially in the anterior bank of sublobule d of the uvula (b). These sites are shown projected onto the diagram of Figure 2B, and their laterality is noted in Table 1. Four of the seven identified electrode locations were in folia 1 and 6 of the nodulus, two were in folia 2 and 5 of the nodulus, and 1 was in folia 1 of sublobule d of the uvula.

FIGURE 2.

(A) Sagittal histological section, stained with cresyl violet, of nodulus and sublobule d of the uvula. Two lesions were made at sites of injection, one in folium 1 and the second in folium 2 of the posterior bank of the nodulus. (B) Diagram of section in part (A). Both the folia of the nodulus and of sublobule d of the uvula are numbered from 1 to 6. Sites of injection are labeled with letters and are marked by the open and closed circles. a, d, and f lay in this section, b was located more medially, and c, e, and g were located more laterally. All sites are projected onto this diagram (see Table 1 for laterality). Abbreviations: MVN, medial vestibular nucleus; IV, fourth ventricle; FN, fastigial nucleus; cp, choroid plexus.

TABLE 1.

Effects of Stimulation and Muscimol Injections in Nodulus and Uvula

Cerebellar Structure	Location of Injection		Orientation of Evoked Response	Change in Vertical Component of OKAN After Muscimol Injection (%)				Change in Horizontal aVOR Tc (%)
	Site	Distance from the Midline (mm)		Tilt Ipsi SD		Tilt Contra SD
				Ipsi OKAN	Contra OKAN	Ipsi OKAN	Contra OKAN	Ipsi aVOR	Contra aVOR
Right Side
Post. Nod. Folium 1	a	2.1	HC	−58	−85	−93	−96	−71	−44
Post. Nod. Folium 1	d	2.2	SO-lSD	−52	−59	−10	−73	−67	−54
Post. Nod. Folium 2	f	2.2	SO-lSD	−47	−74	−20	−61	−43	+22
Left Side
Ant. Uvula Folium 1	b	0.75	SO-rSD	−59	−18	−54	−61	−21	−38
Ant. Nod. Folium 6	c	2.2	SO-rSD	−33	+23	−8	−64	−2	−39
Ant. Nod. Folium 6	e	2.2	SO-rSD	−12	−18	+50	−41	−8	−29
Ant. Nod. Folium 5	g	3.15	Not done	−66	−17	−45	−54	−8	−19
			Means ± (SD)	−47 ± 19	−35 ± 38	−26 ± 45	−64 ± 17	−31 ± 29	−29 ± 25

Note: Columns 1–3: Sites of stimulation and injection with laterality; column 4: Orientation of evoked nystagmus, head-centered (HC) or spatially oriented (SO), right or left side down (rSD or lSD); columns 5–8: Percent increase (+) or decrease (−) in vertical cross-coupled components of OKAN after muscimol injection; columns 9–10: Percent increase (+) or decrease (−) in horizontal aVOR time constant after muscimol. Means and ± 1 standard deviation below.

Stimulation

Nystagmus with horizontal, vertical, and torsional components was commonly induced by electrical stimulation in and around the nodulus. From some sites, the vectors of the slow phase eye velocity of the induced nystagmus were relatively fixed in head coordinates regardless of head position with respect to gravity. We termed these responses as “head centric.” Generally, there was no after-nystagmus associated with head-centric responses, and eye velocity promptly fell toward zero after stimulation. Similar responses have previously been induced from the uvula.^23,27 A sample response, induced from site a in the first folium of the nodulus, 2.1 mm to the right of the midline (Fig. 2B), is shown in Figure 3. With the animal upright (Fig. 3A), stimulation caused nystagmus whose slow phase eye velocity was predominantly along the yaw axis to the right, first with an upward and then a downward component. There was also a small counterclockwise roll component. The direction of the mean slow phase velocity vector relative to the head is shown by the arrow under the head of the upright monkey in the inset and by the arrow drawn in the two-dimensional (yaw and pitch) phase plane plot on the right. The vector was in the downward direction, being predominantly in yaw to the right. When the animal was on its left (Fig. 3B) or right side (Fig. 3C), the characteristics of the trajectory with respect to the head were similar to those of the upright trajectory. That is, the eye velocity vector was initially down with respect to the head along the yaw right and pitch up directions, and then curved back toward the yaw axis. Despite the downward eye velocity induced when the animal was left side down (Fig. 3B) and an upward velocity induced when it was right side down (Fig. 3C), the mean vectors for all three conditions were maintained within ±45° of the vector obtained with the animal upright. Thus, in each case the dominant vector of the induced nystagmus moved with the head, regardless of the animal’s position with respect to gravity.

In contrast, at other sites in the nodulus, spatially dependent responses were evoked. These had a slow rise in slow phase velocity, followed by a substantial period of after-nystagmus at the end of stimulation. The per-stimulus response induced from site c in folium 1, 3.3 mm to the left of the midline, will be considered first (Fig. 4). With the animal upright (Fig. 4A), the slow phase velocity was predominantly to the left during stimulation, but a downward velocity was also induced that declined rapidly toward the end of stimulation. As a result, there was an upward velocity vector that was close to the yaw axis (inset Fig. 4A, and phase plane plot). When the animal was tilted onto its left side (Fig. 4B), eye velocity was also predominantly to the left along the yaw axis during stimulation. Thus, the per-stimulus velocity vector was again upward with respect to the head and it shifted with respect to gravity. In contrast, when the animal was stimulated in the right-side-down-position (Fig. 4C), the yaw-axis component disappeared, and the induced nystagmus was predominantly in the pitch and roll directions. As a result, the mean-eye velocity vector in the pitch/yaw plane was spatially upward. Thus, it was maintained relatively constant in space for the upright and right-side-down positions (compare insets, Fig. 4A, and 4C). We termed per-stimulus responses that tended to maintain their orientation in space within ±45° of the vector obtained with the animal upright, irrespective of the animal’s head position with regard to gravity, as “spatially oriented.” In all cases, per-stimulus, spatially oriented vectors were obtained only when animals were on one side, and in no case was spatial orientation maintained for both the right- and left-side-down positions. In some instances, there was no spatial constancy, but rather the eye-velocity vector changed as a function of head orientation, although it was not maintained along a constant direction. We called these responses “spatially dependent.”

At 11 sites in the nodulus, responses were elicited with sufficient magnitude (> 3°/s) for analysis in at least two positions. For stimulation at five sites on the left side of the nodulus and uvula, the four responses from the nodulus were spatially oriented during right-side-down tilts and the response from the uvula was spatially oriented for left-side-down tilt. For the responses to stimulation at six sites on the right side, one was spatially oriented and two had spatially dependent responses when tilted left side down. Therefore, spatial constancy was mainly elicited when the animal was on the side contralateral to stimulation. Head-centric responses to either side were elicited from the remaining three sites on the right side. Spatially oriented and spatially dependent per-stimulus nystagmus accounted for 72% of the responses induced from the region of the nodulus and 39% of the responses induced from the region of the fastigial nucleus. In contrast, only 28% of responses induced from the region of the nodulus were classified as head centric, whereas 61% of the responses from the region of the fastigial nucleus had such characteristics. Thus, electrical stimulation of the nodulus most commonly induced per-stimulus responses that were either spatially oriented or had spatial dependence.

As in a previous study,²⁷ after-nystagmus was elicited by nodulus stimulation with the characteristics of OKAN. This included its spatial orientation properties, as well.^6,7 With the animal upright (Fig. 4A), the slow phase velocity of the after-nystagmus was predominantly to the left, producing a velocity vector that was upward along the yaw and gravitational axes, which in this case were coincident. In the LSD position (Fig. 4B), yaw velocity to the left was induced during stimulation, and a cross-coupled upward vertical velocity appeared during the after-nystagmus. Consequently, the velocity vector of the after-nystagmus swung toward alignment with the gravity vector. An orientation vector of the after-nystagmus was calculated from the declining yaw and pitch velocities (y = −3 – 1.2*x; r = 0.71, n = 85).^6,7 The resultant angle of the vector (−51°) was close to the angle of the orientation vector computed for this animal’s left OKAN in the LSD position (−49°; Fig. 5C, I). In the RSD position (Fig. 4C), no yaw velocity was induced, and there was no cross-coupling from vertical and roll to yaw. The yaw-axis time constant in the upright position was 23 s, and it fell to 14 s in the LSD position. The orientation of the eye-velocity vector to the GIA in tilted positions, as in Fig. 4B, the absence of cross-coupling from vertical or roll eye velocities to yaw, as in Fig. 4C, and the reduction of the yaw-axis time constant in side-down positions, as in Fig. 4A and, 4B, are essential features of the orientation properties of velocity storage.^6,7 This strengthens the conclusion that the afternystagmus induced by stimulation of the nodulus was similar to OKAN,²⁷ and that it was due to activation of velocity storage.

FIGURE 5.

Effects of muscimol injection at site a in nodulus, 2.1 mm to right of midline on OKN and OKAN with left slow phase velocities. (A)–(C) Horizontal and vertical OKN and OKAN velocities before injection with the animal (A) upright (B) RSD, and (C) LSD. Note the reduction in the horizontal time constant with the animal in side-down positions, and the appearance of vertical cross-coupled velocities during OKAN. (D)–(F) After muscimol, the horizontal time constant was reduced in the upright position and the cross-coupled OKAN velocities disappeared. (G)–(J) Phase-plane plots of the declines in yaw and pitch eye velocity from the onset to the end of OKAN, before (G, I) and after (H, J) muscimol. Each plot starts farthest from the origin at the onset of OKAN and approaches the origin along the orientation vector of the system.^1,2,^6,7 Horizontal eye velocity is on the abscissa, and vertical eye velocity on the ordinate of each graph. The equations, which describe the least-square linear approximation of the data points, are given above each graph together with the correlation coefficient and number of data points. The slope of the linear approximation is in parentheses. Before injection, the velocity vector of the OKAN was tilted 50° (G) and −49° (I) in side-down positions. After injection the slope was reduced (H) or abolished (J).

Muscimol Injections

Optokinetic Stimulation and Cross-Coupling

Seven muscimol injections and a control injection of saline were made into the region of the nodulus. Typical OKN and OKAN in upright and tilted positions before muscimol are shown in Figs. 5A-5C and 6A-6C. Only horizontal and vertical slow-phase velocities (HEV and VEV) are displayed. With the animal upright (Figs. 5A, and 6A), the OKN and OKAN were predominantly horizontal. A small upward slow phase velocity developed at the onset of OKAN when the animal went into darkness, reflecting the weak upward spontaneous nystagmus present in darkness in most monkeys.⁷ Similar to previous results, ^1,2,7,12 horizontal OKAN time constants were shorter in tilted than upright positions, and a prominent cross-coupled vertical component appeared (VEV, Fig. 5B, 5C; Fig. 6B, 6C). It was upward during OKAN to the left when the animal was in the LSD position (Fig. 5C), and during OKAN to the right when the animal was RSD (Fig. 6B). A smaller downward component appeared during leftward OKAN in the RSD position (Fig. 5B) and during rightward OKAN in the LSD position (Fig. 6C).

FIGURE 6.

Scheme as in Figure 5 except that OKN and OKAN had right slow phase velocities. Muscimol caused a reduction in the horizontal time constant (D), a reduction in upward cross-coupling (E), and a loss of down cross-coupling (F). Note the maintained slope in (H) and the loss of slope in (J) for the orientation vectors.

With the animal in the upright position, muscimol inactivation of the nodulus did not affect yaw-axis OKN (Figs. 5D, 6D). The nystagmus remained purely horizontal during OKN, and the orientation vector derived from the subsequent OKAN remained vertical, although the time constant was generally reduced (see below). With the animal in a tilted position, however, cross-coupled OKAN velocities were substantially reduced or abolished. The largest reduction occurred after injection at site a, 2.1 mm to the right of the midline (Fig. 2B). Cross-coupling was essentially abolished for left OKAN, with the animal either left (Fig. 5F) or right side down (Fig. 5E). In this instance, it was also lost for right OKAN with the animal LSD (Fig. 6F), and reduced for right OKAN with the animal RSD (Fig. 6E).

Changes in cross-coupled eye velocities for this site before and after muscimol are summarized in the pitch/yaw phase plane plots of OKAN (Figs. 5G-5J; 6G-6J). In the control condition for OKN and OKAN with upward cross-coupled eye velocity before muscimol (Fig. 5I, 6G), the eye-velocity vector was tilted −49° and −42°. For OKN and OKAN that induced downward cross-coupled velocities (Figs. 5G, 6I), the orientation vectors were tilted 50° and 36°. After muscimol injection into the right nodulus, there was no significant slope for the orientation vector when the animal was left side down for either left (Fig. 5J) or right (Fig. 6J) OKAN velocities. The RSD vector was diminished from 50° to 13° when the slow phase velocity was to the left (Fig. 5H), but the OKAN continued to orient to gravity for velocities to the right (Fig. 6H). Thus, after injecting muscimol at this site in the nodulus on the right, the vector stayed with the body during LSD tilts, regardless of the direction of the nystagmus velocity. This also occurred for the RSD position when the eves moved to the right.

Maximal upward and downward cross-coupled slow phase velocities before (abscissa) and after (ordinate) muscimol injections were calculated for all sites (Fig. 7) as well as the percent of the cross-coupling after muscimol relative to the original cross-coupling (Table 1). The heavy lines in Figure 7 show where points would have fallen if there had been no effect of injection. Following saline injections, the values lay on or close to this line (filled circles). In contrast, cross-coupled velocities were significantly reduced after muscimol injections. Upward cross-coupling was reduced for both right- (A) and left-side-down (B) positions, whereas downward coupling was diminished more for right- (C) than left-side-down (D) positions. In comparisons made of cross-coupled velocities before and after injection, the largest reductions (≈ 65%) occurred when the side contralateral to injection was down and the OKAN slow eye velocity was to the contralateral side, as in Figure 5F (Table 1; contra OKAN, contra SD). Reductions in other sites were less, ranging between 40 and 47%. Despite the variability, which is probably related, at least in part, to the different sites of injection, there was some reduction in cross-coupled slow phase velocity at most sites, including one in folium 1 of sublobule d of the uvula.

FIGURE 7.

Graphs of cross-coupled vertical velocity before (abscissa) and after (ordinate) muscimol injections at the 7 sites shown in Figure 2B. Values from each experiment are shown by a single dot for (A) right OKAN, RSD; (B) left OKAN, LSD; (C) left OKAN, RSD, and (D) right OKAN, LSD. The heavy diagonal line shows where the values would fall if there were no effect of the injections. Values obtained after a saline injection at one of the positive sites (filled circles) lay close to this line. Reduction in cross-coupling was greatest in (A)–(C).

OKAN and Vestibular Horizontal Time Constants

Muscimol injection also affected the horizontal OKAN time constant, which was reduced at most of the injection sites. Injection at site a, in folium 1, 2.1 mm to the right, for example, caused the horizontal OKAN time constants to become shorter on both sides with the animal upright (Figs. 5A, 5D and 6A, 6D). The horizontal OKAN time constants were also shorter in the side-down positions after muscimol at this site. At other sites, there was less reduction or no effect. Changes in time constant of vestibular nystagmus, induced with velocity steps of 60°/s with the animal upright in darkness, had similar characteristics to those described for OKAN in that the time constants generally became shorter after muscimol injection (Fig. 8A, 8B; Table 1). Before injection, mean vestibular time constants from all sites were 31 ± 6 s for left slow phase velocity and 39 ± 8 s for right velocities. After injection, time constants fell to 23 ± 8 s and 23 ± 5 s for left and right slow phase velocities, respectively.

FIGURE 8.

Horizontal (HEV) and vertical (VEV) slow phase velocity during per- and post-rotatory nystagmus in response to angular rotation at 60°/s, recorded in darkness. (A) Before injection at site a, 2.1 mm to the right of the midline, the time constants of per- and postrotatory nystagmus were 26 s (left) and 54 s (right) for the per- and post-rotatory nystagmus, respectively. (B) After injection, time constants of the per- and postrotatory nystagmus were reduced over those obtained before injection. Spontaneous contralateral (left) horizontal nystagmus appeared, rising to a steady state after the animal had been in light with a time constant of 14 s. There was no change in the animal’s spontaneous vertical nystagmus before and after injection.

Spontaneous Nystagmus

Spontaneous contralateral horizontal nystagmus was induced after each of the muscimol injections (Fig. 8B). Slow phase velocities ranged from 4°/s to 45°/s (mean 21 ± 15°/s). Rising time constants of the horizontal nystagmus, tested after the animal was put into darkness, ranged from 10 s to 33 s. An example of a 14-s rising time constant to a steady-state spontaneous nystagmus level of 20°/s to the left is shown in Fig. 8B. At all six sites in the nodulus, the slow phase velocity was contralateral to the side of injection. It was ipsilateral from the site b in lobule d of the uvula. At all sites, the nystagmus was present only in darkness, and was suppressed in light. Upward vertical components were less than 10°/s, and were most likely a reflection of the animal’s spontaneous upward nystagmus in darkness.

There was a marked effect of head position re gravity on the horizontal slow phase velocity of the spontaneous nystagmus (Fig. 9). After muscimol, head positions to the left caused an increase in slow phase velocity to the left and a decrease in slow phase velocity to the right, and vice versa. This could cause a reversal of the direction of spontaneous nystagmus recorded in the upright position, so that the animal had an apogeotropic, direction-changing positional nystagmus, if considered with regard to the quick phase direction. The nystagmus was most symmetrical at site g, but there was reversal of slow phase velocity at other sites as well (a, b, c, and e). The velocities induced by static tilt at each 15° for 6/7 sites after muscimol injection could be approximated by straight lines with high correlation coefficients (r ≥ 0.95). Interestingly, the slopes from the 6/7 sites were approximately parallel, with a mean change in slow phase velocity of ≈ 18°*s*g⁻¹ ± 2°/s (SD). There was no increase in vertical or torsional eye velocity during the horizontal nystagmus in side-down positions. Therefore, the vector or the positional nystagmus was not spatially oriented. No positional nystagmus was induced in control tilts (Fig. 9, dotted line) or after saline injection (dashed line).

FIGURE 9.

Alterations in the velocity of the horizontal spontaneous nystagmus produced by injection of muscimol at various sites in the nodulus and uvula as a function of head position with regard to gravity. Filled symbols are for left-sided and open symbols for right-sided injections. The sites are identified in the box on the right. With the exception of the results at site f, a straight line approximated the data in each set well. The average slopes for these six sites was 18.3°/g ± 2.2 (n = 6) with a mean correlation coefficient of 0.98 ( 0.01. No positional nystagmus was induced in tilted positions in controls or after NaCl injection.

Off-Vertical Axis Rotation

The animals were rotated in yaw at 60°/s about axes tilted at angles from 0° to 90° in 15° steps. The steady-state (bias) velocities, the depth of modulation as a function of head position with regard to gravity, and the phase of these modulations were analyzed for both the horizontal and vertical components. Bias velocities were induced both before and after muscimol injection. When the slow phases of the nystagmus induced by OVAR and the spontaneous nystagmus were oppositely directed, the bias velocity was reduced by the amount of the spontaneous nystagmus. Before injection, horizontal bias velocities decreased slightly as the angle of tilt was increased. After injection, this tendency was retained for some sites and lost at others. The amplitude of the modulations in horizontal slow phase velocity during each cycle of OVAR increased after injection. This increase in modulation could be attributed to the reduction in the horizontal time constant associated with muscimol injection (Fig. 8, Table 1 ), similar to the effects of habituation.⁵² The phase of modulation of horizontal slow phase velocity was unaffected by muscimol injections.

Single-Unit Recording

Ten units, which were sensitive to head orientation with regard to gravity, were recorded in the rostral 1/3 of the nodulus. The mean coefficient of variation of the units was 0.33 ± 0.05, identifying them as “irregular” units⁵⁸ (Table 2). The mean spontaneous activity of these cells was 55.4 ± 11.3 imp*s⁻¹. All of these units received convergent input from neck-muscle proprioceptors, determined by causing oscillations in unit activity while pressing on the back of the neck. Unit sensitivity to tilts of the GIA, determined during static tilt, dynamic tilt, and OVAR are listed in Table 2.

TABLE 2.

Directions of Response Vectors for Ten Otolith-Related Units Obtained with Static Tilts, Dynamic Tilts, and with OVAR^a

Unit No.	CV	Spontan. (imp*s−1)	Static Tilt		Dynamic Tilt		OVAR Phase			Response Plane
			Gain (imp* s−1*g)	Phase (deg)	Gain (imp*s−1*g)	Phase (deg)	CW (deg)	CCW (deg)	Aver. (deg)
Otolith Only—Related
4	0.44	56.8 ± 6.6	11.4	93	17.5	69	147 ± 45	8 ± 42	78	Contra-SD
Otolith & Vertical Canal Related
3	0.33	59.5 ± 3.3	6.8	124	42.0	38	—	—	—	Contra-AC
6	0.41	32.4 ± 3.1	—	—	54.2	145	—	—	—	Contra-PC
12	0.42	42.5 ± 7.5	—	206	12.4	139	—	—	—	Ipsi-PC
13	0.32	62.7 ± 2.4	19.0	222	20.1	226	—	—	—	Contra-PC
14	0.41	54.9 ± 3.8	—	—	91.2	68	—	—	—	Ipsi-SD
15	0.31	73.1 ± 3.2	—	—	11.3	131	290 ± 32	226 ±41	258	Contra-SD
Spatial-Temporal Convergence
2	0.44	57.8 ± 6.5	1.0	335	7.7	311	228	45	316	Ipsi-AC
16	0.39	51.2 ± 8.7	13.8	244	7.6	277	305 ± 27	210 ± 31	258	Contra-AC
18	0.38	62.6 ± 10.3	33.2	306	25.0	341	205 ± 29	23 ± 5	294	Contra-AC

^aThe directions of the response vectors (phases) obtained with the different techniques were generally comparable.

Abbreviations: Ipsi, Ipsilateral; contra, contralatral; SD, side down; AC, anterior canal; PC, posterior canal; imp., impulse; deg., degree.

Recordings from a typical cell during static tilt are shown in Figure 10A. The unit was recorded while the head was tilted 60° about a horizontal axis in various orientations relative to the plane of tilt. Each time the animal was tilted, there was a phasic increase in activity (downward arrows). We attribute this to neck-muscle activation due to the body-position change. The mean firing rate in tilt was determined for each head orientation in the period shown by the heavy horizontal lines (Fig. 10A). The sensitivity of the unit (imp*s⁻¹*g⁻¹) varied as a function of the orientation of the head to the plane of tilt (Fig. 10B). The data were well fit by a cosine function (r=0.760) that had a peak sensitivity of about 14 imp*s⁻¹*g⁻¹ with a head orientation of 244° and a zero crossing at 154°. Thus, this unit was polarized close to the plane of the left posterior canal (225°).

FIGURE 10.

Effect of static tilts (A)–(C) and dynamic tilts (D), (E) on single units recorded in the nodulus. Coordinate system: tilt forward (0°/360°), LSD (90°), tilt back (180°), RSD (270°). (A) Variations in firing frequency produced by tilting the animal 60° with the head oriented differently to the plane of tilt (top trace). The arrows point to the phasic increases in firing, presumably caused by activation of neck proprioceptors. Heavy horizontal lines indicate the mean activity during the periods used for analysis. (B) Sensitivity of this unit plotted as a function of head orientation with regard to the plane of tilt. The data were fitted with a sinusoid whose maximum represents the direction of the response vector, which in this instance, was 244°. (C) Polar plot representing direction of response vectors for all recorded units. One unit was oriented along a 90° axis, and two other units fell within ± 22.5° of 90°/270° plane. The other units fell close to the LARP (135°, 315°) or RALP (45°, 225°) planes. (D) Spatial responses of otolith-related nodulus units tested with dynamic tilt. (D) Non-STC unit; (E) STC unit. Gains (top graph) and phases (bottom graph) are plotted as a function of head orientation with regard to the plane of tilt.

All units were tested with sinusoidal oscillation around a spatial horizontal axis with the head in similar orientation relative to the plane of tilt as for static tilt. The orientation of the sensitivity vectors for these 10 units was similar for static tilt as for dynamic oscillation about a spatial horizontal axis. One of the 10 units had activity modulated in phase with head position, indicating no convergent input from the vertical canals (Table 2). Six of the tested units had modulation in phase with head velocity. The spatial response of one of them is shown in Figure 10D. This unit had a spatial phase of 145°, suggesting convergent input from the right posterior canal. Its peak sensitivity was about 54 imp*s⁻¹*g⁻¹. Three of the 10 units had temporal phases that changed as a function of orientation of the head to the plane of tilt (Fig. 10E). These units have been described in the vestibular nuclei as “spatial-temporal convergence units.”^59,60 This could also indicate possible convergent input from the vertical canals.⁶¹ Five of the 10 recorded units were also tested during OVAR. Response vectors led head orientation in all cases. If the average response vector was defined based on the sum of the responses during clockwise and counterclockwise rotation, then the orientation of the response vectors for each unit was close to that based on static or dynamic tilt data.

The average direction of the response vectors for the 10 units are shown in Figure 10C. For seven units, the response vectors were determined from static tilts. For three units, not tested with static tilt, response vectors were based on dynamic data. We determined that units were related to particular canal planes if the direction of the response vector was within ± 22.5° from the canal plane. Maximal activation of 7 of the 10 units occurred for tilts that were close to the planes of the right (135°) and left (225°) posterior canals and the left (315°) and right anterior (45°) canal, respectively. Three units were ≈ 90° from the midline. Thus, a majority of nodulus units that were recorded tended to have their maximal activation vectors aligned with vertical-canal planes, and some of them also received vertical-canal input.

DISCUSSION

This study supports the postulate that the nodulus is a critical structure for determining alignment of the eye-velocity vector with the GIA, which we define as spatial orientation of the aVOR. When the head is tilted relative to the GIA, this spatial orientation is achieved by an alteration of the horizontal, vertical, and roll time constants of velocity storage and by the generation of cross-coupled vertical and/or roll velocities.^1,2,6-8,12 While the nodulus has been shown to play a critical role in controlling the horizontal time constant of the aVOR from stimulation^23,27 and lesion studies,^1,2,21,62 the present study demonstrates that the nodulus can also produce spatially oriented cross-coupled responses when electrically activated. The spatially oriented eye-velocity responses induced from the nodulus behaved as if the animal were tilted, and the nodulus was generating a response to an altered orientation in space. When the head was actually tilted, generally to the contralateral side, the response was held constant in space, despite the change in the head position with respect to gravity. Consistent with previous data,²⁷ poststimulus eye velocities that had the appropriate temporal and spatial characteristics of OKAN and of velocity storage, including orientation to the GIA, were also elicited from these sites.

Stimulation sites that caused cross-coupled components of eye velocity produced a concomitant reduction of yaw-to-vertical cross-coupling of horizontal OKAN when inactivated with muscimol. Although the effect was variable, inactivation occurred at all seven sites. This is strong support for the postulate that cross-coupling is controlled by the nodulus, and is consistent with effects of lesions that caused a loss of spatial orientation of the aVOR for eye velocity around all axes.^1,2,63,64 Since there is direct input from the labyrinth to the nodulus, the effects of stimulation could conceivably have been mediated by back activation of collateral branches through the vestibular nuclei. The injection data rule out this possibility, because it was local inactivity of cerebellar circuits that caused the reduction in cross-coupling.

The per-stimulus eye velocities that were oriented in space did so with the head in only one lateral direction, and moved with the head when the animal was tilted in the opposite direction. In general, it appeared that the direction of spatial constancy was for contralateral eye velocities. This is probably a reflection of the fact that the predominant pathways from the nodulus to the vestibular nuclei are ipsilateral,⁴⁶ to the side that produces contralateral slow phase velocities when excited. Taken together, with the results of stimulation, these findings extend the conclusions from bilateral lesion experiments,² in that they suggest that sites in the nodulus control spatial orientation in a specific direction. This is compatible with findings that there was only unilateral, never bilateral, dumping after nodulus stimulation.²³ The muscimol injections that abolished cross-coupling with or without changes in horizontal time constant are consistent with the idea that the processes that control cross-coupling and the aVOR time constant are separately controlled by different portions of the cerebellar cortex.²

The idea that spatial orientation should be represented in the cerebellum is not new. It was implicit in the findings that the cerebellum is more highly developed in Cetacea than in terrestrial animals.⁶⁵ Since whales have little or no fine limb coordination, it was surmised that their well-developed cerebella could be used for spatial orientation and navigation. The cerebellum of the weakly electric fish, Eigenmanni, is also involved in location in space.⁶⁶ A similar conclusion was reached in the rat.⁶⁷ More recently, Pompeiano and coworkers have come to a similar conclusion.⁶⁸ Investigating spatiotemporal-response properties of Purkinje cells in the anterior vermis (lobules 1–3), they found a population of Purkinje cells that coded the direction of head tilt in space. “For each selected time in the tilt cycle, the direction of the population vector closely corresponded to that of the head tilt, while its amplitude was related to that of the stimulus.” They conclude that this could provide a substrate for the “spatial organization of vestibulospinal reflexes induced by otolith receptors,” and that “the Purkinje cells of the cerebellar cortex are expected to show prominent responses to head rotation, which could affect the spatially organized postural responses by utilizing vestibular and reticular targets.”^{67 72}

The regions that were explored in this study lay within a relatively restricted range, and the MRI proved to be of value in pinpointing the sites of recording in the cerebellum. Particularly when recording from regions that do not have characteristic activity, as from the nodulus, it may be difficult to determine depth and laterality during electrode penetrations in the live animal. It was of interest that the anatomic structure defined by the MRI was consistent with the postmortem histology: the mean separation of the electrode tracks on the right and left side of the nodulus were 5.4 mm (Table 1), which is close to the 5-mm separation of the electrodes shown in Figure 1B. Thus, MRI provided an efficient technique for establishing electrode placement. Initially, there was concern that the stainless-steel bolts would become overheated when subjected to the magnetic fields, and that the artifacts from the bolts would obscure the MRI. Neither of these concerns proved relevant, nor was it necessary to have special equipment to hold the head in a stereotaxic plane when the MRI was taken, since the plane of section was readily adjusted in the computer reconstruction of the images.

The regions that were injected in the nodulus extended from 2.1 mm to 3.3 mm from the midline (mean 2.7 ± 0.5 mm, n = 6). This encompasses the lateral border of Zone 2 at 2.4 mm, which contains input from the vlo nucleus, the rostral medial accessory olive (rMAO), and the rostral β nucleus. In the rabbit these regions carry climbing fiber activity from the ipsilateral anterior canal, the contralateral posterior canal, and the utricle. Zone 3 receives climbing fiber input from cdc of Kooy, which in turn gets information from the nucleus of the optic tract (NOT) about ipsilateral horizontal retinal slip for movements about a vertical axis. This is a response to movement in the same plane as the lateral canals.⁷³ Assuming that the spread of the electrical current for a 40-μA current²³ and for a muscimol injection of about 1 μL was approximately 1.0 mm,^74,75 then a 1.2–1.4-μL injection would extend somewhat farther. Thus, we presume that the vertical rotatory responses that were elicited by stimulation and the cross-coupling that was abolished by injection were probably largely due to activation or inactivation of Zone 2, and alterations in horizontal time constant were probably due to activation or inactivation of Zone 3. The nodulus is a large structure, approximately 7–8 mm wide, 2 mm thick, and 5 mm from dorsal to ventral surface, and only limited regions were explored in our study. As a result, conclusions about nodulus function must necessarily be tentative.

An interesting aspect of muscimol inactivation of the nodulus was the occurrence of contralateral spontaneous nystagmus from all sites of injection. The origin of this nystagmus was likely due to a reduction in GABAergic inhibition of neurons in the ipsilateral vestibular nuclei. The fact that the nystagmus had contralateral slow phase velocities is probably explained by the fact that the output connections of the nodulus are predominantly to the ipsilateral vestibular nuclei. Muscimol would reduce inhibition in this vestibular nucleus, thereby producing contralateral velocities. Two findings suggest that the relation between head position and velocity of spontaneous nystagmus is a general property of the vestibular system. The alteration in eye velocity is similar to the effect of head position on spontaneous nystagmus that was observed after muscimol injections in the vestibular nuclei (Yokota and Cohen, unpublished observations). Second, the slope of the response was similar after all injections, and was independent of the site of injection.

The positional nystagmus after muscimol was apogeotropic. That is, the slow phases were toward gravity and the quick phases were away from gravity, and this reversed when the monkey’s head position was reversed. Similar nystagmus is often observed after cerebellar and/or brainstem lesions (unpublished observations). Since no vertical or roll spontaneous nystagmus was generated as a function of head orientation, the vector of spontaneous eye velocity was directed along the head yaw and remained invariant. Therefore, this aspect of the response was not related to spatial orientation, apparently precluding involvement by velocity storage. On the other hand, the rising time constants of the slow phase velocity were in the range of 10 to 33 s, which was in the range of the aVOR-dominant time constant, and the peak velocities ranged up to 40°/s, which is close to the saturation velocity of velocity storage,^{57 76} as well as to the saturation level of lateral canal-related, horizontal vestibular-only (VO) neurons to optokinetic stimulation.^77,78 Thus, the nystagmus could have been produced by reduction of Purkinje cell inhibition of VO neurons in MVN, which are believed responsible for production of velocity storage.^54,79,80 The source of the nystagmus as well as the process that rather precisely sets the level of spontaneous nystagmus, as a function of head position with regard to gravity, are subjects for further study.

There were only relatively minor effects of nodulus inactivation on the nystagmus associated with OVAR, namely an increase in the modulation in horizontal eye velocity associated with head position with regard to gravity, and a slight decrease in the peak steady-state yaw velocity. Illis is consistent with the previous lesion studies,² in which OVAR nystagmus was produced with relatively normal characteristics after the nodulus and uvula were removed. The implication of this is that processing of otolith information to produce slow phase velocity during OVAR is probably not done in these regions of the vestibulocerebellum.

Single-unit recording also supported the idea that the nodulus subserves a spatial orientation function by the broad presence of otolith-related activity. Such activity would be a prerequisite to determining the tilt of the head with regard to the GIA. Interestingly, the sensitivity vectors of the nodulus units were polarized close to the same anterior- and posterior-canal planes that are represented in Zone 2 of the vermis of the nodulus,³¹ where injections of muscimol caused a loss of cross-coupling. Recently, we demonstrated that otolith-related units in the vestibular nuclei had their maximal sensitivity close to the planes of the vertical canals from which they had convergent input.⁶¹ This otolith–canal convergence is likely to have specific functional significance: it provides a mechanism for sensing head orientation along particular planes of movement.

From the clustering of the unit sensitivities in or close to canal planes in the nodulus, we utilize a previous model^1,2 to propose how system time constants that account for spatial orientation of the aVOR might be dynamically altered. If the incoming otolith activity to individual zones were organized along canal planes, as is suggested in Figure 10C and Table 2, and if the canal-related Purkinje cells located in these canal-related zones were activated by otolith neurons whose sensitivity vectors lay close to these same canal planes, then the Purkinje cells could provide a feedback mechanism to accomplish the changes in time constants and the production of cross-coupling that are the basis for spatial orientation. This is shown schematically in Figure 11. A broad distribution of spatial vectors in the vestibular periphery, that is, in the utricle and saccule, is mapped onto the nodulus in three functional canal planes, which then influence the eye movements produced around the axis of that canal plane. Purkinje cells in the nodulus canal-related zones project to regions of MVN and SVN where eye movements in these canal planes are produced.⁴⁶ The result of otolith activation due to a tilt of the head with regard to gravity would first be to activate the otolith neurons that sensed the tilt. These in turn would project to the canal-related zones in the nodulus and possibly sublobule d of the uvula via mossy fiber input, in a canal-related organization. In turn, Purkinje cells in these zones would project to specific canal-recipient areas of the vestibular nuclei to produce eye velocity along the axes of each of the reciprocal canal pairs that were associated with that tilt. Presumably, this would be done by inhibition of lateral canal-related VO neurons in MVN and by disinhibition of vertical canal-related VO neurons in SVN.^1,2 If the subject was in angular motion, this input could then alter the axis of eye rotation to align with the tilt of the GIA with regard to the head.

FIGURE 11.

Schematic representation of scheme whereby units with polarization vectors in all directions in the labyrinth are mapped into the semicircular-canal coordinate scheme in the nodulus. See text for details.

Thus, the neural machinery to implement spatial orientation functions during a wide range of motions could simply be a honing of otolith-related polarization vectors that control spatial orientation, which are widely distributed in the utricular and sacular maculae, to more confined planes related to the semicircular canal organization in the nodulus. Such a mechanism has previously been suggested as a means for estimating three-dimensional head velocity during rotation about axes that are not aligned with the GIA.⁸¹ This mechanism could then be utilized for controlling spatial orientation of semicircular canal-based systems utilized for estimating spatial orientation during circular locomotion,¹⁵ as well as during passive circular motion.