INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The vestibuloocular reflex (VOR) reduces the slip of visual images on the retina by rotating the eyes in the direction opposite to head rotation. The vestibular sensors are the semicircular canals, which are stimulated by angular acceleration, and the otolith organs, which are stimulated by linear acceleration. When a person is rotated in darkness at a constant velocity about an axis lying in the earth's horizontal plane, a sustained compensatory eye velocity is produced that is believed to be generated by the changing direction of the pull of gravity on the otolith organs (Guedry 1965). During such constant velocity horizontal or off-vertical axis rotations (OVAR), the sinusoidally modulated activity of otolith afferents with different preferred directions must be converted at central neurons into a bias signal that is proportional to head angular velocity (Goldberg and Fernández 1981, 1982; Raphan et al. 1981, 1983) in a process called "angular velocity estimation" (Raphan and Schnabolk 1988). It has been proposed (Angelaki and Hess 1996) that the same mechanism is responsible for the superior low-frequency performance of horizontal axis VOR as compared with vertical axis VOR.

Both the vestibular nuclei and the cerebellum have been implicated as important parts of a network involved in velocity estimation and the central transformation of otolith signals. Tracer studies show that both areas receive large, direct input from primary otolith afferents (Dickman and Fang 1996; Naito et al. 1995). Velocity estimation can be abolished either by midline medullary lesions of the vestibular commissure as measured during constant velocity OVAR (Katz et al. 1991) or by lesions of nodulo-uvular cerebellum as measured during constant velocity horizontal axis rotations (Angelaki and Hess 1995b). In addition, recording studies demonstrate that neurons in vestibular nuclei carry a compensatory velocity estimator signal in decerebrate cats during parallel swing rotation (Benson et al. 1970) and in alert monkeys during constant velocity OVAR (Reisine and Raphan 1992). These results suggest that the neural circuitry for velocity estimation is shared between the vestibular nuclei and the cerebellum.

In this study we recorded eye movements in wild-type and mutant mice as we rotated the animals about a horizontal axis to test the obligatory role of the nodulus and uvula in velocity estimation. Purkinje cell degeneration (pcd/pcd) mice have an unspecified autosomal recessive mutation on chromosome 13 (Campbell and Hess 1996; Southard and Eicher 1977) resulting in a moderate ataxia associated with the rapid loss between postnatal days 15 and 45 of virtually all Purkinje cells (PCs) (Landis and Mullen 1978; Mullen et al. 1976). Purkinje cell death occurs after normal innervation of PCs by climbing fibers (Ghetti et al. 1987; Triarhou and Ghetti 1991) and of deep cerebellar nuclei by PC axons (Roffler-Tarlov et al. 1979). After PC degeneration, up to 50% of inferior olive neurons die (Ghetti et al. 1987; Shojaeian et al. 1988; Triarhou and Ghetti 1991), and granule cell degeneration is severe (Ghetti et al. 1978). It is believed that any remaining PC axons do not make synapses with other neurons (Bäurle and Grüsser-Cornehls 1994; Bäurle et al. 1998), which may account for the milder ataxia pcd/pcd mice have in comparison with other mutants such as weaver, whose surviving output from cerebellar cortex is dysfunctional (Grüsser-Cornehls et al. 1999) and may disrupt operation of brain stem circuits. The relatively uncomplicated nature of the pcd genetic lesion makes these mutants good subjects for exploring the vestibular-cerebellar network.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

All procedures followed the principles of laboratory animal care set forth by the National Institutes of Health in the Guide for Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee at Northwestern University.

Nineteen mice were used (9 pcd/pcd mutants, 4 pcd/+ wild-type littermates, 6 C57BL/6J controls; Jackson Laboratory). The pcd/pcd mice and littermates were about 6 mo old at the time of experiments. Mice were prepared for repeated eye movement recordings using sterile surgical procedures under isoflurane/O₂ anesthesia. The mouse was held in a stereotaxic apparatus with the use of pressure bars against the sides of the skull so that lambda was level with bregma. A midline incision exposed the dorsal cranial surface, and small stainless steel screws (000 × 3/32 self tapping, Small Parts) were placed in the skull at four locations to anchor an implant. Dental acrylic was used to attach a small plastic head-holder with a surface parallel to the lambda-bregma line, for restraining and positioning the mouse during experiments. Two C57BL/6J control mice were equipped for electrolytic lesions. A small craniotomy was made on the midline over the nodulus, 6.5 mm posterior to bregma. A chamber consisting of 10 mm of 25 gauge hypodermic tubing was oriented vertically and cemented with dental acrylic over the craniotomy.

After the mouse had recovered for several days, a small search coil was acutely attached to the cornea to monitor eye position. Coils had a resistance of about 15  and consisted of 100 turns of varnished 20-µm copper wire (California Fine Wire CFW-051-0008-SPU) wound on a form 0.6 mm diam. The eye was anesthetized with tetracaine HCl (0.5%, one drop/10 min), and the lids were retracted with small blunt hooks. A minimal amount of cyanoacrylate adhesive was applied to one end of the coil. Centered over the pupil, the coil was glued to the cornea so that its long axis was approximately aligned with gaze.

The mouse's head-holder was fixed to a carrier in a clamp angled at a 45° nose-down pitch to approximate the normal head posture. We called this the upright posture. Horizontal VOR was tested in this posture by 0.2-Hz sinusoidal rotations about a vertical yaw axis at a peak stimulus velocity of 100°/s. Head angular velocity estimation was tested by positioning the mouse at a 90° nose-up pitch from the upright posture and rotating it about a dorsal-to-ventral horizontal yaw axis at a constant velocity of 16, 32, 64, 128, or 200°/s to the left or right; data collection started 30 s after the beginning of rotation to allow canal signals time to decay. Low-frequency (0.05 Hz) and velocity step (100°/s) rotations about a vertical yaw axis in the upright posture were used to evaluate VOR dynamics.

All eye movements were detected in darkness by a search coil system. Transmitting coils were attached to the animal carrier to fix them with respect to the animal's head and were used to set up three mutually perpendicular magnetic fields of equal strength, oscillating at 67, 83, and 100 kHz in the rostrocaudal, interaural, and dorsoventral directions, respectively. Eye position and platform angular displacement analog signals were low-pass filtered at 100 Hz and sampled at 1,000 Hz by a computer that also delivered stimulus waveforms. It was necessary to subtract offset signals recorded by the eye coil connector and preamplifier cables, and these signals were isolated by momentarily short-circuiting the leads as close to the eye coil as possible. Horizontal eye position traces were then calculated from the arctangent of the ratio of the eye coil signal induced by the rostrocaudal field to the eye coil signal induced by the interaural field. Eye and head position traces were differentiated, and saccades were manually removed. For sinusoidal rotations the nonsaccadic portions of the eye and head velocity traces were fit by sinusoids. VOR gain was calculated as the ratio of the eye to head velocity fit amplitudes, and VOR phase as the difference in eye and head velocity fit phases, with perfect compensation expressed as a phase error of 0°. For most experiments, eye movements were calibrated by mounting the eye coil on an earth-fixed pole before placing it on the cornea. The eye coil was positioned in the magnetic field where the eye of the mouse would be in an experiment, and the apparatus rotated ±18°. The arctangent trace obtained represented a VOR of ±18°.

Nodulus lesions in the chamber-equipped mice were made by a dorsal approach. The mouse was rotated sinusoidally about a vertical yaw axis in the upright posture and also placed in a nose-up position and rotated at constant velocity about a horizontal yaw axis. Eye movements were recorded to establish baseline VOR. The rotation was stopped and the dura was punctured with a guide tube. A microwire was advanced through the guide tube into the vermis, and a lesion was made. The mouse was rotated as before, and eye movements were recorded. We continued to advance the microwire and lesion at approximately 0.5-mm intervals until we observed an effect on the eye velocity bias during constant velocity rotations. Lesions were made with a straightened, 25µm-diam, insulated stainless steel microwire with an approximately 200-µm exposed tip, passing 100-200 µA of current, electrode tip positive, for 30-60 s. Two to 13 days after the lesion, the VOR was measured again, and then the animals were anesthetized and perfused through the heart with buffered saline followed by 4% paraformaldehyde. The brains were embedded in gelatin, cut in 50-µm coronal sections, and stained with cresyl violet to examine the extent of the lesions. A pcd/pcd mouse was also perfused and its brain similarly processed for microscopic examination of nodular cerebellar cortex.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

All control mice (4 pcd littermates, 6 nonlittermate C57BL/6J) exhibited basic features of the VOR observed in larger mammals, including horizontal slow-phase eye movements during sinusoidal rotation, and steady-state eye velocity bias during long-duration constant velocity rotations about a horizontal axis. Top traces in Fig. 1, A and B, show horizontal eye movements in a wild-type, pcd-littermate mouse, during constant velocity rotation at 64°/s about a horizontal yaw axis. Nystagmus continued for the duration of rotation and is seen as many sawtooth deflections of the eye position trace with slow phases to the left (A, downward slopes of trace) or right (B, upward slopes of trace), depending on the direction of head rotation. Differentiation of these eye position traces shows the compensatory bias in eye velocity (bottom traces) and a modulation of eye velocity that depended on the orientation of the head with respect to gravity.

View larger version (47K): [in this window] [in a new window]   Fig. 1. Horizontal eye movements in response to constant velocity rotations about a horizontal yaw axis at 64°/s in wild-type and Purkinje cell degeneration (pcd/pcd) mutant mice. Top trace is a potentiometer record of platform/head position (H). Jump in trace approximately indicates nose down position. A: rightward rotation, wild-type mouse. B: leftward rotation, wild-type mouse. C: rightward rotation, pcd/pcd mouse. D: leftward rotation, pcd/pcd mouse. Top traces in each panel are horizontal eye position (E). Bottom traces show horizontal eye velocity (), obtained by digitally differentiating the eye position traces. Horizontal line marks zero eye velocity. For all eye position and velocity traces, upward deflections correspond to rightward eye rotation. For display purposes only, eye position traces have been smoothed with a 33-point box smooth, and eye velocity traces have been smoothed with 2 passes through a 33-point box smooth (Igor Pro, Wavemetrics). R, rightward; L, leftward.

All pcd/pcd mice (n = 9) also showed a compensatory eye velocity bias during constant velocity horizontal axis rotations. Top traces in Fig. 1, C and D, show eye movements in a pcd/pcd mutant during 64°/s horizontal axis yaw. The pcd/pcd mouse eye velocity traces in Fig. 1 (C and D, bottom traces) show a larger bias than observed in the control mouse, corresponding to the steeper slow-phase slopes in the eye position traces. Compensatory eye velocity continued for as long a period as was tested, up to 3 min, in both wild-type and pcd/pcd mice.

Eye position traces like those shown in Fig. 1 were differentiated and edited manually to remove quick phases of nystagmus, and the average of the remaining slow phase data was computed as a measure of the bias in eye velocity. Figure 2 compares wild-type (n = 10) and pcd/pcd (n = 9) mouse average bias during constant velocity yaw rotation at head velocities of 16-200°/s, with data averaged by animal. Eye velocity bias in unlesioned wild-type mice and pcd/pcd mice was always compensatory to head velocity. Bias data for leftward and rightward rotations have been averaged for display in Fig. 2. There was no significant difference in bias between pcd/pcd and wild-type mice at 64°/s (t-test, P = 0.74). Below 64°/s bias was slightly larger in pcd/pcd than wild-type mice (significantly larger at 32°/s, P < 0.01), but as head velocity increased the reverse was true. Bias increased with rotation velocity in both pcd/pcd and wild-type mice, but the bias in pcd/pcd mutants appeared to saturate between 128 and 200°/s, and was significantly lower than in the controls at those velocities (P < 0.01).

View larger version (27K): [in this window] [in a new window]   Fig. 2. Means of the bias in horizontal eye velocity in wild-type (; n = 10), pcd/pcd (; n = 9) and 2 lesioned wild-type ( and ) mice during constant velocity rotations about a horizontal yaw axis at stimulus velocities of 16-200°/s. Error bars represent standard errors. Bias was computed as the average eye velocity after removal of fast phases. Data for leftward and rightward rotations have been averaged.

The two wild-type, C57BL/6J mice that were subjected to electrolytic lesions of the nodulus both showed severe attenuation of eye velocity bias during constant velocity rotations about a horizontal yaw axis (Fig. 2, bottom 2 traces), a result first reported in primates (Angelaki and Hess 1995b). Figure 3 shows sample traces from one mouse before (A and C) and after (B and D) the nodulus lesion (E). Before the lesion, constant velocity rotation about a horizontal yaw axis (A) elicited a vigorous nystagmus (A, 2nd trace) with a compensatory eye velocity bias (A, 3rd trace). After the lesion, the eye velocity bias was reduced substantially (B). However, eye velocity in response to 0.2-Hz sinusoidal rotations (C and D) was not decreased by the lesion (see also Table 1). The extent of the lesions in the two mice (Fig. 3, E and F) is shown in tracings of representative brain sections from posterior to anterior (left to right). Both mice had damage to medial nodulus (most ventral cerebellar structure in tracings), as well as damage to overlying cerebellar cortex (lobules 4-6) and minimal invasion of the medial fastigial deep cerebellar nucleus. No brain stem damage was detectable in either case.

View larger version (54K): [in this window] [in a new window]   Fig. 3. Effects in a normal mouse of a medial cerebellar lesion including the nodulus on velocity estimation and sinusoidal horizontal vestibuloocular reflex. A and B: eye movements during yaw rotation about a horizontal axis at constant velocity. Head position and eye traces as in Fig. 1. A: prelesion data. B: postlesion data. C and D: eye movements during sinusoidal yaw rotation about a vertical axis. Gaps in sinusoidal head velocity traces represent breaks in the potentiometer signal. C: prelesion data. D: postlesion data. E and F: lesions of normal mice M091 (A-D, this figure) and M092 from Fig. 2. The extent of each lesion is shown projected onto brain section tracings from an atlas (Franklin and Paxinos 1997) at approximately 6.9 (left) to 6.1 mm (right) posterior to bregma. H, potentiometer record of head position; , head velocity; E, eye position; , eye velocity.

Table 1 summarizes data related to the dynamics of the horizontal VOR of the various mouse groups. Compensatory horizontal VOR eye velocity led head velocity during vertical axis sinusoidal rotations at 0.05 Hz in control mice, pcd/pcd mutants, and the two control mice subjected to nodulus lesions. Phase lead was greatest in pcd/pcd mice and was increased in normal animals by medial nodulus lesions. VOR gain was nearly the same for all groups at 0.05 Hz, but at 0.2 Hz it was higher in pcd/pcd mutants and the lesioned animals than it was in controls. Exponential fits to the decay in nonsaccadic eye velocity were used to assess time constants of horizontal VOR responses to 100°/s vertical axis velocity steps in control mice, two pcd/pcd mutants, and the two mice with nodulus lesions. Time constants for the two pcd/pcd mice tested were slightly shorter than the average for control mice, while nodulus lesions in wild-type mice resulted in the shortest VOR time constants of all.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Models of angular velocity estimation feature a convergence of delayed and undelayed signals from otolith afferents with incremental differences in functional polarization vectors (Raphan and Schnabolk 1988), differentiation of otolith afferent activity to obtain a "jerk" signal (Hain 1986), a pair of otolith signals in spatial and temporal quadrature (Angelaki 1992), or internal models of sensor dynamics (Merfeld et al. 1993). In all cases, processing of neural signal timing or dynamics is required, and such processing has been proposed as fundamental to cerebellar function (Fujita 1982; Ivry et al. 1988; Kettner et al. 1997). The presence of a compensatory eye velocity bias in the pcd/pcd mutant during constant velocity horizontal axis rotations shows that the required neural processing can be accomplished by neural substrates beneath the cerebellar cortex, most probably by the vestibular nuclei or the medial cerebellar fastigial nucleus, the rostral portion of which carries vestibular signals (Gardner and Fuchs 1975), including those of otolith origin (Siebold et al. 2001).

Nodulus lesions that decrease the bias in eye velocity associated with angular velocity estimation also either lengthen or shorten the dominant VOR time constant. Large nodulus lesions in primates abolish velocity estimation (Angelaki and Hess 1995b) and lengthen the horizontal VOR time constant (Angelaki and Hess 1995a,b; Waespe et al. 1985; Wearne et al. 1998). However, in the two normal mice of this study, medial nodulus lesions impaired velocity estimation and shortened the horizontal VOR time constant. An earlier study of nodulus lesions in mice found VOR phase leads suggestive of a shortened VOR time constant (Koekkoek et al. 1997). Medial nodulus lesions may also slightly shorten the horizontal VOR time constant in primates, although the shortening effect is larger on the vertical VOR time constant (Wearne et al. 1998). A better understanding of the relation beween the length of the dominant VOR time constant and velocity estimation will require studies that report both the VOR time constant and eye velocity bias and that take into account the placement and extent of cerebellar lesions.

Unlike the lesioned animals, pcd/pcd mutant mice are nearly normal in both velocity estimation and the length of the VOR time constant, although the large phase lead at 0.05 Hz is consistent with a shorter time constant. It is not likely that the 0-10% remaining Purkinje cells in the nodulus of pcd/pcd mice (Mullen et al. 1976) can maintain cortical functions; they are believed to be abnormal and mostly without corticonuclear connectivity (Bäurle and Grüsser-Cornehls 1994; Bäurle et al. 1998). We were unable to confirm the presence of any Purkinje cells in the pcd/pcd mouse nodulus that we examined. It is unknown to what extent velocity estimation circuits beneath cerebellar cortex in the pcd/pcd mouse differ from the corresponding circuits in normal animals. On the one hand, they may be very similar. Purkinje cell degeneration in the pcd/pcd mutant occurs late enough that neural circuitry outside cerebellar cortex may have developed normally, and loss of Purkinje cell inhibition could stimulate late developmental compensation or adaptation that simply reduces activity in this circuitry to near normal values. The pcd/pcd mouse has an increase in glycine-immunopositive neurons in deep cerebellar nuclei (Bäurle and Grüsser-Cornehls 1997) and a decrease in resting rate and vestibular sensitivity of vestibular nucleus neurons (Bäurle et al. 1997), both of which may be developmental compensations for the loss of Purkinje cell inhibition. On the other hand, it is possible that a modulated signal must be generated to replace the missing control signal from nodulus Purkinje cells, and there may be unknown, early developmental compensations beneath cerebellar cortex in pcd/pcd mice that result in substantially different neural circuits from those that exist in normal mice.

The significantly weaker compensatory bias in slow-phase eye velocity in pcd/pcd mice as compared with controls during high-velocity rotation suggests that cerebellar cortex is required for the optimal processing of rapidly fluctuating otolith signals to produce fast smooth eye rotations. This is consistent with a view of the cerebellum as a structure well suited for the control of rapid movements that require complex computations (Spoelstra et al. 2000). The limits of developmental compensation for cerebellar loss and the role of the cerebellum in ocular control could be clarified by studies of animals with other mutations that have selective effects on cerebellar development, and by time course studies after nodulus lesions in developing and adult normal mice.