HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) All Versions of this Article: 95/3/1588    most recent 00318.2005v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (2) Citing Articles via Google Scholar Google Scholar Articles by Stahl, J. S. Articles by De Zeeuw, C. I. Search for Related Content PubMed PubMed Citation Articles by Stahl, J. S. Articles by De Zeeuw, C. I.

Eye Movements of the Murine P/Q Calcium Channel Mutant Tottering, and the Impact of Aging

¹Department of Neurology, Case Western Reserve University, Cleveland; ²Department of Neurology, Louis Stokes Cleveland Department of Veterans Affairs Medical Center, Cleveland, Ohio; and ³Department of Neuroscience, Erasmus Medical Center, Rotterdam, The Netherlands

Submitted 28 March 2005; accepted in final form 2 December 2005

	ABSTRACT

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Mice carrying mutations of the gene encoding the ion pore of the P/Q calcium channel (Cacna1a) are an instance in which cerebellar dysfunction may be attributable to altered electrophysiology and thus provide an opportunity to study how neuronal intrinsic properties dictate signal processing in the ocular motor system. P/Q channel mutations can engender multiple effects at the single neuron, circuit, and behavioral levels; correlating physiological and behavioral abnormalities in multiple allelic strains will ultimately facilitate determining which alterations of physiology are responsible for specific behavioral aberrations. We used videooculography to quantify ocular motor behavior in tottering mutants aged 3 mo to 2 yr and compared their performance to data previously obtained in the allelic mutant rocker and C57BL/6 controls. Tottering mutants shared numerous abnormalities with rocker, including upward deviation of the eyes at rest, increased vestibuloocular reflex (VOR) phase lead at low stimulus frequencies, reduced VOR gain at high stimulus frequencies, reduced gain of the horizontal and vertical optokinetic reflex, reduced time constants of the neural integrator, and reduced plasticity of the VOR as assessed in a cross-axis training paradigm. Unlike rocker, young tottering mutants exhibited normal peak velocities of nystagmus fast phases, arguing against a role for neuromuscular transmission defects in the attenuation of compensatory eye movements. Tottering also differed by exhibiting directional asymmetries of the gains of optokinetic reflexes. The data suggest at least four pathophysiological mechanisms (two congenital and two acquired) are required to explain the ocular motor deficits in the two Cacna1a mutant strains.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
The exact role the vestibulocerebellum plays in eye movement control remains unresolved. Many investigators have commented on the unusual anatomical, cytoarchitectural, and electrophysiological features of the cerebellum and reflected that the degree to which these features have been conserved through evolution indicates they are vital to the structure’s overall computational role. One of these striking features is the Purkinje cell’s interdependent calcium and potassium conductances, which enable even the most distal inputs to the enormous dendritic arbor to influence the axonal firing rate (De Schutter and Bower 1994; Jaeger et al. 1997; Llinas and Sugimori 1980), while at the same time providing a mechanism whereby the spread of synaptic potentials can be tightly restricted (Llinas and Moreno 1998). There is a consensus that these currents are critical to the Purkinje cell’s function. Nevertheless, the currents’ contribution, in computational terms, remains elusive (Yuste and Tank 1996). We have previously observed that a potential approach to the problem is to study mice carrying mutations predicted to alter Purkinje cell electrophysiology, and we chose to study strains bearing mutations of Cacna1a, the gene encoding the ionophore subunit of the P/Q-type calcium channel (Stahl 2002, 2004a). This channel is the primary determinant of calcium currents in the Purkinje cell dendrite (Llinas et al. 1989), participates in synaptic transmission at several points in the vestibulocerebellar circuitry, and mouse strains bearing Cacna1a mutations all suffer varying degrees of motor impairment customarily interpreted as cerebellar in origin (Ashcroft 2000; Zwingman et al. 2001).

The inherent problem in studying such mutants, however, is that the P/Q calcium channel is expressed at several points within regions of the brain participating in control of eye movement. As such, one cannot attribute any specific behavioral finding a priori to deranged Purkinje cell dendritic integration. Even if one assumes the dysfunction is localized to the cerebellar cortex (and there are reasons to accept this assumption as a working hypothesis; see DISCUSSION), it would remain unknown whether the dysfunction stems directly from altered P/Q calcium currents or from secondary changes induced by the abnormal currents. For instance, the Cacna1a mutation in the tottering strain results in an increased dependency on N-type, rather than P/Q-type, calcium channels at the synapses of parallel fibers on Purkinje cells (Zhou et al. 2003), and the shift was predicted to render these synapses more susceptible to the inhibitory effects of cortical interneurons. A relative sparing of inhibitory compared with excitatory neurotransmission (Caddick et al. 1999) and climbing fiber as opposed to parallel fiber synaptic efficacy (Matsushita et al. 2002) have also been reported in this strain. Such secondary effects would ultimately disrupt the normal balance of inputs to the Purkinje cell. Reduced P/Q channel conductivity may also trigger alterations in mechanisms of intracellular calcium homeostasis and thereby alter processes in which calcium acts as a second messenger, such as long-term depression or nitric oxide generation (Cicale et al. 2002; Dove et al. 2000; Murchison et al. 2002; Rhyu et al. 2003).

Because the Cacna1a mutation is expressed throughout the life of the animal, it could also potentially alter the development and maintenance of the ocular motor circuits. Detailed anatomical studies of the ocular motor circuitry in Cacna1a mutants have not been undertaken, although even if such studies were to confirm the gross integrity of these pathways, they still could not exclude the possibility that there are quantitative changes at the ultrastructural level (e.g., shape of dendritic arbors or density and distribution of synaptic contacts) that influence signal processing. An instance of such ultrastructural changes has, in fact, been described (Rhyu et al. 1999). Finally, it should be recognized that Cacna1a mutations can have diverse effects even at the level of the P/Q calcium channel itself. This point is demonstrated by studies of CACNA1A mutations associated with a human channelopathy disorder, familial hemiplegic migraine (FHM). When expressed in cultured granule cells, P/Q calcium channels incorporating FHM mutations exhibit decreased unitary calcium currents, but more hyperpolarized activation voltages, resulting in increased open-channel probability and thus increased single-channel calcium influx over a large voltage range (Tottene et al. 2002). At the same time, the mutation results in lower channel density, which reduces the total maximal P/Q calcium current. The multiplicity of effects on the P/Q calcium channel renders it more difficult to prove causal relationships between a specific biophysical aberration, alterations of neuronal signal processing, and a given behavioral abnormality.

One approach to unraveling these uncertainties involves studying each Cacna1a mutant strain over a broad range of ages. Abnormalities that appear and progress through life should originate in structural or electrophysiological properties that are similarly progressive. Conversely, progressive behavioral abnormalities are unlikely to stem directly and/or exclusively from a stationary change in channel biophysics. A related strategy involves studying several different Cacna1a mutants, with special attention to identifying behavioral abnormalities whose severity varies across strains. Provided that qualitatively similar behavioral abnormalities stem from the same pathophysiological mechanism, the severity of the behavioral abnormalities should covary with the severity of that causal biophysical, structural, or electrophysiological derangement. Conversely, biophysical, structural, or electrophysiological aberrations that do not covary with the severity of the behavioral abnormality are less likely to be directly responsible for that behavioral abnormality. Thus comparisons across allelic strains can be used to generate hypotheses regarding the causes of ocular motor abnormalities in Cacna1a mutants, and in turn, lead to insight into the connection between a particular biophysical, structural, or systems property and the normal function of the vestibulocerebellum.

To date, geneticists have identified seven murine Cacna1a mutants. Listed in rough ascending order of ataxia severity, the strains are: rocker, tottering, tottering-4J, tottering-5J, rolling-Nagoya, tottering-3J, and leaner (Zwingman et al. 2000, 2001). Although some of these strains exhibit prominent age-related cerebellar degeneration, histology in others (i.e., rocker, tottering, tottering-4J, and tottering-5J) is relatively normal (Zwingman et al. 2000, 2001), and thus behavioral abnormalities may stem from congenitally altered channel biophysics rather than from loss of neuronal elements. Rocker (rkr) and tottering (tg) are particularly convenient for these studies because of the topographical similarity of their mutations (single amino acid substitutions in homologous extracellular loops adjacent to the mouth of the ionophore) and the fact that the animals are robust, with normal life span and fecundity. Our previous study of rocker revealed ocular motor abnormalities that are consistent with interference with functions that have been attributed to the cerebellar flocculus of afoveate mammals (Stahl 2004a). Some of these abnormalities appeared to be congenital, such as subnormal gains of the vestibuloocular reflex (VOR) and visually augmented vestibuloocular reflex (VVOR), abnormally large VOR phase leads at low stimulus frequencies, deficient plasticity of VOR geometry, and slowing of fast phases of vestibular nystagmus. Other abnormalities appeared or clearly progressed with age, including subnormal gain of the optokinetic response (OKR) and static hyperdeviation of the eyes. In the current study we assessed the integrity of the same ocular motor functions in tottering, comparing this strain to both rocker and control animals of the reference inbred strain C57BL/6, on which background the rocker and tottering mutations have been maintained. In addition, we conducted analyses in all three strains of additional functions not assessed in the previous report, including the time constant of the brain stem neural integrator, axis of the ocular response to rapid pulses of head velocity ("thrusts"), and speed tuning of the vertical OKR. Preliminary reports have been published (Stahl 2004b; Stahl and James 2003, 2004, 2005).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Animals and animal preparation

Experimental use of mice was approved by the Institutional Animal Care and Use Committee at Case Western Reserve University and conformed to the National Institutes of Health guidelines for the use and care of vertebrate animals. Homozygous tottering mutants were obtained by crossing animals doubly heterozygous for the closely linked tottering and oligosyndactylism mutations (i.e., +Tg/Os+ x +Tg/Os+), resulting in only new double heterozygotes and tottering homogyzotes (TgTg/++), the ++/OsOs genotype being lethal in utero (Isaacs and Abbott 1992). The tottering homozygote progeny are recognizable by their ataxic gaits and normal (nonfused) digits. Colony founders were originally generated by crossing tottering homozygotes maintained on a C57BL/6 background with leaner/Os double heterozygotes (B6.Cg-Os +/+ Cacna1a^tg-la/J) obtained from The Jackson Laboratory (Bar Harbor, ME). Comparative data for rocker and C57BL/6 control animals were drawn from our earlier study of rocker (Stahl 2004a), augmented by new data from C57BL/6J animals purchased from The Jackson Laboratory and studied concurrently with the tottering mutants. Tottering animals were aged 109–808 days at the time of eye movement recording. Overall, data described herein were collected from 28 tottering, 14 rocker, and 41 control animals. As in the previous study (Stahl 2004a), animals were divided into three age groups: 2–8 mo, 8–14 mo, and >14 mo. These groupings correspond roughly to breeding adult, late-breeding adult, and elderly periods for the laboratory mouse. Tottering recordings were planned so that the average ages of the animals within each age group were separated by about 6 mo. Some animals were recorded repeatedly and contributed to the database for more than one group (usually two). All animals did not undergo all types of testing. Numbers of animals tested in each condition and age group, as well as the average age of the animals tested in a particular condition, are summarized in Table 1.

Eye movement recording and calibration

Eye movement recordings were obtained from restrained, head-fixed animals using video-tracking methods as previously introduced (Stahl et al. 2000) and subsequently refined (Stahl 2002, 2004a). We regularly ensured that the videooculography vertical corresponded to Earth vertical by imaging a machinist’s square or a miniature plumb bob arrangement mounted on the turntable, and adjusted the camera roll until the image of the reference fixture aligned perfectly with a vertical reference generated by the oculography system. This adjustment was particularly important for measuring the angle of the response to the thrust stimulus described below. Video images of either the left or right eye were processed by a commercial pupil tracker (ETL-200, ISCAN, Burlington MA), which extracted signals proportional to the horizontal and vertical (linear) positions of the pupil and reference corneal reflection (CR), as well as horizontal pupil diameter. Sampling rates were 240 Hz during measurements of nystagmus fast phases and 120 Hz for all other purposes. Note that the corresponding sampling rates for the control and rocker data derived from the previous study were 120 and 60 Hz, respectively. Analog outputs of the eye tracker were passed through four-pole Bessel low-pass filters (corner frequency 100 Hz), resampled at 200 Hz, and recorded on a PC-type computer. Other analog signals, including turntable position, optokinetic drum velocity, and planetarium velocity were similarly filtered, sampled, and stored.

The positions of pupil and CR were converted off-line to angular position as previously described (Stahl 2002, 2004a; Stahl et al. 2000). Briefly, we recorded the pupil and CR positions as the video camera was rotated over a 20° arc about an Earth-vertical axis passing approximately through the recorded eye. The positions obtained at the extremes of the camera rotation were used to calculate R_P, the linear distance between the center of the pupil and the center of corneal curvature in the Earth-horizontal plane passing through the pupil. Note that this distance decreases as the pupil assumes greater vertical elevations. We also determined the elevation-independent great circle radius R_P0, the distance between the center of the pupil and the centers of both vertical and horizontal corneal curvatures. Both R_P and R_P0 were determined at various pupil diameters and linear regression performed to generate formulae relating these radii to pupil size. Horizontal or vertical eye-in-head angles (E_H, E_V) could then be determined in subsequent records by trigonometric formulae taking into account the linear pupil and CR positions and either R_P or R_P0, adjusted for pupil size. The equation with R_P0 requires both the horizontal and vertical pupil and CR positions, and thus generates eye angles that are noisier than those generated by the R_P equation, which is based on the horizontal channels alone. Thus the R_P0 equation was reserved for stimuli that generated appreciable vertical movements of the eye (i.e., the responses to the cross-axis and vertical OKR stimuli), which take the pupil away from the horizontal plane it occupied during calibration. The simpler formula based on R_P and the horizontal pupil and CR positions was used for all other data. During assessments of absolute eye elevation at rest, we could assume the pupil lay at the same position that it occupied during the measurement of R_P. Thus the elevation above the ocular equator could be calculated using R_P and the vertical linear distance between the pupil and CR centers (Y) according to the equation E_V = arctan(Y/R_P). During this measurement the reference IR emitter was moved from its customary position directly above the camera lens to the side of the camera lens. In this position there was no need to correct for the vertical separation of the reference emitter and the camera optical axis, which requires assumptions about the radius of corneal curvature (Stahl 2004a).

Stimulus apparatus and conditions

All stimulus instruments were identical to those previously detailed (Stahl 2004a). Briefly, vestibular and optokinetic stimuli about an Earth-vertical axis were generated by a servo-controlled turntable and optokinetic drum, respectively. The calibration procedure guarantees that the rotational axis of the turntable runs through the center of the recorded eye, and the rotational axis of the drum is approximately aligned with that of the table. During cross-axis adaptation and assessment of vertical OKR (see following text), optokinetic stimuli about a rostrocaudal (roll) axis were generated by a projection planetarium (Leonard et al. 1988) mounted over the animal’s head and projecting onto a featureless gray fabric background.

The majority of stimulus conditions used in this study were detailed previously (Stahl 2004a). The determination of average absolute vertical position was based on approximately 60 s of recording with the animal held motionless in the light. Horizontal nystagmus fast phases for analysis of peak velocity–amplitude relationships were generated by slowly rotating the animal under manual control in the light. VOR and VVOR were tested by rotating the animal in darkness and light, respectively, at 0.1 Hz, ±10° amplitude; 0.2 Hz, 10°; 0.4 Hz, 10°; 0.8 Hz, approximately 4.6°; and 1.6 Hz, approximately 3.9°. Corresponding velocity amplitudes were ±6, 13, 25, 23, and 39°/s. Both horizontal and vertical (roll axis) OKR were assessed during constant-velocity rotations at ±2.5, 5, 10, 20, and 40°/s. Constant-velocity periods alternated direction, lasted 4 s each, and were separated by 3.5-s periods of darkness during which the drum/planetarium reversed direction and reaccelerated to the next test speed. The ability of the animal to alter the direction of its compensatory eye movements (cross-axis adaptation) was tested by oscillating the animal in the light at 0.4 Hz, 10° amplitude while simultaneously rotating the planetarium about an Earth-fixed rostrocaudal (roll) axis at the same frequency and amplitude. The phase of the planetarium rotation was adjusted with respect to the turntable so as to promote downward eye movements in association with nasally directed horizontal compensatory eye movements. VOR direction was assessed (in darkness) before adaptation, and then for a brief period every 10 min during a 50-min adaptation period. Samples of the response in the presence of the adaptation stimulus were also obtained at 10-min intervals.

Two types of measurements were not included in the previous study and thus are described here in more detail. The alignment of VOR response and stimulation axes was assessed in the light using 15°, steplike rotations of the turntable with peak velocities of approximately 110°/s and accelerations of 2200°/s² (a so-called thrust stimulus; Walker and Zee 1999). The speed was sufficient to guarantee that the ocular response was largely attributed to the VOR while not being so fast as to agitate the animals. In some control animals it was necessary to reduce the stimulus amplitude by 25% because at the standard stimulus amplitude the response was consistently interrupted by a fast phase. This study also assessed the time constant of the neural integrator. Animals were rotated slowly in the light to generate eccentric eye positions. Once an eccentric eye position had been achieved, rotation was halted for 2 s to allow the semicircular canals to partially reequilibrate. The lights were then extinguished and the subsequent centripetal drift of eye position was recorded.

Data analysis

Analysis procedures for fast phases of vestibular nystagmus, VOR, OKR, and cross-axis adaptation were identical to those previously detailed (Stahl 2004a). Briefly, fast phases of vestibular nystagmus (henceforth loosely termed "fast phases") were quantified by linear regression of peak eye velocity versus fast phase amplitude, with the regression forced to pass through the origin. Horizontal eye velocities used in this analysis were calculated off-line by numerically differentiating the eye position signal after smoothing it by convolution with a Blackman window whose cutoff frequency was either 40 or 80 Hz. The lower value was used whenever data were to be compared with data collected in the previous study (Stahl 2004a), in which the lower video sampling rate also mandated a lower smoothing cutoff frequency. VOR and VVOR data were processed by Fourier analysis to extract gain and phase values with respect to head velocity. Gain versus frequency and phase versus frequency (Bode) plots were compiled for each recording session, curves from two or more sessions were averaged to generate a single Bode plot for each animal, and these single-animal curves were in turn averaged to generate curves for each genotype and age group. Horizontal and vertical OKR data were processed to extract gain values, where gain at each drum velocity was defined as the average slow-phase eye velocity divided by the drum/planetarium velocity. Plots of gain versus stimulus frequency (speed-tuning curves) were compiled for each recording session. As with the VOR/VVOR data, curves from at least two sessions were averaged to yield single-animal curves, and multiple single-animal curves were averaged to generate a curve for each genotype and age group. In descriptions of horizontal and vertical OKR results, positive drum rotations signify, respectively, temporal-nasal and upward rotations with respect to the recorded eye. Cross-axis adaptation data were processed to determine horizontal VOR gain, as well as vertical VOR gain defined as the amplitude of the vertical velocity divided by the amplitude of the horizontal stimulus. The progression of the horizontal and vertical gain changes through the adaptation period was compiled for each animal/age, and average curves from multiple animals were in turn averaged to generate a single curve for each age group.

In addition to the full-cycle Fourier analyses of VOR and VVOR gain described above, data were reanalyzed to determine whether there were differences in the gains of nasally and temporally directed eye movements. The analysis of each data sample (consisting of several cycles of stimulus and response at a single stimulus frequency) proceeded as follows: First, the multiple cycles were pointwise averaged to generate a single averaged cycle of stimulus and response. Second, Fourier analysis was used to fit the eye signal with a sinusoid. Third, the head and eye signals were each separated into two hemicycles, based on where the fitted eye velocities were positive or negative. Fourth, multiple linear regression was performed on the hemicycles of the eye and head movements, according to the regression equation E(t) [or H(t)] = A sin 2ft + B cos 2ft + C, where A, B, and C are regression coefficients, f is stimulus frequency, and t is time. In the case of the nasally directed eye movement (and its associated oppositely directed hemicycle of head movement), t was restricted to those values where the fitted full-cycle eye velocity was positive. Conversely, for the temporally directed eye movement and the associated head movement, t was restricted to values for which the fitted full-cycle eye velocity was negative. Fifth, amplitudes of the eye (|E|) and head (|H|) for each hemicycle were determined from (A² + B²), and gain was then quantified as |E|/|H|.

Responses to thrust stimuli were analyzed by determining, for each rotation, the direction of the ocular response in the plane of the video image, which was approximated as the arctangent of the ratio of the horizontal and vertical eye position changes from the beginning to the end of the head impulse. Response gain was calculated as the ratio of the radial change in eye position to the change in horizontal head position. The angles and gains from 10 to 30 impulses in each direction were then averaged to yield a pair of angles/gains for each animal/age.

Time constants (T) of the neural integrator were determined by fitting each centripetal drift with the exponential E(t) = A exp(–t/T) + C, where t is time, E(t) is eye position, and A and C are fitting constants. We discarded epochs with duration <2.5 s, A <3°, or C >15° from the average equilibrium position, or if the generalized correlation coefficient (r²) was <0.80. We calculated median time constants in each animal separately for abducting and adducting centripetal drifts and averaged across animals. Note that an "abducting centripetal drift" is a drift made from an initial adducted eccentric position.

All results are reported as means ± SD. Significance of interage group and intergenotype comparisons was tested as described in RESULTS. In general, curves were compared using repeated-measures ANOVA, whereas point measurements were compared using two-tailed t-test. Significance of differences in x-intercepts was assessed by t-test, making use of the intercept SE generated in the linear regression. Significance of regression analyses was tested using the t-test for r². Tests of regression significance were not corrected for multiple comparisons. Given the large number of parameters we desired to explore and practical limitations in the number of animals that could be tested, such correction would have generated an excessive risk of type II statistical error (the incorrect conclusion that a correlation is absent when, in fact, the two variables are related), thereby defeating the exploratory goals of the study. As usual, the reduction in the risk of type II statistical error is accomplished at the expense of an increased risk of the converse type I error (false conclusion that two variables are related), and thus the inferences regarding mechanisms drawn from the correlation analyses should be considered provisional in nature.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Static elevation

Figure 1 plots static pupil elevation in the light as a function of animal age. Regression lines have been superimposed. All three genotypes exhibited wide variations in elevation within any one age group, and there was extensive overlap between genotypes. Nevertheless, in rocker and tottering, elevations tended to be greater than that in controls. For tottering, the difference from controls was significant at P < 0.001 or better (two-tailed t-test) in all age groups, and the elevation trended upward with age (elevation vs. age r² = 0.08, P = 0.114, t-test for r²). Elevations for tottering and controls averaged 25.6 ± 6.1° versus 17.8 ± 4.8°, 27.0 ± 3.2° versus 19.5 ± 4.8°, and 28.4 ± 5.1° versus 17.5 ± 3.2° in the young, middle, and old age groups, respectively. As already reported (Stahl 2004a), rocker’s elevation was initially normal, but increased significantly with age (r² = 0.33, P = 0.002), the group average becoming significantly different from controls (P = 0.0014) in the oldest age group. Human P/Q calcium channelopathies and other cerebellar disorders are frequently associated with downbeat nystagmus. However, as was the case in rocker (Stahl 2004a), tottering animals exhibited largely static (although elevated) vertical positions in light and darkness. In occasional animals downbeats followed by slow upward drifts did occur, particularly in the light. When present they were rare and irregular, rarely exceeding five beats in a 40-s record.

View larger version (25K): [in this window] [in a new window]   FIG. 1. Average vertical position of the eye in light plotted as a function of animal age. Each symbol represents a single animal. Dashed vertical lines represent the boundaries between the arbitrarily defined young, middle, and old age groups. Regression fits for control (solid line), tottering (dash–dot line), and rocker animals (dashed line) are superimposed. Tottering exhibited abnormal elevation at all ages.

Figure 2 shows Bode plots for VOR and VVOR of controls and mutants in each age group. Tottering exhibited a similar pattern of gain abnormalities at all ages. The most prominent abnormality was the failure of VOR gain to increase with increasing stimulus frequency as rapidly as in controls, resulting in a large gain disparity at 1.6 Hz, although gain was subnormal at the low stimulus frequencies as well. VVOR gain was deficient at all frequencies. At the lowest frequencies, where VOR gain is closest to normal, the VVOR attenuation is attributed to a reduced optokinetic (visual) response, an attribution supported by the narrowing of the gap between the VVOR and VOR gain curves in tottering compared with controls. At the highest stimulus frequencies, the visual contribution to gain is normally small (note the small gap between VOR and VVOR gain curves in control animals), and thus tottering’s deficiency in VVOR gain at high stimulus frequencies is largely attributable to the subnormal VOR gain. The difference between gain values (both VOR and VVOR) of tottering and controls was significant at all frequencies and ages, usually at P 0.0001. The lowest degrees of significance (highest P values) were obtained for 0.1 and 0.2 Hz in the dark in the youngest age group, in which the P values were 0.0046 and 0.0017, respectively. In contrast to tottering, rocker exhibited VOR gain abnormalities only at the highest stimulus frequencies. Rocker’s augmentation of gain in the presence of vision was normal in the youngest age group and only minimally impaired with aging, and thus VVOR gain curves in rocker exhibited prominent declines with increasing stimulus frequency, merging with the tottering gain curves at 1.6 Hz.

View larger version (30K): [in this window] [in a new window]   FIG. 2. Bode (phase and gain vs. stimulus frequency) plots for vestibuloocular reflex (VOR, filled symbols) and visually augmented vestibuloocular reflex (VVOR, open symbols) for mutants and controls. Data for young, middle, and old age groups are plotted separately. Data points are staggered horizontally and error bars (1 SD in length) are plotted unidirectionally for graphic clarity. Tottering mutants exhibited subnormal VOR and VVOR gains at all stimulus frequencies, whereas in rocker the attenuation was limited to higher stimulus frequencies. Both mutants strains also exhibited abnormally large VOR phase leads at low stimulus frequencies.

To generate insight into whether the abnormalities of VOR/VVOR dynamics described above were congenital or acquired, we regressed the most abnormal parameters versus animal age. (Of note, the conclusion that any feature is "congenital" is a provisional one; see DISCUSSION.) Scatterplots of VOR and VVOR gains versus age are shown in Fig. 3. In tottering, VOR gain at 1.6 Hz declined slightly with age [linear regression slope (m) = –1.9 x 10^–4 day^–1, r² = 0.12, P = 0.049], but most of the subnormality may still be congenital because the regression lines of mutants and controls were so nearly parallel and so widely separated that they did not converge when extrapolated backward to birth (y-intercepts differ significantly, P < 0.0001). Likewise, the gain deficit at 0.1 Hz may be congenital, as the regression curve failed to reach significance (P = 0.32) and roughly paralleled the control curve (y-intercepts differ, P = 0.0005). The VVOR gain of tottering at 0.1 and 1.6 Hz declined gradually with age and, unlike the case for VOR gains, the correlation coefficients were statistically significant (at 0.1 Hz: m = –3.5 x 10^–4 day^–1, r² = 0.36, P < 0.001; at 1.6 Hz: m = –2.4 x 10^–4 day^–1, r² = 0.17, P = 0.015). Again, the y-intercepts for tottering and control VVOR gains differed significantly for 0.1 Hz (P = 0.0004) and 1.6 Hz (P < 0.00001). However, the regression lines for tottering and control gains at 0.1 Hz did converge to within 0.15 at the y-intercept, whereas the regression lines for 1.6 Hz were closer to parallel, and were still separated by a gain of 0.27 at the y-intercept. These results suggest that the VVOR gain deficit exhibits both acquired and congenital components at both frequencies, but the acquired component dominates at low frequencies and the congenital component is more pronounced at high frequencies. These differences in the VVOR gain versus age relationships at different stimulus frequencies are consistent with the differing visual and vestibular contributions to the response at each frequency, the magnitude of the mutation-related VOR gain deficits, and the age-related decline in optokinetic performance described below. Specifically, at 1.6 Hz there is a large initial VOR gain deficit that increases only slightly with age, and a minimal optokinetic contribution to VVOR gain, and thus despite the tendency of OKR to diminish through life, the VVOR gain deficit at 1.6 Hz is initially large and progresses minimally. Conversely, at 0.1 Hz the VOR gain deficit is small and largely age independent, whereas the optokinetic contribution to the VVOR is large. Thus the VVOR gain deficit at 0.1 Hz is initially small, but increases with age in step with the OKR. The other consistent difference between the dynamics of tottering and control animals was the mutant’s larger VOR phase lead at the 0.1-Hz stimulus frequency. Both tottering and control exhibited trends toward increasing 0.1-Hz phase lead with age, but the correlation coefficients were low and statistically nonsignificant (r² = 0.01, P = 0.58 and r² = 0.06, P = 0.11), consistent with the larger phase lead in tottering reflecting a congenital defect.

View larger version (26K): [in this window] [in a new window]   FIG. 3. Scatterplots of VOR and VVOR gains vs. animal age. Each symbol represents a single animal. Regression fits for control (solid line), tottering (dash–dot line), and rocker animals (dashed line) are superimposed.

The increased phase lead at low stimulus frequencies could be explained by an abnormally short time constant of the brain stem neural integrator, which synthesizes the tonic command signals required to compensate for the elasticity of the ocular motor plant (Skavenski and Robinson 1973), and as such is responsible for much of the required phase lag between the modulation of primary vestibular afferents and motoneurons. Neural integrator time constants were determined in control, tottering, and rocker animals by fitting single-exponential decays to the centripetal drifts that occurred when the lights were extinguished after bringing the eye to an eccentric position. Figure 4 shows the time constants for abducting and adducting centripetal drifts, plotted versus animal age. Linear regression on the data for all three strains suggested a weak tendency for time constants to increase with age, but the interanimal variability was wide, and the only regression that was significant at P < 0.05 or better was that of the abducting drifts in rocker (r² = 0.19, P = 0.024). Because the regression lines of mutants approximately paralleled controls, the deficient integrator time constant may be provisionally considered a congenital abnormality. Averaging animals of all ages, adducting and abducting time constants were 5.2 ± 3.1 and 4.5 ± 2.5 s for controls, 3.2 ± 1.3 and 2.0 ± 1.1 s for tottering, and 3.9 ± 1.8 and 2.6 ± 1.2 s for rocker. Tottering differed from controls at P < 0.001 or better, and rocker differed from controls at P = 0.001 for abducting and P = 0.065 for adducting drifts. Assuming that the neural integrator can be modeled as a one-pole lag element whose time constant is the average of our measured abducting and adducting values, the shorter time constants of tottering and rocker would translate to an increase in phase lead (decrease in lag imparted by the integrator) with respect to controls of 13.3° for tottering and 7.9° for rocker, at a stimulus frequency of 0.1 Hz. The actual differences obtained from the VOR data of all animals were 23.0 and 18.2°, indicating that our measures of integrator dysfunction can explain approximately half of the increased phase lead of the mutant strains.

View larger version (20K): [in this window] [in a new window]   FIG. 4. Time constant of the neural integrator plotted as a function of animal age. Each symbol represents a single animal. Regression fits for control (solid line), tottering (dash–dot line), and rocker animals (dashed line) are superimposed. "Abduct" and "adduct" refer to the direction of the drift (opposite the direction of initial eye displacement). Both mutant strains exhibited reduced time constants (poorer gaze stability in darkness) in comparison to controls, particularly for abducting drifts.

Figure 5 depicts horizontal OKR speed-tuning curves for each of the three age groups. Tottering exhibited mild reductions in horizontal OKR gain, and the tottering and control tuning curves differed significantly by repeated-measures ANOVA for all three age groups (P = 0.011, P < 0.001, P < 0.001, respectively). The deficit was most pronounced at the lowest drum speeds and for the temporal-nasal direction of drum rotation. Rocker OKR gains tended to be intermediate between those of controls and tottering and, in fact, the rocker curve became significantly different from control only in the oldest age group (Stahl 2004a). Inspection of the speed-tuning curves in Fig. 5 suggests the possibility of a combination of congenital and acquired deficits in OKR gain in tottering. We investigated this possibility by regressing OKR gains versus animal age, concentrating on the responses to 10°/s rotation, the speed at which the disparity between tottering and control gains was particularly pronounced. Figure 6 shows the gain versus age plots for 10°/s rotation in the nasal-temporal and temporal-nasal directions. Regression fits for control animals were flat, indicating that OKR gain is normally stable over a broad range of animal age. The y-intercepts of the tottering curves were significantly lower than those of control for both nasal-temporal (P = 0.03) and temporal-nasal (P = 0.0018) directions, consistent with a congenital gain deficit. The intercepts were also significantly lower for temporal-nasal rotations at 2.5°/s (P = 0.0012) and 5°/s (P = 0.0012) (plots not shown). The gain versus age plots also suggested an acquired component to the gain deficit because gains trended downward for both directions of rotation at 2.5, 5, and 10°/s. None of these trends reached statistical significance. For 10°/s, the correlation coefficients and associated P values were r² = 0.09 (P = 0.078) for nasal-temporal rotation and r² = 0.10 (P = 0.065) for temporal-nasal rotation. Regression significance would certainly have been depressed by the large interanimal variability. The possibility that the downward trend is meaningful despite the statistical nonsignificance is supported by two observations. First, OKR gains clearly declined with age in rocker (Stahl 2004a). Assuming that allelic mutant strains share abnormal features, the clear age-related decline in rocker makes it more likely that the decline in tottering, although shallow, still reflects a true relationship between gain and age. Second, an acquired decline in OKR gain of Cacna1a mutants would explain the observation that the low-frequency VVOR gains of tottering and rocker diverge from control values with increasing animal age (see above).

View larger version (18K): [in this window] [in a new window]   FIG. 5. Plots of horizontal optokinetic response (OKR) gain vs. stimulus velocity for mutants and controls. Data for young, middle, and old age groups are plotted separately. Rocker data are reproduced from previous study (Stahl 2004a) and error bars are suppressed for graphic clarity. Other error bars are 1 SD and plotted unidirectionally for clarity. Dashed line marks 0°/s. N-T: nasal-temporal; T-N: temporal-nasal. With age, tottering gain became progressively subnormal and the divergence from the symmetry of control gains became more pronounced.

View larger version (25K): [in this window] [in a new window]   FIG. 6. Gain of the horizontal OKR for 10°/s rotations of the optokinetic drum, plotted vs. animal age. Results are plotted separately for nasally and temporally directed drum rotations with respect to the recorded eye (indicated on the plots as "T-N" and "N-T"). Each symbol represents a single animal. Regression fits for control (solid line), tottering (dash–dot line), and rocker animals (dashed line) are superimposed. Both mutant strains diverged from controls with increasing age.

View larger version (22K): [in this window] [in a new window]   FIG. 7. Average hemicycle gains for tottering (8- to 14-mo and >14-mo age groups, only) and control animals, plotted as functions of stimulus frequency. A slight increase in the gain of nasally directed ("T-N") compared with temporally directed ("N-T") eye movements is present in both VOR and VVOR in mutants and controls. Note that the gain asymmetry in tottering is opposite to that seen in its optokinetic speed-tuning curves (Fig. 4). Error bars are 1 SD and plotted unidirectionally for clarity.

The most straightforward interpretation of the horizontal OKR speed-tuning curve is that the sensitivity to optokinetic stimulation is reduced, and more so for the temporal-nasal direction of drum rotation. However, a complication is introduced by the fact that the optokinetic stimulation delivered to each eye was not perfectly matched; the recorded eye’s view of the optokinetic drum was mildly obstructed by the presence of animal-fixed apparatus (the camera and infrared illuminators) in the foreground. If the OKR in mutants were driven to an extent greater than that in controls by the eye receiving temporal-nasal stimulation, then the asymmetry in optokinetic stimulation would reduce mutant OKR gain during temporal-nasal rotation. We explored this possibility by manipulating the optokinetic stimulation to each eye in a subgroup of five middle-aged mutants and four young controls. For the purposes of the following description, "obstruction" refers to a partial obstruction of the view of the optokinetic drum by animal-fixed apparatus. In the case of the recorded eye, the obstruction was generated by the camera and illuminators. In the case of the opposite eye, the obstruction was generated by a dummy copy of the recording apparatus that was positioned to approximate the visual effects of the actual recording apparatus. The "mask" was a gray cloth drape that could be positioned on either side of the turntable to entirely block the view of the drum on one side. In the case of the recorded eye, the mask was perforce placed outside the recording apparatus. Figure 8 shows average speed-tuning curves for controls and tottering. As in the tuning curves generated from the larger sample (see Fig. 5), the mutants evinced attenuated and asymmetrical gains. In controls, adding the obstruction to the contralateral, nonrecorded eye (i.e., balancing the optokinetic stimuli) generated a mild reduction in gain for both nasally and temporally directed eye movements. Masking the nonrecorded eye moderately reduced gains in both directions, slightly more so for temporally directed eye movements. This result is consistent with the possibility that the ability of the contralateral eye to drive the ipsilateral eye is greater for temporal-nasal than nasal-temporal stimulation (re the contralateral eye).

View larger version (23K): [in this window] [in a new window]   FIG. 8. Effect of various asymmetrical visual stimulation conditions on horizontal OKR speed tuning in 4 young control animals (left) and 5 middle-aged tottering mutants (right). All eye movements were recorded from left eye. OS, left eye; OD, right eye. Clear: indicated eye has completely unobstructed view of optokinetic drum. Obstruct: view partially obstructed by actual or mock oculography apparatus. Mask: view fully obstructed by a cloth drape. "OS obstruct, OD clear" condition (or its converse, when recording from the right eye) corresponds to the usual recording arrangement in this study.

OKR speed tuning was also assessed for optokinetic stimulation about the roll axis, as a response to roll visual motion is a prerequisite for the ability to undergo cross-axis adaptation tested below. (It should be noted that the two-dimensional [2D] oculography provides only an approximation of the actual response because the stimulus and response axes are rotated approximately 45° to the optical axis of the camera. In contrast, the horizontal gain measurements are more accurate because the stimulus/response axis and the camera’s optical axis are perpendicular.) The roll (vertical) OKR speed-tuning curves are shown in Fig. 9. Control data were collected from only the youngest and oldest age groups, but the tuning curves were similar at those two ages (P = 0.39, repeated-measures ANOVA), indicating that the vertical optokinetic response, like the horizontal optokinetic response, is normally stable through life. Tottering of the youngest age group exhibited mildly, but significantly, reduced gains compared with those in controls (repeated-measures ANOVA, P = 0.006). With increasing age, the response declined, particularly during downward optokinetic stimulation. Regressing gain versus age for 10°/s rotations yielded correlation coefficients and associated P values of r² = 0.24 (P = 0.0054) for downward rotation and r² = 0.33 (P = 0.0007) for upward rotation, consistent with the acquired (or at least, progressive) nature of the vertical gain deficit in this strain. Gain versus age plots were also negatively sloped for both directions at 2.5°/s (down: r² = 0.09, P = 0.10; up: r² = 0.15, P = 0.03) and 5°/s (down: r² = 0.32, P = 0.001; up: r² = 0.41, P = 0.0001). Average vertical OKR gains for young rockers varied around the curve for the control animals (P = 0.74, repeated-measures ANOVA), and in the oldest age group the rocker and control curves overlapped extensively (ANOVA, P = 0.68). Because of the decline in vertical OKR in tottering, analysis of the cross-axis data from this strain was restricted to the youngest age group (see following text). The relative stability of the vertical OKR in rocker and controls supports the decision to pool animals of all ages in the analysis of the cross-axis data in this and the previous study (Stahl 2004a). Furthermore, the mild degree of reduction in rocker’s vertical OKR argues against attributing its lifelong deficiency in cross-axis adaptation (Stahl 2004a) to an inability to respond to the adapting stimulus.

View larger version (16K): [in this window] [in a new window]   FIG. 9. Plots of vertical OKR gain vs. stimulus velocity for mutants and controls. Data for young, middle, and old age groups are plotted separately. Error bars are 1 SD and plotted unidirectionally. Dashed line marks 0°/s. Rocker error bars in the youngest and oldest age groups are suppressed for graphic clarity. Tottering vertical OKR gains declined with age, particularly for downward rotation of the planetarium with respect to the recorded eye.

Ideally, the axis of a compensatory eye movement should align perfectly with the axis of the vestibular or visual stimulus. However, VOR stimulus–response misalignments are exhibited by humans with cerebellar disorders and form the basis of a bedside test in which the clinician observes aberrant vertical eye movements elicited by small-amplitude, brisk horizontal head rotations (so-called thrusts) (Walker and Zee 1999). Mice were tested using analogous brisk, low-amplitude rotations in the light, and the 2D response angle and gain were determined from the changes in eye position. The effects of age on angle and gain were assessed by linear regression. There was no significant effect of age on either parameter for rocker and control (regression P 0.23), or for the angle of tottering’s abducting eye movements. The correlation was significant for the angle of tottering’s adducting eye movements, but the slope was shallow (slope = –0.006° day^–1, r² = 0.11, P = 0.049). More notable was tottering’s significant decline in gain with age for both abducting (slope = –2.2 x 10^–4 day^–1, r² = 0.15, P = 0.021) and adducting movements (slope = –2.5 x 10^–4, r² = 0.20, P < 0.001), which accords with this strain’s slight but significant decline in VOR gain at 1.6 Hz described above. Because the effects of aging on angle were absent or slight, we pooled animals of all ages and plotted the angles and gains in polar format in Fig. 10. Abducting eye movements are plotted to the left and adducting movements to the right. Normal animals exhibited a moderate misalignment, such that, on average, the eye tended to move downward during adduction and upward during abduction. The magnitude of the deviation from the ideal trajectory was small: 2.4 ± 2.6 and 3.4 ± 2.4° for adducting and abducting movements, respectively. The error is unlikely to be related to a misalignment of the camera roll axis because this adjustment was carefully verified (see METHODS). Moreover, we recorded variously from the right or left eye; if the camera axis had been tilted, the apparent VOR misalignments would have been opposite for the two eyes. Rocker and tottering mutants tended to exhibit lower-amplitude responses, in keeping with their lower high-frequency VOR gains described above. Rocker mutants whose gains most closely approximated control animals exhibited misalignments that, likewise, approached those of the controls. Decreasing gain was associated with an increasing degree of misalignment. The averages of the tottering responses, in contrast, were centered on the stimulus axes, and this alignment appeared to be preserved irrespective of gain. Response angles for rocker differed from control for both abduction (P = 0.009, one-way ANOVA with genotype as factor) and adduction (P = 0.001). Tottering differed from control for abduction (P = 0.001) but not adduction (P = 0.438).

View larger version (24K): [in this window] [in a new window]   FIG. 10. Polar plot summarizing results of thrust experiment, which were assessed for concordance of VOR response and stimulus axes. Gain of eye movement is plotted vs. angle of movement, as projected onto the plane of the video image. Each plotting symbol represents data from one animal. Small degrees of misalignment were demonstrated in control and rocker, but not tottering, strains.

Figure 11 displays the progress of gain changes for rotation in the light (response to the adaptation stimulus) and dark during the course of the 50-min cross-axis adaptation period. Control and rocker curves, reproduced from the previous study, were created by pooling animals of all ages (Stahl 2004a). In contrast, the tottering curves are based exclusively on data drawn from the 12 animals in the youngest age group because older animals developed a considerable reduction in response to the roll optokinetic stimulus, increasing the possibility that deficiencies in cross-axis adaptation reflect an inability to respond to the adapting stimulus. Despite young tottering’s ability to generate a response to the roll optokinetic stimulus (see above), there was no appreciable vertical response at any time during the cross-axis paradigm. The result contrasted markedly with that of rocker, which, as a group, exhibited a detectable vertical response at first exposure to the adapting stimulus, a mild (although subnormal) increase in that light response over the course of the adaptation period, and a slight but detectable increase in vertical motion in darkness by the end of the experiment. Rocker’s deficient but detectable vertical responses were enhanced by adjusting for its decline in horizontal VOR gain across the adaptation period (Stahl 2004a). In contrast, Fig. 11 demonstrates that tottering’s horizontal responses in light and dark were quite stable and thus there was no reason to speculate that a developing vertical response was being masked by habituation of the entire vestibular system. Correcting vertical gain in the style of the previous study (plots not shown) succeeded only in emphasizing the contrast between rocker’s minimal preservation and tottering’s absolute lack of cross-axis adaptation.

View larger version (20K): [in this window] [in a new window]   FIG. 11. Alteration of horizontal and vertical gains in light (top) and darkness (bottom) over the course of the 50-min cross-axis adaptation experiment. Tottering data restricted to the 2- to 8-mo age group, whereas control and rocker data pools animals of all ages. Vertical and horizontal gains are plotted as filled and open symbols, respectively. "Light" gains are obtained in the presence of the adapting optokinetic stimulus. All error bars are 1 SD and are plotted unidirectionally for graphic clarity. Tottering failed to develop cross-coupled vertical movements in darkness, whereas rocker demonstrated some ability to augment its vertical movements to the adapting stimulus.

Fast phase dynamics

As was the case for rocker and C57BL/6 controls (Stahl 2004a), tottering exhibited a strong linear relationship between fast phase peak velocity and amplitude, and abducting fast phases tended to be slower than adducting fast phases. Averaging tottering animals of all ages, these abduction and adduction velocity–amplitude slopes were respectively 22.2 ± 1.2 and 25.3 ± 1.4 s^–1, and the difference was statistically significant (paired t-test, n = 31, P < 0.0001). Figure 12 plots the slopes for mutants and controls versus age for the abducting (top) and adducting (bottom) directions. For all genotypes, velocity–amplitude slopes for adducting fast phases were stable through life (r² 0.04, P 0.17). For abducting fast phases, there was a moderate decline in both tottering (–0.0024 s^–1 day^–1, r² = 0.18, P = 0.016) and rocker (–0.0036 s^–1 day^–1, r² = 0.12, P = 0.089), which contrasted with the slight increase in slopes for controls (+0.0025 s^–1 day^–1, r² = 0.08, P = 0.035). Considering only the youngest age group, velocity–amplitude slopes were very similar for tottering versus controls for both abducting (22.9 ± 0.89 vs. 23.5 ± 1.5, P = 0.18) and adducting (25.8 ± 0.97 vs. 26.9 ± 1.2, P = 0.012) fast phases. Because of the divergent slope versus age relationships, tottering fast phases of both directions became considerably slower than controls in the middle and older age groups (all comparisons, P 0.001). In contrast, rocker fast phases of both directions were significantly slower than controls in all age groups (P 0.001) and slower than tottering (P 0.01) in all age groups, except for the case of adduction at 8–14 mo, where significance reached only P = 0.075.

View larger version (25K): [in this window] [in a new window]   FIG. 12. Slopes of velocity–amplitude relationships for nystagmus fast phases plotted vs. animal age. Abducting and adducting fast phases are plotted separately. Each symbol represents data from one animal. Regression fits for control (solid line), tottering (dash–dot line), and rocker animals (dashed line) are superimposed. Slopes were initially normal in young tottering, but moderately declined with age. Rocker fast phases were slowed at all ages.

Relationship of performance measures

For many of the ocular motor indices quantified above, the degree of abnormality varied across animals of one mutant strain. If these variations originate from a common source (such as a variable degree in the reduction of dendritic calcium currents), then one might expect the variations in behavior to be correlated. For instance, an animal with a particularly depressed OKR gain might also have a more depressed VOR gain. This prediction was assessed in the previous study of rocker and control animals (Stahl 2004a), and linkages were demonstrated between static elevation and other abnormalities, and between VOR phase lead at 0.2 Hz and VOR gain at 1.6 Hz. Similar pairwise linear regressions were conducted between the most striking (but also variable) abnormalities of tottering’s ocular motor behavior (i.e., static elevation), VOR gain at 1.6 Hz, VOR phase at 0.1 Hz, average of abducting and adducting neural integrator time constants, absolute nasal-temporal ("N-T") OKR gains at 10°/s, ratio of T-N and N-T OKR gains at 10°/s, absolute upward OKR gains at 10°/s, and ratio of downward:upward OKR gains at 10°/s. Note that in the case of the OKR data, it was necessary to test correlations only with the T-N:N-T ratio and one of the absolute gains because testing correlations with the ratio and both absolute gains would be redundant. Based on our analysis of the OKR asymmetries above, the N-T absolute gain reflects the best performance of tottering’s optokinetic system, whereas the T-N:N-T ratio captures the asymmetry. For similar reasons, we assessed only the upward gain of the vertical OKR and its downward:upward ratio. It should also be noted that, although the complete lack of cross-axis plasticity was a strong abnormality, it was also invariant, and thus inappropriate for an investigation of the covariation of abnormal features. In the discussion below, correlation coefficients (r²) are signed according to the sign of the regression slope to facilitate recognizing the sense of the relationships. As discussed in METHODS, significance tests were not corrected for multiple comparisons, so the results must be considered provisional in nature.

In control animals, only two of the pairwise correlations were significant at P < 0.05. Greater N-T OKR gain was associated with a smaller T-N:N-T ratio, and greater upward OKR gain associated with a smaller down:up OKR gain ratio. Both of these relationships are expected (in each case a larger value of the absolute gain in the denominator results in a lower gain ratio), provided that sources of variability of the pairs of absolute OKR gains (i.e., downward and upward, or T-N and N-T) are independent.

In tottering, there were four significant correlations. Greater ocular elevation was associated with reduction of the down:up OKR gain (signed r² = –0.311). Although this result raises the possibility that an imbalance in vertical optokinetic tone causes the greater static elevations, rocker did not exhibit the same relationship, despite sharing the static elevation abnormality (signed r² = +0.17, P = 0.18). The second significant association in tottering was between the T-N:N-T and down:up OKR ratios (signed r² = +0.289), suggesting a connection between the OKR asymmetries in different axes. Finally, both N-T and upward OKR gains were positively associated with VOR gain at 1.6 Hz (respective r² values of +0.186 and +0.249), indicating that animals exhibiting poorer OKR gains were more likely to be impaired in their VOR gain as well. Although the corresponding regression analyses in rocker did not reach significance, the regression slopes had the same sign (N-T OKR gain vs. VOR gain r² = +0.151, P = 0.051; upward OKR gain vs. VOR gain r² = +0.241, P = 0.11). Thus a similar relationship between OKR and VOR gains could be present in this mutant. It may be masked in part by the smaller number of rocker animals. One correlation that might have been predicted was absent (i.e., an association between shorter integrator time constants and larger VOR phase leads at low stimulus frequencies). An explanation for this nonassociation as well as implications of the correlations for underlying pathophysiological mechanisms are addressed in the DISCUSSION.

	DISCUSSION

TOP ABSTRACT