P/Q-type voltage-dependent Ca2+ channels (VDCCs) are highly expressed in the cerebellum, and mutations of these channels are associated with disrupted motor function. Several allelic variants of the α1A pore-forming subunit of P/Q-type VDCCs have been described, and mice homozygous for these mutations exhibit gait ataxia, as do α1A knockout mice. Here we report that heterozygous α1A mutants also have a motor phenotype. Mice heterozygous for the leaner and α1A knockout mutations exhibit impaired motor learning in the vestibulo-ocular reflex (VOR), suggesting that subtle disruption of P/Q Ca2+ currents is sufficient to disrupt motor function. Basal VOR and optokinetic reflex performance were normal in the heterozygotes but severely impaired in the leaner and α1A knockout homozygotes.
Voltage-dependent Ca2+ channels (VDCCs) contribute to the regulation of Ca2+ levels in neurons. The P/Q-type VDCCs play a key role in the function of the motor circuitry of the brain, as evidenced by high levels of expression of P/Q-type VDCCs in the cerebellum and by the motor phenotypes of mutants with disrupted P/Q signaling. A number of point mutants of α1A, the pore-forming subunit of P/Q-type VDCCs, have been described, including tottering, leaner, rolling Nagoya, and rocker (Green and Sidman 1962; Oda 1973; Sidman and Green 1965; Zwingman et al. 2001). In addition, two α1A knockout strains have been generated (Fletcher et al. 2001; Jun et al. 1999). These mutants exhibit varying degrees of ataxia, dystonia, tremor, seizure, and episodic dyskinesia (for review, see Pietrobon 2005). In addition to reduced P/Q signaling, the different allelic variants exhibit a range of different structural changes and compensatory biophysical changes within the cerebellum, both of which may contribute to the motor phenotypes. The structural changes include degeneration of Purkinje cells and/or granule cells (Fletcher et al. 2001; Herrup and Wilczynski 1982), changes in dendritic morphology (Miyazaki et al. 2004; Rhyu et al. 1999), axonal swelling (Heckroth and Abbott 1994; Jun et al. 1999; Zwingman et al. 2001), changes in the morphology of parallel fiber-Purkinje cell synapses (Rhyu et al. 1999), and altered expression patterns of tyrosine hydroxylase (Austin et al. 1992; Fletcher et al. 1996; Hess and Wilson 1991). Some allelic variants also exhibit compensatory changes such as the upregulation of N-type VDCCs (Pagani et al. 2004; Qian and Noebels 2000).
Because different α1A gene allelic variants exhibit considerable variability in their phenotypes, a detailed comparison of the allelic variants should help to reveal the causal relationships among the many different biophysical, structural, and behavioral deficits associated with altered expression of P/Q-type VDCCs (Stahl et al. 2006). For example, the severity of gait ataxia in the different allelic variants is not well correlated with the extent to which P/Q conductance is reduced, but it is roughly proportional to the amount of cerebellar degeneration, suggesting that the ataxia results primarily from such degeneration rather than directly from reduced P/Q conductance (Fletcher et al. 2001). However, locomotion is a complex movement that is difficult to quantify, and gait ataxia can arise from any of a number of more elemental deficits. Therefore some investigators have turned to the oculomotor system for a more detailed, quantitative analysis of the motor deficits in α1A mutants. Although rocker and tottering mice exhibit similar, mild levels of gait ataxia, Stahl and colleagues distinguished four, mechanistically distinct groups of oculomotor deficits, including some that were shared by rocker and tottering and some that were unique to each strain (Stahl 2004; Stahl et al. 2006). Here we extend this approach by analyzing oculomotor performance and learning in additional α1A gene allelic variants—homozygous and heterozygous leaner and α1A knockout mice.
Homozygous leaner and α1A knockout mice have a 65–70% reduction in total Ca2+ current in cerebellar Purkinje cells and granule cells (Dove et al. 1998; Fletcher et al. 2001; Jun et al. 1999) and exhibit more severe ataxia than rocker or tottering mutants, which prevents their survival past three weeks postnatally without special care. We were particularly interested in determining whether the heterozygous leaner and α1A knockout mice exhibited an oculomotor phenotype because previous studies have reported the surprising finding of significantly reduced P/Q Ca2+ signaling in the cerebellum of these mice with no apparent gross anatomical deficits or motor consequences (Fletcher et al. 2001). Purkinje cells from leaner heterozygotes exhibit a 30% reduction in Ca2+ conductance (Lorenzon et al. 1998). In a study of one α1A knockout line, no reduced P/Q Ca2+ signaling was found in the heterozygotes (Jun et al. 1999), but in the other knockout line, a 50% reduction of P/Q Ca2+ conductance was found in cerebellar granule cells of the heterozygotes (Fletcher et al. 2001). Despite reduced Ca2+ conductance, no motor phenotype has been found in these or any other heterozygous α1A mutants on examination of complex motor behaviors, such as gait or performance on a rotorod (e.g., Fletcher et al. 2001; Plomp et al. 2000). Therefore after preliminary study, heterozygous α1A mutants have been largely overlooked or used as control mice for comparison with the homozygous mutants (e.g., Walter et al. 2006).
The lack of an obvious motor phenotype in α1A heterozygous mice has supported the idea that a moderate reduction in P/Q Ca2+ current density alone cannot cause the disease pathology associated with α1A mutations (Fletcher et al. 2001). Indeed a comparison of gait ataxia across α1A allelic variants suggests that there is a threshold for some of the pathological effects of reduced P/Q signaling (Fletcher et al. 2001) and that much of the motor pathology arises from the secondary, structural effects in the cerebellum. Nevertheless, given the ample biophysical evidence that the precise regulation of Ca2+ is critical for neuronal processes such as the generation of dendritic action potentials, transmitter release, pacemaking activity, and synaptic plasticity, it would be surprising if the reduced P/Q signaling in heterozygotes had absolutely no consequences for the function of the affected neurons and circuits. Moreover, although previous reports have suggested that all of the available mouse α1A mutants are autosomal recessive for motor phenotypes, human neurological disorders caused by α1A mutations are autosomal dominant or semi-dominant, indicating that at least some heterozygous α1A mutations can disrupt the function of motor circuits. Finally, a sensory phenotype has been reported in heterozygous leaner and α1A knockout mice, namely reduced nociception (Luvisetto et al. 2006; Ogasawara et al. 2001). Therefore we reexamined the α1A heterozygous mice for a motor phenotype, using the oculomotor system, which, because it is readily quantified, might provide a more sensitive measure of motor dysfunction than the measures of locomotion and gross motor coordination used in previous studies.
Experiments were performed on adult mice (10–20 wk old). α1A gene knockout mice were provided by Dr. Richard Tsien and backcrossed to C57BL/6 for six generations. Homozygous mutants (α1A−/−), heterozygous mutants (α1A+/−) and wild-type littermates (α1A+/+) were then obtained by intercrossing the heterozygotes and were identified by genotyping using PCR (Jun et al. 1999). The C57BL/6-congenic leaner mutant strain with an oligosyndactylism marker gene (Os+/+ Cacna1atg-la) was obtained from Jackson Laboratory. Leaner homozygous mice were obtained by intercrossing Os+/+ Cacna1atg-la. Offspring that carry 2 Os alleles (Os+/Os+) die in utero (Van Valen 1966); offspring that are genotypically Os+/+ Cacna1atg-la have fused digits due to the Os allele; therefore surviving offspring with normal digits were identified as leaner homozygous (+ Cacna1atg-la/+ Cacna1atg-la). No mice with the Os mutation were used in the experiments. Leaner heterozygous mice without the Os mutation were obtained by crossing Os+/+ Cacna1atg-la mice with C57BL/6 mice; offspring with normal digits were identified as leaner heterozygotes (+ Cacna1atg-la/+ +). C57BL/6 mice (Jackson Laboratory) were used as controls for the leaner homozygotes and heterozygotes. We found no significant difference between the C57BL/6 mice and the α1A+/+ mice from the crosses of α1A+/− mice on eye tracking performance or motor learning in the VOR (P > 0.15, for each t-test), therefore the results from these two control, genotypically normal groups were pooled and are referred to as “wild-type”. To improve the survival of the homozygous mutant mice, they were hand-fed with infant formula via oral gavage up to four times a day.
Surgical methods were identical to those described previously (Boyden and Raymond 2003). Briefly, while the mouse was under anesthesia, a head post was attached to the top of the skull using anchor screws and dental acrylic, and a scleral search coil (IET, Marly, Switzerland) was implanted on the temporal side of the right eye underneath the conjunctiva. The search coil leads were run subcutaneously to a two-pin connector. Mice were allowed to recover from surgery for 5–7 days before oculomotor testing.
All animal protocols were approved by the Stanford University Administrative Panel for Laboratory Animal Care.
For experiments, the head of the mouse was immobilized by attaching the implanted head post to a restrainer and was oriented so that lambda and bregma were in the same earth-horizontal plane. The restrainer was attached to a turntable (Carco IGTS, Pittsburgh, PA), which delivered a vestibular stimulus by rotating the mouse about an earth-vertical axis. Visual motion stimuli were delivered by a moving optokinetic drum made of a white translucent plastic half-dome with black vertical stripes, each of which subtended 7.5° of visual angle. The optokinetic drum was back-lit by two fiber optic lights (JH Technologies, San Jose, CA).
The eye coil method (Judge et al. 1980; Koekkoek et al. 1997; Robinson 1963) was used to measure eye movements. The eye coil method was used because it is particularly reliable for measuring learning- and memory-related changes in the vestibulo-ocular reflex (VOR) because it allows stable and repeatable precision in the measurement of mouse eye movements, over time scales from milliseconds to days (Boyden and Raymond 2003; Stahl et al. 2000).
After recovery from surgery, oculomotor performance was tested on two consecutive days, and the results from the two days were averaged. The VOR was tested with vestibular stimuli consisting of rotation about an earth-vertical axis. VOR measurements were made in total darkness to isolate the VOR from visually driven eye movements. The optokinetic reflex (OKR) was tested with rotation of the optokinetic drum about an earth-vertical axis. The combined VOR and OKR response was measured during head rotation in the light with the optokinetic drum fixed relative to the earth (x1 VORL). Vestibular stimuli or optokinetic drum movements were delivered at frequencies of 0.5, 0.75, 1.0, 2.0, and 5.0 Hz, with a fixed peak velocity of ±10°/s, and at peak velocities of ±5, 10, 15, 20, and 25°/s, with a fixed frequency of 1 Hz.
For illustration purposes (Figs. 1 and 2), representative raw eye- and head-velocity traces from individual stimulus cycles were obtained by averaging eye- and head-velocity within a sliding window of 100 ms. Sliding window averaging was not used in any of the statistical analyses.
For data analysis, any cycle containing a saccade or motion artifact was deleted. Multiple (5–50) cycles of head- and eye-velocity traces were aligned on the zero crossings of head velocity and then averaged. Fourier analysis was used to extract the amplitude and phase of the eye movement from the averaged traces. The VOR gain was calculated as the ratio of the eye- to head-velocity amplitudes. The VOR phase was calculated as the difference between the peak eye-velocity phase and the peak head-velocity phase in the opposite direction. A perfectly compensatory VOR would thus have a phase of zero. The OKR gain was calculated as the ratio of the eye- to drum-velocity amplitude. The OKR phase was calculated as the difference between the peak eye-velocity phase and the peak drum-velocity phase, with a perfectly compensatory OKR having a phase of zero.
Two or more days after the preceding tests of oculomotor performance, motor learning was evaluated with 30 min of inverted vision x(−1) training. x(−1) training consisted of sinusoidal vestibular stimulation at 1 Hz, ±10°/s paired with movement of the optokinetic drum in phase with the vestibular stimulation but at double the amplitude (±20°/s) relative to the earth. Therefore to stabilize the image of the optokinetic drum on the retina, the eye movement would need to be ±10°/s in phase with head motion. Because by convention a VOR exactly out of phase with head motion is defined as having a phase of 0°, the ideal response in the presence of the x(−1) training stimulus would thus have a phase of 180°. Animals were exposed to the training stimulus for three 10-min blocks, each followed by three 30-s tests of the VOR with ±10°/s sinusoidal vestibular stimulation at 1 Hz in darkness. A bell was rung loudly to maintain animal alertness, followed by an 8-s pause before beginning each 30-s VOR measurement block. VOR-gain data acquired during and after training were normalized to the initial VOR gain measured immediately before training. Learned changes in VOR phase were measured as the difference between the phase after versus before training.
To evaluate statistical differences between groups, we used ANOVA with post hoc Scheffe’s tests, performed using StatView (SAS, Cary, NC). For the oculomotor performance data (Figs. 1 and 2), we performed a two-factor ANOVA with genotype and stimulus frequency or genotype and stimulus amplitude as the factors. For the learning data (Fig. 3), we performed a one-factor ANOVA on genotype. The results of the ANOVAs are summarized in Tables 1 and 2. If, and only if, ANOVA identified a main effect of genotype at a significance level of P < 0.05, we conducted post hoc Scheffe’s tests. To examine the correlation between learning and the baseline oculomotor performance of individual mice (Fig. 4 and associated text), we performed linear regressions.
Several measures of oculomotor function were tested in α1A mutants and compared with wild-type mice. In particular, we focused on the VOR and OKR. The VOR is a reflexive eye movement that stabilizes images on the retina by using vestibular signals to generate compensatory smooth eye movements in the opposite direction from head motion. The OKR is a reflexive eye movement in the direction of a moving visual stimulus. We measured the baseline performance of the VOR and OKR and also examined their interaction during combined vestibular and visual stimulation. In addition, we measured the adaptive modification of the VOR by motor learning.
Severe oculomotor performance dysfunction in homozygous leaner and α1A knockout mice
The oculomotor performance of the homozygous leaner (tgla/tgla) and α1A knockout (α1A−/−) mice are shown in Fig. 1 and compared with that of genotypically wild-type mice. Wild-type mice exhibited higher VOR gains and smaller VOR phase leads as the frequency of the vestibular test stimulus increased (Fig. 1A), and the values were consistent with those reported in most previous studies of the VOR in mice (Boyden and Raymond 2003; Feil et al. 2003; Iwashita et al. 2001; Katoh et al. 1998; Kimpo et al. 2005; Koekkoek et al. 1997; Shutoh et al. 2006; Stahl et al. 2006), although higher VOR gains have been reported in a few studies (Faulstich et al. 2004; Hoebeek et al. 2005; Katoh et al. 2005; Stahl 2004; Stahl et al. 2006). In wild-type mice, the OKR gain decreased as the stimulus frequency increased, with a large increase in phase lag at higher stimulus frequencies (Fig. 1B). Finally, the eye movement responses were measured during head movements in the light with the optokinetic drum fixed relative to the earth and hence moving relative to the mouse (Fig. 1C; x1 VORL). In this condition, the eye movements reflect a combination of both the VOR and OKR, and the mean gains in wild-type mice were therefore higher than those during VOR or OKR alone.
When tested with these same measures of oculomotor performance, tgla/tgla and α1A−/− mice exhibited severe motor performance deficits, which paralleled their severe gait ataxia and gross motor dysfunction. Both homozygous mutants had significantly impaired eye-movement gains on all three tests of oculomotor performance relative to wild-type (Fig. 1; P < 0.0001 for tgla/tgla vs. wild-type and for α1A−/− vs. wild-type, for VOR, x1 VORL, and OKR, Scheffe’s post hoc test). A previous report indicated that these two strains have similarly impaired gaits (Jun et al. 1999), yet there was a difference in their oculomotor performance. None of the α1A−/− animals studied showed a VOR, OKR, or x1 VORL response with a gain greater than 0.07. By contrast, tgla/tgla mice had some residual oculomotor function, with VOR gains ranging from 0.06 to 0.18 in individual animals. Unlike wild-type mice, the gain of the VOR in tgla/tgla mice did not vary with test frequency but was uniformly low across frequency (Fig. 1A). In effect, there was a more pronounced attenuation of VOR gain at high stimulus frequencies similar to what has been reported previously in rocker and tottering mutants (Stahl et al. 2006). The VOR of tgla/tgla mice also had less phase lead than in wild-type mice (P < 0.001, Scheffe’s post hoc test). Thus there was not only a decrease in gain but also a change in the dynamics of the oculomotor system. There was no evidence of visually driven eye movements in tgla/tgla mice. Their OKR gain was less than 0.06 on average, and their x1 VORL gain and phase were similar to their VOR gain and phase (P > 0.90 for gain and P > 0.08 for phase, Scheffe’s post hoc test), indicating that visual inputs did not contribute to the eye movements made by the tgla/tgla mice in response to the x1 VORL stimulus. Thus our data indicate that residual function of the mutant P/Q-type channels in tgla/tgla is sufficient to support limited VOR performance but is not capable of supporting visually driven eye movement.
Normal basal motor performance but impaired motor learning in α1A heterozygous mutant mice
Previous studies had not reported any motor phenotype in heterozygous α1A mutant mice. Consistent with these previous studies, we found no significant difference among heterozygous leaner (tgla/+), heterozygous α1A knockout (α1A+/−), and wild-type mice in any of our measures of basal oculomotor performance. Gain and phase were both normal across all frequencies and velocities tested for VOR, OKR, and x1 VORL (Fig. 2, Table 1).
Although basal performance was normal, the heterozygous mutants both exhibited a deficit in motor learning. Motor learning can be induced in the VOR by pairing head movements with the movement of a visual stimulus. In our experiments, a 1 Hz, ±10°/s head movement was paired with movement of the optokinetic drum at ±10°/s relative to the head and in the same direction as the motion of the head. This training condition is called x(−1) because under such conditions, the ideal VOR is “inverted” so that a head movement elicits an eye movement in the same rather than the opposite direction from head motion.
The learned changes in the VOR induced by the x(−1) training paradigm are shown in Fig. 3. In these polar plots, the ratio of the posttraining VOR gain to the pretraining VOR gain is represented by the distance from the origin, and the learned change in VOR phase (posttraining minus pretraining phase) is represented by the angle. Thus the dashed vector with normalized gain of 1.0 and phase of zero represents the prelearning state, or the condition of no learning, and the distance from this vector reflects the amount of learning in both individual animals (symbols) and the population average (bold vectors). In wild-type mice, 30 min of x(−1) training induced a VOR gain decrease and an increase in phase lead (Fig. 3, left). These results are similar to what has been reported previously for inverted gain training paradigms in mice and other species (Miles and Eighmy 1980; Pastor et al. 1992; Robinson 1976; van Alphen and De Zeeuw 2002). The decrease in gain reflects an adaptive suppression of the normal VOR response in the opposite direction from head motion. Also the learned change in VOR phase moves the phase closer to the ideal for this training paradigm. The ideal eye-movement response for this training paradigm has a phase of 180° (the ideal VOR response under normal viewing conditions is defined as having a phase of 0°). Because the initial, basal VOR has a phase lead of ∼30°, the increase in phase lead induced by the x(−1) training moves the VOR phase closer to the ideal of 180°.
We summarized the effect of training by a single number, the vector distance from the prelearning state, and compared the results for the heterozygous mutant mice (Fig. 3, tgla/+: blue, filled triangles; α1A+/−: red, filled circles) with those from the wild-type mice. Significantly less motor learning, as measured by this vector difference, was observed in both of the α1A heterozygous mutant populations than in wild-type mice (P < 0.001 for α1A+/−, P < 0.01 for tgla/+, Scheffe’s post hoc test). The learned changes in VOR gain were similar in the mutants to the wild-type (WT: 19.5 ± 5.7%; tgla/+: 26.0 ± 3.8%; α1A+/−: 17.8 ± 4.7%; means ± SE); however the learned changes in VOR phase were considerably smaller in the mutants (WT: 29.6 ± 2.1°; tgla/+: 16.1 ± 3.0°; α1A+/−: 10.6 ± 3.0°).
The deficit in the α1A heterozygous mutants appears to be an impairment in learning per se rather than a secondary effect of sensory or motor performance deficits. There was no significant difference in VOR, OKR, or x1 VORL performance relative to wild type (Fig. 2, Table 1). Figure 2 suggests that although not significant, the baseline oculomotor performance gains in the α1A knockout heterozygotes (red, filled circles) tended to be slightly lower than those in leaner heterozygotes or wild-type mice. Yet several observations suggest that this trend for a performance deficit cannot account for the impaired learning in the mutants. First, the leaner heterozygotes had, if anything, slightly higher average baseline VOR gains, whereas the heterozygous α1A knockouts had, if anything, slightly lower baseline VOR gains than the wild-type mice, nevertheless both heterozygous mutants exhibited significantly less learning than the wild-type mice. Second, there was no significant difference in the tracking of the x(−1) training stimulus (Table 2), which means that the mutants and wild-type mice had similar sensory and motor experiences during training. Third, we examined the correlation between the amount of learning and the baseline oculomotor performance of individual animals and found no significant correlation for baseline VOR gain (P > 0.98; Fig. 4A), OKR gain (P > 0.09), x1 VORL gain (P > 0.28) or tracking of the x(−1) training stimulus (P > 0.16; Fig. 4B). Indeed, Fig. 4 illustrates that the distributions of the amount of learning in the individual α1A knockout heterozygotes (red circles) and wild-type mice (black diamonds) were nonoverlapping (ordinate) even though their distributions for the baseline oculomotor performance measures were highly overlapping (abscissa). Therefore the significant differences in learning in the heterozygous mutants cannot be accounted for by any trends for their baseline performance measures to differ from genotypically wild-type mice.
Homozygous α1A mutants all exhibit ataxia (Pietrobon 2005). Here, we report that homozygous leaner and α1A gene knockout mice also have VOR and OKR performance deficits. Previous studies have suggested that α1A mutations in mice function as strictly recessive mutations since the gross motor coordination and rotorod performance of α1A heterozygotes was indistinguishable from wild-type mice (Fletcher et al. 2001). In the current study, we found no basal oculomotor performance deficits in the heterozygotes; however, our analysis revealed a clear deficit in motor learning in both leaner and α1A knockout heterozygotes. The normal tracking of the x(−1) stimulus during training, together with the normal basal VOR and OKR performance, indicate that the different amounts of learning cannot be explained by different sensory or motor experience during training and is likely to reflect a difference in the learning capacity per se rather than a secondary effect of an impairment in sensory or motor performance. This oculomotor learning impairment is the first experimental evidence that the subtle disruption of P/Q Ca2+ channel expression in α1A heterozygotes is sufficient to disrupt motor function.
Comparison of homozygous and heterozygous α1A gene allelic variants
Consistent with their more severe ataxia, we found that leaner and α1A knockout homozygous mice exhibit more severe oculomotor performance deficits than the modestly reduced VOR and OKR gains reported previously in tottering and rocker homozygotes (Stahl 2004; Stahl et al. 2006). In addition to the overall reduction in oculomotor gains, there was a change in the dynamics of the oculomotor system in the leaner mutants as evidenced by a reduced frequency dependence of the VOR gain and reduction in VOR phase lead. Rocker homozygotes also show a trend for reduced VOR phase leads at stimulus frequencies in the range we tested, although the most pronounced effect of the tottering and rocker mutations on VOR phase was an enhanced phase lead at frequencies of less than or equal to 0.2 Hz (Stahl et al. 2006).
The impairment of oculomotor performance in homozygous leaner and α1A knockout mice was more severe than that observed after lesions of the cerebellum in wild-type mice (Katoh et al. 2005; Koekkoek et al. 1997), suggesting that there could be extracerebellar pathology in the oculomotor circuitry, possibly in the extraocular motor neurons, since these neurons also express the α1A subunit (Miles et al. 2004). Alternatively, the output from the cerebellar cortex of these mutants could be abnormal in a way that is more disruptive to downstream components of the oculomotor circuitry than a simple loss of this output (Katoh et al. 2005).
In contrast, the oculomotor deficits in the leaner and α1A knockout heterozygotes were considerably milder than observed after cerebellar lesions and in tottering or rocker homozygotes, indicating a more subtle disruption of the motor circuitry. Stahl et al. (2006) proposed the existence of at least four distinct pathological processes in tottering and rocker, one of which affected both the basal VOR, the basal OKR, the neural integrator and motor learning in the VOR. The current experiments suggest a further subdivision of pathological processes associated with disrupted α1A expression because the leaner and α1A knockout heterozygotes exhibit a very selective disruption of motor learning in the VOR with sparing of basal VOR and OKR performance. Thus motor learning is particularly sensitive to subtle disruption of P/Q expression.
Previous studies had suggested that a moderate reduction of Ca2+ conductance itself does not cause most of the pathology associated with α1A mutations. Our finding of normal oculomotor performance in leaner heterozygotes provides further support for this idea. A 30% reduction in total calcium conductance has been reported in acutely dissociated Purkinje cells from leaner heterozygotes (Lorenzon et al. 1998). This reduction is similar in amplitude to that reported in Purkinje cells from tottering (40%) (Wakamori et al. 1998) and rocker homozygotes (30%) (Itsukaichi et al. 2002). Nevertheless, the leaner heterozygotes share only the motor learning deficit of tottering and rocker homozygotes, but do not share any of the motor performance deficits. Thus the reduced Ca2+ conductance itself does not account for the majority of oculomotor deficits in tottering and rocker homozygotes. Rather, such performance deficits may arise from subtle, secondary electrophysiological or ultrastructural consequences of the tottering and rocker mutation that were overlooked previously.
Selective motor learning deficit in heterozygous α1A mutant mice
Our results suggest that the most direct consequence of reduced P/Q signaling is a motor learning deficit, which is shared by the homozygous tottering and rocker (Stahl et al. 2006) and the heterozygous leaner and α1A knockout mice (Fig. 3). Our finding of a motor learning deficit in the leaner heterozygotes raises the possibility that reduced P/Q Ca2+ signaling in the Purkinje cells could itself disrupt the function of these neurons in motor learning. It is conceivable that the expression of one normal and one mutant copy of the P/Q channel protein in the leaner heterozygotes alters other properties of the P/Q currents in addition to amplitude, such as a change in the average voltage sensitivity or the kinetics of activation or inactivation. On the other hand, all of the P/Q channels in the α1A knockout heterozygotes should be composed of normal subunits, transcribed from the one remaining copy of the α1A gene. Our finding that these latter mice have impaired motor learning suggests that transcriptional processes do not compensate for the missing copy of the gene to produce normal expression levels, and that reduced levels of otherwise normal P/Q channels are sufficient to produce a motor learning phenotype.
Surprisingly, in vitro electrophysiological analysis of the α1A knockout heterozygotes used in our experiments revealed no reduction in Ca2+ signaling in Purkinje cells (Jun et al. 1999). In fact, no biophysical or ultrastructural abnormalities of any kind have been described in the α1A knockout heterozygotes used in this study, although our finding of a motor-learning deficit indicates that there must be some altered P/Q signaling and/or secondary consequences somewhere in the oculomotor circuit that affect learning. One possibility is that the Purkinje cells of the heterozygous α1A knockout mice do have some reduced Ca2+ signaling in vivo that was not detected in the cultured neuron preparation used to measure Ca2+ conductance in these mice (Jun et al. 1999) but that may be detectable in an acutely dissociated Purkinje cell preparation, such as that used to measure Ca2+ conductance in the leaner heterozygotes (Lorenzon et al. 1998). Indeed, a reduction in P/Q Ca2+ current has been described in cerebellar granule cells of another line of α1A knockout heterozygotes (Fletcher et al. 2001). Alternatively, the leaner and α1A knockout heterozygotes may have other deficits in the electrophysiological, biophysical, or ultrastructural properties of the cerebellum or related oculomotor circuitry.
Calcium levels are critical in controlling the induction of LTP and LTD at many synapses in the brain, including the synapses from parallel fibers to Purkinje cells in the cerebellar cortex (Coesmans et al. 2004), which are a hypothesized site of motor memory storage for the VOR. The relative contribution of P/Q versus other calcium channel subtypes to plasticity at the parallel fiber—Purkinje cells synapses has not been evaluated. At most synapses, L-type channels seem to be the major contributor of calcium controlling the induction of LTP and LTD (e.g., Kreitzer and Malenka 2005; Moosmang et al. 2005). Nevertheless, there is some evidence that P/Q signaling can also contribute to plasticity (Komatsu and Yoshimura 2000; Liu et al. 2004).
There is evidence that motor learning in the VOR involves changes in both the cerebellar cortex and vestibular nuclei. P/Q channels are expressed at high levels in the cerebellar cortex (Stea et al. 1994; Tanaka et al. 1995). Expression of P/Q channels is low in the vestibular nuclei (Tanaka et al. 1995); however, it has been hypothesized that the induction of learning-related changes in the vestibular nuclei depends on instructive signals carried by their Purkinje cell inputs (for review, see Boyden et al. 2004). Recording, lesion, and electrical stimulation experiments in vivo and in vitro have suggested that disruption of normal P/Q signaling in the Purkinje cells can reduce the regularity of their spiking and render these neurons impotent in affecting downstream neurons (Hoebeek et al. 2005; Walter et al. 2006). Such an effect could disrupt plasticity in the vestibular nuclei by depriving them of instructive signals from the cerebellar cortex.
Most studies of motor learning in the VOR have focused on adaptive changes in the gain of the VOR. In the heterozygous leaner and α1A knockout mice, the main deficit was a reduction in the changes in VOR phase induced by x(−1) training, with a relative sparing of the changes in VOR gain induced by this training. Previous studies have suggested that different components of motor learning in the VOR, such as increases versus decreases in VOR gain, or changes in gain induced using different sinusoidal stimulus frequencies, are supported by different plasticity mechanisms (Boyden and Raymond 2003; Kimpo et al. 2005; Li et al. 1995; Raymond and Lisberger 1996). The current results raise the possibility that the gain and phase of the VOR response are also regulated by different mechanisms.
Thus α1A heterozygous animals are worthy of additional close study at both the electrophysiological and ultrastructural levels. Unlike other α1A mutants, which exhibit multiple deficits, with apparently different underlying pathophysiological processes, the heterozygotes appear to be very selectively impaired in motor learning. Therefore these mice provide a new resource for systematically analyzing the consequences of disrupted expression of P/Q channels. In particular, these mice provide an important tool for analyzing the most direct consequences of disrupted P/Q signaling on the function of the cerebellar circuit in motor learning.
This research was supported by National Institute on Deafness and Other Communication Disorders R01 DC-04154 and an EJLB Foundation Award to J. L. Raymond. A. Katoh was supported by a long-term fellowship from the Human Frontier Science Program and a Zaffaroni fellowship.
We thank R. Tsien for providing us with α1A gene knockout mice. We also thank P. Louderback, Y. Cao, and K. Oh for technical help and A. Bristol, M. Ke, R. Kimpo, and G. Zhao for helpful comments on the manuscript.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 by the American Physiological Society