Purkinje Cell-Specific Ablation of CaV2.1 Channels is Sufficient to Cause Cerebellar Ataxia in Mice
- 1.2k Downloads
The Cacna1a gene encodes the α1A subunit of voltage-gated CaV2.1 Ca2+ channels that are involved in neurotransmission at central synapses. CaV2.1-α1-knockout (α1KO) mice, which lack CaV2.1 channels in all neurons, have a very severe phenotype of cerebellar ataxia and dystonia, and usually die around postnatal day 20. This early lethality, combined with the wide expression of CaV2.1 channels throughout the cerebellar cortex and nuclei, prohibited determination of the contribution of particular cerebellar cell types to the development of the severe neurobiological phenotype in Cacna1a mutant mice. Here, we crossed conditional Cacna1a mice with transgenic mice expressing Cre recombinase, driven by the Purkinje cell-specific Pcp2 promoter, to specifically ablate the CaV2.1-α1A subunit and thereby CaV2.1 channels in Purkinje cells. Purkinje cell CaV2.1-α1A-knockout (PCα1KO) mice aged without difficulties, rescuing the lethal phenotype seen in α1KO mice. PCα1KO mice exhibited cerebellar ataxia starting around P12, much earlier than the first signs of progressive Purkinje cell loss, which appears in these mice between P30 and P45. Secondary cell loss was observed in the granular and molecular layers of the cerebellum and the volume of all individual cerebellar nuclei was reduced. In this mouse model with a cell type-specific ablation of CaV2.1 channels, we show that ablation of CaV2.1 channels restricted to Purkinje cells is sufficient to cause cerebellar ataxia. We demonstrate that spatial ablation of CaV2.1 channels may help in unraveling mechanisms of human disease.
KeywordsP/Q-type Ca2+ channels Conditional Cell-specific knockout Cacna1a Ataxia
CaV2.1 Ca2+ channels are plasma membrane multimeric protein complexes that play an important role in neurotransmitter release at most central brain synapses . CaV2.1 channels are characterized by the presence of the pore-forming α1A subunit that is encoded by the Cacna1a gene and are widely expressed throughout the nervous system [2, 3]. CaV2.1-α1A-knockout (α1KO) mice that lack CaV2.1 channels exhibit a severe phenotype of ataxia and dystonia with a strong cerebellar component that starts around postnatal day (P)12, and die, if not given very special care, around P20 [4, 5, 6, 7]. Histological analysis in the few “survivors” indicated that a progressive gradual loss of Purkinje cells started between P45 and P100 .
Cerebellar ataxia and dystonia can also be part of the phenotype of certain naturally occurring Cacna1a mouse mutants. For instance, leaner, tottering, rolling Nagoya, but not rocker, are characterized to different extent by granule and/or Purkinje cell death and chronic ataxia [8, 9, 10, 11]. For tottering and leaner mice, it was shown that the chronic ataxia is related to irregular Purkinje cell simple spike firing caused by the loss of precision in intrinsic activity of Purkinje cells and aberrant synaptic input [12, 13, 14, 15]. In addition to the cerebellar ataxia phenotype, tottering mice also exhibit episodic dystonia, which is related to transient low-frequency oscillations in the cerebellar cortex . Notably, when tottering mutants lose their Purkinje cells on a Purkinje cell degeneration mutant background, the dystonic phenotype disappears . Together these findings indicate that mutations that cause a reduction of CaV2.1-mediated Ca2+-influx affect the information processing in the cerebellar cortex and, thereby, induce chronic ataxia and episodic dystonia.
Apart from the aberrant activity in the cerebellar cortex in Cacna1a mutants, also the activity of cerebellar nuclear (CN) neurons is subject to abnormal firing patterns, as was recently shown in tottering mice . Both Purkinje cells and CN neurons receive input (direct and indirect through local interneurons and/or granule cells) from mossy fibers and climbing fibers. While granule cell transmission to Purkinje cells in α1KO, tottering, leaner, and rolling Nagoya mice is abnormal, the response to climbing fiber activation is seemingly normal in these mutants [15, 19, 20]. Besides the excitatory input from climbing and mossy fibers, the neurons in the CN receive input from local interneurons and from the cerebellar cortex through Purkinje cells. Apart from the impact of these inputs, the firing pattern of CN neurons is determined by their intrinsic activities. Since Purkinje cells and CN neurons, as well as the neurons and fibers that innervate them, express CaV2.1 channels , the exact origin of the aberrant cerebellar activity in CaV2.1 mutants that causes cerebellar ataxia could not be resolved using the existing Cacna1a mouse mutants.
Therefore, we used the Cre-lox system to ablate CaV2.1 channels exclusively in Purkinje cells by crossing conditional Cacna1a mice that carry a “floxed” allele  with transgenic mice that express Cre recombinase under the control of the Purkinje cell-specific Pcp2 promoter . Hereby, we were not only able to rescue the early lethality seen in α1KO mice, but also to demonstrate that lack of CaV2.1 channel function in Purkinje cells is sufficient to cause ataxia and that progressive Purkinje cell loss starts well after the onset of the ataxia.
Materials and Methods
Reverse Transcription PCR
Total RNA was isolated from freshly dissected cerebellum and cerebrum of mice. For RT-PCR, first-strand cDNA was synthesized using random primers. Subsequently, Cacna1a-specific PCRs were performed using primers P7 (5′-gatgacacggaaccatac-3′) and P8 (5′-attgtagaggagatcagtcc-3′) that are located in exon 3 and 6, respectively.
Mouse brains (20 PCα1KO and 20 WT) were obtained at various ages after cardiac perfusion with PBS followed by 4% buffered paraformaldehyde (PFA). Post-fixation was performed for 1 h in 4% PFA followed by overnight incubation in 10% sucrose in 0.1 M phosphate buffer at 4°C. Subsequently, the sheet of dura mater separating cerebrum and cerebellum was removed, tissue was embedded in 10% sucrose in 10% gelatine, and the gelatin was fixed with 30% sucrose in 10% PFA for 2.5 h at room temperature. This was followed by an overnight incubation in 30% sucrose in 0.1 M phosphate buffer at 4°C. Forty-micrometer sagittal sections were made using a freezing cryotome (Leica, Bensheim, Germany). Part of the sections was processed for silver staining, which selectively stains dying neurons . The remainder of the sections was processed directly for free-floating immunohistochemistry. Sections were incubated in 10% heat-inactivated normal horse serum (NHS) with 0.5% Triton-X100 in TBS, for 2 h and then incubated either with primary rabbit polyclonal anti-CaV2.1-α1A antibody (ACC-001, 1:100; Alomone Labs, Jerusalem, Israel), or with rabbit polyclonal anti-Calbindin antibody (CB38, 1:10,000; SWANT, Bellinzona, Switzerland) diluted in 1% NHS with 2% heat-inactivated fetal calf serum and 0.4% Triton-X100 in TBS at 4°C. Next, sections were washed and incubated for 2 h at room temperature with a secondary biotin-labeled goat anti-rabbit antibody (1:200; Vector Laboratories, Burlingame, CA). For protein detection, sections were incubated with the avidin-biotin-peroxidase complex (ABC kit, 1:100, Vector Laboratories) for 2 h at room temperature, and developed in 0.1 mg/ml diaminobenzidine in 0.005% H2O2. Sections were Nissl stained, air dried, and cover slipped. Paraffin-embedded cerebellar sections (5 μm) were processed for standard haematoxylin and eosin (HE) staining.
Brains from 30-day-old mice (two PCα1KO and two WT) were isolated after decapitation and processed for Golgi impregnation. Brains were rinsed briefly in Milli-Q water. Tissue was further processed for Golgi-Cox staining using FD Rapid GolgiStain™ Kit (MTR Scientific, Ijamsville, MD) with 120-μm thick sections. Light microscopy images were used to assess the dendritic arborisation of Purkinje cells using the linear Sholl analysis [24, 25, 26]. In short, 7 to 11 concentric circles increasing in size (with 15 μm intervals) were aligned with the Purkinje cell soma. The number of Purkinje cell dendrites intersecting with each circle was counted. Statistical differences between groups were determined by use of repeated measures ANOVAs with one “between-subjects” factor (genotype) and one “within-subjects” factor (i.e. number of Purkinje cell dendrites intersecting with the circle) with 11 levels. The change in the dendritic branching with increasing distance within the groups was subsequently tested by use of separate repeated measures ANOVAs for each genotype with one “within-subjects” factor with 11 levels. The assumption of normality of the distributions (as tested by Kolmogorov–Smirnov tests), and equality of error variances (as tested by Levene’s tests) were not violated. When the assumption of sphericity (as tested by Mauchly’s tests) was violated, Greenhouse–Geisser corrections were used. All statistical analyses were performed using SPSS 17.0 software (IBM corporation, Chicago, IL, USA).
Quantification Number of Purkinje Cells and Degeneration of Purkinje Cell Axons
Using 40-μm thick, calbindin-stained, transversally sectioned tissue from two PCα1KO mice and three WT littermates, we calculated the average number of Purkinje cells per 100 μm of cerebellar cortex. Five to ten pictures were taken at ×20 magnification from either the vermal parts of lobule 2, 6, or 9, or from the paramedian lobule, paraflocculus, or flocculus. In each picture, the number of Purkinje cells per 100 μm of Purkinje cell layer was determined using Metamorph (Universal Imaging, Downing Town, PA, USA). In absence of many Purkinje cells the difference in structure between the molecular and granular layers guided the observer to accurately estimate the location of the Purkinje cell layer.
The extent of degeneration of Purkinje cell axons was qualitatively examined in sagittal cerebellar sections stained with silver impregnation as described elsewhere . For each mouse, photographs of white matter were taken with a ×5 magnification for overview and a ×40 magnification for detailed examination of degenerating fibers.
Neuronal Volume and Density Measurements
Volume measurements were bilaterally taken from the medial cerebellar nuclei (MCN), anterior interposed nuclei (AIN), posterior interposed nuclei (PIN), and lateral cerebellar nuclei (LCN) using transversal, Nissl/calbindin-stained sections from three WT and three PCα1KO mice by the investigator blinded to the genotype. Contours were drawn around the four nuclei in every other section (e.g. 80-μm intervals) using NeuroLucida (MBF Bioscience, Williston, VT, USA) to construct a 3D image. Volumes (in mm3) were then calculated using NeuroExplorer (Nex Technologies, Littleton, MA, USA). The dorsolateral hump was included in the AIN and the “interstitial cell group” was included in the PIN. Statistical differences were determined by an independent samples Student’s t test or an independent samples Mann–Whitney U test (depending on the normality of the distributions as tested by Kolmogorov–Smirnov tests). With regard to t tests, if the assumption of equality of variances was violated (as tested by Levene’s tests), corrected degrees of freedom were used.
Neuronal surface density in the cerebellar nuclei was bilaterally determined in transversal paraffin-embedded, HE-stained cerebellar sections (4 μm) of two WT and two PCα1KO mice. For each CN, sections with a 60-μm interval were photographed at ×25 magnification (five to ten pictures per nucleus). Two independent investigators, each blinded to the genotype, counted the number of cells in each picture using Meta Imaging Series 4.6 (Molecular Devices Inc., Downingtown, PA, USA). Counting was done by drawing a contour around each CN and subsequently counting all somata with a clearly visible nucleolus by contouring the neuron’s nucleus. The surface density (number of cells/mm2) was calculated for each section, after which the average density per CN was calculated. The total numbers of “small” and “large” cells were subsequently determined using a threshold of 8.5-μm diameter (see also Ref. ). The 8.5-μm cut-off value was constructed using a scatter plot of the neuron nucleus’ diameter from which two groups could be distinguished. Statistical differences were determined by an independent samples Student’s t test or Mann–Whitney U test (depending on the normality of the distributions as tested by Kolmogorov–Smirnov tests). With regard to t tests, if the assumption of equality of variances was violated (as tested by Levene’s tests), corrected degrees of freedom were used.
Motor Behavior Analyses
The accelerating Rotarod (UGO Basile S.R.L., Commerio VA, Italy) test was performed on a 3 cm diameter horizontal rotating rod. The rod initially rotated at a constant speed of 4 rpm and accelerated in a stepwise manner (i.e. 4 rpm every 30 s) to a maximum of 40 rpm after 270 s. The maximum time on the rod was 300 s. The latency to fall was recorded on the moment the mouse fell down or stopped walking (i.e. clinging to the rod) for three consecutive turns and subsequently converted to the “maximum rotating speed” by calculation of the rotating speed at the moment of the fall. The test was performed in a semi-dark room with a light source placed at the bottom to discourage the mice from jumping off the rod. Mice of ∼P18 (n = 3 per genotype) were subjected to three trials separated by 90 min on 1 day without pre-training. Mice of P30 and P200–250 (n = 5 per genotype, per age group) were trained on the first day (by placing the mice on the rod that rotated with a constant speed of 5 rpm for 5 min) and then subjected to two trials separated by 30 min. On days 2 to 4, the P30 and P200–250 mice were directly subjected to two trials without pre-training. The results are presented as maximal rotating speed with mean ± SD of five mice each with two trials per day over 5 days. Statistical differences between groups were determined by use of repeated measures ANOVAs with one “between-subjects” factor (genotype) and one “within-subjects” factor (i.e. rotarod performance) with three (for ∼P18 mice) or five levels (for P30 and P200–250 mice). Rotarod performance within the groups was subsequently tested for significant changes over time by use of separate repeated measures ANOVAs for each genotype with one within-subjects factor with again three or five levels. The assumption of normality of the distributions (as tested by Kolmogorov–Smirnov tests), sphericity (as tested by Mauchly’s tests) and equality of error variances (as tested by Levene’s tests) were not violated.
Motor function was also assessed in P30 and P200–250 WT (n = 4 and 5, respectively) and PCα1KO (n = 4 and 5, respectively) mice using a dyskinesia rating scale that was previously described and validated in several mouse mutants [28, 29], including mice carrying a Cacna1a mutation. At the start of the experiment, each mouse was transferred to a new cage. Mice were scored for 1 min at 10-min intervals over a 60-min period. Abnormal motor behavior within the head/neck, trunk, and limbs were recorded in terms of type, including tremor (trunk, limbs, neck, and face), tonic flexion, extension (trunk, limbs, neck, and face), or twisting (trunk and neck) and clonus (limbs and face). Uncoordinated or ataxic motor behavior not attributable to a single-body part was also recorded, including falling, listing, and stumbling. Assessments of the abnormal motor behavior were then calculated by summing the dyskinesia scores during the 60-min observation period. Mice were observed on at least two separate occasions and mean dyskinesia scores were generated for each mouse. Statistical differences were determined by Mann–Whitney U tests, because the data were non-parametric in nature.
Generation of Mice That Lack CaV2.1 Channels in Purkinje Cells
Homozygous PCα1KO mice were generated by crossing Cacna1a flox/flox mice with transgenic mice expressing Cre recombinase driven by the Purkinje-specific Pcp2 gene promoter (i.e. Pcp2 Cre mice—see “Materials and Methods”) (Fig. 1a). Successful genomic deletion of exon 4 of the Cacna1a gene was assessed at the RNA level by RT-PCR on cerebral and cerebellar total RNA (Fig. 1b). Whereas in the cerebrum only non-recombined PCR product of 580 bps was observed, in the cerebellum both non-recombined and recombined (480 bps) products were present. This indicated that Cre recombination only had occurred in part of the cerebellar neuron population. Immunohistochemical staining for α1A protein showed that Cre recombination was Purkinje cell-specific as no protein signal was present in the soma of virtually any Purkinje neurons of PCα1KO mice (Fig. 1d, f). However, α1A protein expression in other cerebellar structures, e.g. the CN, appeared not different between PCα1KO and WT mice. These findings are in line with the reported Purkinje cell specificity of the Pcp2 Cre mice .
Motor Coordination Is Severely Affected in Purkinje Cell CaV2.1-Knockout Mice
Because it is difficult to determine the specific deficit that causes impaired rotarod performance, motor function was also assessed using a dyskinesia rating scale (see “Materials and Methods” for details). There was significantly more ataxia compared to tremor, tonic movements, or clonic movements in both P30 PCα1KO (n = 4) mice (ataxia vs. tremor: U = 16.00, p < 0.01; ataxia vs. tonic: U = 16.00, p < 0.01; ataxia vs. clonic: U = 16.00, p < 0.01) and P200–250 PCα1KO mice (n = 5; ataxia vs. tremor: U = 25.00, p < 0.01; ataxia vs. tonic: U = 25.00, p < 0.01; ataxia vs. clonic: U = 25.00, p < 0.01) (Fig. 2d, e). Ataxia, but not other movements, was significantly increased in young (P30) PCα1KO mice compared with age-matched WT mice (n = 4; U = 16.00, p < 0.01), whereas all abnormal movement types were significantly increased in old (P200–250) PCα1KO mice compared with age-matched WT littermates (n = 5; ataxia: U = 25.00, p < 0.01; tremor: U = 25.00, p < 0.01; tonic: U = 25.00, p < 0.01; clonic: U = 25.00, p < 0.01). In PCα1KO mice, there was a slight but significant increase in tremor with age (U = 20.00, p < 0.01) and a trend for an increase in ataxia and clonus with age (U s = 18 and 17, respectively), but their relative contribution to the abnormal motor behavior in mutant mice remained similar irrespective of their age.
Progressive Cell Death of Purkinje Cells in the Absence of CaV2.1 Channels
Neurodegenerative Changes in Cerebellar Cortex
Neurodegenerative Changes in Cerebellar Nuclei
Volume and surface density in the cerebellar nuclei (CN)
0.125 ± 0.016
494.3 ± 83.4
0.057 ± 0.011
842.6 ± 185.2
0.196 ± 0.026
370.0 ± 81.1
0.087 ± 0.015
696.5 ± 102.9
0.089 ± 0.030
504.8 ± 41.1
0.046 ± 0.008
668.0 ± 103.0
0.198 ± 0.004
394.1 ± 26.0
0.089 ± 0.015
624.5 ± 106.8
0.608 ± 0.071
0.280 ± 0.039
Here we used conditional gene targeting to ablate CaV2.1 calcium channels selectively in Purkinje cells in vivo in mice. This approach allowed us to study the specific contribution of these channels in Purkinje cells to the development of ataxia that is associated with mutations in the Cacna1a gene. Our cell type-specific knockout strategy using conditional Cacna1a-knockout mice  succeeded in circumventing early postnatal lethality that is a very prominent feature in Cacna1a-knockout mice that lack CaV2.1 channels in all neurons [4, 5, 6, 7]. In line with the well-defined expression of Cre recombinase in Purkinje cells in Pcp2 Cre mice , Cre recombination affected α1A expression in almost all Purkinje neurons, as is evident by the lack of anti-α1A staining in these cells at P30 while α1A expression seemed preserved in the rest of the brain (data not shown). We could demonstrate that lack of CaV2.1 channels in Purkinje cells is sufficient to cause an ataxic phenotype since the ataxia occurred prior to structural abnormalities. This finding adds to existing lines of evidence [8, 9, 10, 11], which indicate that the cerebellar ataxia associated with abnormal CaV2.1 channel function can be caused by Purkinje cell dysfunction.
Interestingly, our PCα1ΚΟ mutants exhibited a predominantly ataxic phenotype without episodic motor attacks during the first months of life. Notably, recently developed Cacna1a purk(−/−) mice, which lack functional CaV2.1 channels due to a Purkinje cell specific deletion of exon 1 of the Cacna1a gene, were reported to have dyskinesia attacks from 3 weeks on . Intrinsically generated low-frequency oscillations in the cerebellum of tottering mice are thought to be the source of the episodic motor attacks in these mutants . Further evidence for this hypothesis came from microinjections of the non-selective voltage-sensitive K+-channel blocker 3,4 DiAP in the cerebellum of tottering mice that abolished the episodic attacks of dyskinesia . Future experiments will have to show whether similar mechanisms in the cerebellum of PCα1KO underlie the episodic nature of dyskinesia.
Visual inspection of the mice revealed that the ataxia in PCα1KO mice started at P10–12, around the same age as in conventional α1KO mice that lack CaV2.1 channels in all neurons [4, 5, 6, 7]. Likewise, loss of CaV2.1 function in Purkinje cells did not affect normal development of the cerebellar cortex, similar to what was shown for conventional CaV2.1KO mice [4, 5] and Cacna1a purk(−/−) mice . These findings are in line with the fact that during the first postnatal week CaV1 (L-type) channels are the dominant voltage-gated Ca2+ channel type in Purkinje cell somata . At this time, however, we cannot rule out the possibility that other high-voltage-gated calcium channels, such as CaV2.2 (N-type) and CaV2.3 (R-type), are upregulated during later stages in PCα1KO mice and thereby rescue partly the detrimental effects of CaV2.1 absence. Similar compensatory mechanisms involving CaV2.2, CaV2.3, and CaV1 have been described in other Cacna1a mutant mice at the level of neurotransmission [15, 17, 32, 35].
The first signs of neurodegenerative changes in PCα1KO mice were observed around P45. At this stage, silver impregnation revealed clear argyrophylic staining in the Purkinje cell and molecular layers. From P60 on, Purkinje cell loss was clearly visible and many of the remaining Purkinje cells showed abnormalities such as torpedo’s or axonal swellings. At older ages (P100 and P200–250), Purkinje cell loss became progressively worse and ultimately also affected the survival of neurons in the granular layer and molecular layer. In lurcher and Purkinje cell degeneration (Pcd) mice, which both show near-complete Purkinje cell loss in adulthood, a progressive postnatal Purkinje cell degeneration is also followed by large scale death of granule cells after the loss of Purkinje cells [36, 37, 38]. Thus, the fact that three completely different mutations lead to the same sequence of cerebellar events (for lurcher see Ref. ; for Pcd see Ref. ) indicates that the degeneration of granule cells and molecular layer interneurons is probably a common, secondary effect to the loss of Purkinje cells.
Downstream Effects of Widespread Purkinje Cell Loss
The loss of Purkinje cells also affected the cerebellar nuclei in PCα1KO mice. The extent of volume reduction in PCα1KO mice was similar to that reported in lurcher and Pcd mouse mutants [27, 41]. In lurcher mice deafferentation of the CN resulted in a ∼60% reduction in the volume of all individual CN . We also found a widespread increase in neuronal surface densities in all CN. However, it appeared that the increase in neuronal density is more pronounced in small cells than in large cells. Despite the fact that our results do not allow a firm statement about the total number of cells per CN, our data seems to indicate a possible reduction in the number of large neurons. The data presented here shows a somewhat different pattern from lurcher mutants with regard to cell density . Future experiments should provide a more detailed picture of the total number of neurons per cell type per CN.
Behavioral Effects of Widespread Purkinje Cell Loss
The abnormal motor behavior in both young and old PCα1KO mice was dominated by ataxic movements. Since the behavioral abnormalities were already visible before the degeneration occurred (i.e. in young P18 and P30 PCα1KO mice), this indicates that the lack of functional CaV2.1 channels in Purkinje cells can induce the same behavioral aberrations as an absence of Purkinje cells (i.e. in old P200–250 PCα1KO mice). The severity of the ataxic symptoms in PCα1KO mice was also reflected by the fact that, unlike the WT controls, the mutants did not show an increase in their performance on the rotarod over time. Whether there is cerebellar motor learning deficit in the PCα1KO mice could not be assessed since motor learning (i.e. showing improved performance on the rod over time) may have been obscured by the severity of the ataxic phenotype. The severe effect on motor coordination in PCα1KO mice is likely to be due to the disrupted Purkinje cell output and the subsequent deregulation of activity patterns downstream [18, 42], which might be relatively normal, at least initially, as was described for lurcher mice .
This report represents a cell type-specific ablation of CaV2.1 channels leading to a disease phenotype. We demonstrate that specific ablation of these channels in Purkinje cells is sufficient to cause cerebellar ataxia that precedes progressive Purkinje cell loss. Conditional Cacna1a mice appear especially useful for dissecting the cell-specific role of CaV2.1 channels in neuronal circuitries such as in the cerebellum and in revealing the mechanisms underlying human disease.
The authors are grateful to E. Galliano for constructive discussion and help with the rotarod experiments and T.J.H. Ruigrok for the excellent assistance with the quantification of the histological data. This work was supported by a grant from the Centre for Medical Systems Biology within the framework of the Netherlands Genomics Initiative/Netherlands Organisation for Scientific Research (NWO; A.v.d.M.), Erasmus Fellowship (F.E.H.), NWO-ALW (F.E.H., C.D.Z.), NWO-ZON-MW (F.E.H., C.D.Z.), Neuro-BSIK (C.D.Z.), Prinses Beatrix fonds (C.D.Z.), and National Institutes of Health R01 NS33592 (E.J.H.).
Conflicts of interest
The authors declare that there are no potential conflicts.
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.
- 6.Kaja S, van de Ven RC, Broos LA, Frants RR, Ferrari MD, van den Maagdenberg AM, et al. Characterization of acetylcholine release and the compensatory contribution of non-Ca(v)2.1 channels at motor nerve terminals of leaner Ca(v)2.1-mutant mice. Neuroscience. 2007;144:1278–87.PubMedCrossRefGoogle Scholar
- 10.Sawada K, Kalam Azad A, Sakata-Haga H, Lee NS, Jeong YG, Fukui Y. Striking pattern of Purkinje cell loss in cerebellum of an ataxic mutant mouse, tottering. Acta Neurobiol Exp. 2009;69:138–45.Google Scholar
- 42.Tara E, Khodakhah K (2010) Altered purkinje cell activity during dyskinesia attacks in a mouse model of Episodic Ataxia. FENS [abstract 047.4]Google Scholar