Introduction
The spontaneous mouse mutant tottering (
tg) suffers from recessive neurological disorders including both permanent ataxia as well as episodes of dystonia, paroxysmal dyskinesia, and behavioral absence seizures. The absence seizures in
tg mice resemble human
petit mal seizures in that they are marked by abnormal electroencephalography (EEG) patterns [
1] and that they respond to common antiepileptic drugs [
2], whereas the attacks of paroxysmal dyskinesia cannot be correlated reliably to a clear EEG pattern and do not respond to antiepileptic therapeutics [
3,
4]. The episodes of paroxysmal dyskinesia can be reliably triggered by environmental challenges [
4‐
6] and are distinguished from the permanent ataxia by the sequence of three stages of behavioral abnormalities, which start at the hind limbs, then gradually spread to the front limbs, and eventually also reach the head and neck. These motor attacks, which typically occur approximately one to two times per day and can last up to 40 min, include irregular jerky movements, slow writhing motions, and involuntary stretching of the muscles [
1,
3,
7]. In between these episodes,
tg mice are mildly ataxic in that they show an abnormal gait and decreased motor performance and learning [
8‐
10].
The genotype of
tg mice is characterized by an autosomal recessive mutation in the gene located on chromosome 8 that encodes the α
1A-subunit of P/Q-type Ca
2+-channels in nerve, muscle, and secretory cells [
11,
12]. Since P/Q-type calcium channels are abundantly present in cerebellar Purkinje cells and gate ∼90% of their high-voltage-activated Ca
2+-influx [
13,
14], it is parsimonious to explain the juvenile onset of ataxia in
tg mutants by deficits in their Purkinje cells. Indeed, the calcium influx in
tg Purkinje cells is decreased by ∼45% [
15], their responses to parallel fiber stimulation are reduced by ∼50% [
16], and their simple spike firing patterns show enhanced irregularities with periods of pauses and bursts [
9]. Moreover, Purkinje cells in
tg show morphological aberrations in that their dendritic spines make relatively frequently multiple contacts with individual parallel fiber varicosities [
17] and that their somata have elongated nuclei and are reduced in size [
18,
19]. Interestingly, many of the morphological and physiological changes in the Purkinje cell dendrites and somata precede or coincide with the occurrence of the ataxia at the end of the first month of age suggesting that these changes form a major cause of the behavioral deficits [
9,
17]. However, it remains possible that changes at the level of the terminals of the Purkinje cells also contribute to the ataxia in
tg mutants, because the density of P/Q-type Ca
2+-channels is particularly high in terminals [
20,
21] and because chronically altered firing in Purkinje cells can lead to pathological alterations in their terminals [
22]. We therefore investigated the Purkinje cell terminals of
tg mutants at both the morphological and electrophysiological level. To correlate possible morphological aberrations to the behavioural changes, we investigated the distribution and ultrastructure of Purkinje cell terminals in the cerebellar nuclei before (2–3-week-old animals) and after (5-week and 6-month-old animals) the onset of the ataxia. In addition, we investigated whether the contacts of the Purkinje cell terminals with their postsynaptic neurons in the cerebellar nuclei in the adult mice were functionally intact by recording the extracellular activities of the cerebellar nuclei neurons following stimulation of the Purkinje cells.
Materials and Methods
Animals
Data were collected from 18 tg mice and 17 wild-type littermates (both male and female mice were included; C57BL/6J background; originally ordered from Jackson laboratory, Bar Harbor, ME, USA). The presence of the tg mutation in the Cacna1a gene on chromosome 8 was confirmed by PCR using 3′-TTCTGGGTACCAGATACAGG-5′ and 5′-AAGTGTCGAAGTTGGTGCGC-3′ primers (Eurogentech, The Netherlands) and subsequent digestion using restriction enzyme NSBI at the age of p9–p12. No oligosyndactyly was used. All preparations and experiments were done according to the European Communities Council Directive (86/609/EEC) and were reviewed and approved by the national ethics committee. For light microscopy, we restricted ourselves to animals of 6 months (tg, N = 3; wild type, N = 3), while for electron microscopy, we examined animals at the age of 2–3 weeks (i.e., for both 14 and 20 days tg, N = 2 and wild type, N = 2), 5 weeks (tg, N = 3; wild type, N = 2), and 6 months (tg, N = 3; wild type, N = 3). The electrophysiological recordings were conducted in mice at the age of 6 months (tg, N = 5; wild type, N = 5).
Light Microscopy
In this study, the Purkinje cell terminals were identified by immunocytochemical labeling using anticalbindin labeling [
23,
24]. To do so, three adult wild-type and three
tg littermates were anesthetized with an overdose of Nembutal (i.p.) and transcardially perfused with 0.12 M phosphate buffered saline (pH = 7.4) followed by 4% paraformaldehyde in phosphate buffer (PB) at room temperature. The cerebellum and brainstems were carefully removed, postfixed in 4% paraformaldehyde for 2 h, placed in 10% sucrose in PB at 4°C overnight, and subsequently embedded in gelatin in 30% sucrose. The blocks were cut on a cryotome into coronal sections of 40 μm. Sections were washed in blocking solution containing 10% normal horse serum (NHS) with 0.5% triton for 1 h and incubated in rabbit anticalbindin (1:7,000, Swant) with 2% normal horse serum and 0.5% Triton for 48 h [
25]. Subsequently, the sections were incubated for 2 h in biotinylated goat–antirabbit IgG at room temperature (1 to 500; Vector) followed by 2 h in avidin-biotinylated horseradish peroxidase complex (ABC-HRP; Vector). Sections were rinsed in PB and stained with 0.5% 3,3-diaminobenzidine tetrahydrochloride and 0.01% H
2O
2 for 15 min at room temperature. Sections of
tg mutants and wild-type littermates were processed simultaneously to avoid artificial differences due to the staining procedures. For quantification of terminals in the lateral cerebellar nuclei and interposed nuclei, we framed 500 × 500 μm with a ten times objective and used Neurolucida systems software (MicroBrightField, Colchester, VT, USA) for analyses, which were done blind to the genotype. The terminal numbers were averaged per animal and nucleus.
Electron Microscopy
Wild-type (N = 9) and tg mice (N = 10) were anesthetized with an overdose of Nembutal (i.p.) and transcardially perfused with 4% paraformaldehyde and 0.5% glutaraldehyde in cacodylate buffer. Brains were removed, kept overnight in 4% paraformaldehyde, and cut into 80 μm thick coronal sections on a vibratome. The vibratome sections were subsequently washed and blocked for 1 h in 10% NHS followed by 48 h of incubation in rabbit anticalbindin 4°C (1:7,000, Swant) and 2% NHS. Subsequently, the sections were incubated overnight 4°C in biotinylated goat–antirabbit IgG (1 to 500; Vector) and ABC-HRP (Vector). At the end of the immunostaining, the sections were stained with 0.5% 3,3-diaminobenzidine tetrahydrochloride and 0.01% H2O2 for 15 min at room temperature. Ultimately, the sections were osmicated with 2% osmium in 8% glucose solution, dehydrated in dimethoxypropane, and stained en block with 3% uranyl acetate/70% ethanol for 60 min and embedded in Araldite (Durcupan, Fluka, Germany). Guided by findings in semithin sections, we made pyramids of the medial cerebellar nucleus, lateral cerebellar nucleus, interposed cerebellar nucleus, and superior vestibular nucleus. Ultrathin sections (70–90 nm) were cut using an Ultramicrotome (Leica, Germany), mounted on copper grids, and counterstained with uranyl acetate and lead citrate. Purkinje cell terminals were photographed and analyzed using an electron microscope (Philips, Eindhoven, The Netherlands). Electron micrographs were taken at magnifications ranging from ×1,500 to ×30,000 from single hole EM grids and analyzed with the use of commercially available software (SIS) to study diameters and surface areas of labeled terminals and their surrounding structures in the neuropil. The surface area measurements were deduced automatically by drawing the circumference of all profiles (IBAS systems). Terminals of the lateral cerebellar nucleus and interposed cerebellar nucleus were each quantified per 25,000 μm2 in each animal by a researcher who was blind to the genotype of the mice. Since no significant differences were observed among the two cerebellar nuclei, the data were pooled. Statistics were done with the use of unpaired Student’s t tests assuming equal variances. p values equal or smaller than 0.05 were considered significant.
Electrophysiology
Five
tg mutants and five wild-type littermates of ∼6–8 months were anesthetized with ketamine (50 mg/kg body weight) and xylazine (8 mg/kg body weight) and subjected to extracellular single unit recordings of neurons in the cerebellar nuclei. Borosilicate pipettes (OD 2 mm, ID 1.16 mm, 4–10 MΩ, ∼1–2 μm tip diameter) filled with 2 M NaCl solution were positioned stereotactically using an electronic pipette holder (Luigs & Neumann, Ratingen, Germany). Signals were sampled at 10 KHz (Digidata 1322A, Axon Instr., Foster City, CA, USA), amplified, filtered, and stored for offline analysis (Multiclamp 700A, Axon Instr.). Purkinje cells in the cerebellar cortex were stimulated using custom-made urethane-insulated tungsten electrodes with two tips (separated ∼25 μm). A single negative 100-μs pulse of 100–400 μA (Cornerstone BSI-950, Dagan, Minneapolis, MN, USA) was used to activate the surrounding cerebellar cortical tissue. Stimulus locations were never deeper than 0.5 mm and were positioned in Lobule VI or paramedian lobule. Neurons of the cerebellar nuclei were identified by recording their characteristic activities [
26]. Once a responsive area within a cerebellar nucleus was found, multiple tracks were made to record both stimulus response activity and spontaneous activity. Evoked activity was recorded for at least 70 trials of 2 s each (at a frequency of 0.5 Hz) before or after which spontaneous activity was recorded for >2 min. Histological verification of the location of recordings was done by injection of 4% Alcian blue dye.
Analysis of Electrophysiological Data
Off-line analysis of neuronal firing rates was performed in Matlab (Mathworks Inc. Natick, MA, USA) as previously described by Goossens and colleagues [
27]. Firing frequency, coefficient of variance (CV; standard deviation (SD) interspike interval/mean interspike interval), and peristimulus histograms of the extracellularly recorded neuronal activities in the cerebellar nuclei were constructed using custom made routines in Matlab (Mathworks). To identify statistically significant responses to electrical stimulation of the cerebellar cortex from peristimulus histograms, we constructed an analog representation of each spike train using Gaussian local rate coding [
28]. The sum of these Gaussians represents the instantaneous firing frequency, which we normalized. Poststimulus excursions of the mean instantaneous frequency that exceeded three times the standard deviation were marked as statistically significant responses [
26] and were used to specify the latency of the inhibition. We used a Gaussian width of 1 ms to determine the occurrence of the spike rate change, typically at <6 ms after the stimulus onset. Any spiking activity that occurred during the stimulus artifact was not included in the analysis. Statistical analysis was done using unpaired Student’s
t tests (two tailed) assuming equal variances. Differences were considered to be significant when the
p value ≤ 0.05. Data are presented as mean ± standard error of the mean.
Discussion
The main finding of the present study is that Purkinje cell terminals in the cerebellar nuclei in tg show signs of structural damage such as an increase in size, swelling of mitochondria, presence of pathological vacuoles, and formation of large whorled bodies, while their synapses appear functionally intact. These morphological observations are corroborated by the finding that the activity patterns of their postsynaptic neurons in the cerebellar nuclei are faster and more irregular than those of their wild-type littermates. As will be discussed below, the morphological and physiological findings each have their own implications, but together they suggest that the pathology in Purkinje cell terminals in tg may contribute to a suboptimal neurotransmission in their cerebellar nuclei and thereby to their behavioural deficits.
The observation that the number of Purkinje cell terminals and synapses were not affected in
tg agreed with the fact that we found normal latencies and duration values for inhibition in the cerebellar nuclei neurons following artificial stimulation of the Purkinje cell input. These findings are in line with the findings in
tg that electrical stimulation of floccular Purkinje cells in their vestibulocerebellum can evoke short latency eye movements [
9] and that cortical lesions in their anterior vermis can have a positive impact on the occurrence of intermittent myoclonus-like movements [
30,
31]. Thus, neurotransmission appears possible at the synapses formed by the Purkinje cell terminals and their target neurons in the cerebellar and vestibular nuclei, but the question remains to what extent the pathology in the Purkinje cell terminals impairs signal coding.
The occurrences of swollen mitochondria and pathological vacuoles and to a lesser extent also those of the whorled bodies form the most prominent pathological changes that can be found in the Purkinje cell terminals of
tg mice. The exact mechanisms by which these three phenomena can be explained remain to be shown, but several possibilities should be addressed. First, the Purkinje cell terminals in
tg contain the mutated P/Q-type Ca
2+-channels and thereby they will most likely directly show altered dynamics and kinetics of their vesicle release, which in turn may influence the constitution of the organelles inside them [
32‐
35]. Second, the increased irregularity of Purkinje cell firing in
tg contributes to the occurrence of high frequency bursts of simple spikes [
9]. Increased simple spike firing frequencies have been shown to affect the formation of vacuoles, mitochondria, smooth endoplasmatic reticulum, and cause the formation of whorled bodies, e.g., in response to lesions of the inferior olive [
22,
36,
37]. Although the changes in simple spike firing in
tg are not as profound as seen in wild-type animals after lesioning the inferior olive, the effects in
tg are chronic and could therefore amount to a similar effect on Purkinje cell terminals. For example, Rossi and colleagues showed that the formation of vacuoles, mitochondria, smooth endoplasmatic reticulum, and whorled bodies were all affected in a dynamic fashion in particular time frames after lesioning the olive. Presumably, the changes in smooth endoplasmic reticulum that were observed by Rossi and colleagues, but not by us, were directly related to those of the whorled bodies [
38,
39], which were in fact also more substantial in their study than in the current one [
22]. Therefore, the chronic occurrence of high frequency bursts in Purkinje cell activity in
tg may trigger multiple intracellular mechanisms, which in turn could lead to the increase of the number and volume of mitochondria as well as the formation of vacuoles and whorled bodies within the same terminals.
The findings described and discussed above raise the question to what extent the pathological aberrations in the Purkinje cell terminals in
tg interact with those in their dendrites and cell bodies and to what extent they both contribute to the cerebellar movement disorders [
9,
16,
17]. The possibility that the pathological process at the terminals interacts with that at the cell body and dendritic arbor and that they both contribute to the behavioral deficits is supported by the observation that the period in which the morphological aberrations in the terminals start to occur, i.e., between the third and fifth postnatal week, coincides with the period in which the dendrites show their first abnormalities and in which the first signs of ataxia start to occur [
8,
17]. Moreover, it should be noted that abnormalities occurring in the axons themselves may also interact with those in the dendrites and terminals. In older
tgs (>6 months), the axons also show signs of swelling with accumulations of cytoplasmic organelles, irregularly arranged microtubules, and inclusions of a lysosomal origin [
17,
18], raising the possibility that propagation of action potentials down the Purkinje cell axons can also be affected in these
tg mice. Such a deficit may be especially detrimental, because during burst activity, the simple spike frequency in
tg mice can even exceed the maximum frequency that can be transmitted down the Purkinje cell axon in a healthy rodent [
9,
40,
41]. Thus, since the swelling and abnormalities that occur in the axons and terminals may further reduce this maximum frequency in
tg, the synaptic efficacy of the high frequency simple spike bursts at their cerebellar nuclei neurons will be even lower. This reduced efficacy may add to the more direct cell physiological deficits caused by the mutated P/Q-type voltage-gated calcium channels in Purkinje cell terminals that will affect their machinery of neurotransmitter release, as has also been shown for other cerebellar GABAergic inhibitory synapses [
42‐
44]. Taken together, our previous and present results provide ample evidence that the information relayed by the Purkinje cells in
tg mice is scrambled due to the altered synaptic input and decreased calcium influx in their dendrites and somata, and we propose that the ultrastructural aberrations in the axons and terminals further scramble their pathological spiking pattern. Thus, we conclude that the ultrastructural aberrations in the axonal terminals of the Purkinje cells in
tg described in the current study are likely to contribute to their cerebellar movement disorders.
The question remains to what extent the neurons in the cerebellar nuclei show intrinsic abnormalities in
tg mice. Our current results show that in vivo these neurons fire action potentials more irregular and faster and one may argue that this manifestation of aberrant information processing could be due to the fact that P/Q-type calcium channels are also expressed in cerebellar nuclei neurons themselves [
11,
45,
46]. However, recent evidence shows that the impact of P/Q-type calcium channels on the intrinsic excitability of cerebellar nuclei neurons is minimal [
47], which stands in sharp contrast to their impact on the excitability of Purkinje cells [
48]. Still, the cerebellar nuclei are formed by various types of excitatory and inhibitory neurons, which all have different electrophysiological characteristics [
47,
49,
50]. Thus, in order to further clarify the origin of cerebellar movement disorders in calcium mutants such as the
tg we do not only need to address the transmission of the Purkinje cell to cerebellar nuclei neuron synapse but also the intrinsic excitability of each type of neuron in the cerebellar nuclei.