Single Tottering Mutations Responsible for the Neuropathic Phenotype of the P-type Calcium Channel*

Recent genetic and molecular biological analyses have revealed many forms of inherited channelopathies. Homozygous ataxic mice, tottering ( tg ) and leaner ( tg la ) mice, have mutations in the P/Q-type Ca 2 1 channel a 1A subunit gene. Although their clinical phenotypes, histological changes, and locations of gene mutations are known, it remains unclear what phenotypes the mutant Ca 2 1 channels manifest, or whether the altered channel properties are the primary consequence of the mutations. To address these questions, we have character-ized the electrophysiological properties of Ca 2 1 channels in cerebellar Purkinje cells, where the P-type is the dominant Ca 2 1 channel, dissociated from the normal, tg, and tg la mice, and compared them with the properties of the wild-type and mutant a 1A channels recombinantly expressed with the a 2 and b subunits in baby hamster kidney cells. The most striking feature of Ca 2 1 channel currents of mutant Purkinje cells was a marked reduction in current density, being reduced to ; 60 and ; 40% of control in tg and tg la mice, respectively, without changes of cell size. The Ca 2 1 channel currents in the tg Purkinje cells showed a relative increase in non-inacti-vating component in voltage-dependent inactivation. subunits. Electrophysiological measurements and Northern blot anal- ysis were employed to identify functional expression of the a 1A channel in BHK-BI- tg . To transiently express tg la mutant a 1A channels, BHK6 cells were transfected with the recombinant plasmid pK4KBI- tg la (long) or pK4KBI- tg la (short) plus pH3-CD8 containing the cDNA of the T-cell antigen CD8 (28). Transfection was carried out using SuperFect Transfection Reagent (QIAGEN, Hilden, Germany). Cells were trypsinized, diluted with Dulbecco’s modified Eagle’s medium containing 10% fetal bovine serum, 30 units/ml penicillin, and 30 m g/ml streptomycin, and plated onto Celldesk (Sumitomo Bakelite, Tokyo, Japan) 18 h after transfection. Then cells were subjected to measurements 48–66 h after plating on the coverslips. Cells expressing a 1A channels were selected through detection of CD8 coexpression using polystyrene microspheres precoated with antibody to CD8 (Dynabeads M-450 CD8; DYNAL, Oslo, Norway). Whole Cell Recordings— Electrophysiological measurements were performed on Purkinje cell and BHK cells. BHK cells, stably expressing wild-type or tg - a 1A channel, were seeded onto plastic coverslips, Celldesk, and incubated in culture medium for 5–8 days. Currents were recorded at room temperature (22–25 °C) using whole cell mode of the patch-clamp technique (29) with an Axopatch 200B patch-clamp ampli- fier (Axon Instruments, Foster City, CA) or an EPC-7 (List Medical, Patch pipettes were made from borosilicate glass capillaries (1.5 mm outer diameter, 1.1 mm inner diameter; Na-rishige, Tokyo, Japan) using a model P-87 Flaming-Brown micropipette puller Instrument The patch electrodes were fire-polished. Pipette

Ca 2ϩ controls diverse cellular processes, which include muscle contraction, neurotransmitter release, and other forms of secretion, gene expression, and cell proliferation (1). To evoke these cellular responses, Ca 2ϩ influx across the plasma membrane makes a major contribution to augmenting the cytosolic free Ca 2ϩ concentration. In neurons, voltage-gated Ca 2ϩ channels are one of the major transmembrane pathways, together with Ca 2ϩ -permeable ligand-gated channels. Electrophysiological and pharmacological studies have defined several types of Ca 2ϩ channels in neurons. There are at least five types of high-threshold Ca 2ϩ channels (L, N, P, Q, and R) and a low-threshold Ca 2ϩ channel (T) (2)(3)(4)(5). At the molecular level, the voltage-gated Ca 2ϩ channels are composed of the main pore-forming ␣ 1 subunit and the accessory ␣ 2 and ␤ subunits (6 -8), and optional subunits, such as the ␥ subunit in the skeletal muscle Ca 2ϩ channel (9,10). Molecular biological analyses have found a gene family of ␣ 1 subunits (A, B, C, D, E, S, and G) (11,12). It is generally accepted that the ␣ 1A Ca 2ϩ channel corresponds to the functionally defined P-and Q-type isoforms, which are produced from the same gene by alternative splicing (13) and/or have different subunit compositions (14). The P-type was first described in cerebellar Purkinje cells as a Ca 2ϩ channel that is not blocked by -conotoxin GVIA or dihydropyridines (15), but is highly sensitive to -agatoxin (-Aga) 1 IVA (16). The Q-type is identified in cerebellar granule cells and is similar to the P-type channel, but has a lower sensitivity to -Aga IVA, and exhibits faster inactivation kinetics (17). Several types of these Ca 2ϩ channels are co-localized in a single neuron and are believed to contribute to fine tuning of neuronal activity, because each type of Ca 2ϩ channel is modulated in a different manner. Although the critical role of Ca 2ϩ channels, particularly the P-and N-types, for transmitter release in the synaptic terminals has been well established (18,19), the roles of Ca 2ϩ channels in integration of signals or synaptic plasticity have been poorly understood.
Recent genetic and molecular biological analyses have identified that mutations of the gene encoding the Ca 2ϩ channel ␣ 1A subunit cause cerebellar ataxia and other forms of neurological disorders. A missense mutation was found in the tottering (tg) mice, which display a delayed-onset, recessive disorder consisting of ataxia, motor seizure, and absence seizure resembling petit mal epilepsy (20). The tg mutation causes substitution of leucine for proline at a position close to the conserved porelining region (P region) in the extracellular segment in the second repeat. Mice with an allelic tottering mutation leaner (tg la ), which causes severer symptoms, were found to have a single nucleotide substitution at an exon/intron junction, which results in skipping the exon/intron, or results in failure to splice out the succeeding intron. In both cases, the tg la mutation causes truncation of the normal open reading frame and expression of aberrant C-terminal sequences. Furthermore, in the human ␣ 1A Ca 2ϩ channel gene, missense mutations were found in familial hemiplegic migraine, mutations disrupting the reading frame in episodic ataxia type-2 (21), and CAG repeat extension in autosomal dominant spinocerebellar ataxia (SCA6) (22). Here, in order to disclose the causative relationship among the tottering mutations, the affected Ca 2ϩ channel properties, and the neurological disorders, we made comprehensive comparison of the mutant Ca 2ϩ channel properties in native Purkinje cells of tottering tg and tg la mice, where many other factors can affect the channel phenotype, and in the recombinant expression system, where direct effects of the mutations can be evaluated precisely.

EXPERIMENTAL PROCEDURES
Animals-C57BL/6Jtg and C57BL/6J-Osϩ/ϩtg la strains were introduced from the Jackson Laboratory (Bar Harbor, ME) to the Eisai Tsukuba Research Laboratories. The mice were provided with a commercial diet (CE-2, Nihon Clea, Tokyo, Japan) and water ad libitum under conventional conditions with controlled temperature, humidity, and lighting. The oligosyndactylism (Os) heterozygous phenotype is a marker for the tg la heterozygote of tg.
Polymerase Chain Reaction-restriction Enzyme Fragment Length Polymorphism (PCR-RFLP)-A PCR-RFLP method was developed to distinguish between genotypes of tg and tg la mice. In the case of tg la , its genotypes can be inferred by the Os phenotype which is tightly linked with tg la . However, since recombination between tg la and Os, which leads to an error in tg la genotyping, is possible, we developed a PCR-RFLP method also for tg la genotyping. We first obtained the 5Ј-and 3Ј-flanking regions of the tg and tg la mutations using GenomeWalker TM kit for Mouse (CLONTECH, Palo Alto, CA) according to the manufacturers instruction. First PCR was done with an adapter primer 1 (5Ј-GTAATACGACTCACTATAGGGC-3Ј) and a gene-specific primer, with the parameters consisting of 7 cycles of 25 s at 94°C and 4 min at 72°C, 32 cycles of 25 s at 94°C and 4 min at 67°C with a final extension for 4 min at 67°C in MiniCycer TM (MJ Research, Watertown, MA). The resulting PCR products were amplified using a nested adapter primer 2 (5Ј-ACTATAGGGCACGCGTGGT-3Ј) and a nested gene-specific primer. The conditions of nested PCR consisted of 5 cycles of 25 s at 94°C and 4 min at 72°C, 22 cycles of 25 s at 94°C and 4 min at 67°C with an additional 4 min at 67°C. The oligonucleotides used for the cloning are summarized in Fig. 1. The nested PCR products were TA-subcloned to pT7Blue(R)T-vector (Novagen, Madison, WI). Nucleotide sequences were determined by a Dye Terminator Cycle Sequencing kit (Perkin-Elmer, Foster City, CA) using T7 and U19 primers with an ABI Prism TM 377 DNA Sequencer (Perkin-Elmer). PCR conditions for sequencing were 25 cycles of 10 s at 96°C, 5 s at 50°C, and 4 min at 60°C in GeneAmp PCR System 9600 (Perkin-Elmer). The nucleotide sequences of 1633-and 2099-bp genomic fragments including the tg and tg la mutation were determined (GenBank/EMBL accession numbers, AB011275 and AB011276, respectively).
PCR primers for amplifying genomic fragments necessary for tg and tg la genotyping were designed based on the above mentioned sequence results. Primers designed were: 5Ј-GGAAACCAGAAGCTGAACCA-3Ј (sense) and 5Ј-GAAATGGAGGAATTCAGGG-3Ј (antisense) for tg, and 5Ј-ACGAAGGCGGCATGAAGGAGA-3Ј (TGLA-1; sense) and 5Ј-TTC-CATGGGGAGGTAGTGTT-3Ј (TGLA-5R; antisense) for tg la . Genomic DNA was extracted from the tail as follows: a mouse tail tip about 2 mm in length was cut and put into a 0.5-ml tube with a safety lock. Forty l of distilled water and a drop of mineral oil (Sigma) were added. The sample was heated at 96°C for 10 min, and then treated with proteinase K at 55°C for 90 min. One l of the DNA-extracted solution was used as a template for PCR. The PCR parameters consisted of 45 cycles of 94°C for 30 s, 60°C for 30 s, and 72°C for 30 s, and a final extension of 72°C for 2 min. The resulting PCR products were digested with AciI (New England Biolabs, Beverly, MA) and BsaJI (New England Biolabs) for tg and tg la genotyping, respectively. The digested products were subjected to electrophoresis on a 2.0% agarose/Et-Br gel and typed. Digestion with the enzymes yielded the following fragments: 295 bp in tg/tg, 127 and 168 bp in ϩ/ϩ, and 127, 168, and 295 bp in tg/ϩ; 229 bp in tg la /tg la , and 91 and 138 bp in ϩ/ϩ (Fig. 1C).
Preparation of Dissociated Purkinje Cells-Purkinje cells were freshly dissociated from 18-to 30-day old mice under ether anesthesia. The procedure for obtaining dissociated cells from mice is similar to that described elsewhere (25). Coronal slices (400-m thick) of cerebellum were prepared using a microslicer (DTK-1000, Dosaka, Kyoto, Japan). After preincubation in Krebs solution for 40 min at 31°C, the slices were digested: first in Krebs solution containing 0.01% Pronase (Calbiochem-Novabiochem, La Jolla, CA) for 25 min at 31°C and then in solution containing 0.01% thermolysin (type X, Sigma) for 25 min at 31°C. The Krebs solution used for preincubation and digestion contained the following (in mM): 124 NaCl, 5 KCl, 1.2 KH 2 PO 4 , 2.4 CaCl 2 , 1.3 MgSO 4 , 26 NaHCO 3 , and 10 glucose. The solution was continuously oxygenated with 95% O 2 , 5% CO 2 . Then the brain slices were punched out and dissociated mechanically by the use of fine glass pipettes having a tip diameter of 100 -200 m in a standard perfusing solution. Dissociated cells settled on tissue culture dishes (Primaria number 3801, Nippon Becton Dickinson, Tokyo, Japan) within 30 min. Purkinje cells were identified by their large diameter and characteristic pear shape because of the stump of the apical dendrite. To make a sufficient space-clamp of the Purkinje cell body, Purkinje cells lacking most of dendrites were used throughout the present experiments.
Whole Cell Recordings-Electrophysiological measurements were performed on Purkinje cell and BHK cells. BHK cells, stably expressing wild-type or tg-␣ 1A channel, were seeded onto plastic coverslips, Celldesk, and incubated in culture medium for 5-8 days. Currents were recorded at room temperature (22-25°C) using whole cell mode of the patch-clamp technique (29) with an Axopatch 200B patch-clamp amplifier (Axon Instruments, Foster City, CA) or an EPC-7 (List Medical, Darmstadt, Germany). Patch pipettes were made from borosilicate glass capillaries (1.5 mm outer diameter, 1.1 mm inner diameter; Narishige, Tokyo, Japan) using a model P-87 Flaming-Brown micropipette puller (Sutter Instrument Co., San Rafael, CA). The patch electrodes were fire-polished. Pipette resistance ranged from 1 to 2 megohm when filled with the pipette solutions described below. The series resistance was electronically compensated to Ͼ70% and both the leakage and the remaining capacitance were subtracted by ϪP/6 method. Currents were sampled at 100 kHz after low pass filtering at 10 kHz (Ϫ3 db) using the 8-pole Bessel filter (Model 900, Frequency Devices, Haverhill, MA) in the experiments of activation kinetics, otherwise sampled at 10 kHz after low pass filtering at 2 kHz (Ϫ3 db). Data were collected and analyzed using the pCLAMP 6.02 software (Axon Instruments). Ba 2ϩ currents were recorded in an external solution that contained (in mM): 3 BaCl 2 , 155 tetraethylammonium chloride, 10 HEPES, 10 glucose (pH adjusted to 7.4 with tetraethylammonium-OH). The pipette solution contained (in mM): 85 Cs-aspartate, 40 CsCl, 2 MgCl 2 , 5 EGTA, 2 ATPMg, 5 HEPES, 10 creatine phosphate (pH adjusted to 7.4 with CsOH). In the experiments with -Aga IVA, the external solution was always supplemented with 0.1 mg/ml cytochrome c. Cytochrome c at 0.1 mg/ml had no effect on Ba 2ϩ currents. Rapid application of drugs were made by a modified "Y-tube" method. Details of this technique have already appeared (30). The external solution surrounding a cell recorded was completely exchanged within 200 ms. Statistical comparison between normal and mutant mice or mutant channels was performed by Student's t test (*, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001).
Single Channel Recordings-The cell-attached patch recording technique was used to measure single-channel currents (29). Patch electrodes were coated with silicone (KE-106, Shinetsu Silicone, Tokyo, Japan) and had resistance of 5-8 megohms. The bath solution contained 140 mM KCl, 5 mM HEPES, and 0.2 mM EGTA, pH 7.4, with KOH. The composition of the pipette solution was 110 mM BaCl 2 and 10 mM HEPES, pH adjusted to 7.4 with Ba(OH) 2 . Voltage steps with duration of 150 ms were given every 2 s from a holding potential of Ϫ100 mV. The data, low-passed filtered at 1 kHz (Ϫ3 db, 8-pole Bessel filter), were digitized at 10 kHz and analyzed using the pCLAMP 6.02 software. The records were corrected for capacitative and leakage currents by subtraction of the average of records without channel activity.

Comparison of Ca 2ϩ Channel Currents in Purkinje Cells from
Normal, tg, and tg la Mice-Mouse ␣ 1A genotypes were determined prior to electrophysiological characterization of Ca 2ϩ channel currents in Purkinje neurons. PCR and subsequent digestion of the resulting products by the restriction enzymes AciI and BsaJI, whose sites are only present in normal genomes, distinguished tottering (tg) and leaner (tg la ), respectively, from normal genotypes (Fig. 1).
Cerebellar Purkinje neurons were freshly dissociated from 18-to 30-day old normal (C57BL/6J, n ϭ 13), tg (n ϭ 9), and tg la mice (n ϭ 7). We first compared maximum amplitudes of currents evoked by step pulses from a holding potential (V h ) of Ϫ80 mV in 3 mM Ba 2ϩ solution (Fig. 2). In normal mice, maximum amplitudes varied from 1.21 to 12.54 nA, whereas they were reduced in tg mice and tg la mice; their maximum amplitudes were at most 5.39 and 4.67 nA, respectively. The mean amplitudes for tg mice (2.95 Ϯ 0.25 nA, n ϭ 27) and tg la mice (2.09 Ϯ 0.17 nA, n ϭ 25) were significantly smaller than that for normal mice (5.09 Ϯ 0.29 nA, n ϭ 67). The tg la mice grow up slowly presumably due to malnutrition, which might cause a smaller cell size, consequently resulting in the smaller amplitude. However, the cell capacitance was very similar among the three groups. The mean capacitance was 16.7 Ϯ 0.5 pF (n ϭ 67) for normal mice, 18.0 Ϯ 0.8 pF (n ϭ 27) for tg mice, and 17.4 Ϯ 0.8 pF (n ϭ 25) for tg la mice, indicating that cell bodies of cerebellar Purkinje neurons grow up similarly in regard to morphology. The current density, current amplitude divided by cell capacitance, is also an index for comparison of channel activity among the three groups. The current density for both tg mice (184.2 Ϯ 18.1 pA/pF, n ϭ 27) and tg la mice (122.8 Ϯ 9.0 pA/pF, n ϭ 25) is significantly lower than that for normal mice (333.4 Ϯ 18.1 pA/pF, n ϭ 67). During preparation of this article, reduction in Ca 2ϩ channel currents in tg la Purkinje cells was reported (31).
Do the natural mutant mice exhibit diminished P-type Ca 2ϩ channel activity? We examined this question with a P-type Ca 2ϩ channel blocker, -Aga IVA. Application of 100 M -Aga IVA reduced the current density to 25.5 Ϯ 2.5 pA/pF (n ϭ 7) in normal mice, indicating that more than 90% of high threshold current in cerebellar Purkinje cells flow through the P-type Ca 2ϩ channel. This result confirms the previous observation in cerebellar Purkinje neurons (32,33) showing that -Aga IVA FIG. 1. Detection of tg and tg la mutations in the mouse ␣ 1A subunit gene. Genomic DNA was subjected to genotype determination by PCR-RFLP method. Genotypes ϩ/ϩ (wild-type), tg/ϩ (tottering heterozygote), and tg/tg (tottering homozygote) were distinguished by cutting the PCR product with AciI (A and C), and genotypes ϩ/ϩ (wildtype), and tg la /tg la (leaner homozygote) were distinguished by cutting the PCR product with BsaJI (B and C). Exons are indicated by open boxes, and introns by thin lettered lines. The nucleotide residues substituted are underlined.
inhibits around 90% of total Ca 2ϩ current. The current density insensitive to -Aga IVA was 21.7 Ϯ 5.9 pA/pF (n ϭ 4) in tg mice and 29.8 Ϯ 2.6 pA/pF (n ϭ 12) in tg la mice. The residual current density in tg la mice was higher than that in normal mice, but the values were not different substantially among the three groups. It is concluded that current amplitude and current density of P-type Ca 2ϩ channel are reduced in tg and tg la mice. Furthermore, -Aga IVA-insensitive Ca 2ϩ channels, whose amplitude is not affected by the mutations, do not compensate cerebellar Purkinje cells for low channel activity. Fig. 3 shows Ca 2ϩ channel currents and their current-voltage (I-V) relationships in Purkinje cells from normal and mutant mice. Ba 2ϩ currents were elicited with 30-ms depolarizing pulses from a V h of Ϫ80 mV to test potentials from Ϫ50 to 50 or 60 mV with increments of 10 mV in 3 mM Ba 2ϩ solution. The current density was significantly lower at test potentials between Ϫ20 and 20 mV for tg mice and between Ϫ40 and 20 mV for tg la mice. The Ba 2ϩ currents in normal and tg mice were similar in the I-V relationship with currents first detectable at voltages near Ϫ40 mV, grew to reach the maximal amplitude around Ϫ20 mV and then declined toward a zero current asymptote with further depolarization. However, the I-V relationship for tg la mice was shifted in the depolarizing direction by about 10 mV. To describe activation parameters more accurately, we measured tail currents evoked by clamp-back to the fixed potential of Ϫ60 mV after 5-ms step depolarization from Ϫ50 to 30 mV for normal and tg mice or from Ϫ45 to 30 mV for tg la mice with 5 mV increments (Fig. 4A). Peak amplitude of tail currents, which should reflect Ca 2ϩ channel activation, could be fitted by a single Boltzmann function, where the voltages for half-maximal activation and slope factors were Ϫ28.0 Ϯ 1.1 and 4.9 Ϯ 0.5 mV (n ϭ 11) for normal mice, Ϫ28.3 Ϯ 1.1 and 4.7 Ϯ 0.3 mV (n ϭ 13) for tg mice, and Ϫ19.2 Ϯ 1.3 and 5.4 Ϯ 0.3 mV (n ϭ 13) for tg la mice, respectively ( Table  I). The activation curve for tg la mice was significantly shifted in the depolarizing direction without changing the slope factor (Fig. 4A). When we ignore the negative curvature at membrane potentials over 30 mV, apparent reversal potentials, calculated from the I-V relationships, were 37.6 Ϯ 1.3 mV (n ϭ 52) for normal mice, 35.2 Ϯ 0.7 mV (n ϭ 25) for tg mice, and 37.9 Ϯ 1.0 mV (n ϭ 20) for tg la mice, indicating that apparent permeability was not changed in mutant mice.
The voltage dependence of inactivation was measured by the use of 2-s prepulses. Peak current amplitude induced by the test pulse to Ϫ20 or Ϫ10 mV from various prepulse voltages was normalized to the amplitude induced by the test pulse from a prepulse potential of Ϫ100 mV and was plotted against the prepulse potentials (Fig. 4B). Voltage dependence of inactivating components was fitted with the Boltzmann's equation. The inactivating components induced by the 2-s voltage displacements were significantly smaller for tg mice (0.53 Ϯ 0.04, n ϭ 9) and tg la mice (0.53 Ϯ 0.03, n ϭ 9) than for normal mice (0.70 Ϯ 0.04, n ϭ 15) ( Table I). In addition, the Boltzmann's parameters for tg la mice were different from normal mice. The midpoint of inactivation curve was Ϫ23.7 Ϯ 2.0 mV with the slope factor of 10.6 Ϯ 1.6 mV (n ϭ 9) for tg la mice, Ϫ36.9 Ϯ 2.9 mV with the slope factor of 5.3 Ϯ 0.5 mV (n ϭ 15) for normal mice, and Ϫ33.9 Ϯ 2.2 mV with the slope factor of 5.4 Ϯ 0.7 mV (n ϭ 9) for tg mice.
Because mutant mice have different voltage dependence of activation, we next examined the activation kinetics. Ba 2ϩ currents were elicited by 5-ms test pulses to various potentials from Ϫ45 to 30 mV with 5 mV increments (Fig. 5A). The time course of the rising phase was well described by a single time constant. The time constant versus test potential relationship was "bell shaped" for normal mice and two mutant mice (Fig.  5B). At a test potential of Ϫ30 mV, the time constants for normal mice and tg mice reached a maximum (1.83 Ϯ 0.23 ms, n ϭ 9 for normal mice, and 2.35 Ϯ 0.37 ms, n ϭ 13 for tg mice). At test potentials positive to Ϫ30 mV, the time constant decreased with increasing test pulse voltage. In accordance with the shift of activation curve for tg la mice, voltage dependence of the time constant was shifted in the depolarizing direction by 10 mV with a maximum of 2.16 Ϯ 0.20 ms (n ϭ 13 at Ϫ20 mV). The time constants of activation phase were 0.23 Ϯ 0.02 ms (n ϭ 10), 0.28 Ϯ 0.03 ms (n ϭ 13), and 0.46 Ϯ 0.03 ms (n ϭ 13) at a test potential of 10 mV for normal mice, tg mice, and tg la mice, respectively. The activation of tg la mice was statistically slower than that of normal mice and tg mice at potentials positive to Ϫ20 mV.
It is hard to find the difference in the rate of inactivation in Purkinje cells among the three types of mice, because as shown in Fig. 3, Ba 2ϩ currents decayed little within 30-ms test pulses. However, the mutant mice have different inactivation curves from normal mice (Fig. 4B). To further examine the inactivation rate of Ca 2ϩ channel in Purkinje cells from the three types of mice, the decay phase of Ba 2ϩ currents evoked by 2-s test pulses was analyzed. The decay phase was well fitted by two exponential functions with a non-inactivating component (Fig.  6). The two exponential time constants for tg mice and tg la mice were similar to those for normal mice (about 50 ms for the fast component and about 1600 ms for the slow component) except the fast exponential time constant for tg la mice at test potentials of Ϫ10 and 0 mV. The mean values of the fast and slow  (Fig. 6C). The ratios of fast and slow inactivating components for tg mice and tg la mice were smaller than those for normal mice at all potentials between Ϫ20 and 20 mV, although the difference was not statistically significant. The non-inactivating components were significantly larger for tg mice and tg la mice at test potentials between Ϫ10 and 10 mV, being in agreement with the voltage-dependent inactivation curves in Inset shows Ba 2ϩ currents evoked by 20-ms test pulse to Ϫ20 mV after the 10-ms repolarization to Ϫ100 mV following 2-s V h displacement from Ϫ80 to 0 mV with 10-mV increments in a normal Purkinje cell. Amplitude of currents elicited by the test pulses was normalized to the current amplitude elicited by the test pulse after a 2-s V h displacement to Ϫ100 mV. The mean values were plotted against potentials of the 2-s V h displacement, and fitted to the Boltzmann's equation with a half-inactivation voltage (V 0.5 ) of Ϫ36.7 mV and a slope factor (k) of 8.2 mV for normal mice (E), a V 0.5 of Ϫ34.4 mV and a k of 6.3 mV for tg mice (⌬), and a V 0.5 of Ϫ24.7 mV and a k of 11.4 mV for tg la mice (Ⅺ). The inactivating component, induced by the 2-s V h displacement, was 70, 53, and 52% of total inward currents for normal, tg and tg la mice, respectively. Each point represents an average value of 11, 13, and 13 Purkinje neurons from normal, tg, and tg la mice in activation curves (A), and 15, 9, and 9 Purkinje neurons from normal, tg, and tg la mice, respectively, in inactivation curves (B). Vertical bars show mean Ϯ S.E. if they are larger than symbols.

Fig. 4.
Comparison of Wild-type and Mutant ␣ 1A Channels Expressed in BHK Cells-The tg mutation is a nucleotide (C to T) substitution ( Fig. 1) that replaces Pro-601 with Leu in the extracellular S5-S6 linker region in repeat II (20). The S5-S6 linkers contain the P regions that form a pore structure in Ca 2ϩ channels. On the other hand, the tg la mutation is a nucleotide (G to A) substitution (Fig. 1) at a splice donor site that would cause two aberrant splicing patterns: failure of splicing out an intron and skipping of one exon (20). Inclusion of an intron results in translation of the intron sequence and an out-offrame read through of subsequent exons, substituting an abnormal 99-amino acid sequence for the C-terminal sequence from Gln-1967 (tg la (long)). The exon skip results in an out-offrame splice, exchanging the C-terminal sequence from Met-1922 with an abnormal 57-amino acid sequence (tg la (short)). Because we did not know which is the major product in the tg la mice, we constructed both candidates, tg la (long) and tg la (short).
We attempted to establish the stable cell lines of mutant ␣ 1A channels in BHK6 cells, which stably express the ␣ 2 /␦ and ␤ 1a subunits. The stable cell line of the tg-␣ 1A channel was successfully established, however, for unknown reasons, we could not obtain cell lines stably expressing tg la mutant channels. Therefore, we transiently expressed the tg la mutant channels into BHK6 and compared their channel activities with those of the wild-type ␣ 1A channel transiently expressed in BHK6. Fig. 7A shows typical Ca 2ϩ channel currents in BHK cells, and Fig. 7B compares current density for the three mutant ␣ 1A channels with that for the wild-type ␣ 1A channel. In the stable expression system, current density ranged from 27.8 to 240.3 pA/pF with an average of 109.4 Ϯ 11.3 pF/pA (n ϭ 26) for the wildtype ␣ 1A channel, and from 5.7 to 164.2 pA/pF with an average of 48.1 Ϯ 6.1 pF/pA (n ϭ 38) for the tg-␣ 1A channel. In the transient expression system, current density was distributed from 27.9 to 130.0 pA/pF with an average of 66.3 Ϯ 7.2 pF/pA (n ϭ 20) for the wild-type ␣ 1A channel, from 17.4 to 205.0 pA/pF with an average of 83.2 Ϯ 12.6 pF/pA (n ϭ 18) for the tg la (long)-␣ 1A channel, and from 9.3 to 117.3 pA/pF with an average of 32.9 Ϯ 10.4 pF/pA (n ϭ 10) for the tg la (short)-␣ 1A channel. The activity of the tg-␣ 1A and the tg la (short)-␣ 1A channel was significantly lower than control values. On the other hand, the activity of the tg la (long)-␣ 1A channel was larger than that of the wild-type ␣ 1A channels expressed transiently, but the difference was not statistically significant.
As shown in the Fig. 7B, current densities of the wild-type ␣ 1A channel in the transient expression system were smaller than those in the stable cell line, however, kinetics of activation and inactivation and parameters of voltage dependence were similar between the two expression systems (data not shown). Therefore we used the stable expression system for more detailed analysis of the recombinant wild-type ␣ 1A channels.
Depolarizing pulses for 30 ms from a V h of Ϫ100 mV evoked inward currents that first appeared at Ϫ30 mV and grew with increments of depolarization, reached a peak in the I-V relationship around 0 mV, and then declined with progressively more depolarized voltage steps (Fig. 7C). To draw activation curves, tail currents were recorded at a potential of Ϫ50 mV following the termination of 5-ms test pulses to various potentials between Ϫ30 and 50 mV with 5-mV increments (Fig. 8A) as in Fig. 4. Tail current amplitude was normalized to the tail current amplitude following a test pulse to 50 mV and plotted against test potentials. The tail current activation shows "Sshaped" curves reaching half-maximal and maximal near Ϫ5 and 30 mV, respectively. The curves were symmetric around the half-maximal voltage and thus could be fitted to a singlecomponent Boltzmann equation, in which the voltages for halfmaximal activation and slope factors were Ϫ7.6 Ϯ 1.0 mV and 5.0 Ϯ 0.4 mV (n ϭ 8) for the wild-type ␣ 1A channel, Ϫ8.2 Ϯ 0.5 mV and 5.9 Ϯ 0.3 mV (n ϭ 6) for the tg-␣ 1A channel, Ϫ2.4 Ϯ 0.7 mV and 7.0 Ϯ 0.6 mV (n ϭ 9) for the tg la (long)-␣ 1A channel, and Ϫ7.0 Ϯ 1.4 mV and 7.0 Ϯ 0.5 mV (n ϭ 6) for the tg la (short)-␣ 1A channel, respectively (Table II). The activation curve for the tg la (long)-␣ 1A channel was shifted in the depolarizing direction by about 5 mV. In addition, the voltage dependence of the activation was slightly broader in the tg la (long)-␣ 1A and the tg la (short)-␣ 1A channels than in the wild-type ␣ 1A channel.
The voltage dependence of inactivation was determined by measuring the amplitude of the peak currents evoked by 20-ms  test pulses to 0 mV following 2-s prepulses to potentials between Ϫ110 and 10 mV and 10-ms interval at a V h of Ϫ100 mV (Fig. 8B). Peak current amplitude induced by the test pulse to 0 mV from various prepulse voltages was normalized to the peak current amplitude induced by the test pulse from a prepulse potential of Ϫ110 mV and was plotted against the prepulse potentials. Voltage dependence of inactivation was fitted with the Boltzmann's equation. The estimated half-inactivation potential and the slope factor were Ϫ52.9 Ϯ 1.4 and 8.7 Ϯ 0.2 mV (n ϭ 12) for the wild-type ␣ 1A channel, Ϫ52.4 Ϯ 2.7 and 8.3 Ϯ 0.3 mV (n ϭ 7) for the tg-␣ 1A channel, Ϫ42.2 Ϯ 1.3 and 7.6 Ϯ 0.6 mV (n ϭ 7) for the tg la (long)-␣ 1A channel, and Ϫ53.5 Ϯ 2.7 and 7.8 Ϯ 0.4 mV (n ϭ 7) for the tg la (short)-␣ 1A channel, respectively (Table II). These results indicate that the tg la (long)-␣ 1A channel suffers less voltage-dependent inactivation at holding potentials between Ϫ70 and Ϫ30 mV.
Because the activation kinetics of Ca 2ϩ channel currents in Purkinje cells of tg la mice was statistically slower than those of normal and tg Purkinje cells at potentials positive to Ϫ20 mV, we next examined the activation kinetics of the wild-type and the three mutant ␣ 1A channels in BHK cells. Ba 2ϩ currents were elicited by 5-ms test pulses to various potentials from Ϫ25 to 45 mV with 5 mV increments. Current activation was well fitted by a single exponential. The time constant versus test potential relationship showed a "bell shape" for all the channels (Fig. 9). Around the voltages for half-maximal activation, the time constant reached a maximum (2.85 Ϯ 0.25 ms, n ϭ 8 for the wild-type ␣ 1A , 2.00 Ϯ 0.16 ms, n ϭ 7 for the tg-␣ 1A , 2.19 Ϯ 0.38 ms, n ϭ 9 for the tg la (long)-␣ 1A and 1.46 Ϯ 0.18 ms, n ϭ 7 for the tg la (short)-␣ 1A channels at 10 mV). At test potentials positive to Ϫ5 mV, the time constant for the wild-type and the three mutant ␣ 1A channels decreased asymptotically toward 0.3 ms with increasing test pulse voltage. In our expression system in BHK cells, we could not see the slow activation, which was observed in Purkinje neurons dissociated from tg la mice (Fig. 5).
In contrast to the P-type Ba 2ϩ currents recorded in Purkinje neurons (Fig. 3A), substantial inactivation occurred in the ␣ 1A channel expressed in BHK cells during 30-ms step pulses to potentials positive to Ϫ10 mV (Fig. 7A). To compare inactivation kinetics of the three mutant ␣ 1A channels with that of the wild-type ␣ 1A channel, test pulses lasting 300 ms were given to FIG. 6. Voltage dependence of inactivation time constant. A, Ba 2ϩ current evoked by 2-s test pulse to 0 mV from a V h of Ϫ100 mV in a Purkinje neuron of a normal mouse. Current decay was fitted by a sum of two exponential functions with time constants of 59 and 1369 ms, whose amplitude was Ϫ0.21 and Ϫ0.63 nA, respectively. Its sustained component was Ϫ0.38 nA. B, voltage dependence of the two inactivation time constants, fast (a) and slow (b). The mean inactivation time constants were plotted as a function of test potential from Ϫ20 to 10 mV. C, voltage-dependence of the fraction of the three components, fast (a), slow inactivation (b), and sustained components (c). The fractions of the components were plotted against test potentials. Data are expressed as mean Ϯ S.E. of 15, 9, and 9 Purkinje neurons from normal (E), tg (⌬), and tg la mice (Ⅺ), respectively. Vertical bars show mean Ϯ S.E. if they are larger than symbols.

FIG. 7. Current-voltage (I-V) relationships and current density of wild-type and three mutant, tg-, tg la (long)-and tg la (short)-␣ 1A channels recombinantly expressed with ␣ 2 and ␤ subunits in BHK cells.
A, families of Ba 2ϩ currents evoked by 30-ms depolarizing pulses from Ϫ30 to 40 mV with increments of 10 mV from a V h of Ϫ100 mV. Wild-type and tg-␣ 1A channels were expressed stably in BHK6. tg la (long)-and tg la (short)-␣ 1A channels were expressed transiently in BHK6. External solution contained 3 mM Ba 2ϩ as the charge carrier. B, distribution of current density (E) and mean (open box) Ϯ S.E. The numbers of recorded BHK cells are 26, 38, 20, 18, and 10 for stable wild-type, tg-␣ 1A , transient wild-type, tg la (long)-, and tg la (short)-␣ 1A channels, respectively. C, current density-voltage relationships of wildtype (E), tg-(⌬), tg la (long)-(Ⅺ), and tg la (short)-␣ 1A channels (f). Data are expressed as mean Ϯ S.E. of 26, 38, 18, and 10 BHK cells expressing wild-type, tg-, tg la (long)-, and tg la (short)-␣ 1A channels, respectively. different voltages from a V h of Ϫ100 mV. The decay phase was well fitted by a two-exponential function with a non-inactivating component (Fig. 10). The two exponential time constants and their fractions of the tg-␣ 1A and the tg la (short)-␣ 1A channels were not significantly different from the corresponding values of the wild-type ␣ 1A channel at all test potentials. In the tg la (long)-␣ 1A channel, these values did not deviate from the control values at test potentials positive to Ϫ5 mV, whereas deviation was observed at some test potentials negative to Ϫ10 mV. This may be due to small voltage-dependent inactivation below Ϫ10 mV. The mean values of the fast and slow time constants at a test potential of 10 mV are 21.0 Ϯ 1.4 and 94.7 Ϯ 5.6 ms (n ϭ 11) for the wild-type ␣ 1A channel, 23.1 Ϯ 5.8 and 152.5 Ϯ 42.6 ms (n ϭ 6) for the tg-␣ 1A channel, and 25.2 Ϯ 3.0 and 123.5 Ϯ 13.9 ms (n ϭ 13) for the tg la (long)-␣ 1A channel, and 20.3 Ϯ 3.0 and 95.9 Ϯ 7.3 ms (n ϭ 7) for the tg la (short)-␣ 1A channel, respectively. The ratio of the three components, fast, slow, and non-inactivating components, were 0.28 Ϯ 0.04, 0.69 Ϯ 0.04, and 0.03 Ϯ 0.01 for the wild-type ␣ 1A channel, 0.28 Ϯ 0.09, 0.65 Ϯ 0.06, and 0.07 Ϯ 0.03 for the tg-␣ 1A channel, 0.24 Ϯ 0.02, 0.71 Ϯ 0.02, and 0.05 Ϯ 0.01 for the tg la (long)-␣ 1A channel, and 0.33 Ϯ 0.06, 0.63 Ϯ 0.05, and 0.04 Ϯ 0.01 for the tg la (short)-␣ 1A channel at a test potential of 10 mV, respectively (Fig. 10).
Single Channel Properties of Wild-type and Mutant ␣ 1A Channels-Because the tg mutation is close to the P region, it is possible that the tg mutation affects the single-channel properties. Therefore we compared the single-channel properties of the wild-type and the tg mutant ␣ 1A channels expressed in BHK cells. The ␣ 1A channel exhibited several conductance levels, which made it difficult to detect subtle changes in singlechannel conductance. As shown in current traces in Fig. 11A, the unitary current amplitudes of well resolved long openings were the same in the wild-type and the tg mutant channels. Fig. 11B shows the amplitude histograms constructed from the currents at a test potential of 10 mV. The single channel conductance was 12.7 Ϯ 0.7 pS (n ϭ 4) for the wild-type ␣ 1A channel and 13.8 Ϯ 0.3 pS (n ϭ 3) for the tg-␣ 1A channel. Together with the unaltered reversal potential of the macroscopic I-V relationship, we conclude that the tg mutation does not affect the channel pore properties. DISCUSSION In this study we evaluated the properties of the Ca 2ϩ channels in acutely dissociated cerebellar Purkinje cells of the normal and the tottering (tg) and leaner (tg la ) ataxic mutant mice, and compared them with the properties of the wild-type and mutated Ca 2ϩ channels recombinantly expressed in BHK cells, in order to determine which aspects of the Ca 2ϩ channel functions are affected as the primary consequences of the mutations. We used the technique of acute dissociation of the Purkinje cells, because we can obtain quantitative properties of the voltage-gated Ca 2ϩ channels most precisely with the voltageclamp method. Comparison of the native and recombinant systems was particularly useful for mutations affecting RNA splicing, like tg la , because these mutations can cause multiple gene products, resulting in complicated phenotypes in native Purkinje cells.
Functional Alterations of Mutant Ca 2ϩ Channels-Comparison of the peak current amplitudes in the Purkinje cells of the normal, tg, and tg la mice showed a significant reduction in the tg cells (ϳ60% of control) and a severer reduction in the tg la cells (ϳ40% of control). Reduction in the current amplitude was not likely due to the result of nonspecific developmental or nutritional effects, because the size of Purkinje cells of mutant mice was not different from the normal control, at least in the range of age used for measurements. Comparison of the current density thus revealed 45 and 63% reduction in the tg and tg la cells, respectively. It is interesting to note that the -Aga IVA-insensitive component at a test potential of Ϫ10 mV was not increased in the tg and tg la mutant mice, indicating that no compensatory mechanism operates to restore the reduced Ca 2ϩ influx through the P/Q channel.
Reduction of current density was also observed in mutant channels expressed in the BHK cells. Although the level of expression of stably or transiently transformed BHK cells was variable, current density of the cells expressing the Ca 2ϩ channel ␣ 1A subunit with the tg mutation or with the tg la (short) mutation was significantly smaller than the normal control. Because the tg mutation is a single nucleotide substitution, it is unlikely that the mutation affected the efficiency of transcription and translation. The cDNA constructs encoding the tg la (long) and tg la (short) mutant channels contains sequences derived from the mouse Ca 2ϩ channel ␣ 1A subunit gene. The observation that the expression level of the tg la (long) construct, which has a longer mouse genomic insert, slightly increased indicates that the insertion by itself did not affect functional expression strongly. Thus the reduced current density in BHK cells expressing the tg and tg la (short) mutants was not caused by altered efficiency of gene expression, but likely resulted from the same post-translational mechanism that im- FIG. 8. Voltage-dependence of activation and inactivation of wild-type ␣ 1A and three mutant ␣ 1A channels expressed in BHK cells. A, comparison of activation curves. Inset represents superimposed tail current elicited by repolarization to Ϫ50 mV after a 5-ms test pulse from Ϫ25 to 35 mV with 5-mV increments in wild-type ␣ 1A channel. Currents were filtered at 10 kHz and digitized at 100 kHz. Amplitude of tail currents was normalized to the tail current amplitude obtained with a test pulse to 50 mV. The mean values were plotted against test pulse potentials, and fitted to the Boltzmann's equation with a V 0.5 of Ϫ7.7 mV and a k of 5.3 mV for wild-type ␣ 1A channel (E), a V 0.5 of Ϫ8.3 mV and a k of 5.9 mV for tg-␣ 1A channel (⌬), a V 0.5 of Ϫ2.4 mV and a k of 6.5 mV for tg la (long)-␣ 1A channel (Ⅺ), and a V 0.5 of Ϫ7.0 mV and a k of 6.5 mV for tg la (short)-␣ 1A channel (f). B, comparison of inactivation curves. Inset shows Ba 2ϩ currents evoked by 20-ms test pulse to 0 mV after the 10-ms repolarization to Ϫ100 mV following 2-s V h displacement from Ϫ110 to 10 mV with 10-mV increments in a BHK cell expressing wild-type ␣ 1A channel. Amplitude of currents elicited by the test pulses was normalized to the current amplitude induced by the test pulse after a 2-s V h displacement to Ϫ110 mV. The mean values were plotted against potentials of the 2-s V h displacement, and fitted to the Boltzmann's equation with a V 0.5 of Ϫ52.9 mV and a k of 9.2 mV for wild-type ␣ 1A channel (E), a V 0.5 of Ϫ52.4 mV and a k of 9.0 mV for tg-␣ 1A channel (⌬), a V 0.5 of Ϫ42.3 mV and a k of 8.3 mV for tg la (long)-␣ 1A channel (Ⅺ), and a V 0.5 of Ϫ53.5 mV and a k of 8.9 mV for tg la (short)-␣ 1A channel (f). Each point represents an average value of 8, 6, 9, and 6 BHK cells expressing wild-type, tg-, tg la (long)-, and tg la (short)-␣ 1A channels in activation curves (A), and 12, 7, 7, and 7 BHK cells expressing wild-type, tg, tg la (long)-, and tg la (short)-␣ 1A channels, respectively, in inactivation curves (B). Vertical bars show mean Ϯ S.E. if they are larger than symbols.
pairs the channel activity in the mutant Purkinje cells.
The tg mutation is located in the extracellular region close to the pore-forming P region of the second repeat (20). Replacement of proline for leucine might conceivably cause a substantial conformational change and alter the ion-conducting pathway, because a proline residue has its side chain fixed to the main chain and stabilizes the protein structure. However, neither the apparent reversal potential nor the single-channel conductance changed in the tg mutant Ca 2ϩ channel (Figs. 3 and 11), suggesting that the reduction in current density is not due to decreased ion conductance caused through modification of the pore structure by the tg mutation.
The tg la mutations result in truncation of the normal open reading frame expression of aberrant C-terminal sequences (20). The C-terminal portion of the Ca 2ϩ channels (34,35), together with the linker portion connecting the repeats I and II (36), is involved in interaction with the ␤ subunit. Because association with the ␤ subunits affects availability of the functional Ca 2ϩ channels, judged from enhanced opening probability or dihydropyridine binding (23,(37)(38)(39), the changes in the C-terminal sequence would alter the density of functional channels. Furthermore, it is possible that altered interaction with the ␤ subunit affects the gating properties of the mutant channel. Alternatively, alteration in the C-terminal region may cause impaired incorporation into the membrane or accelerated degradation of the channel protein. The tg mutation may cause conformational changes to affect the availability of the func-tional Ca 2ϩ channel or the gating mechanism, or alternatively to modify metabolism of the channel protein.
The tg and tg la mutations also affect macroscopic gating. The Ca 2ϩ channel in Purkinje cells from tg la mice showed a shift in current-voltage relationship in the depolarizing direction by ϳ10 mV. This change can be attributable to the shifts in voltage dependence of activation and inactivation in the depolarizing direction. Of the two forms of tg la mutant channels in expressed BHK cells, only the tg la (long) form showed the similar depolarizing shifts of voltage dependence, while the voltage dependence of the tg la (short) form, whose current density was markedly reduced, remained normal. The larger slope factor of voltage dependence of inactivation in the tg la Purkinje cells (Fig. 4B) could be explained by a mixture of the tg la (long) type and the tg la (short) type channels. Taken together with the reduced current density of the tg la (short) form, these results indicate that both tg la (long) and tg la (short) forms are expressed in the tg la Purkinje cells. The Ca 2ϩ channel current in the tg Purkinje cells showed no changes in the voltage dependence in activation or inactivation, and the voltage dependence of the tg mutant channel in BHK cells was also normal.
The tg la mutation affected gating kinetics. Activation of the tg la channel in Purkinje cells was slower than the normal control. This change was not reproduced in the BHK cells; activation of the tg la (long) form was normal and that of the tg la (short) was slightly faster. The slower activation in the tg la Purkinje cells may be mainly due to the -Aga IVA-insensitive component, which shows slower activation kinetics. Other kinetic changes in the tg and tg la Purkinje cells were the increased proportion of the non-inactivating component. This altered property must be the consequence of mutations, because this result cannot be explained by contamination of the -Aga IVA-insensitive component which inactivates faster. This mutant phenotype was not reproduced in the BHK cells, primarily because of the nature of the ␤ 1 subunit we used in this study, which confers the fast inactivating character on the ␣ 1A channel. We used the ␤ 1 subunit because we used this subunit in our previous studies. On the other hand, cerebellar Purkinje cells express multiple isoforms of the ␤ subunits (40). Such a subtle difference may have caused functional differences between the Purkinje cells from ataxic mice and the BHK cells transfected with mutant ␣ 1A subunits, although involvement of other proteins is also possible.
The present data described above include the most comprehensive description of the properties of Ca 2ϩ channel current in Purkinje cells of tg and tg la mutant mice and the thorough comparison between the native and recombinant systems. The reduction of the current density and most of the changes in gating were reproduced in the recombinant system using BHK cells. These supportive results in BHK cells strongly suggest that the neuropathic phenotype of the tg and tg la Purkinje cells is the primary consequence of the mutations of the ␣ 1A Ca 2ϩ channel. Reduced Ca 2ϩ Influx as the Cause of Cerebellar Atrophy-Impaired inactivation is a frequently observed mechanism of dysfunction for other ion channels. For example, mutations of the skeletal muscle Na ϩ channel can cause slower inactivation, leading to hyperkalemic periodic paralysis (41). Based on the histological data that shrinkage and apoptosis of Purkinje cells are observed in tg and tg la mutant mice (42,43), our initial guess guided by the "Ca 2ϩ -overload hypothesis" (44) was that mutations would affect the inactivation process, leading to prolonged opening of the channel, and that chronic excess entry of Ca 2ϩ into neurons and excess activation of Ca 2ϩ -dependent intracellular signaling pathways would ultimately lead to cell death. Contrary to the prediction, the present results unambiguously demonstrate that the Ca 2ϩ conductance of the Purkinje cells of tg and tg la mutant mice is markedly reduced. Thus our observation indicates that insufficient Ca 2ϩ influx leads to shrinkage and apoptosis of the Purkinje cells and ultimately to cerebellar atrophy. Consistent with this idea is the recent report that a mutation near the P region of the first repeat of the human Ca 2ϩ ␣ 1A subunit causes autosomal dominant progressive ataxia (45). The mutation is located very close to the selectivity filter, and substitutes arginine for glycine, presumably reducing the channel conductance because of the electrorepulsive force between Ca 2ϩ and the positively charged arginine residue. This idea is borne out by the morphological studies that a decrease in the levels of intracellular free Ca 2ϩ , induced by organic Ca 2ϩ antagonists or by low extracellular K ϩ , triggers the apoptotic process, which is prevented by the application of Bay K8644, L-type Ca 2ϩ channel agonist (46,47). Furthermore, altered P-type channel activity is suggested to cause the tg-like symptoms. Lethargic (lh) mice, which have the mutation of the ␤ 4 subunit, suffer from neurological symptoms, including ataxia, absence epilepsy, and spontaneous focal motor seizure (48 -50), while histopathological changes are not observed (51). The ␤ 4 subunit is expressed most exclusively in neuronal tissues, with a high level in cerebellum (40,52,53). The affinity of ␤ 4 subunit binding to the ␣ 1A subunit interaction domain is highest among the four ␤ subunits (54). Because these four different ␤ subunits change both activation and inactivation of ␣ 1A channel differently (14), and because omission of the ␤ subunit significantly reduces the functional activity of the P/Q type channel (23), it is most likely that the Ca 2ϩ channel activity in the Purkinje cells of lh mice is decreased. Together with the Ca 2ϩ -overload hypothesis, neuronal cell death triggered by low levels of intracellular free Ca 2ϩ demonstrates that strict Ca 2ϩ homeostasis is essential for normal functions of neurons.
Molecular Mechanism of Cerebellar Dysfunction-The mechanism how reduced Ca 2ϩ conductance causes cerebellar dysfunction and atrophy is unknown at present, but we can raise several possibilities on the basis of current knowledge of functional roles of Ca 2ϩ in neurons. Cerebellar Purkinje cells have unique properties. Branches of the dendritic tree can generate spikes independently, contributing to local processing and integration of information. Because the Ca 2ϩ channel plays a critical role in generating action potentials in dendrites (55), impaired Ca 2ϩ conductance would cause failures in spike generation, and consequently in signal processing. Alternatively, the Purkinje cells in the mutant mice may not receive proper transsynaptic inputs strong enough to maintain their activity. Because the P/Q type is probably the main presynaptic Ca 2ϩ FIG. 10. Voltage dependence of inactivation time constant. A, Ba 2ϩ current evoked by 300-ms test pulse to 10 mV from a V h of Ϫ100 mV in a BHK cell expressing wild-type ␣ 1A channel. Current decay was fitted by a sum of two exponential functions with time constants of 19.9 and 92.4 ms, whose amplitude was Ϫ0.79 and Ϫ1.70 nA, respectively. Its sustained component was Ϫ0.11 nA. B, voltage dependence of the two inactivation time constants, fast (a) and slow (b). The mean inactivation time constant was plotted as a function of test potential from Ϫ20 to 40 mV. C, voltage dependence of the fraction of the three components, fast (a), slow inactivation (b), and sustained components (c). The components were plotted against test potentials. Data are expressed as mean Ϯ S.E. of 7, 6, 13, and 7 BHK cells expressing wild-type (E), tg-(‚), tg la (long)-(Ⅺ), and tg la (short)-␣ 1A channels (f), respectively. Vertical bars show mean Ϯ S.E. if they are larger than symbols.
FIG. 11. Single-channel recordings from wild-type and tg-␣ 1A channels. A, currents were recorded in cell-attached configuration from BHK cells expressing wild-type ␣ 1A channel (a and b) and tg-␣ 1A channel (c and d) using 110 mM Ba 2ϩ as charge carrier. Typical singlechannel currents elicited by 150 ms stepping to 0 (a and c) and 10 mV (b and d). Arrowheads indicate beginning and end of test depolarization. B, amplitude histogram at a test potential of 10 mV was constructed from 160 and 176 traces of wild-type ␣ 1A channel (a) and tg-␣ 1A channel (b), respectively. Histograms were fitted with three Gaussian functions. The components had means of Ϫ0.08, Ϫ0.57, and Ϫ1.12 pA for wild-type ␣ 1A channel, and Ϫ0.04, Ϫ0.59, and Ϫ1.25 pA for tg-␣ 1A channel. channel involved in neurotransmitter release from the synaptic terminals of parallel and climbing fibers (56,57), the tg and tg la mutations would impair the Ca 2ϩ influx, resulting in reduction in quantal contents.
Reduced Ca 2ϩ conductance in the Purkinje cells would exert not only immediate effects mentioned above but also long-term effects related to synaptic plasticity. Parallel fiber synapses onto Purkinje cells undergo long-term depression, when parallel fiber inputs coincide with depolarization caused by climbing fiber inputs. Because injecting a Ca 2ϩ chelator into Purkinje cells abolishes long-term depression, elevation of Ca 2ϩ is considered to be a prerequisite for induction of long-term depression (58). Reduced Ca 2ϩ conductance caused by the mutations would certainly impair long-term depression, and consequently prevent coordinated movements.