Spinocerebellar ataxia type 6 mutation alters P-type calcium channel function.

Abnormal CAG repeat expansion in the alpha1A voltage-dependent calcium channel gene is associated with spinocerebellar ataxia type 6, an autosomal dominant cerebellar ataxia with a predominant loss of the Purkinje cell. A reverse transcriptase-polymerase chain reaction analysis of mRNA from mouse Purkinje cells revealed a predominant expression of the alpha1A channel lacking an asparagine-proline (NP) stretch in the domain IV (alpha1A(-NP)). Human alpha1A channels carrying various polyglutamine length with or without NP were expressed in HEK293 cells, and channel properties were compared using a whole-cell voltage clamp technique. alpha1A(-NP), corresponding to P-type channel, with 24 and 28 polyglutamines found in patients showed the voltage dependence of inactivation shifting negatively by 6 and 11 mV, respectively, from the 13 polyglutamine control. Contrarily, the alpha1A channel with NP (alpha1A(+NP)), corresponding to Q-type channel, with 28 polyglutamines exhibited a positive shift of 5 mV. These results suggest that altered function of alpha1A(-NP) may contribute to degeneration of Purkinje cells, which express predominantly alpha1A(-NP), due to the reduced Ca(2+) influx resulting from the negative shift of voltage-dependent inactivation. On the other hand, other types of neurons, expressing both alpha1A(-NP) and alpha1A(+NP), may survive because the positive shift of voltage-dependent inactivation of alpha1A(+NP) compensates Ca(2+) influx.

Spinocerebellar ataxia type 6 (SCA6) 1 is one of the autosomal dominant neurodegenerative diseases, which is characterized by late-onset slow-progressive cerebellar ataxia and Purkinje cell predominant degeneration in the cerebellum (1, 2). Zhuchenko et al. (3,4) demonstrated abnormal CAG repeat expansion in the coding region of the ␣1A voltage-dependent calcium channel gene (CACNA1A), which maps to chromosome 19p13.1. All the polyglutamine diseases identified so far are progressive neurodegenerative disorders such as spinal and bulbar muscular atrophy (5), Huntington's disease (6), SCA1 (7), dentatorubral-pallidoluysian atrophy (8,9), SCA3/ Machado-Joseph disease (10), SCA2 (11)(12)(13), and SCA7 (14). Although their clinical and pathological features are widely diverse, a common mechanism is presumed to underlie the pathogenesis in which some cytological abnormalities such as nuclear inclusion are claimed to induce cytotoxicity. On the other hand, several unique characteristics have been demonstrated in SCA6 compared with other polyglutamine diseases. The CAG repeat expansion is smaller (4 -20 in normal alleles and 21-33 in mutated alleles), the repeat is more stable, and the anticipation is much milder than other polyglutamine diseases (3,4,15). Furthermore, other mutations in CACNA1A are known to be associated with both human and mouse diseases such as familial hemiplegic migraine (FHM), episodic ataxia type 2 (EA-2) (16), tottering and leaner mice (17). These diseases often have a progressive ataxia like SCA6. Considering these together with the fact that the CACNA1A encodes an ␣1A subunit of the P/Q (P and/or Q)-type calcium channel which plays a crucial role in the brain function, especially for the Purkinje cells (18 -20), we hypothesized that, unlike the other polyglutamine diseases, functional alterations of the ␣1A calcium channel are causally related to pathophysiology of SCA6.
An alteration of the P/Q (P and/or Q)-type calcium channel property by the polyglutamine stretches in the rabbit ␣1A channel expressed in baby hamster kidney (BHK) cells was reported for the channels with 30 or 40 polyglutamines but not with 24 polyglutamines (21). However, it is still obscure whether the observed change reflects functional abnormalities in SCA6 brains, since the authors did not observe the alteration for the channel with 24 polyglutamines, which falls exactly within the range in SCA6 patients. These observations and our hypothesis prompted us to explore the conditions that ensure matching the functional alteration to the disease without such discrepancy. Furthermore, the presence or absence of an NP insertion in the domain IV S3-S4 in the ␣1A channels is recently postulated as the essential determinant of the P/Q chan-nel subtype distinction (22). In SCA6, degeneration occurs dominantly in Purkinje cells, which possess mostly the P-type channel, whereas other neurons such as granule cells, which express Q-type together with P-type, are basically preserved. Therefore, we designed experimental conditions in which the human full-length ␣1A channels are expressed in human cells to allow an in situ operation of mutated channel functions, and we compare their properties between channels with or without the NP insertion in order to relate the channel subtypes to the altered mechanism. Here, we report that a moderate elongation caused a negative shift of voltage dependence of inactivation in the presumed P-type channel, whereas the same elongation caused a positive shift in the presumed Q-type channel. These may explain a plausible mechanism of the selective Purkinje cell degeneration.

Sequence Analysis
Total RNAs were obtained separately from frozen cerebellar tissues of control and a SCA6 cases. RT-PCR of CACNA1A mRNA was performed as described previously (23). The nucleotide sequences of a coding region of CACNA1A cDNA from a control and a patient with SCA6 were analyzed by directly sequencing the PCR products using ABI 310 Prism Sequencing Analysis (PE Applied Biosystems).

RT-PCR
Single Purkinje cells from mouse cerebella were obtained under microscopic observation using a glass micropipette and were directly subjected to reverse transcription in a tube containing random hexamer (Perkin Elmer, Pomona, CA) and First Strand Buffer (Life Technologies, Inc.). Total RNA was obtained from mouse cerebella. Young male C57BL/6N (Clea Japan Inc., Japan) mice (20 days old) were used. Primers for amplifying the region spanning the AATCCG insertion site were 5Ј-GGAATGTGTGCTGAAAGCCATGG-3Ј for a sense primer and 5Ј-(6-FAM)-TGGTGTAACCCTGACGGAGAAGT-3Ј for an antisense primer. Quantification was performed using ABI 310 Prism Gene Scan (PE Applied Biosystems) as described (23).
BI 11E-S13 and BI 11E-L24 -In these rabbit-human chimeric ␣1A cDNA, the 2-kb 3Ј region of the rabbit ␣1A cDNA (BI-1) (24) was substituted with human ␣1A 3Ј-terminal region (1.7 kb) flanking the CAG repeat. The 3Ј region of the human cDNA was amplified from control human cerebellar cDNA and subcloned into pCRII. The two primers were 5Ј-TGGTCACACCTCACAAGTCCACGGA-3Ј for a sense primer, which corresponded to nucleotides 5863-5887 in the CACNA1A cDNA, and 5Ј-GCTCTAGATTAGCACCAATCATCGTCAC-3Ј for an antisense primer, which corresponded to nucleotides 7576 -7595 in the CACNA1A cDNA. The antisense primer contained an XbaI site. Amplification with the primer pair yielded an approximately 1.7-kb fragment that contained a BglII site in nucleotides 5902-5907. The BglII and XbaI fragment of the 3Ј region of CACNA1A cDNA was finally subcloned into the BglII and XbaI sites of BI11E vector (rabbit BI-1 in pcDNA I/Amp) to yield BI 11E-S13. To generate the construct BI 11E-L24, the KpnI fragment flanking the CAG repeats was replaced with the fragment amplified from genomic DNA obtained from an SCA6 patient harboring 24 repeats.
PKCR␣2 and PKCR ␤1a that encode ␣2/␦ and ␤1a subunits, respectively, were described previously (25). pCAGS65A encoding a mutated form of green fluorescence protein (GFP), in which serine 65 has been changed to an alanine and a 20-amino acid transmembrane region from GAP-43 has been inserted to the N-terminal region, was a gift from Dr. C. Akazawa.

Cell Cultures and Transfections
Human embryonic kidney (HEK) 293 cells were obtained from the American Type Culture Collection (CRL-1573) and were grown at 34°C in a humidified atmosphere containing 5% CO 2 . The culture medium contained 90% Dulbecco's modified Eagle's medium/F-12 (Life Technologies, Inc.), 10% fetal bovine serum (Life Technologies, Inc.), 50 g/ml gentamycin (Life Technologies, Inc.). Every 2-3 days, the cells were briefly trypsinized and replated. Native HEK293 cells lack endogenous Ca 2ϩ channel activities. One day prior to transfection, cells were replated onto poly-D-lysine-coated glass coverslips in a 35-mm culture dish. The calcium phosphate precipitation method (Mammalian Transfection Kit; Stratagene, La Jolla, CA) was used to transfect the seeded cells with expression plasmids (1.5 g of plasmid DNA per dish) encoding the ␣1A subunit, in a combination with or without ␣2/␦ and ␤1a subunits. The transfection mixture also included pCAGS65A or pEGFP-C2 (CLONTECH, Palo Alto, CA) encoding GFP to discriminate successfully transfected cells. Twenty four to 72 h after transfection, cells were used for the electrophysiological analysis.

Electrophysiology
Patch pipettes were pulled from borosilicate micropipettes (Narishige, GD-1.5) and were filled with a solution containing (in mM) 140 CsCl, 2 MgCl 2 , 10 EGTA, 10 HEPES, 3 ATP-Mg. pH was adjusted to 7.4 with CsOH. Aliquots of pipette solution were stored at Ϫ80°C and kept on ice after thawing. Pipettes were coated with paraffin to reduce capacitance and then fire-polished. Pipette resistance ranged from 1.5 to 3 M⍀ when filled with the pipette solution.
Calcium currents were recorded at room temperature (23-27°C) using a whole-cell mode of the patch clamp recording (26) with an amplifier (EPC-9, HEKA, Germany). The holding potential was Ϫ80 mV unless otherwise stated. The series resistance was electronically compensated by 30 -80%. All illustrated and analyzed currents have been corrected for remaining capacitance and leakage currents using the ϪP/4 method. Data were filtered at 3 kHz (4-pole Bessel filter) and sampled at 10 kHz. Software (PulseϩPulseFit 8.09, HEKA) was used for data acquisition and analysis. Statistical analysis was performed using Student's t test.

RT-PCR Analysis of NP Insertion in ␣1A
Channel-Two splice variants, with or without insertion of an asparagineproline (NP) stretch in the domain IV, are known in rat ␣1A channel (22). In situ hybridization and the pharmacological studies suggested that native P-type channels lack NP and that Q-type channels contain these residues (22).
To investigate the difference in the levels of mRNAs spanning the AATCCG insertion site, we conducted RT-PCR analysis of mouse cerebella and Purkinje cells. In a whole mouse cerebellum, the amount of RT-PCR product of ␣1A channel without NP (␣1A(ϪNP)) was not more than twice as large as that of ␣1A channel with NP (␣1A(ϩNP)) (Fig. 1A). On the other hand, mRNA for ␣1A(ϪNP) was predominant in single mouse Purkinje cells (Fig. 1B). These results suggest that, like the rat ␣1A channel, ␣1A(ϪNP) generally corresponds to the P-type, whereas ␣1A(ϩNP) to the Q-type channels, respectively.
Elecrophysiological Study of Rabbit-Human Chimeric ␣1A Channel in HEK293 Cells-To investigate the effect of polyglutamine expansion on the function of ␣1A voltage-dependent calcium channel, we first made an electrophysiological analysis using rabbit-human chimeric ␣1A, BI11E-S13 and BI11E-L24. These chimeric constructs carry cDNAs for rabbit ␣1A subunit, which has the C-terminal tail, replaced with the tail region of human ␣1A containing 13 or 24 polyglutamines. Since Purkinje cells were suggested to express predominantly ␣1A channel without NP as mentioned above, we used rabbit ␣1A which lacked the NP insertion.
BI11E-S13 and BI11E-L24 were expressed with ␣2/␦ and ␤1a subunits in HEK293 cells, and their electrophysiological properties were studied using a whole-cell patch clamp technique with the same protocol as described below. No statistically significant differences were detected between BI11E-S13 and BI11E-L24 in the current density, activation and inactivation kinetics, and voltage dependence of activation and inactivation (data not shown).
Sequencing Analysis of Human Full-length ␣1A Calcium Channel Subunit-Because we did not observe any noticeable functional alteration for rabbit-human chimeric ␣1A channels as described above, we speculated that residual regions of the human ␣1A channel might be needed for alteration of the channel function by polyglutamine expansion. In order to examine this possibility, we directly sequenced the amplified fragments of the full-length CACNA1A cDNA from a control and a SCA6 cerebella. The sequence of human ␣1A channel was, indeed, different from BI in several regions. Therefore, we constructed full-length CACNA1A with various CAG repeats using the control clone. The amino acid sequence deduced from the cDNA in an SCA6 cerebellum was essentially the same as that in a control cerebellum, although there were several polymorphic alterations between them. The nucleotide sequence data reported in this paper will appear in the DDBJ/EMBL/ GenBank nucleotide sequence data bases with the accession numbers AB035727 (control) and AB035726 (patient).
Effect of Polyglutamine Expansion on Human ␣1A(ϪNP)-We compared the properties of Ca 2ϩ current in HEK293 cells transiently expressing human ␣1A(ϪNP) of various numbers of polyglutamine (HS13(ϪNP), HL24(ϪNP), and HL28(ϪNP)), which was suggested to be dominantly expressed in Purkinje cells. Whole-cell Ca 2ϩ currents recorded from a HEK293 cell expressing ␣1A(ϪNP) coexpressed with ␣2/␦ and ␤1a subunits are shown in Fig. 2A. Calcium currents were elicited by 400-ms depolarizing test pulses from a holding potential of Ϫ80 mV to test potentials (Ϫ30 to ϩ50 mV) in 15 mM Ca 2ϩ solution. Inward currents of all the three channels were first activated at the threshold potentials of approximately Ϫ20 mV and gradually developed to a peak amplitude at approximately 20 mV. To compare the activation and inactivation kinetics of these currents, the time courses of activation and inactivation were quantified. The activating phase of the currents was fitted to a single exponential function, and the obtained time constants for activation at 20 mV are shown in Table I. There is no significant difference in time constants among these channels. Then, the decaying phase of the currents was fitted to a double exponential function, and the time constants and their fractions of the components at 20 mV are also shown in Table I. In HL28(ϪNP), the fraction of the fast inactivating component increased significantly, and oppositely the fraction of the slow inactivating component decreased compared with those of HS13(ϪNP). However, no difference was observed between HL24(ϪNP) and HS13(ϪNP). As for the current density of these channels, there was no noticeable difference among the three types (data not shown).
The voltage dependence of activation for the expressed Ca 2ϩ channels was then compared (Fig. 3A and Table II). Conductance was calculated from the peak current (see legend to Fig.  3). The normalized values of conductance were then fitted to a single Boltzmann function. The half-maximal voltage of activation (V1 ⁄2 ) and the slope factor (k) revealed no difference among ␣1A(ϪNP) of various numbers of polyglutamine stretch.
To examine the voltage dependence of inactivation of the Ca 2ϩ channels, a two-pulse protocol was used. A long prepulse (10 s) to different voltages (from Ϫ120 to ϩ40 mV with an increment of 20 mV) was applied to allow inactivation to reach a steady state, then a test pulse (20 or 30 mV, the voltage which yielded a peak amplitude) followed. The values of normalized current amplitude were plotted against holding potentials (Fig.  3B). Compared with HS13(ϪNP), the half-maximal voltage for the voltage-dependent inactivation of HL24(ϪNP) and HL28(ϪNP) shifted to a significantly more negative potential by 6 and 11 mV, respectively. As for the slope factor, the value was significantly smaller in HL24(ϪNP) than in HS13(ϪNP).
Effect of Polyglutamine Expansion on Human ␣1A(ϩNP)-We analyzed the properties of Ca 2ϩ current in HEK293 cells transiently expressing human ␣1A(ϩNP) of different numbers of polyglutamine, HS13(ϩNP) and HL28(ϩNP). Whole-cell Ca 2ϩ currents of ␣1A(ϩNP) were recorded with the same protocol as that for ␣1A(ϪNP) (Fig. 2B). The activation and inactivation kinetics of these currents were not different from those of HS13(ϪNP) ( Table  I). The peak current densities of these channels were not significantly different either (data not shown).
Voltage dependence of activation for ␣1A(ϩNP) was illustrated  ; open triangle (n ϭ 8)) and were plotted against test potentials. Continuous curves were obtained by fitting data with a Boltzmann function. B, voltage dependence of inactivation for Ca 2ϩ currents. Currents elicited by a two-pulse protocol: a conditioning prepulse with 10 s duration to the voltage ranging from Ϫ120 to 40 mV with a 20-mV increment was delivered before the test pulse (50 ms) which was set to the voltage yielding a peak amplitude (20 or 30 mV). The values of current amplitude normalized to the maximum amplitude in the series were plotted against potentials of the conditioning prepulse. Means and S.E.s were obtained from 7, 13, and 6 cells of HS13(ϪNP), HL24(ϪNP), and HL28(ϪNP), respectively.

TABLE II Parameters of voltage dependence of activation and inactivation of ␣1A channels in HEK293 cells
Values are means Ϯ S.E. n, number of cells recorded; V1 ⁄2 , half-maximal voltage of activation and inactivation; k, slope factor. Data were fitted with a single Boltzmann function: for inactivation, where G Ca /G max is the relative conductance, I Ca /I max the relative current, and V m , the membrane potential. Statistical analysis was performed using Student's t test. in Fig. 4A. The slope factors of HS13(ϩNP) and HL28(ϩNP) were significantly smaller than that of HS13(ϪNP) ( Table II).
Comparing voltage dependence of inactivation between ␣1A channels with and without NP, there was no difference in the half-maximal voltage of inactivation between HS13(ϩNP) and HS13(ϪNP) (Fig. 4B and Table II). However, this value for HL28(ϩNP) shifted to significantly more positive potential by 5 mV from that of HS13(ϩNP), which was in opposite direction to the shift for HL24(ϪNP) and HL28(ϪNP) from HS13(ϪNP).

DISCUSSION
In the present study, we employed a human cell (HEK293) expression system to study the effect of polyglutamine expansion on calcium channel properties of the rabbit-human chimeric ␣1A channel and the full-length human ␣1A channel. First, rabbit-human chimeric ␣1A channel with 13 or 24 polyglutamines was expressed to conduct an electrophysiological study, but no significant differences in channel properties were detected. These findings are consistent with the previous report (21), which showed that rabbit ␣1A with 24 polyglutamine expressed in BHK cells had a normal channel function. Thus, we speculated that regions other than 1.7-kb 3Ј tail were needed to yield a functional difference due to expansion of the CAG repeat. It is well known that two molecular forms with or without an insertion of NP in the domain IV S3-S4 region are expressed in rat ␣1A channel (22). In situ hybridization to rat brain sections revealed a predominant expression of the variant without NP in the Purkinje cell (22). This gives a good agreement with our data of RT-PCR showing a predominance of the ␣1A(ϪNP) in the mouse Purkinje cell. Moreover, a pharmacological study determined by -AgaIVA sensitivity in rat suggested that native P-type channels contain ␣1A subunit lacking NP (22). Therefore, it is most likely that ␣1A(ϪNP) is the main component in Purkinje cells and also in human. To study the mechanism of the dominant degeneration of Purkinje cells expressing this form of ␣1A channel in SCA6, we evaluated the properties of the wild-type and mutated ␣1A channels with or without NP expressed in HEK293 cells.
In ␣1A(ϪNP), the most striking change caused by abnormal expansions of the polyglutamine was hyperpolarizing shift of the half-maximal voltage for the voltage-dependent inactivation in proportion to the increase in the length of polyglutamine. Furthermore, ␣1A(ϪNP) with 24 polyglutamines had a steep slope factor, consequently its voltage-dependent inactivation stays at a similar level to that with 13 polyglutamines at hyperpolarized holding potentials, whereas it gets closer to that with 28 polyglutamine at more depolarized holding potentials. Thus ␣1A(ϪNP) with 24 polyglutamines is characterized as an intermediate between those with 13 and 28 polyglutamines. As a result of the shift to the hyperpolarizing direction in the voltage-dependent inactivation, available Ca 2ϩ channels near the resting membrane potential are predicted to be reduced in the cells expressing ␣1A(ϪNP) harboring extended polyglutamine. The more the polyglutamine expands, the less Ca 2ϩ influx is induced in the cells. Our PCR analysis proved that Purkinje cells predominantly express ␣1A(ϪNP). Therefore, we would present a hypothesis that the alteration of inactivation in ␣1A(ϪNP) with an abnormal expansion of polyglutamine, inducing the reduction of Ca 2ϩ influx, finally contributes to Purkinje cell death, although there is no molecular linkage suggested between the reduced Ca 2ϩ influx and the degeneration at present. This would also suggest that other cell types expressing both forms of polyglutamine-stretched ␣1A channels, i.e. ␣1A(ϪNP) and ␣1A(ϩNP), can be rescued by compensated Ca 2ϩ influx due to the opposite direction of channel dysfunction of ␣1A(ϩNP). In other words, Purkinje cells may be selectively damaged because of the lack of this compensation.
Another alteration observed in this study was a slight increase of the fast inactivation component without a change in inactivating time constant for ␣1A(ϪNP) with 28 polyglutamines. Although this finding may also contribute to the reduction of Ca 2ϩ influx, we cannot conclude that it is associated with the pathogenesis of SCA6 because ␣1A(ϪNP) with 24 polyglutamines did not show a similar tendency.
Among the wide variety of degenerative diseases, SCA6 has valuable features. First, in SCA6, Purkinje cell degeneration is predominant. Therefore, it will be a suitable model for pure cerebellar ataxia from a clinical point of view. Second, SCA6 is one of the calcium channelopathies. In addition to SCA6, the ␣1A channel is linked to tottering (tg) and leaner (tg la ) mice. The tg la mice have a severe progressive ataxia, and this mutated channel was reported to exhibit a considerably reduced Ca 2ϩ influx into Purkinje cells (27)(28)(29), whereas tg, which expresses mild phenotype, showed a smaller functional alteration (29). Accordingly, the reduced Ca 2ϩ influx by these mutations may account for their ataxic phenotype. On the other hand, although shrinkage and apoptosis of Purkinje cells were revealed in tg and tg la mutant mice (30 -32), the degeneration was not strictly specific to the Purkinje cell. Therefore, the selective degeneration of the Purkinje cell, which is seen in SCA6, cannot be explained by the reduced Ca 2ϩ influx alone. Cytoplasmic aggregates were observed in more than half of the remaining Purkinje cells (23) and may participate in the cell death, but this mechanism still remains to be confirmed. Furthermore, other mutations in the ␣1A channel are associated with human diseases of FHM and EA-2 (16). FHM is caused by four independent missense mutations. Although alterations of channel function by these mutations were indicated (33,34), no agreeable direction of functional changes has been shown. We have demonstrated in the present study a consistent direction of functional alteration of ␣1A channel in proportion to the increase of the polyglutamine in SCA6. Since the expansion is inversely correlated with the age of onset (4,15), the extent of possible reduction in Ca 2ϩ influx may account for the seriousness of SCA6 phenotype. Recently, the methodology of investigating Ca 2ϩ channel functions being considerably refined and means to modulate them being increasingly established, an insight into their dysfunction will potentially lead to the treatment of this disease in the near future. Finally, SCA6 is a member of the polyglutamine disease, which shares several common features. Understanding the pathogenesis of SCA6 as one of polyglutamine disease is very important for elucidation of the common mechanism of these neurodegenerative diseases. In this respect, we would emphasize that both the molecular biological approaches to polyglutamine and the physiological analysis of channel function are essential for clarifying the pathogenetic mechanism of this disease.