Novel Cav2.1 splice variants isolated from Purkinje cells do not generate P-type Ca2+ current.

The alpha(1)2.1 (alpha(1A)) subunits of P-type and Q-type Ca(2+) channels are encoded by a single gene, Cacna1a. Although these channels differ in the inactivation kinetics and sensitivity to omega-agatoxin IVA, the mechanism underlying these differences remains to be clarified. Alternative splicings of the Cacna1a transcript have been postulated to contribute to the respective properties, however, the splice variants responsible for P-type Ca(2+) channels have not been identified. To explore P-type-specific splice variants, we aimed at cloning alpha(1)2.1 from isolated mouse Purkinje cells using single-cell reverse transcription-PCR, because in Purkinje cells P-type currents dominate over the whole currents (>95%) with Q-type currents undetected. As a result, two novel splice variants were cloned. Compared with the previously cloned mouse alpha(1)2.1, two novel variants had additional 48 amino acids at the amino termini, six single amino acid changes, and splicing variations at the exon 46/47 boundary, which produced different carboxyl termini. Furthermore, one variant had one RNA editing site. However, electrophysiological and pharmacological studies indicated that these variants did not generate P-type current in cultured cells. These results suggest that P-type-specific splice variants may exist but that post-translational processing or modification by uncharacterized interacting proteins is also required for generating the P-type current.

the ␣ 1 subunits. Electrophysiologically and pharmacologically, VDCCs are divided into six types (L, N, P, Q, R, and T), and ten genetically different cDNAs, which encode the ␣ 1 subunit, have been identified. They are grouped into three families based on the similarities of deduced amino acid sequences (3). The Ca v 1 family (Ca v 1.1 through Ca v 1.4) includes channels containing ␣ 1S , ␣ 1C , ␣ 1D , and ␣ 1F , which constitute L-type Ca 2ϩ channels. The Ca v 2 family (Ca v 2.1 through Ca v 2.3) includes channels containing ␣ 1A , ␣ 1B , and ␣ 1E , which constitute P/Q-type, Ntype, and R-type Ca 2ϩ channels, respectively. The Ca v 3 family (Ca v 3.1 through Ca v 3.3) includes channels containing ␣ 1G , ␣ 1H , and ␣ 1I , which mediate T-type Ca 2ϩ currents. P/Q-type Ca 2ϩ channels are expressed mainly in the central nervous system and contribute to neurotransmitter release (4 -6). P-type Ca 2ϩ channels were originally identified in cerebellar Purkinje cells (7), and Q-type Ca 2ϩ channels were first described in cerebellar granule cells (8). Native P-and Q-type Ca 2ϩ channels differ in inactivation kinetics and sensitivity to -agatoxin IVA (-Aga IVA) (8,9). Recently, much attention has been paid to P/Q-type Ca 2ϩ channels, because mutations in the ␣ 1 2.1 subunit were reported to cause several neurological disorders such as familial hemiplegic migraine, episodic ataxia type 2, and spinocerebellar ataxia type 6 (SCA6) (10,11). Clinically, SCA6 is characterized by pure progressive cerebellar ataxia (12). Pathologically, SCA6 is marked by the severe loss of Purkinje cells, where only P-type Ca 2ϩ channels are expressed, and relatively intact cerebellar granule cells, where both P-and Q-type Ca 2ϩ channels are expressed (13,14). Although the relationship between P/Q-type Ca 2ϩ channels' function and the pathophysiology underlying SCA6 has been extensively studied (15)(16)(17), the mechanisms relating Ca 2ϩ channels' function to SCA6 are still unclear. To determine the relationship, it is important to elucidate the mechanism for the generation of Pand Q-type currents.
It has been hypothesized that the differences in the properties between P-and Q-type Ca 2ϩ channels originate from alternative splicings of the pre-mRNA encoding ␣ 1 2.1 subunit (1,18). In fact, the Ca v 2.1 gene encodes both P-and Q-type Ca 2ϩ channels. This was confirmed by two recent findings: 1) antisense oligonucleotide blocked both P-and Q-type currents (19,20); 2) both P-and Q-type currents were eliminated in ␣ 1 2.1deficient mice (21). Many Ca v 2.1 splice variants have been cloned by screening of mammalian cDNA libraries (11,16,18,22). However, these variants have never shown native P-typelike currents when expressed in cultured cells (16, 18, 22, 24 -26). One of the reasons why the possible P-type-specific splice variants have not been cloned may be that the amount of P-type splice variants is quite small in the cDNA libraries. It is difficult to obtain cDNAs from specific cells by screening of a conventional cDNA library, because it is made up of cDNAs * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AB066608 and AB066609.
ʈ To whom correspondence should be addressed: Tel.: 81-3-5803-5167; Fax: 81-3-5803-0122; E-mail: t-tanabe.mphm@tmd.ac.jp. 1 The abbreviations used are: VDCCs, voltage-dependent Ca 2ϩ channels; -Aga IVA, -agatoxin IVA; SCA6, spinocerebellar ataxia type 6; RT, reverse transcription; ACSF, artificial cerebrospinal fluid; 3Ј-RACE, 3Ј rapid amplification of cDNA end; nt, nucleotide(s); HEK, human embryonic kidney. derived from heterogeneous cells. Recently single-cell reverse transcription-polymerase chain reaction (RT-PCR) was developed to investigate the expression of genes in a specific cell and has proven to be a useful tool for the identification of ion channel subunits (27,28). In this study, we applied the singlecell RT-PCR to cloning of ␣ 1 2.1 subunit from isolated mouse single Purkinje cells, where almost all the Ca 2ϩ currents are P-type and no Q-type currents are recorded.

EXPERIMENTAL PROCEDURES
Preparation of RNA-C57BL/6N mice were anesthetized with methoxyflurane and decapitated, and the brains were quickly removed. Then cerebral cortices and cerebella were dissected. Total RNA was obtained from mouse cerebral cortices and cerebella using the standard acid guanidinium thiocyanate-phenol-chloroform extraction method (29).
Single Cell RT-PCR and Sequence Analysis-To obtain a single Purkinje cell or granule cells, a mouse cerebellum was cut into 300-m slices in artificial cerebrospinal fluid (ACSF) saturated with 95% O 2 -5% CO 2 , and then the slices were incubated for 2 h at room temperature in the saturated ACSF to allow recovery from the injury due to slicing. The ACSF consisted of (in millimolar) 137 NaCl, 2.5 KCl, 0.58 NaH 2 PO 4 , 1.2 MgCl 2 , 2.5 CaCl 2 , 21 NaHCO 3 , and 10 glucose. A single Purkinje cell was identified morphologically and then aspirated into a glass micropipette by applying negative pressure. Granule cells were also collected in a group using the same procedure. Aspirated single Purkinje cells or granule cells were directly expelled into thin-walled plastic tubes (PerkinElmer Life Sciences, Norwalk, CT) containing 5ϫ First Strand buffer (4 l), dNTP mixture (4 l, 2.5 mM), RNase inhibitor (0.5 l, 28,000 units/ml), dithiothreitol (2 l, 0.1 M), and random hexamer (0.5 l, 5 ng/l) (PerkinElmer Life Sciences, Pomona, CA). The reaction mixture was incubated at 70°C for 10 min and then placed on ice. SuperScript II reverse transcriptase (1 l, 200 units/l) (Invitrogen, Gaithersburg, MD) was added to the mixture, and then the reaction mixture was incubated at 25°C for 10 min, 42°C for 50 min, and 70°C for 15 min. After the treatment, 1 l of RNase H was added, and the sample was incubated at 37°C for 15 min.
Primers were designed in reference to the sequence of the previously reported mouse ␣ 1 2.1 cDNA (30) and of the 5Ј-upstream region of mouse Ca v 2.1 gene (31) (Table I). To discriminate PCR products derived from the genomic DNA, all the primer pairs were designed so that each primer was located in different exons. Nested PCR was carried out to FIG. 1. Novel splice variants of mouse ␣ 1 2.1 subunit from Purkinje cells. A, single amino acid substitution at six sites and one RNA editing site are indicated as black dots along the proposed topography of mouse ␣ 1 2.1 subunit. These substitutions, including the RNA editing site, are compared among the sequences of previously reported mouse (30), rat (23), rabbit (22), and human (10, 11, 16) ␣ 1 2.1 subunits. The slashes at amino acid positions 886 and 1085 indicate that corresponding amino acid sequences were not determined, because of the diverged sequences among mouse and other species. B, different exon 46/47 boundaries are shown for MPI and MPII. MPI contains a 5-bp (GGCAG) insertion prior to the TAG stop codon, and MPII lacks this TAG stop codon. This 8-bp difference leads to a translational frameshift in the exon 47. The exon numbers used in this figure are deduced from those of the human ␣ 1 2.1 subunit gene (10), because the mouse ␣ 1 2.1 gene structure has not been clarified yet. C, deduced carboxyl-terminal amino acid sequences of MPI and MPII. Identical amino acids between the two variants before exon 47 are shaded. RNA editing causes a single amino acid substitution from arginine (R) to tryptophan (W) at amino acid position 2102 in MPI. The coding region of MPc terminates at the position indicated by an arrowhead. Asterisks indicate stop codons. MPII has a terminal DDWC-COOH motif.
amplify ␣ 1 2.1 cDNA from single Purkinje cells. Single-cell cDNA (2.5 l) was used as a template for the first PCR amplification. The reaction mixture contained: PCR-grade H 2 O (7.5 l), 5ϫ Advantage-GC PCR Buffer (5 l), dNTP mixture (2.5 l, 2.5 mM), GC Melt (2.5 l, 5 M), each primer (2 l, 2.5 M), and 50ϫ Advantage-GC cDNA polymerase mix (1 l) (CLONTECH, Palo Alto, CA). The thermal cycling program for the first PCR was 35 cycles of 96°C for 4 min, 60°C (56°C for some primers) for 1 min, and 72°C for 3 min. The first PCR product (2.5 l) was used as a template for the second PCR. The reaction mixture contained: PCR-grade H 2 O (10 l), 10ϫ Cloned Pfu DNA polymerase reaction buffer (2.5 l), dimethyl sulfoxide (2.5 l), dNTP mixture (2.5 l, 2.5 mM), each primer (2 l, 2.5 M), and PfuTurbo DNA polymerase (1 l, 2.5 units/l) (Stratagene, La Jolla, CA). The second PCR program was the same as the first except that the number of PCR cycles was 20 or 25. To obtain the 3Ј-downstream region, 3Ј rapid amplification of cDNA end (3Ј-RACE) was performed initially with 1 g of poly(A) ϩ RNA from mouse cerebellum using a Marathon cDNA amplification kit (CLONTECH) according to the manufacturer's instructions. The reaction products were subcloned into pCR 2.1 (Invitrogen, San Diego, CA) and sequenced using an ABI Prism 310 genetic analyzer (PerkinElmer Life Sciences). Then reverse primers were designed so that they were located downstream from the termination codons of any reading frames (Ma1A3ЈR3 and Ma1A3ЈR4; Table I). With these primers, nested PCR was also performed with a single Purkinje cell as described above. All PCR products were gel-purified and subcloned into the HincII site of pUC18 and sequenced. We used two independent Purkinje cells to avoid possible PCR errors.
Expression Vectors-Eight independent nested PCR reactions were designed so that the resulting PCR products covered the entire coding region of the mouse ␣ 1 2.1 subunit. The adjacent PCR products possessed overlapping sequences in which unique restriction sites occurred, making it easy to connect all the fragments to construct expression vectors (pcDNAI/Amp, Invitrogen, was used as the backbone vector). With regard to the most 3Ј region, two fragments with different sizes (1364 and 1506 bp) were amplified (Table I). The ␣ 1 2.1 with the shorter 3Ј sequence was designed MPI and the longer MPII. Thus, MPI and MPII differed only in their 3Ј sequence corresponding to the exon 41-47 (Fig. 1C).

Construction of MPc-
We also constructed a plasmid carrying the mouse ␣ 1 2.1 whose coding region terminated in the end of exon 46 and designated this version of ␣ 1 2.1 as MPc. A stop codon (TAG) was artificially introduced immediately after the end of exon 46 by PCR using MPII as a template and primer MalAF4 and MalAR4 (Table I) as a mutagenic primer. Then MPc was constructed by connecting the artificially made exon 41-46 fragment together with the seven same fragments that were used to construct MPI or MPII.
Colony Hybridization for Detection of RNA Editing Events-Fragments with a size of 780 bp containing nucleotide (nt) 6358 were PCR-amplified from cDNAs that were prepared from single Purkinje cells, cerebellar granule cells, or cerebral cortex. The primers used were Ma1A-F4 (Table I) as a forward primer and MalA-RE (5Ј-TGGGC-GAGCGGGACCAGCG-3Ј) as a reverse primer. Each PCR product was subjected to both direct sequencing and subcloning into pCR 2.1. For control experiments, we used two plasmids that contained the exon 41-47 fragment of MPI and MPII. The Escherichia coli colonies were transferred to nylon membranes (Hybond-Nϩ, Amersham Biosciences, Inc., Uppsala, Sweden). The membranes were hybridized with 18-mer oligonucleotide probe 6358C (5Ј-AGAACCAACGGTACCACC-3Ј) or 6358T (5Ј-AGAACCAATGGTACCACC-3Ј). The probes were 32 P-endlabeled using [␥-32 P]ATP (Ͼ5000 Ci/mmol) and T4 polynucleotide kinase (New England BioLabs, Beverly, MA). Hybridization was performed at 37°C for 16 h in 4ϫ SSC containing 0.1% SDS, 1ϫ Denhardt's reagent, and 50 g/ml herring sperm DNA (Roche Molecular Biochemicals, Mannheim, Germany). After hybridization, the membranes were washed in 1ϫ SSC with 0.1% SDS at 37°C for 20 min. Autoradiography was performed with Hyperfilm-ECL (Amersham Biosciences, Inc.) at Ϫ80°C using an intensifying screen.
Cell Culture and Transfection-The procedures of cell culture and transfection were the same as those previously described (16). In brief, human embryonic kidney (HEK) 293 cells were grown in Dulbecco's modified Eagle's medium/F-12 medium (Invitrogen) supplemented with 10% fetal bovine serum (Invitrogen) and 50 g/ml gentamicin (Invitrogen). The ␣ 1 2.1 subunits were transiently co-expressed with ␣ 2 ␦ and one of the ␤ subunits, and pEGFP-C2 (CLONTECH) was used as a transfection marker. Transfection was performed with a calcium phosphate precipitation method (Mammalian Transfection kit, Stratagene). Electrophysiological analysis was performed 36 -72 h after transfection.
Whole-cell Recordings-Patch pipettes were pulled from borosilicate glass (GC150F-4, Warner, Instruments, Hamden, CT) and filled with a solution consisting of (in millimolar) 140 CsCl, 2 MgCl 2 , 10 EGTA, 10 HEPES, 3 ATP⅐Mg (pH 7.4 with CsOH). Pipette resistance ranged from 2 to 4 M⍀. The external solution consists of (in millimolar) 15 BaCl 2 , 145 tetraethylammonium chloride, 10 HEPES, and 10 glucose (pH 7.4 with tetraethylammonium-OH). Barium currents were recorded at room temperature (22-25°C) under a whole-cell mode of the patch clamp recording with an amplifier (EPC-9, HEKA, Lambrecht/Pfalz, Germany). The holding potential was Ϫ80 mV unless otherwise stated. Series resistance was electronically compensated by 50 -80%. All illustrated and analyzed currents were corrected for remaining capacitance and leakage currents using the ϪP/4 method. Data were filtered at 3 kHz (four-pole Bessel filter) and sampled at 10 kHz. The software (PulseϩPulseFit 8.09, HEKA) was used for data acquisition and analysis. In the experiments with -Aga IVA (Peptide Institute, Inc., Osaka, Japan), the external solution was supplemented with 0.1 mg/ml cytochrome c to prevent nonspecific binding of the toxin. Statistical Analysis-All data were presented as mean Ϯ S.E. Statistical analysis was performed using an unpaired Student's t test.

RESULTS
Cloning of Novel Mouse ␣ 1 2.1 Subunit Variants-We cloned two novel splice variants of ␣ 1 2.1 subunit from the mouse Purkinje cell using the single-cell RT-PCR technique. Compared with the sequences of rabbit ␣ 1 2.1 subunit (BI-1) or rat ␣ 1 2.1 (rbA-1), the previously reported mouse ␣ 1 2.1 lacked 46 or 48 amino acids in the amino terminus, respectively (30). However, ␣ 1 2.1 variants we cloned in this study commonly contained the 5Ј sequences homologous to those of BI-1 and rbA-1, suggesting that the amino termini of our newly cloned ␣ 1 2.1 variants are longer by 48 amino acid residues than that of the previously reported mouse ␣ 1 2.1. Together with this difference, two novel ␣ 1 2.1 had single base substitutions at 16 sites located from exon 1 to exon 41 compared with the previously reported mouse ␣ 1 2.1. We confirmed these 16 single base substitutions in two independent Purkinje cells. Therefore, PCR errors are unlikely to be the cause of the substitutions. Ten of the 16 substitutions were silent polymorphisms leading to no amino acid changes, and the rest of them resulted in amino acid changes. Two altered amino acids found in II-III loop (L886P, D1085N) were identical to the sequence of rbA-1, and the remaining four found in repeat I (P79S, F82L) and repeat III (F1409L, F1433L) were conserved in BI-1, rbA-1, and human ␣ 1 2.1 subunit (Fig. 1A).
In the human Ca v 2.1 gene, differential splice acceptor usage at the boundary of intron 46/exon 47 is known to yield an ␣ 1 2.1 variant with a longer carboxyl terminus (11), because the stop codon in the beginning of exon 47 was no longer in-frame. The same kind of variant was also found in rat ␣ 1 2.1 (36). But the nucleotide sequence of mouse ␣ 1 2.1 corresponding to the variant with a longer carboxyl-tail has not been reported so far. This carboxyl-terminal sequence, which is thought to be encoded by exon 47, is expressed abundantly in rat Purkinje cell bodies and dendrites as revealed by an immunohistochemical study (17) and may affect channel properties. Therefore, we tried to clone the mouse exon 47 by 3Ј-RACE. Because a single Purkinje cell has an extremely small amount of RNA, it is expected to be difficult to apply 3Ј-RACE. Therefore, we first applied 3Ј-RACE to mouse cerebellar poly(A) ϩ RNA. As a result, a novel fragment with a size of 1504 bp was cloned, and sequence analysis revealed that there were termination codons in all the reading frames on the sequence. Then we designed reverse primers (Ma1A-3Ј-R3, Ma1A-3Ј-R4; Table I) to perform RT-PCR with single Purkinje cells. Two fragments with different sizes of 1364 and 1506 bp were amplified. We assigned the novel ␣ 1 2.1 with the shorter fragment MPI, and the ␣ 1 2.1 with the longer fragment MPII. The nucleotide sequences of MPI and MPII were completely the same except for these most 3Ј sequences, where several distinctive features were found in the two variants (Fig. 1, B and C). First, a single nucleotide conversion (C to T) was observed at nt 6358 in MPI. This conversion led to an amino acid substitution from arginine to tryptophan (Fig. 1, B and C). Second, there were two variations at the beginning of exon 47. In MPI, 5 nucleotide residues (GGCAG) were inserted, whereas in MPII 3 nucleotide residues (TAG) were deleted, when compared with the previously published sequence (30) (Fig. 1B). These 5-bp insertion and 3-bp deletion steps resulted in different reading frames in exon 47. It may be worth noting that the previously reported variant was cloned from cerebellar cDNA but not from Purkinje cell cDNA. Third, in MPI, a 150-bp sequence corresponding to nt 6831-6980 in MPII was deleted (Fig. 1B). The 3Ј sequence of the MPII (nt 6691-7149) was identical to the corresponding mouse genomic sequences (GenBank TM accession number AC079509), suggesting that this region was encoded by a single exon. This also suggests that the 150-bp deletion in MPI was caused by an alternative splicing. Finally, carboxyl-terminal sequences are different. MPII has an evolutionarily conserved sequence (DDWC-COOH) at the carboxyl terminus but MPI does not have this sequence (Fig. 1C).
Sequence analysis of MPI and MPII revealed an open reading frame of 6981 and 7095 nucleotides encoding a protein of 2327 and 2365 amino acid residues, respectively. Except for the exon 47 sequences, deduced amino acid sequences of MPI and MPII were 89%, 91%, and 99% identical to those of BI-1 (22), human ␣ 1 2.1 (16), and rbA-1 (23), respectively.
RNA Editing in the ␣ 1 2.1 Subunit Occurred Specifically in Purkinje Cells-Sequence analysis of genomic DNA revealed that the nucleotide corresponding to position 6358 of the cDNA was C (Fig. 2A), therefore, the cytidine-to-uridine (C-to-T in cDNA) conversion in MPI was thought to be due to RNA editing. We then directly sequenced cDNA fragments containing this region, which were prepared from single Purkinje cells, granule cells, or cerebral cortex. The sequence analysis revealed that nucleotide position 6358 is C in granule cells and cerebral cortex. On the other hand, in Purkinje cells, sequence analysis at the same position displayed N, suggesting that C and T were simultaneously exhibited ( Fig. 2A). Then colony hybridization was performed to confirm the specificity and efficiency of this conversion. In control experiments, autoradiograms of representative hybridization with the 32 P-end-labeled oligonucleotide probes, 6358C and 6358T, sharply discriminated the colonies carrying plasmids, which contained the exon 41-47 fragment of MPI and that of MPII (Fig. 2B). As for the experiment with Purkinje cells, the number of colonies hybridizing with the probe 6358C and 6358T was almost the same. By contrast, no hybridization of the probe 6358T occurred in the experiments with cerebellar granule cells or cerebral cortex (Fig. 2C). These results suggest that C-to-T conversion occurred specifically and in approximately half of the Ca v 2.1 mRNA in this Purkinje cell.
Electrophysiological and Pharmacological Studies of MPI, MPII, and MPc Channels Expressed in HEK293 Cells-MPI, MPII, or MPc was expressed together with ␣ 2 ␦ and one of the four ␤ subunits (␤ 1a , ␤ 2a , ␤ 3 , and ␤ 4 ) in HEK293 cells, and the electrophysiological properties were studied with a whole-cell patch-clamp technique. Because functional properties of the previously cloned mouse ␣ 1 2.1 were not reported, we needed a control mouse ␣ 1 2.1 to investigate how the carboxyl termini of MPI and MPII influence channel properties. To this end, we constructed MPc whose coding region was terminated in the end of exon 46. Expression plasmids encoding this version of ␣ 1 2.1 of other species have been used for many electrophysiological studies (18,22,25,26,28,41). Although MPc as well as MPI and MPII contained six single amino acid changes from exon 1 to exon 41 compared with the sequence of the previously cloned mouse ␣ 1 2.1, the changed amino acids were conserved among the ␣ 1 2.1 cloned from other species (Fig. 1A). Therefore, these amino acid changes were supposed to affect the channel properties little. In fact, the current recorded from the MPc channel was similar to those recorded from the channels containing ␣ 1 2.1 of other species (Fig. 3).
Barium currents were elicited by 400-ms depolarizing test pulses from a holding potential of Ϫ80 mV to test potential (Ϫ40 to ϩ50 mV) in 15 mM Ba 2ϩ solutions. Whole-cell Ba 2ϩ currents recorded in HEK293 expressing MPI, MPII, or MPc were almost the same. Although ␤ subunits were reported to significantly influence Ca 2ϩ channel properties, we were not able to record precise P-type currents from the cells expressing any combinations of ␣ 1 2.1 and ␤ subunits (Fig. 3, A-D). All channels were activated at a test potential of Ϫ20 mV, and the currents gradually developed to maximal amplitude at about 20 mV. Normalized current-voltage curves were not significantly different among MPI, MPII, and MPc. To compare the activation kinetics, we quantified the time course by fitting the currents to single-exponential function. To compare the inactivation kinetics, we quantified the time course by fitting the currents to double-exponential function in ␤ 1a -associated channels and single-exponential function in the other channels. The obtained time constants for activation and inactivation at 20 mV are shown in Table II. They were not significantly different among these channels. Next, the voltage dependence of activation for the expressed Ca 2ϩ channels was compared ( Fig. 4 and Table III). Conductance was calculated from the peak current. The normalized values of conductance were then fitted to a single Boltzmann function. The half-maximal voltage of activation (V1 ⁄2 ), the slope factor (k), and the percent decaying were not different among the channels. Compared with ␤ 1a -and ␤ 3 -associated channels, ␤ 2a -and ␤ 4 -associated ones demonstrated slower inactivation currents. However, they were still faster than P-type current, which does not decay over 1 s (37). A two-pulse protocol was applied to examine the voltage dependence of inactivation. To reach a steady state, a 2-s prepulse (from Ϫ100 to 0 mV with a 10-mV increment) preceded a test pulse of 20 mV. The values of normalized current amplitude were plotted against holding potentials (Fig. 4). The V1 ⁄2 and k were not significantly different among the channels (Table III).
The major pharmacological difference between P-and Q-type Ca 2ϩ channels is the sensitivity to the spider peptide toxin,  20 mV Values are means Ϯ S.E., n, number of cells recorded; act , time constants for activation; fast and slow , fast and slow time constants for inactivation. Activation kinetics were derived by fitting a single exponential function: I Ca ϭ I S [1 Ϫ exp(Ϫt/ act )], where I Ca is the total current, and I S is the steady-state current, to the activating segment of test currents. Inactivation kinetics were obtained by fitting single-or double-exponential functions: I Ca ϭ ϪI intact Ϫ exp(Ϫt/ slow ) Ϫ NI or I Ca ϭ ϪI fast ⅐ exp(Ϫt/ fast ) Ϫ I slow ⅐ exp(Ϫt/ slow ) Ϫ NI, respectively; I inact , total inactivation component of the total current, I fast and I slow , the fast and slow inactivation component of the total current; NI, the noninactivating component of the total current.
-Aga IVA. Co-expression studies with different ␤ subunits revealed that ␤ 1 -associated channels exhibit high affinity to this toxin (38). Therefore, we chose the ␤ 1a subunit for this study. The currents mediated by MPI, MPII, and MPc channels were halfblocked by 50 nM -Aga IVA (Fig. 5). This sensitivity was close to that of Q-type Ca 2ϩ channels previously described (8). DISCUSSION In this study, we have cloned two novel mouse ␣ 1 2.1 variants, MPI and MPII, carrying different 3Ј sequences. There are several distinctive features between the two variants.
First, a single base substitution at nt 6358 (C or T) is likely caused by RNA editing, because sequence analysis of genomic DNA corresponding to this region revealed C. Interestingly, this conversion occurred specifically in Purkinje cells as far as examined and the efficiencies were variable among Purkinje cells. In some cells, the efficiencies reached to 50%, in other cells, efficiencies were very low (data not shown). These results suggest that MPI, which has this C-to-T conversion, is expressed specifically in Purkinje cells with different abundance. RNA editing is a post-transcriptional modification that results in generation of nucleotides within an RNA transcript that are different from the sequence of the genome. RNA editing may have profound functional consequences in protein functions. For example, the adenosine-to-inosine (A-to-I) conversion within the RNA encoding the GluR-B subunit of ␣-amino-3hydroxy-5-methyl-4-isoxazole propionic acid-subtype glutamate receptors (Q/R site) alters calcium permeability (39). Although channel properties of MPI, which contains an RNAedited site, were not different from those of MPc in the heterologous expression system, the cell specificity of the editing event would suggest important roles of this variant in Purkinje cells.
Second, the two variants differed at the beginning of exon 47, presumably due to different splice acceptor usage. It is known that there are sequence variations at the boundary of exon 46/47 in ␣ 1 2.1 mRNA: two variations with and without GGCAG insertion before the TAG stop codon located at the beginning of exon 47 have been reported in human and rat ␣ 1 2.1 (11,26). MPI seems to correspond to the human and rat version possessing the GGCAG insertion, however, MPI has the deduced amino acid sequence diverged from these variants. We have cloned another variant, MPII, in which the TAG stop codon is deleted and carboxyl terminus is extended with different reading frame from that of MPI. Thus, we have identified only the carboxyl-terminal extended versions of ␣ 1 2.1 in mouse Purkinje cells. An immunohistochemical study revealed that the exon 47 sequence, with the same reading frame as MPI, is expressed intensely in Purkinje cells (17). Interestingly, however, the ␣ 1 2.1 in which the coding region is terminated in the end of exon 46 was cloned from a cerebellar cDNA library, suggesting that these variants for ␣ 1 2.1 are expressed in different cell populations in the cerebellum.
Third, as mentioned, MPI and MPII have different carboxylterminal sequences. In MPII, there is an evolutionarily conserved sequence at the carboxyl-tail: DDWC-COOH. This (E/ D)XWC-COOH motif is conserved in the wide range of animals from Drosophila to human. This suggests that this motif has important biochemical roles. Indeed, it is known that this motif is specifically recognized by the PDZ domain of Mint1, which in association with CASK and Ca 2ϩ channel can modulate synaptic vesicle fusion and neurotransmitter release (40). Therefore, MPI, which does not have this motif, and MPII might have distinct functions in the control of synaptic transmission.
Taken together, our results suggest that MPI and MPII are  differently localized in cerebellar cells with different functions. Therefore, we then investigated the channel properties of MPI, MPII, and MPc. It was reported that ␤ subunits were influential on the Ca 2ϩ channel properties (38,41); however, P-typeassociated ␤ subunit has not yet been identified. Therefore, we expressed MPI, MPII, or MPc with one of the various ␤ subunits in HEK293 cells and compared their channel properties.
As a result, the characteristics of MPI, MPII, and MPc were not significantly different, suggesting that the alternatively spliced form of ␣ 1 2.1 identified in Purkinje cells by itself is not sufficient to generate P-type current when expressed in combination of ␤ and ␣ 2 ␦ subunits in cultured cells. Our data are inconsistent with some results of the recent report that alternative splicing determines properties of P-and Q-type Ca 2ϩ channels. Bourinet et al. (18) reported that a non-inactivating current form, which was similar to a P-type one, was recorded when the ␣ 1 2.1 subunit, containing a single valine insertion in repeat I-II linker, was co-expressed with ␣ 2 ␦ and ␤ 2 subunits. However, it is unlikely that native P-type current is encoded by this combination of subunits for the following reasons. First, our splice variants, MPI and MPII isolated from Purkinje cells, lack the single valine insertion. Second, the inactivation for MPI, MPII, or MPc channel became slower by co-expression of ␤ 2 (Fig. 3B), however, it was still faster compared with the native P-type current (37).
Pharmacologically, P-and Q-type Ca 2ϩ channels differ in the sensitivity to -Aga IVA. The P-type current is significantly blocked by -Aga IVA, with an IC 50 value of 1-2 nM (42). In contrast, the Q-type current is less sensitive to -Aga IVA, with an IC 50 of 89 nM (8). The sensitivity of MPI, MPII, and MPc channels were similar to that of Q-type channels. The sensitivity to -Aga IVA was shown to depend significantly on the type of cells where ␣ 1 2.1 was expressed. For example, the IC 50 value for rat rbA-1 is higher than 200 nM in Xenopus oocytes, whereas it is as low as 16.3 nM in HEK293 cells (18). Recently, it has been reported that splicing of ␣ 1 2.1 also affects the toxin sensitivity: ␣ 1 2.1 without arginine and proline (NP) insertion between segments S3 and S4 in repeat IV showed higher sensitivity (IC 50 ϭ 16.3 nM) than ␣ 1 2.1 with the NP insertion (IC 50 ϭ 146 nM) (18). Because the sensitivity observed in HEK293 cells was still not as high as that in Purkinje cells, high sensitivity observed in Purkinje cells was expected to arise from the lack of NP insertion in ␣ 1 2.1 together with the specific environment of Purkinje cells. Our results support this idea in the pharmacological respect, because MPI and MPII lacked the NP insertion.
In conclusion, our results suggest that MPI and MPII may distribute differently and play some specific roles in Purkinje cells. Further study is necessary to substantiate this point. MPI, MPII, and MPc cannot generate P-type currents in HEK293 cells, suggesting that alternative splice variants of ␣ 1 2.1 subunit expressed in Purkinje cells alone are not sufficient to generate native P-type currents in heterologous expression system. Some specific environmental modifications such as post-translational processing or modulation by putative proteins associated with the channels may be necessary to generate the native P-type current.