Immunochemical identification and differential phosphorylation of alternatively spliced forms of the alpha 1A subunit of brain calcium channels.

Biochemical properties of the α1 subunits of class A brain calcium channels (α1A) were examined in adult rat brain membrane fractions using a site-directed anti-peptide antibody (anti-CNA3) specific for α1A. Anti-CNA3 specifically immunoprecipitated high affinity receptor sites for ω-conotoxin MVIIC (Kd ∼100 pM), but not receptor sites for the dihydropyridine isradipine or for ω-conotoxin GVIA. In immunoblotting and immunoprecipitation experiments, anti-CNA3 recognized at least two distinct immunoreactive α1A polypeptides, a major form with an apparent molecular mass of 190 kDa and a minor, full-length form with an apparent molecular mass of 220 kDa. The 220- and 190-kDa α1A polypeptides were also specifically recognized by both anti-BI-Nt and anti-BI-1-Ct antibodies, which are directed against the NH2- and COOH-terminal ends of α1A predicted from cDNA sequence, respectively. These data indicate that the predicted NH2 and COOH termini are present in both size forms and therefore that these isoforms of α1A are created by alternative RNA splicing rather than post-translational proteolytic processing of the NH2 or COOH termini. The 220-kDa form was phosphorylated preferentially by cAMP-dependent protein kinase, whereas protein kinase C and cGMP-dependent protein kinase preferentially phosphorylated the 190-kDa form. Our results identify at least two distinct α1A subunits with different molecular mass, demonstrate that they may result from alternative mRNA splicing, and suggest that they may be differentially regulated by protein phosphorylation.

The class A calcium channel (also designated BI) was the first non L-type calcium channel to be cloned, sequenced, and expressed (Starr et al., 1991;Mori et al., 1991). ␣ 1A forms high voltage activated calcium channels and Northern blot analysis shows high expression in the cerebellum (Starr et al., 1991;Mori et al., 1991;Sather et al., 1993;Stea et al., 1994a). ␣ 1A currents expressed in Xenopus oocytes are insensitive to dihydropyridines and -conotoxin GVIA, and therefore ␣ 1A subunits may form P-type and/or Q-type channels (Mori et al., 1991;Sather et al., 1993;Stea et al., 1994a). ␣ 1A channels expressed in Xenopus oocytes inactivate slowly or rapidly depending on the ␤ subunit expressed with them, and are blocked by -agatoxin IVA purified from Agelenopsis aperta venom at high concentration and by -conotoxin MVIIC from Conus ma-gus (Hillyard et al., 1992;Sather et al., 1993;Stea et al., 1994a). In contrast, native P-type calcium channels are blocked by low concentrations of -agatoxin IVA and by higher concentrations of -conotoxin MVIIC (Mintz et al., 1992a, Hillyard et al., 1992. The pharmacological properties of ␣ 1A calcium channels in Xenopus oocytes are distinct from P-type channels, but more closely resemble those of calcium channels in cerebellar granule cells, which have been designated Q-type Zhang et al., 1993). In the experiments described in this paper, we used site-directed anti-peptide antibodies against unique sequences in rat brain ␣ 1A to identify the corresponding polypeptides and demonstrated that there are multiple isoforms of ␣ 1A subunits that may result from alternative RNA splicing and are differentially phosphorylated by second messenger-activated protein kinases.
Production and Purification of Peptides and Antibodies-The peptide CNA3 ((KY)SEPQQREHAPPREHV) corresponds to residues 882-896 (Starr et al., 1991) and the peptide CNA1 ((KY)PSSPERAPGREG-PYGRE) corresponds to residues 865-881, which are located in a highly variable site in the intracellular loop between domains II and III of the ␣ 1A subunit of rat brain calcium channels. The NH 2 -terminal lysine and tyrosine are not part of the channel sequence and were added for cross-linking and labeling purposes. The CNA1 and CNA3 peptides were synthesized by the solid phase method (Merrifield, 1963) and then purified by reversed phase high pressure liquid chromatography on a Vydac 281TP10 column. The identity of the purified peptides was confirmed by amino acid analysis.
The purified peptides were coupled through amino groups with glutaraldehyde to bovine serum albumin (BSA), dialyzed against phosphate-buffered saline (10 mM NaH 2 PO 4 (pH 7.4), 150 mM NaCl) and emulsified in an equal volume of Freund's complete (initial injection) or incomplete adjuvant. The coupled peptides were injected into multiple subcutaneous sites on New Zealand White rabbits at 3-week intervals. Antisera were collected after the second injection and tested by enzymelinked immunosorbent assay using microtiter plates with wells coated with 0.5 g of peptide (Posnett et al., 1988). Antibodies were purified by affinity chromatography on the corresponding peptides coupled to CNBr-activated Sepharose. Two ml of the antiserum were bound to the column at 4°C overnight and washed with TBS (10 mM Tris-HCl (pH 7.4), 150 mM NaCl). The bound IgG was eluted with 3.0 M MgCl 2 . The affinity-purified antibodies were then dialyzed against TBS using a Centriprep 30 (Amicon).
Anti-BI-Nt and anti-BI-1-Ct antibodies were generous gifts from Dr. Masami Takahashi (Mitsubishi-Kasei Life Sciences Institute, Tokyo, Japan), and these antibodies were produced against peptides MARFG-DEMPARYGGGGAGAA(C) (Leveque et al., 1994) and (C)RDQRWSR-SPSEGREHTTHRQ, which correspond to residues 1-23 and 2254 -2273 of the BI-1 cDNA clone encoding a rabbit brain ␣ 1A subunit, respectively (Mori et al., 1991). The cysteine residue in each peptide was added to facilitate cross-linking and radiolabeling and is not part of the ␣ 1 subunit sequence.
Membrane Preparation-Brains were dissected from 15 2-month-old Sprague-Dawley rats, obtained from Bantin and Kingman (Bellevue, WA), and calcium channels were solubilized and partially purified as described previously (Westenbroek et al., 1992). Briefly, samples of rat brain were homogenized and subjected to a brief low speed centrifugation to yield supernatant S1 containing mixed brain membranes. The cell surface membranes were collected by high speed centrifugation. Calcium channels were solubilized with 1.2% digitonin, and insoluble material was removed by high speed centrifugation to yield supernatant S3. The calcium channels were then partially purified by the chromatography on wheat germ agglutinin (WGA)-Sepharose as described previously (Westenbroek et al., 1992).
Radioactive Ligand Binding Studies-For [ 3 H]PN200-110 (isradipine) binding studies, 40 ml of S1 fraction were labeled with 10 Ci of [ 3 H]PN200-110 (85.8 Ci/mmol) at a concentration of 2.9 nM for 1 h on ice. The bound radioligand is stable throughout the subsequent purification steps. Calcium channels were purified from 250 l of the S3 fraction (ϳ6000 cpm) containing 300 mM KCl, 150 mM NaCl, 10 mM Tris-HCl (pH 7.4), 1.2% digitonin with 0.2% BSA using 15 g of affinitypurified anti-CNA3, anti-CNC1, or control rabbit nonimmune IgG. After a 1.5-h incubation on ice, 2.5 mg of PAS, prewashed three times with TBS containing 0.1% digitonin and 0.5% BSA, were added to the samples. The samples were mixed on ice for additional 2.5 h, pelleted by centrifugation, and washed three times in TBS, 0.1% digitonin. After the final wash, the antibody-bound PAS complexes were transferred to vials, and the amount of immunoprecipitated [ 3 H]PN200-110 receptors was quantified in a scintillation counter. Total receptor-bound [ 3 H]PN200-110 was determined by filter binding assay. 250 l of the labeled S3 fraction were precipitated by incubation with 4 ml of ice-cold 10% polyethylene glycol (average molecular weight 8000) in 10 mM MgCl 2 and 10 mM Tris-HCl (pH 7.4) for 5 min and poured over Whatman GF/C filters. Samples were washed four times in ice-cold polyethylene glycol solution and quantified in a scintillation counter. The correction factor for ligand-receptor loss in the filter-binding assay was 0.7 (Westenbroek et al., 1992).
Determination of 125 I--conotoxin GVIA binding was done by incubation of 100 l of S3 fraction containing 0.2% BSA with 0.06 Ci of 125 I--conotoxin GVIA (2200 Ci/mmol) at a concentration of 0.27 nM for 30 min on ice. Samples were immunoprecipitated with 15 g of affinitypurified anti-CNA3, anti-CNB2, or control rabbit IgG, and washed four times with TBS, 0.1% digitonin. The matrix was transferred to vials for ␥ counting. Total 125 I--conotoxin GVIA binding was determined using 100 l of the labeled S3 fraction in the filter-binding assay described above for [ 3 H]PN200-110. 125 I--Conotoxin MVIIC binding was determined by incubation of 400 l of samples containing 140 l of WGA extract, 10 mM Tris-HCl (pH 7.4), and 0.2% BSA with 0.1 Ci of labeled toxin (1300 Ci/mmol) at a concentration of 0.15 nM on ice for 30 min. This was added to 15 g of affinity-purified anti-CNA3, anti-CNC1, or control rabbit IgG, coupled to 2 mg of PAS, and incubated for 4 h on ice in a tilting mixer. Samples were washed quickly three times in 10 mM Tris-HCl (pH 7.4), 75 mM NaCl, 0.1% digitonin, 0.2% BSA. 125 I--Conotoxin MVIIC binding in the pellet was counted in a ␥ counter. For the competitive displacement studies, the unlabeled ligands -conotoxin MVIIC or -agatoxin IVA at concentrations ranging from 10 Ϫ14 to 10 Ϫ5 M were added to samples with 125 I--conotoxin MVIIC and incubated with affinity-purified anti-CNA3, and the bound 125 I--conotoxin MVIIC was measured as described above. For peptide block, 20 M test peptide was added to the affinity-purified antibodies and incubated overnight on ice prior to incubation with samples containing WGA extract.
Immunoblotting of Calcium Channels-To concentrate the calcium channels, WGA column fractions containing 0.1 mg of total protein were incubated for 4 h on ice with 40 l of heparin-agarose (Sakamoto and Campbell, 1991). The resin was washed three times with 10 mM Tris-HCl (pH 7.4), 0.1% digitonin and once with 10 mM Tris-HCl (pH 7.4). Calcium channels were extracted for 30 min at 50 -60°C with 30 l of SDS-sample buffer (200 mM Tris-HCl (pH 6.8), 10 mM dithiothreitol, 4 M urea, 8% SDS, 10% glycerol). After separation by SDS-polyacrylamide gel electrophoresis (SDS-PAGE), proteins were transferred onto a nitrocellulose membrane (0.2 m) in a buffer containing 12.5 mM Tris (pH 8.3), 96 mM glycine, 0.1% SDS, 15% (v/v) methanol. Unbound sites on the nitrocellulose were blocked for 2 h at room temperature with TBS containing 10% skim milk powder and incubated with affinity-purified anti-CNA3 (1-10 g/ml), and with protein A-purified anti-BI-Nt, or anti-BI-1-Ct antibodies (50 -250 g/ml) in TBS for 2 h at room temperature. After five 5-min washes at 4°C with TBS, blots were incubated for 1 h with horseradish peroxidase-protein A, diluted 1:2000 in TBS. After another eight 10-min washes at 4°C, the blots were developed with the ECL reagent. For peptide block, the corresponding peptide at 0.2-2 M was added to the affinity-purified antibodies and incubated overnight on ice prior to incubation with the samples.
Biotinylation and Purification by Immunoprecipitation-WGA frac-tions, diluted 1:1000 in 100 mM sodium borate (pH 8.5), 0.1% digitonin, were concentrated to a volume of ϳ500 l in a Centricon-30 microconcentrator to remove the amine in Tris-HCl buffer. One mol of sulfosuccinimidyl-6-(biotinamide) hexanoate (NHS-LC-biotin) was used to biotinylate the partially purified membrane fractions. After 2 h of incubation on ice, the reaction was terminated by addition of 0.2 volume of 2 M glycine (pH 8.5). Samples were diluted in TBS, 0.1% digitonin and concentrated to a volume of approximately 300 l by ultrafiltration. Biotinylated samples were preabsorbed for 1 h on ice with 300 l of Sepharose CL-4B and for 2 h on ice with 10 mg of PAS, which was preincubated with 200 g of control rabbit IgG and washed three times with TBS, 0.1% digitonin in order to remove the nonspecifically binding proteins in the sample. After centrifugation, supernatants were incubated for another 2 h on ice with 10 mg of PAS to adsorb the free IgG dissociated from PAS-control rabbit IgG complex. After centrifugation for 1 min on a table-top centrifuge, the supernatants were collected and incubated with anti-CNA3 (40 g), anti-BI-Nt (30 g), anti-BI-1-Ct (30 g), anti-CNA1 (80 g), or control antibody (80 g) for 1.5 h on ice. The immunoprecipitation was performed as described in the section above, and the pellets were extracted for 30 min at 50 -60°C with 20 l of 1.5% SDS, 50 mM Tris-HCl (pH 7.4), 5 mM dithiothreitol, 1 M pepstatin A, 2 g/ml leupeptin, and 4 g/ml aprotinin, and diluted with 250 l of Triton buffer (1% Triton X-100, 0.5% BSA, 75 mM NaCl, 25 mM Tris-HCl (pH 7.4), 20 mM EDTA). The supernatant was collected and incubated for 1.5 h on ice with the secondary antibodies anti-CNA3 (40 g) or anti-CP(1382-1400) (20 g). Three mg of PAS, pretreated as described above, were added, and the samples were incubated on a tilting mixer for 2.5 h on ice. The immunoprecipitated complexes were pelleted by centrifugation, washed three times with Triton buffer and once in 10 mM Tris-HCl (pH 7.4), and extracted for 30 min at 50 -60°C with SDS sample buffer. After a short centrifugation, the supernatants were loaded onto an SDS-PAGE gel. The proteins were blotted, blocked as described above, and nitrocellulose sheets were rinsed with TBS, 5% BSA, 0.2% Nonidet P-40, and 0.05% Tween 20, and incubated for 1 h at room temperature with streptavidin-biotinylated horseradish peroxidase complex, diluted 1:8000 in TBS containing 0.2% Nonidet P-40 and 0.05% Tween 20. After a 3-h wash with 0.2% Nonidet P-40, 0.05% Tween 20 in TBS (8 -9 changes), the blots were developed with the ECL reagent.
Immunoprecipitation and Phosphorylation of Calcium Channels-Calcium channels in the WGA extract were concentrated by immunoprecipitation with either affinity-purified anti-CNA3 or control rabbit IgG as described above. Prior to phosphorylation of the immunoprecipitated calcium channels, the resin was washed once in the basic phosphorylation buffer (50 mM Hepes (adjusted to pH 7.4 with NaOH), 10 mM MgCl 2 , 1 mM EDTA, 0.1% digitonin). Phosphorylation reactions were performed in 50 l of the reaction mixture containing 0.5-1.0 g of PKA, PKC, or PKG in the basic phosphorylation buffer, along with 1 mM dithiothreitol, 1 M pepstatin A, 1 mM EGTA, and 0.2 M [␥-32 P]ATP. This buffer was supplemented with 1.5 mM CaCl 2 , 50 g of diolein, and 2.5 mg of phosphatidylserine for PKC and 2 M cGMP for PKG. Incubations were at 32-34°C for 30 min with gentle mixing every 2 min. The samples were washed twice with 0.1% digitonin in radioimmunoassay buffer (25 mM Tris-HCl (pH 7.4), 20 mM EDTA, 75 mM NaCl, 20 mM sodium pyrophosphate, 20 mM ␤-glycerol phosphate, 50 mM NaF, and 1 mM p-nitrophenyl phosphate), three times with 1% Triton X-100 in radioimmunoassay buffer, and once in 10 mM Tris-HCl (pH 7.4). The pellets were extracted, and the second immunoprecipitations were performed with affinity-purified anti-CP(1382-1400) as described in the previous section. Samples were analyzed by SDS-PAGE, and autoradiography was performed.

Immunoprecipitation of High Affinity Receptor Sites for Dihydropyridines, -Conotoxin GVIA, or -Conotoxin MVIIC-
The amino acid sequences of the large intracellular loops connecting domains II and III of the neuronal calcium channel ␣ 1 subunits are highly variable and characteristic for each class of channels. For production of specific antibodies against ␣ 1A , a unique sequence in this intracellular loop was selected, and the corresponding peptide (CNA3) was synthesized, coupled to BSA as a carrier, and used as an antigen for antigen production as described under "Experimental Procedures." To determine the pharmacological properties of the ␣ 1A polypeptides recognized by anti-CNA3, we labeled brain calcium channels with [ 3 H]PN200-110, 125 I--conotoxin GVIA, or 125 I--conotoxin MVIIC, and immunoprecipitated labeled brain calcium channels with anti-CNA3 antibodies.
Displacement of specific binding of 125 I--conotoxin MVIIC by unlabeled -conotoxin MVIIC was observed between 10 pM and 1 nM, with half-maximal inhibition at approximately 100 pM (Fig. 2). The K d value of approximately 100 pM is in agreement with that determined in rat synaptosomal membranes (10 -300 pM) (Hillyard et al., 1992;Kristipati et al., 1994;Woppmann et al., 1994). -Agatoxin IVA, a 48-amino acid peptide toxin from funnel web spider venom with no obvious similarity in sequence to -conotoxin MVIIC, blocks P-type calcium currents with IC 50 of 1-2 nM (Mintz et al., 1992a) and ␣ 1A calcium currents at 100 -300 nM , Stea et al., 1994a. To assess whether -conotoxin MVIIC and -agatoxin IVA might bind to the same high affinity sites, the ability of -agatoxin IVA to displace high affinity binding of -conotoxin MVIIC was tested. Displacement of specific binding of 125 I--conotoxin MVIIC to the -conotoxin MVIIC receptor site with unlabeled -agatoxin IVA occurred only at high concentration with half-maximal inhibition at approximately 1-5 M (Fig. 2). This result indicated that -agatoxin IVA can displace -conotoxin MVIIC from its high affinity binding site nonspecifically at high concentration, but not at concentrations at which it inhibits calcium channels containing ␣ 1A . Evidently, -agatoxin IVA does not bind to the same receptor site as -conotoxin MVIIC in inhibiting class A calcium channels.
Identification of ␣ 1A in Rat Brain Membranes-To identify ␣ 1A polypeptides, rat brain glycoproteins were isolated by affinity chromatography on WGA-Sepharose, and calcium channels were enriched by adsorption to heparin-agarose and analyzed by immunoblotting (see "Experimental Procedures"). Affinity-purified anti-CNA3 antibody revealed at least three immunoreactive bands of ␣ 1A subunits with apparent molecular masses of 210 -230, 180 -195, and 160 kDa (Fig. 3). The molecular mass of the largest polypeptide varies between 210 and 230 kDa depending on the concentration of acrylamide used for SDS-PAGE (Fig. 3, lanes 1 and 4). In a 5% acrylamide gel, the polypeptide migrated to a position just above myosin, the 205-kDa marker, whereas in a 7% acrylamide gel its position was between the longer and shorter forms of spectrin, the 240-and 220-kDa markers, respectively. We refer to this band as the 220-kDa form of ␣ 1A polypeptide. In a higher resolution autoradiogram, two distinct but closely spaced bands were often separately visualized within this 220-kDa band (Fig. 3, lane 1). The intermediate size form of ␣ 1A , which we have designated the 190-kDa form, migrated with an apparent molecular mass of 180 kDa in a 5% acrylamide gel, and 195 kDa in a 7% acrylamide gel when compared with the marker proteins. The smallest polypeptide migrated with an apparent molecular mass of 160 kDa in both gels. Anomalous migration was also observed during the determination of molecular masses of the two size forms of the skeletal muscle L-type calcium channel ␣ 1 subunit (De Jongh et al., 1991).
The specificity of the interaction of anti-CNA3 antibodies with these polypeptides was tested with the CNA3 peptide. After preincubation with CNA3 peptide at a concentration of 2 M, no signal could be detected with anti-CNA3 antibody (Fig.  3, lane 2). These specific bands were only observed with affinity-purified anti-CNA3 antibodies; affinity-purified anti-CNC1 antibodies against ␣ 1C revealed distinct bands with apparent FIG. 1. Immunoprecipitation of brain calcium channels labeled with [ 3 H]PN200-110, 125 I--conotoxin GVIA, or 125 I--conotoxin MVIIC. A, rat brain membrane fraction (S1) was incubated with [ 3 H]PN200-110 at the concentration of 2.9 nM, solubilized, immunoprecipitated with anti-CNC1, anti-CNA3, or rabbit control antibodies, and the bound [ 3 H]PN200-110 was counted (see "Experimental Procedures"). Anti-CNC1 was raised against a highly variable site of the class C L-type calcium channel ␣ 1 subunit (Hell et al., 1993a). Total [ 3 H]PN200-110 receptor sites were estimated by filter binding assay. B, S3 fractions were incubated with 125 I--conotoxin GVIA (0.27 nM) and immunoprecipitated with anti-CNB2, anti-CNA3, or control IgG. Anti-CNB2 is specific for class B N-type calcium channel ␣ 1 subunit . C, calcium channels were purified from the S3 fraction by chromatography on WGA-Sepharose, incubated with 125 I--conotoxin MVIIC (0.15 nM), and immunoprecipitated with anti-CNA3, anti-CNC1, or control IgG. The specificity of immunoprecipitation of 125 I--conotoxin MVIIC receptors with anti-CNA3 was tested by peptide block. Anti-CNA3 was preincubated with 20 M CNA3 peptide or with 20 M CNC1 peptide, and immunoprecipitation was carried out as described under "Experimental Procedures." Membrane glycoprotein fractions were isolated from solubilized brain membranes by WGA affinity chromatography, and calcium channels were concentrated by adsorption to heparin agarose, extracted, and analyzed by SDS-PAGE using a 5% acrylamide gel (lanes 1-3) or a 7% acrylamide gel (lane 4). Proteins were transferred onto a nitrocellulose membrane, blocked, incubated with anti-CNA3 (lanes 1, 2, and 4) or anti-CNC1 (lane 3), incubated with horseradish peroxidase-protein A, washed, and visualized with ECL reagent, as described under "Experimental Procedures." Anti-CNA3 antibodies were preincubated overnight on ice with 2 M CNA3 peptide (lane 2). The migration positions of ␣and ␤-spectrin, myosin heavy chain, ␣ 2 -macroglobulin, ␤-galactosidase, and fructose-6-phosphate kinase are indicated at the left side of the gel together with their molecular masses in kDa. molecular mass values of 210 and 190 kDa in the parallel gel (Fig. 3, lane 3), and antibodies directed against the other brain calcium channels also recognized distinct polypeptides (data not shown). Thus, the 220-, 190-, and 160-kDa polypeptides are distinct size forms of ␣ 1A subunits. From these results, we cannot exclude that one or more of the multiple size forms of ␣ 1A is created by in vitro proteolysis. However, it seems unlikely because inclusion of high concentrations of several protease inhibitors and careful handling of all instruments including rotors to keep them at 0°C did not alter the appearance of multiple forms of ␣ 1A . In addition, results presented below indicate that the 220-and 190-kDa forms contain the NH 2 -and COOH-terminal sequences predicted from cDNA sequence, indicating they have not been proteolytically cleaved. The doublet band with an apparent molecular mass of 220 kDa appears smaller than the deduced molecular mass of ␣ 1A from cDNA sequence (250 kDa; Starr et al., 1991), but it may represent the full-length form of ␣ 1A because ␣ 1 subunits migrate anomalously in SDS-PAGE (De Jongh et al., 1991). For example, an apparently full-length form of ␣ 1C (predicted molecular mass of 250 kDa; Snutch et al., 1991) migrated to a similar position in the same gel (Fig. 3, lanes 1 and 3) (Hell et al., 1993a).
We also examined the ␣ 1A polypeptides in immunoprecipitation experiments (Fig. 4). To isolate and detect ␣ 1A in immunoprecipitation, we biotinylated the WGA-purified glycoprotein fraction, isolated ␣ 1A by double immunoprecipitation with affinity-purified anti-CNA3 antibodies, and visualized ␣ 1A by the streptavidin-biotin detection method (see "Experimental Procedures"). When samples were analyzed in 5% acrylamide SDS-PAGE gels, anti-CNA3 antibody revealed at least two distinct immunoreactive polypeptides, which correspond to 220-and 190-kDa forms detected in immunoblotting (Fig. 4, lane 1). The 190-kDa polypeptide was the major form, whereas 220-kDa polypeptide was a minor form and was often visualized as a doublet band as in immunoblotting. Occasionally, an additional band with a molecular mass of 160 kDa was weakly stained in immunoprecipitation experiments. The 160-kDa polypeptide may not be sufficient in quantity to be detected consistently in double immunoprecipitation.
The immunoreactive polypeptides of 220 and 190 kDa were specifically recognized by affinity-purified anti-CNA3 antibod-ies in immunoprecipitation, since preincubation with CNA3 peptide at 50 M eliminated the immunoprecipitation of ␣ 1A polypeptides with anti-CNA3, but the CNC1 peptide at the same concentration had no effect (Fig. 4, lanes 3 and 4). In addition, neither band was recognized when immunoprecipitated with nonspecific antibodies (Fig. 4, lane 6). These specific bands of ␣ 1A polypeptide were only observed with affinitypurified anti-CNA3 antibodies, whereas affinity-purified antibodies against ␣ 1B revealed distinct bands with apparent molecular masses of 230 and 210 kDa in parallel experiments (data not shown).
Calcium channels are multisubunit complexes and may interact with other cellular components such as cytoskeletal proteins and synaptic vesicle proteins in immunoprecipitation. Therefore, it is possible that the proteins immunoprecipitated by anti-CNA3 antibodies under native conditions might be associated proteins of similar size to the ␣ 1 subunit rather than the ␣ 1 subunit itself. To exclude other proteins from the immunoprecipitates, double immunoprecipitation experiments were performed under conditions that should completely dissociate the calcium channel subunits and associated proteins. Anti-CP(1382-1400), which recognizes a segment of the ␣ 1 subunit whose sequence is conserved in all calcium channel ␣ 1 subunits so far characterized, was used as a probe in the second immunoprecipitation. The CP(1382-1400) sequence is accessible to anti-CP(1382-1400) only after solubilization in Triton X-100, which removes the ␣ 2 and ␦ subunits (Ahlijanian et al., 1991). Following the double immunoprecipitation with anti-CNA3 and anti-CP(1382-1400) antibodies, two immunoreactive bands corresponding in size to the 220-and 190-kDa polypeptides were visualized (Fig. 4, lane 1), indicating that these immunoreactive polypeptides are ␣ 1A subunits. Two forms of ␣ 1A polypeptides with apparent molecular masses of approximately 220 and 190 kDa were also recognized by the affinitypurified anti-CNA1 antibody, which is directed against a unique amino acid sequence in the intracellular loop between domains II and III of ␣ 1A (residues 865-881) immediately on the NH 2 -terminal side of the CNA3 sequence. Double immunoprecipitation with anti-CNA1 and anti-CP(1382-1400) antibodies revealed two immunoreactive bands with molecular masses of 220 and 190 kDa, and the 190-kDa polypeptide was the major form of ␣ 1A as detected with anti-CNA3 antibody (Fig. 4, lane 5). These observations in immunoblotting and immunoprecipitation experiments demonstrate that ␣ 1A subunits consist of at least two distinct polypeptides that are specifically recognized by anti-CNA3 antibody, and that the 190-kDa polypeptide is a major form of ␣ 1A whereas the 220-kDa polypeptide is a minor form of this subunit.
Identification of the Predicted NH 2 -and COOH-terminal Sequences of the ␣ 1A Polypeptides-␣ 1 subunits of skeletal muscle calcium channels and class B, C, and D brain calcium channels each have multiple size forms that are truncated at the NH 2 or COOH termini (De Jongh et al., 1989Westenbroek et al., 1992;Hell et al., 1993aHell et al., , 1993bHell et al., , 1994Leveque et al., 1994). To test if this is true for ␣ 1A , we used antibodies that recognize the NH 2 -terminal (anti-BI-Nt) or COOH-terminal (anti-BI-1-Ct) of ␣ 1A subunits deduced from cDNA sequence. Two distinct cDNAs encoding ␣ 1A subunits have been cloned: rbA from rat brain and BI (isoforms BI-1 and BI-2) from rabbit brain (Starr et al., 1991;Mori et al., 1991). The anti-BI-Nt antibody is directed against the NH 2 terminus (residues 1-23) of BI clone (Mori et al., 1991), which is conserved in all known isoforms of ␣ 1A subunits. Two distinct BI cDNA clones, designated BI-1 and BI-2, differ from each other in COOH-terminal sequence (Mori et al., 1991). Anti-BI-1-Ct was raised against the COOH terminus (residues 2237-2254) of the BI-1 clone (Mori et al., 1991), which is conserved in the rbA clone (Starr et al., 1991). We examined the immunoreactive ␣ 1A polypeptides for the presence of the predicted NH 2 -and COOH-terminal sequences using anti-BI-Nt and anti-BI-1-Ct.
Immunoblotting with anti-BI-Nt antibodies revealed four immunoreactive bands with apparent molecular masses of 220, 190, 160, and 95 kDa in a 5% acrylamide gel (Fig. 5, lane 1). These immunoreactive polypeptides were specifically detected with anti-BI-Nt antibodies, since 0.2 M of the BI-Nt peptide blocked the interaction of anti-BI-Nt with the immunoreactive polypeptides (Fig. 5, lane 2). The immunoreactive band of 190 kDa was the major form, and the band at 220 kDa was a doublet as observed with anti-CNA3. In other blots, we stripped the membrane used for immunoblotting with anti-BI-Nt antibodies by incubation at 50°C for 30 min in Tris-HCl buffer (pH 6.7) containing 2% SDS and 20 mM dithiothreitol, and re-probed with anti-CNA3 antibodies. Immunoreactive bands with molecular masses of 220, 190, and 160 kDa detected with anti-BI-Nt or anti-CNA3 were identical (data not shown). Anti-BI-1-Ct antibodies revealed an immunoreactive band with an apparent molecular mass of 190 kDa, which was blocked by preincubation with 2 M BI-1-Ct peptide (Fig. 5, lanes 3 and 4). Anti-BI-1-Ct antibodies did not detect immunoreactive bands with molecular mass values of 220 or 160 kDa in immunoblots, possibly because of insufficient quantity of these polypeptides in situ. Thus, immunoblotting with anti-BI-Nt and anti-BI-1-Ct shows that the 190-kDa form of ␣ 1A polypeptide contains both the predicted NH 2 -and COOH-terminal ends of the ␣ 1A subunits.
It is possible that anti-BI-Nt or anti-BI-1-Ct may recognize different immunoreactive polypeptides with equivalent molecular weights or may recognize ␣ 1B since the BI-Nt sequence is 54% identical to the corresponding ␣ 1B sequence. To exclude this possibility, we performed double immunoprecipitation with anti-CNA3 antibodies and either anti-BI-Nt or anti-BI-1-Ct antibodies (Fig. 6). We used anti-BI-Nt or anti-BI-1-Ct antibodies for the first immunoprecipitation and followed with anti-CNA3 or anti-CP(1382-1400) in the second immunoprecipitation. Double immunoprecipitation with anti-BI-Nt and anti-CP(1382-1400) revealed two distinct immunoreactive polypeptides with molecular masses of 220 and 190 kDa (Fig. 6,  lane 1), and immunoprecipitation with anti-BI-1-Ct and anti-CP(1382-1400) antibodies detected two forms of ␣ 1A polypeptide with equivalent molecular weights (Fig. 6, lane 4). In double immunoprecipitation with anti-BI-Nt and anti-CNA3 antibodies, ␣ 1A polypeptides recognized by anti-BI-Nt were specifically immunoprecipitated with anti-CNA3 antibody, since preincubation of 50 M CNA3 peptide blocked the interaction of anti-CNA3 with ␣ 1A polypeptides (Fig. 6, lanes 2 and  3). Similarly, immunoprecipitation with anti-BI-1-Ct and anti-CNA3 detected immunoreactive polypeptides of 220 and 190 kDa and was blocked by 50 M CNA3 peptide (Fig. 6, lanes 5  and 6). These results demonstrated that anti-CNA3, anti-BI-Nt, and anti-BI-1-Ct antibodies recognized the same immunoreactive ␣ 1A polypeptides with molecular mass values of 220 and 190 kDa, and that 220-and 190-kDa forms of ␣ 1A have both the predicted NH 2 -and COOH-terminal ends of ␣ 1A . These results indicate that these isoforms of ␣ 1A do not result from post-translational proteolytic processing, but may instead be products of alternative RNA splicing.
Phosphorylation of ␣ 1A Subunits by Second Messenger-activated Protein Kinases-Calcium channels are regulated by phosphorylation by multiple protein kinases (Tsien et al., 1986;Levitan, 1988;Miller, 1990;Catterall, 1994b). To examine the phosphorylation of ␣ 1A subunits by second messenger-activated protein kinases, ␣ 1A was purified by immunoprecipitation with affinity-purified anti-CNA3 antibodies. The resulting immune complexes were incubated with different kinases in the presence of [␥-32 P]ATP. After washing, the PAS-antibody-channel complexes were dissociated with 1.5% SDS-sample buffer and diluted with 1% Triton, and ␣ 1A subunits were re-immunoprecipitated with anti-CP(1382-1400) (see "Experimental Procedures"). Following phosphorylation with PKA, two labeled ␣ 1A polypeptides with molecular masses of 220 and 190 kDa were observed. No incorporation of radiolabel was detected with a non-immune rabbit IgG in the first immunoprecipitation (Fig.  7, lane 2), or if 50 M CNA3 peptide was preincubated prior to the first immunoprecipitation (data not shown). The 220-kDa band was observed as a doublet in a high resolution autoradiogram, as demonstrated in immunoblotting and immunoprecipitation experiments (Fig. 3, lane 1, and Fig. 4, lane 1). Immunoblotting and immunoprecipitation experiments showed that the 190-kDa band is a major form of ␣ 1A and the 220-kDa band is minor in quantity. In contrast, PKA phosphorylated the FIG. 5. Detection of ␣ 1A with anti-BI-Nt and anti-BI-1-Ct antibodies by immunoblotting. Membrane glycoprotein fractions were isolated from solubilized brain membranes by WGA affinity chromatography, and calcium channels were concentrated by adsorption to heparin agarose, extracted, and analyzed by SDS-PAGE. Immunoblotting was performed with anti-BI-Nt (lanes 1 and 2) or with anti-BI-1-Ct antibodies (lanes 3 and 4) as described before (Fig. 3). To test for nonspecific labeling, anti-BI-Nt antibody was preincubated overnight on ice with 0.2 M BI-Nt peptide (lane 2), and anti-BI-1-Ct was preincubated with 2 M BI-1-Ct peptide (lane 4). Molecular mass markers are indicated in Fig. 3.  5 and 6). Anti-CNA3 antibodies were preincubated overnight on ice with 50 M CNA3 peptide (lanes 3 and 6) in the second immunoprecipitation. Note that the blot was overexposed to demonstrate the immunoreactive band at 220 kDa (lane 4). Molecular markers are given in Fig. 3. 220-kDa polypeptide much more extensively than the 190-kDa polypeptide. A plausible explanation for this result is that the 220-kDa form contains PKA phosphorylation sites that are not present in the 190-kDa form. PKA phosphorylated these two ␣ 1A forms similarly when disulfide bonds were not reduced, when disulfide bonds were reduced with 20 mM dithiothreitol in SDS-sample buffer, or when samples were solubilized with 2% CHAPS or 0.1% digitonin (data not shown).
The 190-kDa form of ␣ 1A was a substrate for phosphorylation by PKC and PKG (Fig. 7, lanes 3 and 5). Control rabbit IgG was ineffective in precipitating the 190-kDa polypeptides phosphorylated by these enzymes confirming the identification of ␣ 1A (Fig. 7, lanes 4 and 6). The 220-kDa form of ␣ 1A could not be visualized as a phosphorylated polypeptide by PKC or PKG. However, we cannot exclude the possibility that the 220-kDa form of ␣ 1A is also a substrate for these kinases, since 220-kDa polypeptide could be present in insufficient quantity for detection of phosphorylation by these enzymes. Nevertheless, the results show that PKA preferentially phosphorylates the 220-kDa form of ␣ 1A whereas PKC and PKG preferentially phosphorylate the 190-kDa form.

DISCUSSION
High Affinity -Conotoxin MVIIC Binding Site on ␣ 1A -Our pharmacological experiments using radioactive ligands show that anti-CNA3 antibody does not immunoprecipitate either dihydropyridine or -conotoxin GVIA binding sites, but specifically immunoprecipitates the high affinity -conotoxin MVIIC receptor site (K d ϳ100 pM). These results are consistent with the conclusion that ␣ 1A forms Q-type calcium channels because -conotoxin MVIIC blocks ␣ 1A currents expressed in Xenopus oocytes and Q-type currents in the cerebellar granule neurons with IC 50 less than 150 nM Stea et al., 1994a;Zhang et al., 1993;Randall and Tsien, 1995). In contrast, native P-type calcium channels are blocked by higher concentrations of -conotoxin MVIIC (IC 50 of 1-10 M; Hillyard et al., 1992). The K d in our biochemical experiments is substantially lower than the IC 50 obtained in electrophysiological experiments with ␣ 1A channels (IC 50 Ͻ 150 nM). This discrepancy in IC 50 may result from differences in the ionic strength in the various experimental conditions. Na ϩ , Ba 2ϩ , and Ca 2ϩ all inhibit binding of -conotoxin MVIIC at physiological concentrations (Kristipati et al., 1994), but are absent from our binding assay solutions.
Our results are closely correlated with covalent cross-linking experiments identifying the polypeptides that bind -conotoxin MVIIC (Woppmann et al., 1994). Polypeptides with apparent molecular masses of 220, 170, 150, and 140 kDa were observed. Although one or more of these bands may represent ␣ 2 subunits of class A channels or other associated proteins, it seems most likely that the bands of 220 and 170 kDa correspond to the 220and 190-kDa isoforms of ␣ 1A .
␣ 1A subunits have been suggested to be components of both P-type and Q-type calcium channels. ␣ 1A is localized in high density in the cell bodies and dendrites of cerebellar Purkinje cells where P-type calcium currents are recorded, as well as in the cell bodies and nerve terminals of cerebellar granule cells where Q-type calcium currents are recorded (Westenbroek et al., 1995). Coexpression of ␣ 1A with various calcium channel ␤ subunits results in modulation of the amplitude, time course, and the voltage-dependent properties of the ␣ 1A calcium currents (Mori et al., 1991;Sather et al., 1993;Stea et al., 1994a;Soong et al., 1994;De Waard et al., 1994). ␣ 1A calcium currents expressed in Xenopus oocytes inactivate more rapidly than native P-type calcium channels, but coexpression of the rbA-I or rbA-II isoforms of the ␣ 1A subunit with a ␤ subunit (rbA-I with ␤ 2a, or rbA-II with ␤ 1b ) in Xenopus oocytes gives currents with much slower inactivation like a P-type calcium channel (Stea et al., 1994a;Soong et al., 1994). However, the sensitivity of ␣ 1A to -conotoxin MVIIC and -agatoxin IVA is not significantly affected in these coexpression studies. These findings suggest that pharmacological and physiological differences between Q-type and P-type calcium channels may be due to an unidentified isoform of ␣ 1A , which may result from alternative RNA splicing or post-translational modifications or may result from assembly with other auxiliary subunits of calcium channels.
Identification of Multiple Alternatively Spliced Forms of ␣ 1A Subunits-In our immunoblotting, immunoprecipitation, and phosphorylation experiments, affinity-purified anti-CNA3 antibodies identified at least two distinct ␣ 1A polypeptides: a minor doublet of polypeptides with an apparent molecular mass of approximately 220 kDa and a major polypeptide with an apparent molecular mass of 190 kDa. These polypeptides were specifically recognized by anti-CNA3 antibody, since the CNA3 peptide blocked binding of anti-CNA3 antibody to these immunoreactive polypeptides. Multiple size forms of calcium channel ␣ 1 subunits were first described for the skeletal muscle calcium channel (De Jongh et al., 1989, and found for neuronal L-type Hui et al., 1991;Williams et al., 1992b;Hell et al., 1993a) and non-L-type calcium channels (Mori et al., 1991;Starr et al., 1991;Westenbroek et al., 1992;Coppola et al., 1994;Leveque et al., 1994;Williams et al., 1994). In skeletal muscle, the two size forms of ␣ 1 subunits may arise from post-translational processing because only a single mRNA has been characterized. In contrast, sequencing of cDNA clones encoding the neuronal calcium channels has revealed multiple isoforms in each case which vary in the cytoplasmic loops and COOH-terminal regions (Mori et al., 1991;Soong et al., 1994;Coppola et al., 1994;Williams et al., 1992aWilliams et al., , 1992bWilliams et al., , 1994Snutch et al., 1991;Hui et al., 1991;Niidome et al., 1992;Soong et al., 1993). For ␣ 1A subunits in rat central nervous system, four distinct transcripts were identified by Northern blot analysis (Starr et al., 1991), and alternative RNA splicing of a single rat class A gene has been shown to generate isoforms (rbA-I and rbA-II) that have similar molecular size (approximately 250 kDa; Soong et al., 1994). No alternatively spliced mRNAs encoding ␣ 1A subunits of substantially different size have been characterized previously in rat brain. However, our immunochemical experiments demonstrate that class A calcium channels are composed of multiple isoforms of ␣ 1A with different molecular size, and that these multiple isoforms of ␣ 1A may be produced by alternative RNA splicing rather than FIG. 7. Phosphorylation of the class A calcium channel ␣ 1 subunits by PKA, PKC, and PKG. Class A calcium channel ␣ 1 subunits were purified from WGA glycoprotein fractions by immunoprecipitation with anti-CNA3 (lanes 1, 3, and 5), or with rabbit control antibodies (lanes 2, 4, and 6). Immunoprecipitated ␣ 1A was phosphorylated with PKA ( lanes 1 and 2), PKC (lanes 3 and 4), or PKG (lanes 5 and 6) and reprecipitated with anti-CP(1382-1400) antibodies as described under "Experimental Procedures." Molecular mass markers are described in Fig. 3. by post-translational proteolytic processing. These results suggest that additional uncharacterized mRNAs encoding different size forms of ␣ 1A must be present in rat central nervous system.
The BI ␣ 1 subunit cDNA clones from rabbit brain (BI-1 and BI-2) encode Q-type calcium channels when expressed in Xenopus oocytes (Mori et al., 1991;Sather et al., 1993). Analysis of BI clones revealed multiple isoforms differing by insertion/ deletion of 349 amino acids in the loop between domains II and III (residues 772-1,120) and 195 amino acids in COOH-terminal region and by alternative expression of a 28-amino acid substitution in COOH-terminal region (Mori et al., 1991). These differences, which apparently result from alternative RNA splicing, can give rise to at least eight distinct mRNAs encoding multiple size forms of ␣ 1A in rabbit brain. The differences in size caused by either of these large deletions would be sufficient to reduce the apparent size of the ␣ 1A subunit from 220 to 190 kDa. These findings suggest the possibility that ␣ 1A subunits in both rat and rabbit contain multiple splicing cassettes in the loop between domains II and III and in the COOHterminal region and that the two size forms that we have observed in our biochemical experiments are derived from these alternative splicing events. Because the known cDNAs could encode multiple ␣ 1 subunits with a size of approximately 190 kDa, it is possible that multiple ␣ 1 isoforms are contained within the protein bands present in this region of the gel. In addition, because the immunostaining with anti-BI-Ct was weaker than with anti-BI-Nt, it is possible that this band also contains ␣ 1A subunits that have been truncated at the COOH terminus by proteolytic processing.
In immunoblotting experiments, affinity-purified anti-CNA3 and anti-BI-Nt antibodies specifically identified an additional immunoreactive band of ␣ 1A with an apparent molecular mass of 160 kDa (Figs. 3 and 5). The biochemical properties of the 160-kDa form of ␣ 1A polypeptide could not be extensively characterized, since this form was not consistently detected in double immunoprecipitation experiments. However, our results show that the 160-kDa polypeptide contains both the CNA3 sequence and the BI-Nt sequence. It may be an additional spliced isoform of ␣ 1A , or a proteolytic product of the longer forms of ␣ 1A polypeptide, which has a cleaved COOH terminus.
Possible Physiological Significance of Differential Phosphorylation of Class A Calcium Channels-Since P-type calcium currents were first described in the cerebellar Purkinje neurons and the presynaptic terminal of the squid giant synapse (Llinas et al., 1989), intensive studies of P-type and/or Q-type channels have demonstrated their broad distribution in the central and peripheral nervous systems and in the endocrine system (Regan et al., 1991a(Regan et al., , 1991bHillman et al., 1991;Mintz et al., 1992aMintz et al., , 1992bMintz et al., , 1993Uchitel et al., 1992;Usowicz et al., 1992;Swartz et al., 1993;Regehr and Mintz, 1994;Wheeler et al., 1994;Castillo et al., 1994;Artalejo et al., 1994;Stea et al., 1994a, Brown et al., 1994Westenbroek et al., 1995). On the basis of electrophysiological findings, P-type and/or Q-type calcium channels are involved in excitatory and inhibitory synaptic transmission at central synapses (Takahashi and Momiyama, 1993, Luebke et al., 1993Castillo et al., 1994, Wu and Saggau, 1994, Wheeler et al., 1994, Regehr and Mintz, 1994 and at the mammalian neuromuscular junction (Uchitel et al., 1992;Bowersox et al., 1995). -Agatoxin IVA-sensitive and -conotoxin GVIA-resistant calcium channels regulate glutamate release in rat brain synaptosomes (Turner et al., 1992) or in hippocampal slices (Gaur et al., 1994) after potassium-induced depolarization. Immunocytochemical experiments using anti-peptide antibodies clearly demonstrate the subcellular localization of class A calcium channels in the presynaptic terminals of many central neurons (Westenbroek et al., 1995) and in presynaptic terminals at the neuromuscular junction (Ousley and Froehner, 1994;Sugiura et al., 1995). These findings indicate that at least one isoform of P-type and/or Q-type calcium channels is localized in presynaptic terminals, controls the neurotransmitter release at central synapses, and may contribute to synaptic plasticity (Wheeler et al., 1994). However, most P-type and/or Q-type calcium currents identified in the electrophysiological experiments are recorded in the somata or dendrites of neurons such as Purkinje neurons, cerebellar granule cells, neocortical pyramidal cells, and dorsal root ganglia (Llinas et al., 1989;Usowicz et al., 1992;Regan et al., 1991aRegan et al., , 1991bMintz et al., 1992aMintz et al., , 1992bRandall and Tsien, 1995), and ␣ 1A subunits are also observed in these locations by immunocytochemistry (Westenbroek et al., 1995). These results indicate that class A calcium channels are also localized in the postsynaptic membrane. Since class A calcium channels are present in different subcellular locations and participate in different physiological events, the distinct isoforms we have observed in these experiments may be specialized for localization in specific subcellular compartments and for function in different neuronal processes. P-type and/or Q-type calcium channels are modulated by GTP-binding proteins (G protein) and protein phosphorylation. P-type channels in Purkinje neurons and spinal cord interneurons are inhibited by ␥-aminobutyric acid (GABA) through GABA B receptor activation and this inhibition is mediated through G proteins . On the other hand, in hippocampal CA3 pyramidal neurons, P-type calcium channels are potentiated by adenosine through A 2 receptor activation (Mogul et al., 1993). This potentiation involves a PKA-dependent process (Mogul et al., 1993). Phosphorylation by second messenger-activated protein kinases is a well known pathway for functional modulation of neuronal calcium channels. Injection of cerebellar mRNA into Xenopus oocytes leads to the expression of a single type of voltage-gated calcium channels similar to P-type channels, and this calcium current is enhanced by activators of PKA and PKC (Fournier et al., 1993a(Fournier et al., , 1993b. In contrast, I Ba of ␣ 1A channels coexpressed with ␤ subunit in Xenopus oocyte is not affected by the activation of PKC (Stea et al., 1994b). Whereas the functional effects of phosphorylation of P-type and Q-type calcium channel are still incompletely described, our results provide the first evidence that class A calcium channel ␣ 1 subunits are substrates for phosphorylation by PKA, PKC, and PKG, and indicate that the different ␣ 1 subunit size forms may be differentially phosphorylated and differentially regulated. Further work is required to determine whether different isoforms of the class A calcium channels are differentially regulated by PKA, PKC, and PKG in vivo and to evaluate the physiological effect of phosphorylation on ␣ 1A channel function.