Identification of three subunits of the high affinity omega-conotoxin MVIIC-sensitive Ca2+ channel.

N-, P- and Q-type voltage-dependent Ca2+ channels control neurotransmitter release in the nervous system and are blocked by ω-conotoxin MVIIC. In this study, both a high affinity and a low affinity binding site for ω-conotoxin MVIIC were detected in rabbit brain. The low affinity binding site is shown to be present on the N-type Ca2+ channel. Using optimized conditions for specific labeling of the high affinity ω-conotoxin MVIIC receptor and a panel of subunit specific antibodies, the molecular structure of the high affinity receptor was investigated. We demonstrate for the first time that this receptor is composed of at least α1A, α2δ, and any one of the four brain β subunits. Such association of different β subunits with α1A and α2δ components may produce Ca2+ channels with distinct functional properties, such as P- and Q-type.

Voltage-dependent Ca 2ϩ channels play a central role in neurotransmitter release and muscle contraction. Several voltagedependent Ca 2ϩ channels have been identified which are classified by electrophysiological and pharmacological criteria as T-, L-, N-, P-, and Q-type Ca 2ϩ channels (1,2). To date, three different types of these Ca 2ϩ channels (N-, P-, and Q-type) are known to mediate synaptic transmission in the nervous system (3,4). Biochemical characterization of the brain N-type Ca 2ϩ channel revealed that it is composed of ␣ 1B , ␣ 2 ␦, and ␤ subunits and a unique 95-kDa protein (5,6). Among the subunits of brain N-type Ca 2ϩ channels, ␣ 1 is the pore-forming subunit and the receptor for antagonist drugs (7); ␤ is cytoplasmic and regulates the function of ␣ 1 (8 -10); ␣ 2 ␦ is heavily glycosylated and modifies the functional properties of the channel complex (11,12). Molecular cloning has thus far identified six different ␣ 1 genes (S, A, B, C, D, and E), four different ␤ genes (1, 2, 3, and 4), and one ␣ 2 ␦ gene. Several splice isoforms for ␣ 1 , ␤, and ␣ 2 ␦ genes have also been identified (13)(14)(15)(16)(17)(18).
Although P-and Q-type Ca 2ϩ channels seem to play an even greater role than N-type channels in neurotransmission, little is known about the structural aspects of these channels (4,19).
Pharmacologically, both channels can be blocked by -agatoxin IVA and -conotoxin (CTX) 1 MVIIC (20 -22), although both have different sensitivities to -agatoxin IVA (22). However, it has not been established with certainty whether the P-and Q-type Ca 2ϩ channels are encoded by the same ␣ 1 subunit gene; or whether they represent different ␣ 1 gene products (1,12,22,23). Since the activity of ␣ 1A subunits expressed in Xenopus oocytes can be blocked by both -agatoxin IVA and -CTX MVIIC, it has been proposed that this subunit may be the pore-forming subunit of P-and/or Q-type Ca 2ϩ channels (1,23). Lack of specific blockers to distinguish P-and Q-type channels has made it difficult to further characterize these channels by either electrophysiological or biochemical approaches.
The major types of Ca 2ϩ channels controlling neurotransmission (N-, P-, and Q-type) have been shown to be blocked by -CTX MVIIC (24,25). In this study, we identified a high affinity -CTX MVIIC receptor which can be distinguished from the low affinity N-type Ca 2ϩ channel. Our data demonstrate for the first time that the high affinity -CTX MVIIC receptor is composed of at least an ␣ 1A , ␣ 2 ␦, and any one of the four different brain ␤ subunits. The association of different ␤ subunits with the ␣ 1A subunit supports a structural heterogeneity of Ca 2ϩ channels which may be the basis for functional differences observed between P-and Q-type channels.
Ligand Binding Analysis of the -CTX MVIIC Receptor and the N-type Ca 2ϩ Channel-Rabbit brain membranes were prepared as detailed elsewhere (26). Protein concentration was determined by the method of Lowry (27). To obtain enough material of 125 I--CTX MVIIC, 125 I--CTX MVIIC (specific activity, 2200 Ci/mmol) was mixed with unlabeled -CTX MVIIC to reach a specific activity of 440 Ci/mmol. Aliquots (50 g of protein) of membranes were incubated with increasing concentrations (0.05-8 nM) of 125 I--CTX MVIIC at 25°C for 1 h in a total volume of 200 l of buffer A (10 mM HEPES, pH 7.4, 0.2 mg/ml bovine serum albumin, 100 mM NaCl, 0.75 mM benzamidine, 0.1 mM phenylmethylsulfonyl fluoride) containing 0.1 mM EDTA, and 0.1 mM EGTA. The receptor-ligand complexes were collected on GF/B filters and washed rapidly with 4 ml of ice-cold buffer A. The counts remaining on the filter represent total binding. Nonspecific binding was determined by the addition of 0.5 M nonradioactive ligand prior to the addition of radioactive ligand. Specific binding was calculated by sub-* 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  tracting nonspecific binding from total binding. 125 I--CgTX GVIA binding was assayed similarly to 125 I--CTX MVIIC but without EGTA and EDTA in buffer A. Binding of 125 I--CTX MVIIC and 125 I--CgTX GVIA to the solubilized membrane extracts are the same as described above. All binding data were analyzed with Grafit software (Sigma) and all results were presented as value Ϯ S.E. Each sample point represents the average of triplicates for all binding and immunoprecipitation experiments in this study.
Competition of Unlabeled -CTX MVIIC with 125 I--CgTX GVIA Binding-Rabbit brain membranes (4 mg/ml) were solubilized for 1 h on ice in buffer (0.5 M NaCl, 1% (w/v) digitonin, 10 mM HEPES, pH 7.4) plus protease inhibitors mixture A (0.6 g/ml pepstatin A, 0.5 g/ml aprotinin, 0.5 g/ml leupeptin, 0.1 mM phenylmethylsulfonyl fluoride, 0.75 mM benzamidine). Nonsolubilized material was sedimented by centrifugation in a Beckman TL 100 for 10 min at 400,000 ϫ g. The supernatant was diluted 8-fold with buffer (10 mM HEPES, pH 7.4, 0.1 mM EDTA, 0.1 mM EGTA, 0.5 g/ml aprotinin, 0.5 g/ml leupeptin). Aliquots were mixed with increasing concentrations (0.001 nM to 1 M) of -CTX MVIIC for 10 min before addition of 125 I--CgTX GVIA (42.5 pM). After incubating for 1 h at room temperature, receptor-ligand complexes were collected on GF/B filters and washed as described above. For binding to brain membranes, 50 g of brain membranes were incubated with increasing concentrations of -CTX MVIIC for 10 min in 200 l of buffer (10 mM HEPES, pH 7.4, 62.5 mM NaCl, 0.1% (w/v) bovine serum albumin, 0.1 mM EDTA, 0.1 mM EGTA, 0.5 g/ml aprotinin, 0.5 g/ml leupeptin) before addition of 125 I--CgTX GVIA (0.1 nM). The IC 50 was determined by Grafit software using the average of triplicates. Since the concentration of 125 I--CgTX GVIA used should not affect the presentation of IC 50 in the competition curve, data were normalized by setting 125 I--CgTX GVIA binding in the absence of -CTX MVIIC as 100%.
Primers Preparation of Glutathione S-transferase (GST) Fusion Proteins-pGEX vectors were used to produce rabbit ␣ 1A and ␣ 1C , rat ␤ 1b , ␤ 2 , ␤ 3 , and ␤ 4 GST fusion proteins. The II-III loop region of rabbit ␣ 1A cDNA was amplified with ␣ 1A primers. The resulting products were cut with BamHI and EcoR I and ligated into a pGEX 1 vector which had been linearized with BamHI and EcoR I. The II-III loop region of rabbit ␣ 1C cDNA (GenBank™ accession no. X15539) was amplified with ␣ 1C primers. The resulting PCR products were cut with BamHI and EcoR I and ligated into linearized pGEX 2T. pBluescript rat ␤ 1b plasmid was cut at NcoI and HindIII restriction enzyme sites. The fragment containing nucleotide 1343-1987 was isolated and ligated into a pGEX KG vector which had also been linearized with NcoI and HindIII. PCR products were obtained from rat PC12 cDNA with 2T␤ 2 primers. The PCR products were cut with BamHI and EcoR I and ligated into a linearized pGEX 2T vector. The C-terminal region of ␤ 4 cDNA was amplified from ␤ 4 cDNA plasmid with ␤ 4 primers. The resulting PCR products were cut with BamHI and XbaI and ligated into linearized pGEX KG vector. ␤ 3 GST fusion protein was prepared as detailed elsewhere (28). Each fusion protein construct was sequenced to verify the correct insertion and each fusion protein was purified as described previously (28).
Production of Antibodies against GST Fusion Proteins-250 -400 g of purified GST fusion proteins were emulsified in 1 ml of Freund's complete (initial injection) or incomplete adjuvant (subsequent injections), and injected into multiple subcutaneous sites on New Zealand White rabbits at 2-week intervals. Antisera were collected after the second injection and tested for reactivity with the antigen on immunoblots of GST fusion proteins. Listed here are the injected animals and their immunogen: rabbit 141 and 142 (fusion protein (FP) ␤ 1 ), rabbit 143 and 144 (FP ␤ 2 ), rabbit 145 and 146 (FP ␤ 4 ), sheep 49 (FP ␤ 3 ), sheep 37 (FP ␣ 1A ), and sheep 41 (FP ␣ IC ). The monoclonal anti-␣ 1B IgG (41) was secreted by a hybridoma produced from the lymph node B cells of a rat hyperimmunized with FP ␣ 1B (28). ␣ 1S and ␣ 2 ␦ antibodies are as described elsewhere (26,29) Immunoblot Analysis-Fusion proteins or rabbit brain proteins (20 -100 g of protein/lane) were resolved on 3-12% gradient SDS-polyacrylamide gels and transferred to nitrocellulose. Horseradish peroxidase colorimetric detection was performed as described previously (6). An ECL detection system (Amersham Corp.) was used according to the manufacturer's instructions. The ␤ 2 subunit was detected with an ECL immunoblotting technique; all others used horseradish peroxidase colorimetric detection. Since polyclonal antibodies vary in their affinities for different epitopes, immunoblot staining does not necessarily reflect the level of different proteins in a quantitative way. Immunoprecipitation of 35 S-Labeled Full-length ␤ Subunits-The full-length ␤ 1b , ␤ 2 , ␤ 3 , and ␤ 4 subunits were in vitro translated using a coupled transcription and translation system (Promega) as detailed elsewhere (30). Immunoprecipitation conditions were intended to be identical to the immunoprecipitation conditions used for native Ca 2ϩ channels as described in the following section. Brain membranes (4 mg/ml) were solubilized and diluted as indicated under "Competition of Unlabeled -CTX MVIIC with 125 I--CgTX GVIA Binding." Triplicates of 25 l of each ␤ subunit antibody and control ␣ 1C antibody were coupled to protein G-Sepharose beads by overnight incubation in phosphate buffered saline. Subsequently, these antibodies were washed with phosphate-buffered saline extensively and incubated with aliquots of solubilized brain extract (1 ml) in buffer (0.1% digitonin, 10 mM HEPES, pH 7.4, 100 mM NaCl, 0.5 g/ml aprotinin, 0.5 g/ml leupeptin) for 15 min before the addition of 200,000 cpm of 35 S-labeled full-length ␤ subunit per tube for overnight incubation at 4°C. The antibodyprotein G beads were washed extensively with phosphate-buffered saline and subjected to scintillation counting. The nonspecific binding determined from the control ␣ 1C antibody was subtracted from the total binding for each antibody to obtain specific binding. Data were presented as values Ϯ S.E.
Immunoprecipitation of Various Types of Ca 2ϩ Channels-Brain membranes (4 mg/ml) were solubilized and diluted as described above. The solubilized protein was labeled with 45 pM 125 I--CTX MVIIC for 1 h at room temperature. Then aliquots of the labeled extract (0.5-1 ml) were incubated for 3 h with 50 l of antibody-protein G-Sepharose beads. The beads were quickly washed with 1.5 ml of ice-cold buffer A twice, and total binding was quantified by ␥ counting. Nonspecific binding was determined by the addition of 0.5 M unlabeled -CTX MVIIC prior to the addition of the labeled toxin. Immunoprecipitation of L-type Ca 2ϩ channel was as described elsewhere (26). Saturating immunoprecipitation was achieved with 50 l of antibody/tube as compared with that of 100 and 150 l of antibody.
Enrichment of the High Affinity -CTX MVIIC Receptor-Affinitypurified sheep 37 (␣ 1A ) antibody was coupled to Hydrazide Avidgel AX according to the manufacturer's instructions. Protease inhibitors mixture A was present in all buffers during the partial purification. Rabbit brain membranes (1 g) were solubilized in buffer (1% digitonin, 1 M NaCl, 10 mM HEPES, pH 7.4) for 1 h on ice, followed by centrifugation at 140,000 ϫ g for 37 min in a Beckman 45 Ti rotor. The solubilized extracts were diluted 3-fold with ice-cold distilled and deionized water and applied to the heparin-agarose column (50 ml). After washing the heparin column extensively with buffer B (10 mM HEPES, pH 7.4, 0.1% digitonin, 100 mM NaCl), the column was eluted with the buffer B containing 0.7 M NaCl. The eluates from the heparin column were applied to the mAb ␣ 1B (CC18) column. Then the ␣ 1A column was incubated with the void from the ␣ 1B column overnight. The column was washed with 50 ml of buffer B containing 0.6 M NaCl followed by 50 ml of buffer B. Then the column was eluted with glycine buffer (50 mM glycine, pH 2.5, 100 mM NaCl, 0.1% digitonin). 125 l of 2 M Tris (pH 8.0) were added to every milliliter of eluate immediately after the elution. The eluted material was concentrated with Centricon-100 (Amicon) and subjected to SDS-polyacrylamide gel electrophoresis analysis. The protein concentration was determined either by the method of Lowry (27) or a silver binding method by Krystal (31) for measuring nanogram amounts of proteins.
Preparation of Xenopus Oocytes and Injection-Mature Xenopus laevis female frogs were purchased from NASCO (Fort Atkinson, WI). Preparation of Xenopus oocytes, cRNA synthesis, and cRNA injection into oocytes were as described previously (10). Fifty nanoliters of various in vitro transcribed RNA mixtures were injected per oocyte (0.5 g/l ␣ 1A and 0.5 g/l ␣ 2 ␦ b together with either 0.1 g/l rabbit ␤ 2a (16) and 0.1 g/l ␤ 3 in the case of a ␤ 2a :␤ 3 ratio of 1:1, or 0.1 g/l ␤ 2a and 1 g/l ␤ 3 in the case of a ␤ 2a :␤ 3 ratio of 1:10). Electrophysiological recording and data analysis were as described previously (10). Briefly, Ba 2ϩ currents were recorded by a standard two-microelectrode voltageclamp technique using a Dagan amplifier (TEV-200). Voltage and current electrodes (0.5-2 megohms) were filled with 3 M KCl. The bath solution was (in mM): 40 Ba(OH) 2 , 50 NaOH, 2 KCl, 1 niflumic acid, 0.1 EGTA, 5 HEPES, pH 7.4 (methanesulfonic acid). Ba 2ϩ currents were induced by 2-s voltage steps from a holding potential of Ϫ90 mV to a testing potential of ϩ10 mV. The number of inactivating exponential components was determined by the best fitting of the current trace using pClamp software (Axon Instruments) according to the Chebyshev procedure. The decay of the current was fitted by either one single or the sum of two exponential functions. The equation used for two exponentials is I Total ϭ i 1 ⅐exp(Ϫt/ 1 ) ϩ i 2 ⅐exp(Ϫt/ 2 ) ϩ C, where i denotes each monoexponential inactivating component with a time constant . C is the offset.

Rabbit Brain Expresses a High
Affinity -CTX MVIIC-sensitive Ca 2ϩ Channel-Both a high affinity and a low affinity binding site for -CTX MVIIC were detected in rabbit brain (Fig. 1A). Curve fitting of the direct binding curve and Scatchard binding analysis (32) reveal a high affinity binding site with a K d of 0.37 Ϯ 0.09 nM and a B max of 870 Ϯ 92 fmol/mg; and a low affinity binding site with a K d of 4 Ϯ 2 nM and a B max of 290 Ϯ 110 fmol/mg. It is technically difficult to analyze a low affinity site (K d2 Ͼ K d1 ) with a smaller binding capacity because the high affinity site always forms more ligand-bound receptor than the low affinity site (32). The characteristics of the high affinity -CTX MVIIC binding site are in approximate agreement with previous results reported for rat and human brain membranes (24,33,34). In comparison, 125 I--conotoxin GVIA (-CgTX GVIA), a specific blocker for N-type Ca 2ϩ channel, binds to the same rabbit brain membrane with a K d of 45 Ϯ 6 pM and a B max of 207 Ϯ 14 fmol/mg (Fig. 1B). These results demonstrate that the expression level of the high affinity -CTX MVIIC receptor is higher than that of N-type Ca 2ϩ channel in rabbit brain; while the expression of the low affinity receptor is similar to that of N-type Ca 2ϩ channel. In order to further analyze the -CTX MVIIC binding receptor, unlabeled -CTX MVIIC was used to compete with [ 125 I] -CgTX GVIA for binding to rabbit brain membrane and solubilized brain extracts (Fig. 1C). The results reveal an IC 50 of 7.7 Ϯ 0.3 (brain membranes) and 7.9 Ϯ 0.3 nM (solubilized brain extracts) for -CTX MVIIC and a K i of ϳ4.4 nM. All of these data suggest that -CTX MVIIC binds with low affinity (about 4 nM) to the N-type Ca 2ϩ channel. Notably, -CTX MVIIC up to 1 nM did not significantly inhibit 125 I--CgTX GVIA binding to the N-type Ca 2ϩ channel, while it starts to bind to the N-type Ca 2ϩ channel at higher than 1 nM concentration (Fig. 1C). This important property of -CTX MVIIC enabled us to distinguish these two binding sites by specifically labeling only the high affinity receptor with these low toxin concentrations (Ͻ1 nM).
␣ 1A and Any One of the Four ␤ Subunits Are Components of the High Affinity -CTX MVIIC Receptor-A low concentration of 125 I--CTX MVIIC (45 pM) was used to specifically label the high affinity -CTX MVIIC receptor. Polyclonal ␣ 1A subunit antibodies immunoprecipitated 21 fmol of labeled high affinity -CTX MVIIC receptor (84% of total labeled receptors) ( Fig.  2A). In contrast, monoclonal antibodies raised against the ␣ 1S subunit (IIC12) and the ␣ 1B subunit (CC18) together with a polyclonal antibody against the ␣ 1C subunit (sheep 41) did not sediment significant amounts of labeled high affinity -CTX MVIIC receptor (less than 5% of total labeled receptors). In control experiments, however, the mAb against ␣ 1S (IIC12) immunoprecipitated more than 80% of 3 H-PN200 -110 binding to the skeletal muscle L-type Ca 2ϩ channel (26); the mAb against ␣ 1B (CC18) immunoprecipitated more than 80% of brain N-type Ca 2ϩ channel; and the polyclonal antibody against ␣ 1C (sheep 41) immunoprecipitated 80% of cardiac Ltype Ca 2ϩ channel (data not shown). These results demonstrate that the ␣ 1A subunit is a component of the brain high affinity -CTX MVIIC receptor, which is in agreement with the results from rat brain (35). Furthermore, the data confirm that low concentrations of -CTX MVIIC specifically label the brain high affinity -CTX MVIIC receptor but not the N-type Ca 2ϩ channel.
To establish which ␤ subunit is associated with the ␣ 1A pore-forming component of Ca 2ϩ channel, we have produced ␤ subunit subtype specific antibodies. ␤ subunit genes have two highly homologous central domains, but unique N and C termini (Fig. 3A). Although domains I and II of each ␤ subunit share 65 and 78% amino acid identity, respectively, the C termini of all four brain ␤ subunits share only 3% amino acid identity. The C-terminal regions of four ␤ subunits were thus expressed as GST fusion proteins. Excluding the common GST portion of these fusion proteins, the overall amino acid identity of all four fusion proteins is 1.2%. Antibodies were therefore raised against these fusion proteins in order to produce specific antibodies for each ␤ subunit. As shown in Fig. 3B, the purified ␤ 1b , ␤ 2 , ␤ 3 , and ␤ 4 GST fusion proteins are of molecular mass of 52, 42, 39, and 34 kDa, respectively. The purified fusion proteins were used to generate polyclonal antibodies in rabbits (␤ 1b , ␤ 2 , and ␤ 4 ) or sheep (␤ 3 ).
The specificity of each of the four FP antibodies was analyzed by immunoblotting. Crude bacterial lysates expressing the four fusion proteins were loaded on SDS-polyacrylamide gel electrophoresis. Approximately equal amounts of FP␤ 1b , FP␤ 2 , FP␤ 3 , and FP␤ 4 were present on the blots as determined by densitometry (corresponding Coomassie Blue-stained gel not shown). Since all four fusion protein antibodies should also contain antibodies against GST, they were removed by preincubation with bacterial GST lysates. Since an excess of fusion proteins were present in the blots, this assay should be very sensitive to detect any cross reactivity of the antibodies. As shown in Fig. 3C, antibodies to FP␤ 1b , FP␤ 2 , FP␤ 3 , and FP␤ 4 specifically recognized unique antigens and did not cross react with any of the other three fusion proteins in immunoblots. The antibodies to FP␤ 1b , FP␤ 3 , and FP␤ 4 also detected proteolyzed forms of the fusion proteins which were present in low amounts during the production of the fusion proteins.
In addition, the specificity of these antibodies for immunoprecipitation were tested using full-length in vitro translated ␤ 1b , ␤ 2 , ␤ 3 , and ␤ 4 subunits. 35 S-Labeled ␤ 1b , ␤ 2 , ␤ 3 , and ␤ 4 subunits were previously shown to be the only radiolabeled proteins present in the translation products (30). The results show that each antibody was able to immunoprecipitate their respective labeled ␤ subunit but did not cross react significantly with the other three in vitro translated ␤ subunits (Fig. 3D).
In contrast, none of the ␣ 1A or four brain ␤ subunit antibodies were capable of immunoprecipitating the skeletal muscle Ltype Ca 2ϩ channel (Fig. 2C). The C-terminal regions of brain ␤ 1b , ␤ 2 , ␤ 3 , and ␤ 4 subunits share little homology with the ␤ 1a subunit of skeletal muscle L-type Ca 2ϩ channels. The results suggest that none of these antibodies (␤ 1b , ␤ 2 , ␤ 3 , and ␤ 4 ) reacted with the ␤ 1a subunit of skeletal muscle L-type Ca 2ϩ channel. The results also confirm that the nonspecific immunoprecipitation by each of these antibodies is negligible. Copurification of ␣ 1A , ␣ 2 ␦, and Any One of the ␤ 1b , ␤ 2 , ␤ 3 , and ␤ 4 Subunits-The ␣ 1A subunit antibody was raised against a 200-amino acid region, which is located on the cytoplasmic loop between domains II and III of the rabbit ␣ 1A subunit. This region of ␣ 1A shares a 37-amino acid sequence homology with the ␣ 1B subunit, but shares little sequence homology with either ␣ 1S , ␣ 1C , ␣ 1D , or ␣ 1E . After enrichment of the Ca 2ϩ channels on a heparin agarose column (26,28), we utilized the mAb CC18 which was raised against the II-III loop of the ␣ 1B subunit as an immunoaffinity ligand to specifically remove all the N-type Ca 2ϩ channel (6), and then subsequently used an ␣ 1A antibody affinity column to enrich the high affinity -CTX MVIIC receptor. The void from the mAb ␣ 1B column was devoid of 125 I--CgTX GVIA binding, showing that over 95% of the N-type Ca 2ϩ channel was removed by mAb ␣ 1B column (Fig.  4A). Second incubation with mAb ␣ 1B column did not remove any additional -CgTX GVIA binding. Notably, the mAb ␣ 1B column did not remove any -CTX MVIIC binding when brain receptors were labeled with 0.1 nM (data not shown) and 0.5 nM -CTX MVIIC, which only labeled the high affinity receptor as previously shown in Fig. 1C. In contrast, ␣ 1A antibody affinity column bound greater than 95% of the labeled -CTX MVIIC receptor from the ␣ 1B column void. The results demonstrate that the ␣ 1A subunit is a major, if not the only, ␣ 1 present in the high affinity -CTX MVIIC receptor. Since the eluted receptors were inactive in -CTX MVIIC binding, the enrichment of receptor could not be estimated. The recovery of protein from the ␣ 1A column was about 0.3% compared to the total amount of protein loaded to the column (the ␣ 1B column void) and about 0.02% compared to the quantity of starting brain membranes.
Further analysis of heparin column eluate, ␣ 1B column void, ␣ 1A column void and ␣ 1A column eluate by immunoblotting showed that ␣ 1A , ␣ 2 ␦, ␤ 1b , ␤ 2 , ␤ 3 , and ␤ 4 subunits copurified on the ␣ 1A immunoaffinity column. As shown in Fig. 4B, a fraction of the ␣ 2 ␦ subunit was removed by the ␣ 1B column since this subunit is also present in the N-type Ca 2ϩ channel. Importantly, the ␣ 1A column further removed another portion of the ␣ 2 ␦ subunit and thus this subunit was enriched in the ␣ 1A eluate. These results demonstrate that the ␣ 2 ␦ subunit is also a component of the high affinity -CTX MVIIC receptor. In agreement with this result, wheat germ agglutinin-Sepharose sedimented a significant amount of high affinity 125 I--CTX MVIIC receptors probably through its interaction with the heavily glycosylated ␣ 2 ␦ subunit (data not shown). Similar to ␣ 2 ␦ subunit, ␤ 1b , ␤ 2 , ␤ 3 , and ␤ 4 subunits were also enriched in the eluate of ␣ 1A column (Fig. 4C) while they were barely detectable in 100 g of ␣ 1B column void (data not shown). The ␤ 1b subunit has a molecular mass of 76 kDa on immunoblot. There is also a higher band running at 80 kDa which may be a modified form or a ␤ 1 isoform. The ␤ 2 subunit has a molecular FIG. 3. Production of ␤ subunit subtype specific antibodies. A, schematic representation of four fusion proteins containing GST polypeptide and C-terminal ␤ subunits. All known ␤ subunits have two highly homologous domains, Domains I and II. The numbers above the fusion protein inserts are the amino acid positions of each inserted ␤ subunits. B, Coomassie Blue-stained SDS-polyacrylamide gel of four purified fusion proteins of C-terminal ␤ subunits. Molecular mass markers (in kilodaltons) are on the left. These purified fusion proteins were used to produce ␤ subunit antibodies. C, immunoblots stained with ␤ subunit antibodies preadsorbed with GST bacterial lysates. The antibody used is shown on the right side of each immunoblot. D, immunoprecipitation of 35 S-labeled full-length ␤ 1b , ␤ 2 , ␤ 3 , and ␤ 4 subunits (marked inside each gragh) with four ␤ subunit antibodies. Data were normalized by setting the value of the maximum immunoprecipitating antibody as 100% (n ϭ 3). I--CgTX GVIA (shaded bar) to solubilized rabbit brain receptors after elution from heparin (Hep) column, and passage through column of ␣ 1B antibody, followed by ␣ 1A antibody column or a second column of ␣ 1B antibody. Equal amounts of total proteins (50 g) were present in each condition. B and C, immunoblot analysis of enriched high affinity -CTX MVIIC receptor. Affinity purified antibodies against ␣ 2 ␦ (B), ␤ 1b , ␤ 2 , ␤ 3 , and ␤ 4 subunits (C) were used. 100 g of total proteins were present in heparin (Hep) eluate, ␣ 1B column void and ␣ 1A column void. Between 20 -40 g of total proteins were present in the ␣ 1A column eluate, which is one-fourth of total eluted material from ␣ 1A column with a starting 1-g brain membrane. Molecular mass markers (in kilodaltons) are on the left. mass of 84 kDa. The ␤ 3 subunit has a molecular mass of 58 kDa. The ␤ 4 subunits appear to run at 59 and 55 kDa possibly representing differential phosphorylation states or isoforms. These results confirm the immunoprecipitation data with the ␤ subunit antibodies. Furthermore, they demonstrate the specificity of ␤ subunit antibodies on the immunoblot and the copurification of the ␣ 2 ␦ and four ␤ subunits during enrichment of high affinity -CTX MVIIC receptors.
Association of Various ␤ Subunits with the ␣ 1A Channel Contributes to Its Functional Diversity-Previously all four ␤ subunits have been coexpressed separately with the ␣ 1A and ␣ 2 ␦ subunit in Xenopus oocytes. This results in several different modifications of the biophysical properties of the ␣ 1A channel (10,23). Previous results from our laboratory indicate that the current decay of the ␣ 1A ␣ 2 ␦ channel coexpressed with the ␤ 3 subunit is monophasic and fast (10), which resembles the Q-type current found in cerebellar granule cells (22). The current inactivation is voltage-dependent but reaches its fastest decay at ϩ10 mV with a time constant of ϳ120 ms (10). When this channel is coexpressed with ␤ 2 subunit, the current is slowly inactivating with a time constant of ϳ500 ms (10), which resembles the P-type current detected in cerebellar Purkinje cells (20).
Since an 18-amino acid domain in ␣ 1 subunit has been identified to be responsible for ␤ subunit association (36), the affinity of various ␤ subunits binding to ␣ 1A interaction domain has been determined (30). Compared to ␤ 1b , ␤ 2 and ␤ 4 subunits, ␤ 3 has a 10-fold lower affinity for ␣ 1A interaction domain based on in vitro binding assay. To test if ␤ 3 subunit is able to form Q-like channels in the presence of other ␤ subunits in one cell, we coexpressed ␤ 3 subunit with the ␤ 2 subunit. The formation of ␣ 1A ␣ 2 ␦␤ 3 or ␣ 1A ␣ 2 ␦␤ 2 channel can be easily distinguished because these channels exhibit distinct inactivation kinetics. When ␤ 3 and ␤ 2 subunits were coexpressed in Xenopus oocytes with cRNA ratio of 1:1, the kinetics of the ␣ 1A channel current showed a biexponential decay (Fig. 5A). The biphasic decay can be separated into two components with distinct inactivation kinetics ( 1 ϭ 69 Ϯ 12 ms, 2 ϭ 471 Ϯ 32 ms, n ϭ 5) which represents the currents carried by ␣ 1A ␣ 2 ␦␤ 3 or ␣ 1A ␣ 2 ␦␤ 2 channel. The results suggest the formation of two populations of functionally distinct channels. When ␤ 3 and ␤ 2 subunits were coexpressed in Xenopus oocytes with cRNA ratio of 10:1, the inactivation of the ␣ 1A channel is also biphasic with a 1 of 59 Ϯ 7 ms and a 2 of 338 Ϯ 25 ms (n ϭ 4). As expected for the formation of two populations of Ca 2ϩ channel complexes, the time constants of the biphasic decay for both cRNA ratios were comparable to that of ␤ 3 and ␤ 2 expressed separately. While the total biexponential current is the sum of the two monoexponential components, the fast component (␤ 3 ) increased from 52 to 77% of the total current when cRNA ratio of ␤ 3 :␤ 2 changed from 1:1 to 10:1 (Fig. 5B). Although protein levels of ␤ 3 and ␤ 2 subunits could not be determined due to extremely low amounts of expression, these results show that both ␤ 3 and ␤ 2 subunits are able to associate with ␣ 1A ␣ 2 ␦ to form functionally distinct channels for both ␤ 3 :␤ 2 cRNA ratios. Furthermore, these results suggest that the expression level of a ␤ subunit may be one factor involved in ␣ 1A ⅐␤ complex formation. DISCUSSION There has been some controversy concerning the specificity of -CTX MVIIC in the literature. While some researchers found that this toxin blocks N-, P-, and Q-type Ca 2ϩ channels (24,25), others obtained specific blockage of P-and Q-type Ca 2ϩ channels but not the N-type channel (19,37). In this study, we detected both a high affinity and a low affinity binding site for this toxin in rabbit brain. Importantly, there is a critical concentration of -CTX MVIIC, below which the toxin specifically binds the high affinity receptor and above which the toxin starts to bind the N-type Ca 2ϩ channel. Kristipati et al. (34) has reported that -CTX MVIIC binding is sensitive to concentration changes of monovalent (sodium, potassium) and divalent ions (calcium, barium) commonly used in biochemical and electrophysiological studies. It is conceivable that this critical concentration varies depending on the experimental conditions and different investigators were using variable ranges of the -CTX MVIIC concentration, which may explain some of the reported inconsistencies.
Although N-, P-, and Q-type Ca 2ϩ channels are all involved in neurotransmission, P-and Q-type channels seem to be of greater importance than N-type not only in terms of larger Ca 2ϩ currents but also in the efficacy of these Ca 2ϩ currents coupling to neurotransmitter release (4,22,25,38). The high affinity -CTX MVIIC receptor is expressed at a higher level than N-type channel which is in agreement with larger current expression of P-and Q-type channels. -CTX MVIIC has been used as a potent blocker for P-and Q-type Ca 2ϩ channels (19,22). Therefore the high affinity -CTX MVIIC receptor we detected in this study probably represents P-and Q-type Ca 2ϩ channels.
Although it has been generally assumed that ␣ 1 , ␣ 2 ␦, and ␤ subunits may be common subunits for high voltage-activated Ca 2ϩ channels, whether or not this holds true for all high voltage-activated channels has not been tested. Among the voltage-dependent Ca 2ϩ channels, subunit composition of skel- etal muscle L-type, cardiac muscle L-type and brain N-type channels have been extensively characterized, while brain Ltype (␣ 1C -or ␣ 1D -containing), P-type, Q-type, and ␣ 1E -containing Ca 2ϩ channels are poorly understood. Our data demonstrate for the first time that the high affinity -CTX MVIIC receptor is composed of at least an ␣ 1A , ␣ 2 ␦, and any one of the four ␤ subunits. Furthermore, we demonstrate that different ␤subunits are associated with the high affinity -CTX MVIICsensitive Ca 2ϩ channel. This heterogenous association of ␤ subunits results in channels with distinct properties such as inactivation kinetics when studied in Xenopus oocytes. Therefore, the association of various ␤ subunits with the ␣ 1A subunit in brain may produce some of the functional heterogeneity expected of Ca 2ϩ channels sensitive to -CTX MVIIC.
The distinction between P-and Q-type Ca 2ϩ channels is poorly defined. To our knowledge, there is no evidence to suggest that -CTX MVIIC has different affinity for brain P-or Q-type Ca 2ϩ channels. The difference between inactivation kinetics of these channels could result from association of distinct ␤ subunits. The only other known difference between these two channels are their sensitivity to -agatoxin IVA. Unfortunately, we and other investigators have found it difficult to obtain active radiolabeled -agatoxin IVA for biochemical studies. Based on our results, we propose that the high affinity -CTX MVIIC receptor represents both P-type and Q-type Ca 2ϩ channel. It is likely that the combinations of various splice variants of ␣ 1A , ␣ 2 ␦, and their association with different ␤ subunits may underlie the functional differences between P-type and Q-type Ca 2ϩ channels. In support of this speculation, P-type represents the majority of Ca 2ϩ channel current in cerebellar Purkinje cells where ␣ 1A transcripts are most prominent (12,23,39). In our study, the ␣ 1A subunit is the major (Ͼ90%) if not the only ␣ 1 subunit of the high affinity -CTX MVIIC receptors. Since no gene other than the one encoding the ␣ 1A subunit has been shown to encode a high affinity -CTX MVIIC receptor (21,23), it is therefore possible that both P-and Q-type Ca 2ϩ channels are encoded by splice variants of the ␣ 1A gene.
Our data demonstrate that the high affinity -CTX MVIIC receptors in rabbit brain contain a diversity of ␤ subunits in the following rank order, ␤ 4 Ͼ ␤ 3 Ͼ Ͼ ␤ 1b Ն ␤ 2 . Expression studies in Xenopus oocytes show that ␤ subunits are able to associate with the ␣ 1A channel in the presence of other ␤ subunits in a single cell, and the expression level of a ␤ subunit may be one factor involved in ␣ 1A ⅐␤ complex formation. Previously, the affinity of different ␤ subunits binding to the ␣ 1A interaction domain (fusion protein) were determined in the order of ␤ 4 Ϸ ␤ 2a Ϸ ␤ 1b Ͼ ␤ 3 (30). The estimated protein levels of the ␤ subunits in rabbit brain are ␤ 3 Ϸ ␤ 4 Ն ␤ 1b Ͼ ␤ 2 (40). These results suggest that the interaction of the ␣ 1A and ␤ subunits in vivo may be dependent upon the spatial and temporal expression of ␤ subunits and the affinity of ␣ 1A -␤ subunit interaction.
The association of various ␤ subunits with the same ␣ 1 subunit increases the heterogeneity of voltage-dependent Ca 2ϩ channels which are important in carrying out diverse functions. This is the first report that the high affinity -CTX MVIIC receptors are composed of at least ␣ 1A , ␣ 2 ␦, and any one of the four brain ␤ subunits. Heterogeneity of high affinity -CTX MVIIC receptors in brain provides a structural basis for the functionally distinct channels such as P-and Q-type channels. These functional differences are likely to be significant determinants of efficient synaptic transmission.