Beta subunit heterogeneity in N-type Ca2+ channels.

The β subunit of the voltage-dependent Ca channel is a cytoplasmic protein that interacts directly with an α subunit, thereby modulating the biophysical properties of the channel. Herein, we demonstrate that the α subunit of the N-type Ca channel associates with several different β subunits. Polyclonal antibodies specific for three different β subunits immunoprecipitated I--conotoxin GVIA binding from solubilized rabbit brain membranes. Enrichment of the N-type Ca channels with an α subunit-specific monoclonal antibody showed the association of β, β, and β subunits. Protein sequencing of tryptic peptides of the 57-kDa component of the purified N-type Ca channel confirmed the presence of the β and β subunits. Each of the β subunits bound to the α subunit interaction domain with similar high affinity. Thus, our data demonstrate important heterogeneity in the β subunit composition of the N-type Ca channels, which may be responsible for some of the diverse kinetic properties recorded from neurons.

The ␤ subunit of the voltage-dependent Ca 2؉ channel is a cytoplasmic protein that interacts directly with an ␣ 1 subunit, thereby modulating the biophysical properties of the channel. Herein, we demonstrate that the ␣ 1B subunit of the N-type Ca 2؉ channel associates with several different ␤ subunits. Polyclonal antibodies specific for three different ␤ subunits immunoprecipitated 125 I--conotoxin GVIA binding from solubilized rabbit brain membranes. Enrichment of the N-type Ca 2؉ channels with an ␣ 1B subunit-specific monoclonal antibody showed the association of ␤ 1b , ␤ 3 , and ␤ 4 subunits. Protein sequencing of tryptic peptides of the 57-kDa component of the purified N-type Ca 2؉ channel confirmed the presence of the ␤ 3 and ␤ 4 subunits. Each of the ␤ subunits bound to the ␣ 1B subunit interaction domain with similar high affinity. Thus, our data demonstrate important heterogeneity in the ␤ subunit composition of the N-type Ca 2؉ channels, which may be responsible for some of the diverse kinetic properties recorded from neurons.
Voltage-dependent Ca 2ϩ channels are essential for regulating Ca 2ϩ concentrations in many cells. Based upon electrophysiological and pharmacological properties, these channels have been classified into five major groups (L, N, T, R, and P/Q types) (1,2). L-type Ca 2ϩ channels are central to excitationcontraction coupling in skeletal and cardiac muscle, while Ttype channels are involved in pacemaker activity. The N-, P/Q-, and R-type Ca 2ϩ channels are found predominantly in the central and peripheral nervous systems and have major roles in controlling neurotransmitter release. The skeletal muscle L-type and the brain N-type Ca 2ϩ channels have both been purified. Although functionally distinct, these have similar subunit compositions (␣ 1 , ␣ 2 ␦, and ␤), with a variable channelspecific subunit (␥ or 95 kDa, respectively) (3,4). The genes encoding the ␣ 1 pore-forming subunits have been separated into six groups (S, A, B, C, D, and E), each containing multiple splice variants, while the ␤ subunits have been classified into four major classes (namely ␤ 1 , ␤ 2 , ␤ 3 , and ␤ 4 ), also containing several splice variants (5). Recent studies have identified complementary interaction domains on the ␣ 1 and ␤ subunits (6,7).
The ␤ subunit regulates channel activity by binding to a highly conserved portion of the cytoplasmic linker of the I-II loop of all ␣ 1 subunits, known as the ␣ 1 subunit interaction domain (AID). 1 The ␤ subunit's interaction site, known as the ␤ subunit interaction domain, encompasses ϳ30 amino acids in the amino-terminal portion of the second highly conserved domain. In vitro binding studies have shown that an ␣ 1 subunit binds a single ␤ subunit in a 1:1 stoichiometry (8).
N-type Ca 2ϩ channels are involved in regulating neurotransmitter release in the central and peripheral nervous systems and in controlling endocrine secretion. These also serve as autoantigens in paraneoplastic neurologic disorders and may be the target of pathogenic autoantibodies responsible for autonomic dysfunction in the Lambert-Eaton myasthenic syndrome (9). Electrophysiological analysis of the N-type Ca 2ϩ channels in different neurons has revealed an unusual degree of diversity in the rates of inactivation (10 -12). The structural basis of this functional diversity is not understood. Herein, we examine which ␤ subunits are associated with the N-type Ca 2ϩ channels from rabbit brain using a monoclonal antibody specific for the ␣ 1B subunit and polyclonal antibodies for three different ␤ subunit genes. Several lines of evidence are presented that demonstrate that there is heterogeneity in the ␤ subunit of native N-type channels, and this may account for some of the functional diversity of channel kinetic properties recorded from neurons.

EXPERIMENTAL PROCEDURES
Materials-125 I--Conotoxin (CTx) GVIA, [ 35 S]methionine, and the ECL kit were purchased from Amersham Corp. Digitonin was obtained from ICN Biomedicals and purified as detailed elsewhere (3). Other biochemicals used were protein G-Sepharose (Pharmacia Biotech Inc.), horseradish peroxidase-conjugated secondary antibodies (Boehringer Mannheim), and Avidchrom hydrazide gel (Unisyn Technologies). All other chemicals were of reagent grade. The GraFit Version 3.0 curve fitting program was purchased from Sigma.
Production of a Monoclonal Antibody (mAb) to the ␣ 1B Subunit-mAb CC18 was secreted by a hybridoma produced from the splenic B lymphocyte of a rat hyperimmunized with a fusion protein corresponding to the II-III cytoplasmic loop of the ␣ 1B subunit (29). It reacts selectively with high affinity brain receptors for -CTx GVIA.
SDS-PAGE and Immunoblot Analyses-Proteins were analyzed by SDS-PAGE on 3-12 or 5-16% gradient gels using the Laemmli buffer system (13). Gels were transferred to nitrocellulose and immunoblotted as described previously (14). The specific protein bands were detected using either the horseradish peroxidase or ECL detection methods (according to the manufacturers' instructions). Antibodies to fusion proteins containing the diverse C-terminal portion of each of four ␤ subunits (30) were immunoaffinity-purified as described previously (14), using the appropriate fusion protein for ␤ 1b (residues 428 -597; * 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  GenBank accession number X61394), ␤ 2a (residues 462-578; GenBank accession number M80545), ␤ 3 (residues 369 -484; GenBank accession number M88751), and ␤ 4 (residues 419 -519; GenBank accession number L02315).
Immunoprecipitation of N-type Ca 2ϩ Channels-Antibodies were incubated overnight with protein G-Sepharose beads in PBS at 4°C. The beads were then washed three times with PBS prior to resuspension in an equal volume of PBS. An aliquot of rabbit brain membranes (ϳ20 mg) was incubated with 125 I--CTx GVIA (0.5 nM) in 10 mM HEPES/ NaOH, pH 7.5, containing 0.1 M NaCl and 0.2 mg/ml bovine serum albumin in the presence and absence of 1000-fold unlabeled -CTx GVIA for 1 h at 22°C. Then, the membranes were sedimented by centrifugation in a Beckman TL100 centrifuge at 50,000 rpm for 10 min at 4°C. The resulting pellet was resuspended in solubilization buffer (10 mM HEPES/NaOH, pH 7.5, containing 1 M NaCl, 0.23 mM phenylmethylsulfonyl fluoride, 0.64 mM benzamidine, 1 M leupeptin, 0.7 M pepstatin A, 76.8 nM aprotinin, and 1% (w/v) digitonin) to a final volume of 7 ml and incubated at 4°C for 1 h. Particulate material was removed from solution by sedimentation at 100,000 rpm for 30 min, and the supernatant was diluted 3-5-fold with ice-cold distilled, deionized water. Aliquots (1 ml) of the labeled solubilized extract were then incubated with saturating concentrations of each antibody-protein G-Sepharose bead complex at 4°C overnight with agitation. Subsequently, the beads were sedimented by centrifugation and washed twice with ice-cold buffer A (10 mM HEPES/NaOH, pH 7.5, 100 mM NaCl plus protease inhibitors as listed above) containing 0.1% (w/v) digitonin prior to quantification by ␥-counting.
N-terminal Sequence Analysis of the 57-kDa Subunit of the N-type Ca 2ϩ Channel-The N-type Ca 2ϩ channel was purified as described previously (4), and the subunits were resolved by SDS-PAGE and transferred onto an Immobilon PSQ membrane in 10 mM CAPS/NaOH, pH 11, containing 10% (w/v) methanol for 3 h. The 57-kDa subunit was visualized by Coomassie Brilliant Blue staining, excised, and digested with trypsin for 18 h at 37°C. The resulting peptides were separated by reverse-phase HPLC and then subjected to Edman degradation.
Transient Transfection of ␤ Subunits into COS-7 Cells-COS-7 cells were obtained from the American Type Culture Collection and grown in high glucose Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. cDNAs encoding the ␤ 3 and ␤ 4 subunits were subcloned into the vector pcDNA3 (Invitrogen). Plasmid DNA (30 g) was introduced into 1 ϫ 10 7 cells by electroporation at 950 microfarads and 320 V in a Bio-Rad Gene Pulser. To increase cell viability, cells were electroporated in Cytomix buffer (15). After 72 h, cells were washed and harvested in PBS. Crude cell lysate was prepared by homogenization in PBS with 0.23 mM phenylmethylsulfonyl fluoride and 0.64 mM benzamidine. Cell pellets were collected by centrifugation at 14,000 rpm for 10 min and resuspended in Laemmli sample buffer (13).
Immunoaffinity Enrichment of Ca 2ϩ Channels Containing ␣ 1B -Protease inhibitors were included in all buffers at the concentrations indicated above to minimize proteolysis of the receptors throughout the purification. Rabbit brain membranes (ϳ2 mg), prepared as detailed elsewhere (14), were prelabeled with 125 I--CTx GVIA as described above and used as a tracer to detect the channels throughout the purification. Labeled receptors, together with unlabeled protein (ϳ1 g), were extracted from the membrane in solubilization buffer for 1 h at 4°C. After centrifugation at 35,000 rpm for 37 min in a 45 Ti rotor, the detergent extract was diluted 3-fold with ice-cold distilled, deionized water and applied to a heparin-agarose column pre-equilibrated with buffer A at a flow rate of 5 ml/min. The column was washed extensively with buffer A and eluted in the same buffer containing 0.7 M NaCl, collecting 5-ml fractions. Peak fractions were detected by ␥-counting and pooled. The enriched channels were then incubated overnight with mAb CC18 coupled to Avidchrom hydrazide (ϳ2 ml of settled resin) prepared according to the manufacturer's instructions. The resin was washed extensively with buffer A containing 0.7 M NaCl and then eluted with 50 mM glycine HCl, pH 2.5, containing 0.6 M NaCl and 0.1% (w/v) digitonin. Fractions (1 ml) were immediately neutralized with 2 M Tris-HCl, pH 8.0 (125 l). The peak fractions were detected by ␥-counting and were concentrated in an Amicon ultracentrifugation unit using a YM-100 membrane. The subunit composition was analyzed by SDS-PAGE and immunoblotting.
Binding of AID B -Glutathione S-Transferase Fusion Protein to 35 S-Labeled ␤ Subunits-The affinity of each of the in vitro synthesized 35 S-labeled ␤ subunits for AID B was performed as described previously (8). Briefly, various 35 S-labeled ␤ subunit probes (␤ 1b , ␤ 2a , ␤ 3 , and ␤ 4 ) were synthesized by coupled in vitro transcription and translation with the TNT kit (Promega). 0.7-1.3 pM 35 S-labeled ␤ subunit was incubated overnight at 4°C in PBS (1 ml) with increasing concentrations (100 pM to 1 mM) of the AID B -glutathione S-transferase fusion protein (residues 378 -434 of the ␣ 1B subunit (16)) noncovalently coupled to glutathione-Sepharose beads. The beads were then washed four times with PBS and subjected to scintillation counting, and the data were analyzed using the GraFit program.
FIG. 1. Immunoprecipitation of 125 I--CTx GVIA binding with subunit-specific antibodies. Aliquots (ϳ20 mg) of rabbit brain membranes were labeled with 125 I--CTx GVIA in the presence and absence of 1000-fold unlabeled toxin and solubilized as described under "Experimental Procedures." A, to determine the maximum amount of antibody required to sediment a constant amount of solubilized 125 I--CTx GVIA-labeled receptors, increasing quantities of antibody coupled to protein G Sepharose were incubated with an aliquot of the labeled channels (1 ml) overnight at 4°C. The beads were washed three times with 10 mM HEPES/NaOH, pH 7.5, 0.1 M NaCl, and 0.1% (w/v) digitonin containing five protease inhibitors, and immunoprecipitation was quantified by ␥-counting. B, saturating concentrations of each antibody coupled to protein G-Sepharose were incubated with solubilized 125 I--CTx GVIA receptor complexes (ϳ1 ml) overnight at 4°C. The beads were washed as described above and quantified by ␥-counting. The amount of precipitation in each case was determined relative to that of Sheep 46 (Sh46; 100%). IIC12 is a mAb raised against the skeletal muscle dihydropyridine receptor.

RESULTS AND DISCUSSION
The N-type Ca 2ϩ channel subunit composition was investigated using a number of different antibodies. Saturating amounts of four ␤ subunit-specific antibodies were determined by incubating increasing quantities of these antibodies with a constant amount of 125 I--CTx GVIA-labeled receptors to establish the maximum immunoprecipitation of solubilized rabbit brain N-type channels (Fig. 1A). The maximum immunoprecipitation by a variety of antibodies raised against several components of voltage-dependent Ca 2ϩ channels was subsequently compared. Polyclonal antisera against the purified Ntype Ca 2ϩ channel (Sheep 46) precipitated the 125 I--CTx GVIA receptors and provided the 100% control value (Fig. 1B). A second polyclonal antibody (Y006) raised against the II-III loop of the ␣ 1B subunit precipitated 93 Ϯ 1.8% of the 125 I--CTx GVIA receptors with respect to the Sheep 46 antibodies. This confirmed the presence of the ␣ 1B subunit in all N-type Ca 2ϩ channel oligomers. In contrast, a monoclonal antibody (IIC12) raised against the purified skeletal muscle dihydropyridine receptor (3) was included as a negative control and was shown to precipitate negligible amounts (Ͻ3%) of 125 I--CTx GVIA receptors. In addition, a polyclonal antibody specific for the ␣ 1C subunit (Rabbit 140) did not sediment significant amounts of the N-type channels and served as a second negative control.
Saturating amounts of each of the ␤ subunit antibodies were used to precipitate the relative amounts of each ␤ subunit associated with the 125 I--CTx GVIA receptor. Consistent with previous observations (14), a polyclonal antibody specific for the ␤ 3 subunit precipitated the largest amount of toxin binding (56.1 Ϯ 8.3%), although a ␤ 3 subunit antibody was previously shown to sediment a larger fraction of the receptors. In the earlier study, however, the ␤ 3 subunit antibodies (affinitypurified from Sheep 46) may have been cross-reactive with the ␤ 4 subunit, which at that time had not been cloned. In the present study, the ␤ 3 subunit-specific antibody was raised directly against a C-terminal fusion protein (Sheep 49) and was shown to be specifically reactive with the ␤ 3 subunit (30). Interestingly, the ␤ 4 subunit-specific antibody also sedimented a significant proportion of receptors (30.5 Ϯ 2.1%), suggesting that it is a major component of the purified N-type Ca 2ϩ channels. Moreover, coincubation of saturating amounts of the ␤ 3 and ␤ 4 subunit antibodies with the labeled receptor precipitated 84 Ϯ 0.7%, which is approximately equivalent to the sum of precipitation by both sera incubated separately (56.1 Ϯ 8.3% (␤ 3 ) ϩ 30.5 Ϯ 2.1% (␤ 4 )). This further confirmed the specificity of these antibodies and demonstrated that both ␤ subunits were not present in the same oligomer since saturating concentrations of both antibodies incubated together did not immunoprecipitate less than the sum of each antibody incubated separately.
Notably, the ␤ 1b subunit-specific antibody also precipitated a significant amount of the labeled receptors after subtracting the nonspecific precipitation (10.3 Ϯ 1.6%), suggesting that it also associates with the ␣ 1B subunit of the N-type Ca 2ϩ channel in brain. In contrast, the ␤ 2a subunit antibody did not precipitate significant amounts of 125 I--CTx GVIA binding over the nonspecific binding, which was reproducibly Ͻ3%, suggesting either that the ␤ 2a subunit may not be associated with the brain N-type Ca 2ϩ channel or that its expression level in brain is too low to detect (26). These immunoprecipitation data represent the first evidence that three different ␤ subunits associate with the ␣ 1B subunit in the native N-type Ca 2ϩ channel.
To investigate the ␤ subunit heterogeneity of N-type channels further, immunoaffinity-purified Ca 2ϩ channel subunits (4) were separated by SDS-PAGE, electrophoretically transferred onto Immobilon PSQ membrane, and visualized by Coomassie Blue staining. The 57-kDa band was excised and digested with trypsin. This generated seven peptides, which were resolved by reverse-phase HPLC, followed by Edman degradation (Table I). Comparison of the sequences with those in the database showed that peptides 1-3 had Ͼ80% amino acid identity to the ␤ 3 subunit, which confirmed the previous observation that this subunit is present in the purified N-type Ca 2ϩ channel (4). Peptides 4 and 7 showed Ͼ85% amino acid identity to the more recently cloned ␤ 4 subunit (17) and were absent from the ␤ 3 subunit sequence. The remaining two peptide sequences are present in the second conserved domain of each of the ␤ subunits and could therefore not be specifically assigned to any of the ␤ subunits. These data confirmed that both the ␤ 3 and ␤ 4 subunits are associated with the N-type Ca 2ϩ channels.

FIG. 2. Determination of the specificity of the ␤ 3 and ␤ 4 antibodies.
The specificity of the affinity-purified ␤ 3 and ␤ 4 subunit antibodies was determined by transiently transfecting COS-7 cells as described under "Experimental Procedures" with constructs encoding the respective ␤ subunit. After harvesting the cells, aliquots (ϳ150 g) were subjected to SDS-PAGE on a 5-16% gel, followed by either Coomassie Blue staining (CB) or immunostaining with affinity-purified ␤ 3 and ␤ 4 subunit antibodies. Ctl, control untransfected cells.
a Residues in the ␤ 4 sequence.
Since the ␤ 1 subunit has a molecular mass of ϳ72 kDa, it was resolved from the ␤ 3 and ␤ 4 subunits (molecular masses of 57 kDa) by SDS-PAGE prior to sequencing, which explains why no specific sequences for this subunit were detected. The specificity of the polyclonal antisera that were raised against the C-terminal fusion proteins of the ␤ 3 (Sheep 49) and ␤ 4 (Rabbit 145) subunits was tested in immunoblot experiments. COS-7 cells, which do not contain detectable levels of endogenous Ca 2ϩ channel subunits, were transiently trans-fected with constructs encoding the ␤ 3 and ␤ 4 subunits separately. The cells were harvested, and equal amounts of the protein were subjected to SDS-PAGE on a 5-16% gel. The proteins were electrophoretically transferred onto nitrocellulose and probed with affinity-purified ␤ 3 and ␤ 4 subunit-specific antibodies. Neither antibody recognized any proteins in the untransfected cells. The resulting immunoblots demonstrate the specificity of the ␤ 3 and ␤ 4 subunit-specific antibodies since the ␤ 3 antibodies recognized only the protein in the FIG. 3. Analysis of N-type Ca 2؉ channels immunoaffinity-enriched using mAb CC18 against the ␣ 1B subunit. A, shown is the elution profile of 125 I--CTx GVIA receptors from the heparin-agarose column. The column was developed with 0.7 M NaCl in buffer A, collecting 5-ml fractions. B, shown is the elution profile of the mAb CC18 immunoaffinity column. The eluate from the heparin-agarose column was loaded onto this antibody column, and enriched channels were eluted with 50 mM glycine buffer, pH 2.5. C, the mAb CC18 immunoaffinity column eluate (ϳ1-2 g) was resolved on a 3-12% SDS gel and electrophoretically transferred to nitrocellulose. The subunit composition of the enriched channels was then examined by immunoblotting with mAb CC18 and each of the indicated affinity-purified ␤ subunit antibodies. Molecular mass markers (in kilodaltons) are shown to the left.

Specificity of mAb CC18
The mAb specific for the ␣ 1B subunit (mAb CC18) was prepared by hyperimmunizing a rat with a fusion protein encoding the II-III loop of the pore-forming subunit of the N-type channel. To assess the specificity of this antibody, the N-and P/Q-type channels were labeled with either 125 I--CTx GVIA or 125 I--CTx MVIIC, and immunoprecipitation was performed using mAb CC18. The reactivity was also tested by enzyme-linked immunosorbent assay using as the antigen the fusion protein for the II-III loops of the ␣ 1A and ␣ 1B subunits, and serial dilutions of mAbCC18. ␤ 3 -transfected cells and none in those transfected with ␤ 4 . The ␤ 4 antibodies were likewise shown to be specific for the ␤ 4 subunits expressed in COS-7 cells (Fig. 2). The subunit composition of the N-type Ca 2ϩ channel was further established by the development of a purification scheme using a heparin-agarose column, as was previously published (4), followed by immunoaffinity chromatography using the ␣ 1B subunit-specific mAb CC18 coupled to Avidchrom hydrazide. This monoclonal antibody was raised against a fusion protein containing the II-III loop of the ␣ 1B subunit. This bound specifically and with high affinity to 125 I--CTx GVIA receptors, but did not immunoprecipitate 125 I--CTx MVIIC receptors (9) or react with the fusion protein containing the II-III loop of the ␣ 1A subunit (Table II) in enzyme-linked immunosorbent assay and immunoblotting experiments, even though there is a low sequence identity in this region between both ␣ 1 subunits (18,19). Furthermore, this mAb did not immunoprecipitate any significant proportions of [ 3 H]PN 200-110 binding to skeletal muscle triads (data not shown), which was not surprising since there is very little homology between the sequence of the II-III loop of the ␣ 1B subunit and the other ␣ 1 subunit genes (␣ 1S , ␣ 1C , ␣ 1E , and ␣ 1D subunits), so the possibility of cross-reactivity of mAb CC18 in either immunoblotting or immunoprecipitation experiments with any other ␣ 1 subunit was highly unlikely. Rabbit brain membranes were first labeled with 125 I--CTx GVIA, solubilized in high ionic strength buffer containing 1% (w/v) digitonin and a mixture of five protease inhibitors, and applied to a heparin-agarose column. Elution of the column with 0.7 M NaCl resulted in ϳ10-fold enrichment of the channels (Fig. 3A). The peak of channel activity was pooled and subsequently applied to the mAb CC18-Avidchrom column. Development with glycine buffer, pH 2.5, yielded a large peak of radioactivity (Fig. 3B) that was pooled and analyzed by SDS-PAGE followed by Western blot analysis with antibodies to the ␣ 1B subunit and each of the four ␤ subunits. The resulting immunoblots showed the presence of the broad diffusely stained ␣ 1B subunit, with an apparent molecular mass ranging from 190 to 230 kDa. This broad band may contain more than one ␣ 1 species since multiple splice variants of the ␣ 1B subunit have been shown to exist (19). However, mAb CC18 is raised against the II-III loop and therefore would be unable to distinguish between the different C-terminal splice variants. Interestingly, immunoblotting with affinity-purified ␤ 1b , ␤ 3 , and ␤ 4 subunit-specific antibodies (Fig. 3C) demonstrated the presence of each of these ␤ subunits in the preparation. As predicted from the immunoprecipitation data, the ␤ 2a subunit was not detected in the enriched preparation using its specific antibody and the highly sensitive ECL detection method. Although it cannot be excluded that the ␤ 2a subunit may interact with the ␣ 1B subunit, the protein levels of this subunit or its level of association with ␣ 1B is too low to detect in brain.
Recent identification of the interaction domains between the ␣ 1 and ␤ subunits has allowed the development of an assay for studying the specific association of these subunits in vitro. The binding affinity of the AID B fusion protein to in vitro translated 35 S-labeled ␤ subunits from each of the four genes was measured. Interestingly, the AID B fusion protein interacted with the ␤ 1b (K D ϭ 4.7 nM; 90% of total binding capacity), ␤ 2a (K D ϭ 4.8 nM; 98% of total binding capacity), and ␤ 3 (K D ϭ 7.26 nM; 91% of total binding capacity) subunits with similar high affinities (Fig. 4). Unlike ␤ 1b , ␤ 2a , and ␤ 3 , the ␤ 4 subunit appeared to bind to two sites, one with high affinity (8.4 nM; 63% of total binding capacity) and the other with low affinity (444 nM; 55% of total binding capacity). The lower binding affinity may have been due to the binding of some proteolyzed forms of the ␤ 4 subunit that were generated during the synthesis of the probe, as was previously shown (8). These data demonstrate that the AID B fusion protein can interact with each of the ␤ subunits with similar high affinity, unlike the binding to the AID A site, which showed a 20-fold lower affinity for ␤ 3 than for ␤ 4 (8).
Although several in vitro expression studies using recombinant protein have shown that ␣ 1 subunits form functional Ca 2ϩ channels with variable channel kinetics depending upon which ␤ subunit is coexpressed (19 -24), this is the first report to demonstrate directly that different ␤ subunits are associated with the ␣ 1B subunit in the native N-type Ca 2ϩ channel. The initial suggestion that only the ␤ 3 subunit was present in the N-type Ca 2ϩ channel was made prior to the discovery and cloning of the ␤ 4 subtype, which we have demonstrated here to be a significant component of the N-type Ca 2ϩ channel by immunoprecipitation, internal sequence information, and Western blotting analysis. Data presented herein revealed that there is more diversity in the N-type Ca 2ϩ channel subunit composition than was originally indicated. In contrast, the ␣ 1S subunit of the skeletal muscle dihydropyridine receptor appears to associate only with the ␤ 1a subunit (data not shown) since this is the only ␤ subunit known to be expressed in skeletal muscle tissue (25). In neurons, the levels of expression of each ␤ subunit may, in part, determine the apparent specificity of association between the ␣ 1B subunit and various ␤ subunits. Our data support this hypothesis since immunoprecipitation of four different ␤ subunits with specific antibodies correlates well with the relative amounts of each ␤ gene expressed in brain, the most abundant being ␤ 3 and ␤ 4 , with smaller quantities of ␤ 1b and negligible proportions of the ␤ 2 subunit (26). It is thus possible that in different tissue sources, the levels of expression of certain ␤ subunits determine the levels of association with the ␣ 1 subunit.
The association of different subunit combinations as a method of generating subtle differences in channel properties has been reported for other ion channels. The neuronal voltagedependent ␣-dendrotoxin-sensitive K ϩ channels form heterooligomers with various combinations of four ␣ subunits with and without four ancillary ␤ subunits (27). In the ligand-gated ␥-aminobutyric acid type A receptor, particular ␤ subunits have also been shown to associate with several different ␣ subunits in vivo (28), thereby creating more channel heterogeneity. In conclusion, the generation of different N-type Ca 2ϩ channel oligomers that differ in their ␤ subunit composition may account for some of the functional diversity of these channels in the nervous system.