Interaction of SNARE complexes with P/Q-type calcium channels in rat cerebellar synaptosomes.

P- and Q-type calcium channels, which trigger rapid neurotransmitter release at many mammalian synapses, are blocked by omega-conotoxin MVIIC. 125I-omega-Conotoxin MVIIC binding to rat cerebellar synaptosomes was not displaced by omega-conotoxins GVIA or MVIIA (Ki > 1 microM), which are selective for N-type calcium channels. Solubilized 125I-omega-conotoxin MVIIC receptors were specifically recognized by antibodies directed against alpha1A calcium channel subunits, proteins known to constitute a pore with P/Q-like channel properties. Antibodies against syntaxin 1, SNAP 25, and VAMP 2 (synaptobrevin) each immunoprecipitated a similar fraction (20-40%) of omega-conotoxin MVIIC receptors. Immunoprecipitation was not additive, suggesting that heterotrimeric (SNARE) complexes containing these three proteins interact with P/Q-type calcium channels. Immobilized monoclonal anti-syntaxin antibodies retained alpha1A calcium channel subunits of 220, 180 and 160 kDa monitored by immunoblotting with site directed antibodies. Synaptotagmin was detected in channel-associated complexes, but not synaptophysin, Rab 3A nor rat cysteine string protein. Trimeric SNARE complexes are implicated in calcium-dependent exocytosis, a process thought to be regulated by synaptotagmin. Our results indicate that these proteins interact with P/Q-type calcium channels, which may optimize their location within domains of calcium influx.

Neurotransmitter release from axonal terminals is triggered by calcium entry through voltage-gated channels in the presynaptic plasma membrane. The sub-millisecond delay between calcium influx and exocytosis of the contents of synaptic vesicles implies that the calcium channels involved are in close proximity to release sites (Llinas et al., 1981;Adler et al., 1991). Synaptic transmission in mammals is generally insensitive to the antagonists that act at L-type calcium channels, but is inhibited by peptide neurotoxins that block N-or P/Q-type calcium channels (Takahashi and Momiyama, 1993;Wheeler et al., 1994;Mintz et al., 1995;Turner and Dunlap, 1995). Pharmacological evidence therefore suggests that channels containing ␣ 1 A or ␣ 1 B subunits are preferentially coupled to the exocytosis of rapid neurotransmitters.
Recent biochemical evidence suggests that direct interactions between N-type calcium channel proteins and the exocytotic machinery may contribute to excitation-secretion coupling. N-type calcium channels can associate with syntaxin and synaptotagmin (Leveque et al., 1992;Yoshida et al., 1992;Bennett et al., 1992;Leveque et al., 1994), two key proteins that contribute to the targeting and calcium-dependent fusion of synaptic vesicles at the plasma membrane (reviewed by Scheller and Bennett (1994) andSü dhof (1995)). This interaction involves binding of syntaxin to a cytoplasmic loop of the ␣ 1 B subunit (Sheng et al., 1994). In contrast, associations between syntaxin and L-type channels have not been detected (Yoshida et al., 1992;El Far et al., 1995).
AgaIVA and MVIIC inhibit a predominant component of transmitter release at most rapid mammalian synapses, by blocking calcium influx through presynaptic P/Q-type calcium channels (Takahashi and Momiyama, 1993;Wheeler et al., 1994;Mintz et al., 1995). However, biochemical evidence for an interaction between P/Q-type channels and the exocytotic complex has not been reported. We have therefore used 125 I-MVIIC to label calcium channels containing ␣ 1 A subunits solubilized from rat cerebellar synaptosomes, and to explore their association with proteins of the secretory pathway. Our results indicate that a population of calcium channels containing ␣ 1 A subunits can interact with a trimeric SNARE complex (Söllner et al., 1993a) composed of syntaxin 1, SNAP 25, and VAMP 2 (synaptobrevin).  Martin-Moutot et al. (1995) and purified by high performance liquid chromatography on an analytical C18 column (Beckman). Monoclonal antibodies to synaptotagmin (mAb 1D12), syntaxin (mAb 10H5), and SNAP 25 (mAb BR05) and polyclonal antibodies against VAMP2 and calcium channel ␣ 1 A subunits (rbA-1 and rbA-2) were prepared as described previously (Yoshida et al., 1992;El Far et al., 1995;Martin-Moutot et al., 1995).Monoclonal anti-synaptophysin (mAb 171B5) antibodies were provided by Dr. S. Fujita, Mitsubishi Kasei Institute. Polyclonal antibodies were raised in rabbits against the carboxyl-terminal peptide (C)GPQLTDQQAPPHQD of the rat Rab 3A sequence.
Binding Assays-10 g rat cerebellar P2 membranes in 0.1 ml of 25 mM Tris, 150 mM NaCl, 0.1% bovine serum albumin adjusted to pH 7.4 with HCl (TBSA) were incubated with 0.5 nM 125 I-MVIIC for 1 h at 30°C. Membrane-bound radioactivity was measured after rapid filtration and washing on GF/C (Whatman) filters treated with 0.3% polyethyleneimine as described previously .
Immunoprecipitation Assays-Membranes were prelabeled with 125 I-MVIIC in TBSA overnight at 4°C, washed by centrifugation, and solubilized in 1% CHAPS, 10 mM Tris, 0.32 M sucrose, adjusted to pH 7.4 with HCl. Aliquots of 100,000 ϫ g supernatants containing 3 fmol of 125 I-MVIIC-receptor complex were diluted 10-fold in 0.4% CHAPS, TBSA; and antibodies were added to give a final assay volume of 0.1 ml. After 2 h at 4°C, immune complexes were recovered by mixing for 1 h with Protein A-Sepharose 4BCL. Beads were recovered by centrifugation and washed in TBSA, and the immunoprecipitated radioactivity was counted. Aliquots of the initial extract were filtered through GF/B filters (Whatman) treated with 0.3% polyethyleneimine to trap and measure total 125 I-ligand/receptor complex.
Anti-syntaxin antibodies (mAb 10H5) were covalently coupled to Protein A-Sepharose (Leveque et al., 1994). 4 ml of a CHAPS extract of cerebellar synaptosomes were batch-incubated overnight with 1 ml of immunoaffinity resin. Beads were loaded into a column, washed with 20 ml of 10 mM Tris, 0.2 M NaCl, 0.4% CHAPS adjusted to pH 7.5 with HCl, and eluted with 0.1 M triethylamine, 0.2 M NaCl, 0.4% CHAPS adjusted to pH10.5 with HCl. All steps were performed at 4°C in buffers containing Complete protease inhibitor mixture (Boehringer Mannheim). SDS-polyacrylamide gel electrophoresis and Western blotting were performed as described previously , using Protein A-peroxidase, and an ECL detection kit (Amersham Corp.).

RESULTS AND DISCUSSION
-Conotoxins provide useful probes for the biochemical assay of voltage-dependent calcium channels. GVIA and -conotoxin MVIIA (MVIIA) are selective antagonists of N-type channels in which the transmembrane pore is constituted by an ␣ 1 B subunit. In contrast MVIIC, which is highly homologous to MVIIA, blocks P-, Q-, and N-type calcium channels in neurons (Hillyard et al., 1992;Randall and Tsien, 1995), and inhibits currents induced by the heterologous expression of calcium channels containing ␣ 1 A or ␣ 1 B subunits (Sather et al., 1993;Grantham et al., 1994). Although micromolar concentrations of MVIIC display little selectivity and block both P/Qand N-type channels, subnanomolar concentrations of 125 I-MVIIC reveal binding sites pharmacologically distinct from those associated with N-type channels (Hillyard et al., 1992;Kristipati et al., 1994), and which are therefore likely to be constituted by P/Q-type channels . 125 I-MVIIC displayed specific high affinity binding to rat cerebellar synaptosomes (Fig. 1A), which was displaced by native MVIIC with a K i of about 1 nM, but not by the selective N-type calcium channel antagonists GVIA or MVIIA (K i Ͼ 1 M). Conversely, using 125 I-GVIA as a probe, significant inhibition was only obtained with relatively high concentrations of MVIIC (K i Ͼ 30 nM; data not shown), presumably due to MVIIC competing with GVIA at the same binding site on N-type calcium channels. Nevertheless these data suggest that at sub-nanomolar concentrations, 125 I-MVIIC labels binding sites that are distinct from those associated with ␣ 1 B subunits.
Cerebellar membranes were labeled with 125 I-MVIIC, and radioligand-receptor complexes were solubilized using CHAPS. Approximately 50% of these complexes were adsorbed by antibodies directed against a peptide specific to the ␣ 1 A sequence. Only 8% were retained if antibodies were preincubated with the cognate ␣ 1 A peptide (Fig. 1B). Anti-␣ 1 A antibodies reacted with specific 125 I-MVIIC binding sites as the amount of immunoprecipitated radioactivity was strongly reduced if labeling was performed in the presence of 0.3 M MVIIC but not 0.3 M GVIA. No significant immunoprecipitation of 125 I-MVIIC binding sites was detected with antibodies directed against ␣ 1 B (8 Ϯ 1%) or ␣ 1 C (8 Ϯ 1%) subunits, although these antibodies recognize Ͼ50% of 125 I-GVIA receptors (N-type channels) or [ 3 H]PN200 -110 receptors (L-type channels) respectively in analogous experiments (not shown).
High affinity 125 I-MVIIC binding thus provides a specific assay for calcium channels containing ␣ 1 A subunits, and immunoprecipitation of solubilized receptors can be used to examine whether proteins implicated in exocytosis are associated . Membrane-bound radioactivity was measured by filtration and ␥ counting. Nonspecific binding, defined as radioactivity bound in the presence of 0.3 M MVIIC, was subtracted (means Ϯ S.D., n ϭ 6). B, synaptosomes were labeled with 125 I-MVIIC, in the presence or absence of 0.3 M MVIIC or GVIA. CHAPS extracts containing 125 Iligand/receptor complexes were incubated with anti-␣ 1 A subunit antibodies (rbA-1) in the presence or absence of an excess of the ␣ 1 A peptide used to raise the antibodies, and the radioactivity in immune complexes was counted (means Ϯ S.D., n ϭ 6).

SNARE Complex Association with P/Q-type Calcium Channels
with these subunits. Antibodies against three proteins that form the core SNARE complex, syntaxin 1, SNAP 25, and VAMP2, each adsorbed 20 -40% of solubilized 125 I-MVIIC receptors, whereas non-immune IgG trapped Ͻ5% (Fig. 2A). Recovery of 125 I-MVIIC receptors by antibodies against other synaptic proteins implicated in the secretory pathway: synaptophysin, Rab 3A ( Fig. 2A), and cysteine string protein (data not shown) was similar to that obtained with non-immune IgG, although these antibodies did immunoprecipitate their respective antigens (not shown).
Immunoadsorption with increasing concentrations of antibodies against syntaxin, SNAP 25, or VAMP reached a plateau level, indicating that a limited fraction of calcium channels is associated with each SNARE protein ( Fig. 2A). If each SNARE protein interacts individually with a different fraction of calcium channels, immunoprecipitation by two antibodies combined should be additive. Alternatively, if all three SNARE proteins are associated with the same population of calcium channels, immunoprecipitation should not be additive. The results shown in Fig. 2B are consistent with the second hypothesis. When saturating amounts (20 g) of two antibodies were combined in the same assay, the percentage of immunoprecipitation was similar to that obtained when anti-syntaxin antibodies were combined with control antibodies. These results suggest that SNAP 25 and VAMP are only associated with calcium channels that are bound to syntaxin, implying the interaction of a trimeric syntaxin-SNAP 25-VAMP complex with a significant fraction of 125 I-MVIIC-labeled calcium channels.
The experiments illustrated in Fig. 2C were performed to verify that immunoprecipitation of 125 I-MVIIC binding sites associated with N-type channels did not occur. In each case the amount of radioactivity adsorbed by the antibodies was strongly reduced by addition of 0.3 M MVIIC, but not significantly affected by 0.3 M GVIA, indicating that the detected interaction with SNARE complexes is specific for P/Q-type channels that contain ␣ 1 A subunits. Synaptotagmin, a synaptic vesicle transmembrane protein that is thought to function as a calcium sensor in exocytosis, can associate with N-type calcium channels (Leveque et al., 1992(Leveque et al., , 1994Yoshida et al., 1992) and SNARE complexes (Söllner et al., 1993b), by binding to syntaxin (Li et al., 1995). An antibody that recognizes synaptotagmin isoforms I and II also immunoprecipitated P/Q-type channels (Fig. 2C).
In order to confirm the association of ␣ 1 A subunits with syntaxin, CHAPS extracts of cerebellar membranes were loaded onto an immunoaffinity matrix composed of a monoclonal anti-syntaxin antibody covalently coupled to Protein A-Sepharose 4BCL. After extensive washing, the column was eluted by a step to pH 10.5. Western blots of the recovered proteins were probed with anti-␣ 1 A antibodies, revealing several immunoreactive bands (Fig. 3, lane 1). These proteins were not detected when CHAPS extracts were loaded onto a control column (Fig. 3, lane 2) or when the antibodies were preincubated with an excess of the cognate ␣ 1 A peptide (Fig. 3, lane 3). The 220-, 180-, and 160-kDa proteins have been detected in cerebellar homogenates using anti-␣ 1 A (rbA-2) antibodies , and similar size forms have also been reported by Westenbroek and colleagues (1995). Heterogeneity in the size of these ␣ 1 A polypeptides expressed in the brain may result from alternative splicing or proteolytic processing. However, the 140-kDa protein was not detected in homogenates and is likely to be an ␣ 1 A cleavage product, generated during affinity chromatography in spite of the presence of protease inhibitors. These results thus indicate that syntaxin interacts with ␣ 1 A subunits of P/Q-type calcium channels.
FIG. 2. Association of SNARE complexes with P/Q-type calcium channels A, solubilized 125 I-MVIIC receptor complexes were incubated with anti-syntaxin 1 (squares), anti-SNAP 25 (diamonds), anti-VAMP 2 (circles), anti-Rab 3A (triangles), and anti-synaptophysin (asterisks) antibodies or non-immune IgG (crosses), and the radioactivity recovered in immune complexes was counted (means Ϯ S.D., n ϭ 3) B, additivity was examined with 20 g of each of the two indicated antibodies combined in the same assay (means Ϯ S.D., n ϭ 6). C, immunoassays were performed with 125 I-MVIIC receptors labeled in the absence of native -conotoxin (shaded) or in the presence of 0.3 M MVIIC (open) or GVIA (hatched) (means Ϯ S.D., n ϭ 6). SNARE Complex Association with P/Q-type Calcium Channels 6569 Sheng et al. (1994) did not detect binding of fusion proteins containing part of the cytoplasmic loop (residues 723-868) linking homologous domains II-III of the ␣1A subunit to syntaxin 1A. Our data indicate that there are alternative sites of interaction between SNARE complexes and P/Q-type calcium channel subunits, although they have yet to be identified at the sequence level. Current models of exocytosis are based on sequential protein-protein interactions leading to the assembly of a multimolecular complex, located at the interface between a synaptic vesicle and the presynaptic plasma membrane (reviewed by Sü dhof (1995)). The core interaction concerns a vesicle membrane protein VAMP, which binds to two proteins that are predominantly located at the plasma membrane: syntaxin and SNAP 25. A second vesicle protein synaptotagmin can bind to the trimeric SNARE complex, and there is compelling evidence that synaptotagmin acts as a calcium-dependent regulator of transmitter release (reviewed by Littleton and Bellen (1995)).
Our data indicate that a SNARE complex associated with synaptotagmin interacts with native voltage-gated calcium channels containing ␣ 1 A subunits solubilized from nerve terminals. Taken together with reports that N-type (Leveque et al., 1994;Sheng et al., 1994), but not L-type (Yoshida et al., 1992;El Far et al., 1995) channels bind to syntaxin, they are thus compatible with a predominant role for P/Q-and N-type but not L-type channels in rapid synaptic transmission in the mammals. Furthermore, our findings are consistent with recent functional evidence that syntaxin modulates the gating properties of calcium channels formed by the heterologous expression of ␣ 1 A or ␣ 1 B, but not ␣ 1 C subunits in Xenopus oocytes (Bezprozvanny et al., 1995). Molecular interactions that link calcium channels to proteins of the exocytotic complex could contribute to the rapid kinetics of excitation-secretion coupling by optimally locating the calcium sensors that trigger vesicle fusion, and may also regulate the ability of channels to open in response to depolarization.