Synaptotagmins are synaptic vesicle proteins containing two calcium‐binding C2 domains which are involved in coupling calcium influx through voltage‐gated channels to vesicle fusion and exocytosis of neurotransmitters. The interaction of synaptotagmins with native P/Q‐type calcium channels was studied in solubilized synaptosomes from rat cerebellum. Antibodies against synaptotagmins I and II, but not IV co‐immunoprecipitated [125I]ω‐conotoxin MVIIC‐labelled calcium channels. Direct interactions were studied between in vitro‐translated [35S]synaptotagmin I and fusion proteins containing cytoplasmic loops of the α1A subunit (BI isoform). Gel overlay revealed the association of synaptotagmin I with a single region (residues 780–969) located in the intracellular loop connecting homologous domains II and III. Saturable calcium‐independent binding occurred with equilibrium dissociation constants of 70 nM and 340 nM at 4°C and pH 7.4, and association was blocked by addition of excess recombinant synaptotagmin I. Direct synaptotagmin binding to the pore‐forming subunit of the P/Q‐type channel may optimally locate the calcium‐binding sites that initiate exocytosis within a zone of voltage‐gated calcium entry.
Transmitter release from nerve terminals is controlled by calcium influx through voltage‐gated channels. The consequent cytoplasmic calcium elevation triggers the fusion of synaptic vesicles at the active zone and the exocytosis of their contents into the synaptic cleft. Much recent effort has been directed at identifying nerve terminal proteins which couple calcium transients to exocytosis, and compelling mutational evidence indicates a key role for synaptotagmins as calcium sensor proteins in regulated secretion (reviewed by Littleton and Bellen, 1995; Südhof and Rizo, 1996).
Ten synaptotagmin genes have been identified in the rat (Craxton and Goedert, 1995; Li et al., 1995; Vician et al., 1995; Thompson, 1996), of which some appear to be specifically expressed in distinct patterns in neural and neuroendocrine tissues (Ullrich et al., 1994; Marquèze et al., 1995; Berton et al., 1997). Synaptotagmin I, a major membrane glycoprotein of both synaptic vesicles and neurosecretory granules has been the most extensively studied. It contains a short intravesicular N‐terminal segment, a single transmembrane region and a cytoplasmic tail containing two regulatory C2 domains which bind calcium and mediate Ca2+‐dependent processes including phospholipid binding, self‐association (Chapman et al., 1996) and interaction with syntaxin (Li et al., 1995).
Synaptotagmin and synaptic core complexes containing syntaxin, SNAP25 and VAMP can also associate with the voltage‐dependent calcium channels that are involved in triggering exocytosis, including N‐type (Lévêque et al., 1992, 1994; Yoshida et al., 1992; El Far et al., 1995) and P/Q‐type (Martin‐Moutot et al., 1996), but not L‐type calcium channels (El Far et al., 1995). Furthermore, co‐expression of syntaxin with calcium channels induces a modification of current gating properties displaying a similar specificity for N‐ or P/Q‐type channels (Bezprozvanny et al., 1995). These findings are consistent with observations that neurotransmitter release at many central synapses is blocked by antagonists of N‐ or P/Q‐type calcium channels, but is insensitive to inhibitors of L‐type channels (Takahashi and Momiyama, 1993; Wheeler et al., 1994). Neuronal calcium channels are heteromeric proteins constituted by an α1 subunit which forms the transmembrane pore, associated with auxiliary α2δ and β subunits (Birnbaumer et al., 1994). Association with core complexes involves syntaxin and SNAP25 binding to α1 subunits on the cytoplasmic loop that links homologous domains II and III (Sheng et al., 1994, 1996; Rettig et al., 1996).
It has been suggested that interactions between synaptic protein complexes and calcium channels may optimally locate the calcium sensor synaptotagmin within domains of elevated calcium concentration generated close to the channel (Lévêque et al., 1992; Neher and Penner, 1994). However, it is not known whether synaptotagmin binds directly to the calcium channel or is linked indirectly via its association with a component of the core complex such as syntaxin. We have therefore used radiolabelled synaptotagmin I as a probe to detect and characterize binding sites on fusion proteins containing cytoplasmic loops of the α1A (BI) subunit that constitutes the pore of the P/Q‐type calcium channel.
Interaction of synaptotagmins with P/Q‐type calcium channels in cerebellar membranes
Antibodies were prepared against synthetic peptides specific to the sequences of synaptotagmins I, II and IV, which are neuronal synaptotagmin isoforms expressed in cerebellum (Marquèze et al., 1995; Berton et al., 1997). Maltose binding protein (MBP) fusion proteins containing the entire coding sequences of synaptotagmins I, II and IV were prepared (Figure 1A) and the specificity of the antibodies was confirmed by ELISA (Figure 1B). Immunoprecipitation and Western blotting experiments (not shown) confirmed that antibodies coupled to protein A–Sepharose beads were able to trap their respective antigens from detergent extracts of cerebellar membranes. These antibodies were then used to detect synaptotagmin association with native P/Q‐type calcium channels. Cerebellar P/Q‐type channels were specifically labelled with 125I‐labelled ω‐conotoxin MVIIC (ωMVIIC), solubilized with 3‐[(3‐cholamidopropyl)dimethylammonio]‐1‐propanesulfonic acid (CHAPS) (Martin‐Moutot et al., 1996) and the ability of different anti‐synaptotagmin antibodies to capture protein complexes containing [125I]ωMVIIC receptors was assayed. Antibodies directed against calcium channel α1A subunits and synaptotagmins I and II immunoprecipitated significant fractions of P/Q‐type calcium channels, whereas anti‐synaptotagmin IV antibodies did not (Figure 2A). These antibodies reacted with protein bands of 65 and 67 kDa in Western blots of rat brain cerebellar membranes, corresponding to synaptotagmins I and II respectively (Figure 2B). In contrast, synaptotagmin I was strongly stained in membrane preparations from rat cortex, whereas synaptotagmin II was not detected. The expression pattern of synaptotagmins I and II, particularly the very low density of synaptotagmin II in the cortex, is in good agreement with the localization of their mRNAs by in situ hybridization (Marquèze et al., 1995). In all cases, immunoreactivity in Western blots was blocked by including an excess of the peptides used to raise the antibodies.
In view of these results, the possibility of a direct interaction between synaptotagmin I and the α1A subunit of the P/Q‐type calcium channel was examined using recombinant proteins.
Interaction between synaptotagmin I and a cytoplasmic domain of the α1A subunit
35S‐labelled synaptotagmin I was prepared by in vitro translation in the presence of [35S]methionine (Figure 3A). Glutathione S‐transferase (GST) fusion proteins containing sequences from cytoplasmic loops I–II, II–III and III–IV linking homologous domains of the α1A subunit were purified, separated by SDS–PAGE and detected by Coomassie blue staining (Figure 3B). Proteins from a separate gel were transferred to PVDF membranes and probed with [35S]synaptotagmin I. Autoradiography of the overlay revealed binding of [35S]synaptotagmin I to proteins containing loop II–III (Figure 3C), but not to other loops nor to GST, suggesting a specific interaction.
These data were quantified by performing binding assays in solution, then recovering fusion protein–[35S]‐synaptotagmin I complexes on glutathione–Sepharose beads. Significant interaction was detected with the fusion protein containing residues 780–969 of the BI isoform of the α1A subunit (Figure 4A), but not with other intracellular domains, nor the equivalent II–III region of the α1C subunit of the L‐type calcium channel. The ability of this region to associate with native synaptotagmin was examined by incubating GST‐α1A780−969 with solubilized cerebellar membranes. Protein complexes were trapped on glutathione–Sepharose and analysed by SDS–PAGE and Western blotting. Immunoreactive bands containing synaptotagmin were identified (Figure 4B). These results indicate that both recombinant and native synaptotagmins can associate with an intracellular domain of the P/Q‐type calcium channel.
Characterization of the interaction between synaptotagmin I and GST‐α1A780–969
Incubation of [35S]synaptotagmin I with increasing concentrations of GST‐α1A780–969 revealed saturable binding (Figure 5A). This interaction was reversible and did not occur with GST alone, nor when a large excess of MBP–synaptotagmin I fusion protein was included in the assay (Figure 5C). Analysis of the saturation curve (Figure 5B) revealed two sites, with equilibrium dissociation constants (KD) of 70 and 340 nM at pH 7.4 and 4°C.
As synaptotagmin I contains calcium‐binding sites, we examined the effects of increasing calcium concentrations in 10‐fold increments over a range of 10 nM to 1 mM. No significant effects of calcium on [35S]synaptotagmin I binding to GST‐α1A780–969 nor to GST alone were revealed (Figure 6A). In positive control experiments (Figure 6B), MBP–synaptotagmin was found to bind to GST–syntaxin I, but not to GST. The interaction between GST–syntaxin I and MBP–synaptotagmin I, estimated by densitometric scanning of SDS–PAGE gels stained with Coomassie blue, was stimulated 2‐fold by the presence of calcium, in agreement with previous reports (Li et al., 1995).
Neurotransmitter release from axonal terminals is triggered by calcium entry through voltage‐gated channels in the presynaptic plasma membrane and compelling evidence suggests that certain synaptotagmins may act as calcium‐dependent regulators of exocytosis. Synaptotagmin displays calcium binding in the 100 μM range (reviewed by Südhof and Rizo, 1996), which parallels the affinity of the calcium sensor for exocytosis (reviewed by Matthews, 1996), and the introduction of synaptotagmin peptides or anti‐synaptotagmin antibodies into neuronal cells disrupts exocytosis (Bommert et al., 1993; Elferink et al., 1993). Synaptic responses in hippocampal neurones from mice homozygous for a mutation in the synaptotagmin I gene display suppression of synchronous, but not asynchronous or spontaneous neurotransmitter release (Geppert et al., 1994). Certain mutations in the synaptotagmin gene of Drosophila melanogaster affect the calcium dependence of evoked transmitter release at the neuromuscular junction, compatible with the view that synaptotagmin functions as a calcium receptor for initiating exocytosis (Littleton et al., 1994).
The sub‐millisecond delay between calcium influx and neurotransmitter release implies that calcium channels at active zones are closely associated with the calcium sensors for exocytosis (Llinas et al., 1981). We have therefore examined interactions between P/Q‐type calcium channels, which control a predominant fraction of transmitter release at many synaptic fields in the mammalian brain (Takahashi and Momiyama 1993; Wheeler et al., 1994) and the synaptotagmins.
Immunoprecipitation assays demonstrated that antibodies against synaptotagmins I or II but not synaptotagmin IV captured native P/Q‐type channels. This is consistent with the high degree of homology between synaptotagmins I and II. In contrast, synaptotagmin IV (Vician et al., 1995) has relatively low homology to synaptotagmins I and II, and does not bind syntaxin or phospholipids in a Ca2+‐dependent manner (Ullrich et al., 1994), although some controversy surrounds the latter issue (Fukuda et al., 1995; Südhof and Rizo, 1996). However, we cannot eliminate the possibility that our results are a reflection of the relatively low expression of synaptotagmin IV. Given the large number of synaptotagmin isoforms identified to date, it will be interesting to investigate differences in their capacity to interact with P/Q‐type calcium channels more thoroughly.
Although these results demonstrate the association of synaptotagmins with calcium channels in vitro, they cannot inform us as to whether association occurs in vivo and, if so, in which subcellular compartments. Synaptotagmins are predominantly distributed in synaptic vesicles, but the fact that they have also been localized to the presynaptic plasma membrane should be taken into account (Takahashi et al., 1991).
Immunoprecipitation of complexes containing synaptotagmin and P/Q‐type calcium channels does not demonstrate direct interaction between the two proteins, and may occur because both proteins can bind simultaneously to a third partner. Synaptic vesicle exocytosis is thought to require formation of a trimeric synaptic core complex containing two abundant plasma membrane proteins, syntaxin and SNAP25, and a synaptic vesicle membrane protein, VAMP. Multiple synaptic proteins can bind to the core complex, suggesting that it constitutes the hub of a precise sequence of protein–protein interactions that prepare vesicle fusion (reviewed by Südhof, 1995). P/Q‐type calcium channels associate with synaptic core complexes (Martin‐Moutot et al., 1996)—an interaction that involves syntaxin and SNAP25 binding to the cytoplasmic loop connecting homologous domains II and III of the α1A (BI) subunit (Rettig et al., 1996). Synaptotagmin I can bind to syntaxin and trimeric core complexes, an interaction which could thus mediate an indirect association with the calcium channel. However, our experiments with recombinant synaptotagmin I confirmed direct interaction with residues 780–969 of the BI isoform of the α1A subunit, with two KDs of 70 and 350 nM. Detection of two binding sites could be due to the presence of proteolytic fragments of the GSTα1A780–969 fusion protein with different affinities for synaptotagmin I. Alternatively, although self‐association is unlikely at the low [35S]synaptotagmin I concentrations used in our assays, it may reflect distinct binding affinities of monomeric versus multimeric forms of synaptotagmin I.
The synaptotagmin interaction site is thus located within the syntaxin and SNAP25 binding domain (residues 722–1036) reported by Rettig et al. (1996). This suggests that distinct sequential protein–protein interactions may occur in a restricted region of the calcium channel, possibly representing steps in the assembly of a fusion‐competent complex. It will therefore be important to focus future investigations towards the precise mapping of binding sites for synaptotagmin, syntaxin and SNAP25 on the P/Q‐type calcium channel, which may be independent or overlapping, possibly leading to allosteric or competitive interactions. Rigorous analysis of these interactions will however be challenging, given the ability of each component to bind to multiple partners in the complex.
The detection of a synaptotagmin binding site on the calcium channel has several implications for the mechanisms of transmitter release. Firstly, it provides a novel site for interaction between the synaptic vesicle and the plasma membrane that may be relevant to the docking process. Secondly, synaptotagmin alone can mediate a link which would localize calcium‐binding C2 domains within a cloud of voltage‐gated calcium entry close to the channel. Finally, synaptotagmin I binds to the α1A subunit of the calcium channel, a protein which contains four charged helical S4 segments that sense the membrane electric field. This direct coupling of a calcium sensor to a voltage sensor may be relevant to the Ca2+‐voltage hypothesis (reviewed by Parnas and Parnas, 1994; see also discussion by Neher and Penner, 1994) which holds that transmitter release is regulated by both Ca2+ and membrane potential. In this context it is conceivable that the ability of synaptotagmin to trigger release may also be modulated by voltage‐driven conformational transitions corresponding to channel gating.
Materials and methods
A monoclonal antibody (mAb) directed against synaptotagmin I (mAb3F10A, raised against a peptide corresponding to residues 1–20), polyclonal antibodies against synaptotagmin II (residues 1–20) and a monoclonal antibody that recognizes both synaptotagmins I and II (mAb1D12), were prepared as previously described (Shoji‐Kasai et al., 1992). Polyclonal antibodies against synaptotagmin IV were raised in rabbits against peptide (C)PKLFPETEKEAVSPES (residues 123–138) from the sequence of synaptotagmin IV.
Rat synaptotagmins I, II and IV were bacterially expressed as MBP fusion proteins using the New England Biolabs Protein Fusion and Purification System. Plasmids encoding the entire coding sequence of synaptotagmins I, II and IV were constructed by PCR amplification of rat synaptotagmin cDNA using oligonucleotides containing appropriate flanking restriction sites for subcloning into the pMAL‐C2 vector (New England Biolabs). Recombinant proteins were purified by affinity chromatography on an amylose column (New England Biolabs) and stored at −80°C. GST fusion proteins from the rabbit brain calcium channel α1A (BI‐2, Mori et al., 1991) subunit were prepared as described previously (De Waard et al., 1997). GST fusion proteins containing the III–IV loop of α1A and the II–III loop of α1C (Wei et al., 1991) were constructed by amplifying base pairs 4561–4725 and 2350–2757 of the respective genes and subcloning into pGEXkG and pGEX2Tk. A GST fusion protein containing full‐length syntaxin 1A was constructed by cloning into pGEX2T. Fusion protein production was induced in cultures of protease‐deficient Escherichia coli strain BL21, and purification from lysates was performed with glutathione–agarose. [35S]methionine‐labelled (>1000 Ci/mmol, Dupont) synaptotagmin I was synthetized in vitro by coupled transcription and translation using the TNT™ system (Promega).
ELISAs were performed by immobilizing synaptotagmin fusion proteins in 96‐well microplates, blocking with 1% bovine serum albumin in Tris 25 mM, NaCl 150 mM adjusted to pH 7.4 with HCl (Tris‐buffered saline, TBS) and incubating with 1–10 μg/ml IgG for 2 h at 37°C. After washing and incubation with a secondary antibody coupled to alkaline phosphatase, immunoreactivity was quantified by monitoring the cleavage of p‐nitrophenylphosphate. The preparation of 125I‐labelled ω‐conotoxin MVIIC (2200 Ci/mmol), SDS–PAGE, Western blotting and immunoprecipitation assays with solubilized [125I]ω‐conotoxin MVIIC receptors were performed as described by Martin‐Moutot et al. (1996).
Following SDS–PAGE and electroblotting to PVDF membranes, overlay assays were performed by blocking the membrane with 5% non‐fat milk and 5% BSA in TBS and incubation with 30 000 c.p.m./ml of [35S]synaptotagmin I in the same buffer. After washing for 1 h in 5% BSA in TBS, membranes were exposed overnight to Kodak X‐Omat film. Binding assays were performed in solution (De Waard et al., 1995) by incubating GST fusion proteins with in vitro‐translated [35S]synaptotagmin I in 200 μl of EGTA‐Ca2+ buffer (0.5% BSA, 0.1% Triton X‐100, EGTA 50 μM and CaCl2 to give the appropriate free Ca2+ concentration, in TBS) at 4°C in the presence of 50 μl glutathione–agarose beads. Beads were washed three times by centrifugation with the same buffer and [35S]synaptotagmin I binding was quantified by scintillation counting.
This study was supported by a research grant from the International Human Frontiers Science Programme, a joint grant from INSERM and the Japanese Society for the Promotion of Science, and an INSERM post‐doctoral fellowship (poste vert) to D.W.
- Copyright © 1997 European Molecular Biology Organization