Direct interaction of the Rab3 effector RIM with Ca2+ channels, SNAP-25, and synaptotagmin.

To define the role of the Rab3-interacting molecule RIM in exocytosis we searched for additional binding partners of the protein. We found that the two C(2) domains of RIM display properties analogous to those of the C(2)B domain of synaptotagmin-I. Thus, RIM-C(2)A and RIM-C(2)B bind in a Ca(2+)-independent manner to alpha1B, the pore-forming subunit of N-type Ca(2+) channels (EC(50) = approximately 20 nm). They also weakly interact with the alpha1C but not the alpha1D subunit of L-type Ca(2+) channels. In addition, the C(2) domains of RIM associate with SNAP-25 and synaptotagmin-I. The binding affinities for these two proteins are 203 and 24 nm, respectively, for RIM-C(2)A and 224 and 16 nm for RIM-C(2)B. The interactions of the C(2) domains of RIM with SNAP-25 and synaptotagmin-I are modulated by Ca(2+). Thus, in the presence of Ca(2+) (EC(50) = approximately 75 microm) the interaction with synaptotagmin-I is increased, whereas SNAP-25 binding is reduced. Synaptotagmin-I binding is abolished by mutations in two positively charged amino acids in the C(2) domains of RIM and by the addition of inositol polyphosphates. We propose that the Rab3 effector RIM is a scaffold protein that participates through its multiple binding partners in the docking and fusion of secretory vesicles at the release sites.

Secretion of neurotransmitters, polypeptide hormones, and a variety of other proteins occurs by exocytosis, a multistage process including targeting, docking, and, finally, fusion of secretory vesicles with the plasma membrane. During the last few years, a combination of genetic and biochemical approaches has lead to the identification of several proteins involved in this complex cascade of events. Most of these proteins turned out to be specialized components of the evolutionary conserved machinery that governs intracellular vesicular trafficking in eukaryotic cells (1). Exocytosis was found to necessitate the assembly of a ternary complex between the vesicular SNARE 1 VAMP, associated with the secretory vesicle, and the target SNAREs syntaxin-1 and SNAP-25, localized at the plasma membrane (2). The SNARE complex was initially proposed to ensure the docking of secretory vesicles at the plasma membrane (3). However, it is unlikely that SNARE assembly constitutes the sole determinant for the targeting of secretory vesicles at the release sites because SNARE pairing is rather promiscuous (4,5), and the localization of syntaxin-1 and SNAP-25 is not restricted to active zones (6). A current hypothesis, supported by biochemical and structural data, proposes that the assembly of the heterotrimeric complex between VAMP, SNAP-25, and syntaxin-1 provides the driving force for membrane fusion (7).
In most secretory systems, the exocytotic process is initiated by an increase in the intracellular Ca 2ϩ concentration. In some cells, such as neurons, the elevation of Ca 2ϩ ions is due to opening of voltage-gated calcium channels that are clustered at the release sites, whereas in others, Ca 2ϩ ions are mobilized from intracellular stores. Biochemical and genetic studies indicate that synaptotagmins constitute the main Ca 2ϩ sensors for regulated exocytosis (8). Synaptotagmin (Syt)-I, the best characterized member of the synaptotagmin family, is localized on synaptic vesicles and is essential for the fast, Ca 2ϩ -dependent component of neurotransmitter release (9,10). Synaptotagmins possess a short N-terminal tail that resides inside secretory vesicles, a single membrane-spanning region, and a large cytoplasmic sequence containing two so-called C 2 domains designated C 2 A and C 2 B. These two domains cooperate to mediate the binding of synaptotagmins with several proteins and phospholipids. Although most of these interactions are influenced by both C 2 domains, Ca 2ϩ -dependent binding to syntaxin-1 and anionic phospholipids is predominantly associated with C 2 A (11)(12)(13). In contrast, Ca 2ϩ -dependent oligomerization of synaptotagmins and several Ca 2ϩ -independent interactions such as the binding to the II-III cytoplasmic loop or "synprint" (synaptic protein interaction) motif of the pore-forming subunit of Ca 2ϩ channels (14) to inositol polyphosphates (15) and to SNAP-25 (16) are mediated predominantly by C 2 B.
Ras-like GTPases of the Rab family represent additional components of the machinery involved in vesicular trafficking. Rab GTPases are associated with distinct cellular compartments and control different steps in the secretory pathway (17). The four isoforms of Rab3 (Rab3A, -B, -C, and -D) are associated with secretory vesicles of neuronal, endocrine, and exocrine cells and regulate exocytosis (18). It is still unclear whether Rab3 proteins participate in vesicle docking, membrane fusion, or both processes. Constitutively active mutants of Rab3 diminish regulated secretion (19 -22), and in Rab3Adeficient mice, the number of synaptic vesicle fusions elicited by a nerve impulse is increased (23), suggesting that Rab3 functions as a negative modulator of exocytosis. However, amylase secretion is enhanced in transgenic mice displaying mild overexpression of Rab3D in pancreatic acinar cells (24), and inhibition of Rab3B expression in anterior pituitary cells by antisense oligonucleotides results in a decrease in Ca 2ϩ -induced secretion (25). These contradictory observations could be due to differences in the cell systems, or they may reflect the existence of positive and negative regulatory functions of Rab3 proteins differentially highlighted by the experimental approach used. Rabs have been suggested to contribute an additional layer of specificity to vesicular transport by controlling the assembly of the SNARE complex. Ypt1, a Rab protein implicated in the transport between endoplasmic reticulum and Golgi in yeast, has been proposed to favor SNARE assembly by associating with the target SNARE Sed5 (26). However, the binding of other members of the Rab family to target SNAREs was found to be inefficient and nonspecific (4). Rab6, Rab3, and Rab4 have also been shown to bind to the N-ethylmaleimide-sensitive fusion protein, but the physiological meaning of this interaction remains to be established (27). Recently, the Rab5 effector EEA1 has been reported to interact directly with the target SNAREs syntaxin-6 and syntaxin-13 (28,29), indicating that some Rabs may rely on their effectors for controlling the assembly of the SNARE complex.
RIM is a putative effector of Rab3 that associates selectively with the active form of the GTPase (30). RIM contains an N-terminal domain that interacts with Rab3 and Munc13-1 (31) and two C 2 domains located at the C terminus. In rat, the distance between the two C 2 regions of RIM is determined by alternatively spliced sequences that generate a large set of different isoforms (30,32). RIM is concentrated in presynaptic active zones, and overexpression of the Rab3-binding domain of RIM in PC12 and insulin-secreting cells enhances exocytosis elicited by secretagogues (30,33). RIM was also found to bind to the cAMP sensor cAMP-GEFII and to mediate cAMP-induced secretion (34).
In this study, we attempted to elucidate the role of this putative Rab3 effector by searching for possible binding partners of the protein. We found that the C 2 domains of RIM mediate the interaction of this Rab3 target with the poreforming subunit of Ca 2ϩ channels and with Syt-I and SNAP-25. This indicates that RIM constitutes a scaffold protein that binds to different components of the exocytotic machinery and suggests that it coordinates the docking and fusion of secretory vesicles at release sites.

EXPERIMENTAL PROCEDURES
Materials-The plasmids encoding the cytoplasmic domain of syntaxin-1A and a fusion protein between GST and the cytoplasmic domain of rat Syt-I (residues 80 -421) were kindly provided by Dr. R. Scheller (Stanford University, Stanford, CA). The human brain cDNA clone KIAA0340 (GenBank TM accession number AB002338) was obtained from the Kazusa DNA Research Institute. The antibodies against Syt-I and the ␣1B subunit of Ca 2ϩ channels were purchased from Chemicon International. The antibody against SNAP-25 (SMI 81) was from Sternberg Monoclonals Incorporated. Inositol-1,2,3,4,5,6-hexakisphosphate (IP6) was from Calbiochem. The antibody directed against Syt-II has been described previously (35).
cDNA Cloning-The C-terminal region including the two C 2 domains of different RIM isoforms was cloned by RT-PCR from human brain mRNA (1 g) (CLONTECH). The first-strand cDNA was synthesized with an oligo(dT) primer using the omniscript kit (Qiagen). The polymerase chain reaction was performed using forward primer 5Ј-TG-GACGTCCTCGAAATCCCTATG-3Ј and reverse primer 5Ј-ATATGCG-GCCGCTGATCGAATACAGGGAGGCCC-3Ј. The former primer was designed according to the sequence of human brain cDNA clone KIAA0340 from the Kazusa DNA Research Institute (GenBank TM accession number AB002338); the latter primer was designed according to the sequence of the human RIM gene (GenBank TM accession number AL035633). The polymerase chain reaction was performed using the following conditions: 94°C for 2 min, followed by 30 cycles of 94°C for 30 s, 52°C for 30 s, and 72°C 2 min. The polymerase chain reaction products were subcloned in the pGEM T-easy system (Promega) for sequencing. Sequence analysis was performed by MWG Biotech Company.
The cDNA sequences encoding the cytoplasmic loop between domain II and III of the ␣1B, ␣1C, and ␣1D subunits of Ca 2ϩ channels were cloned by RT-PCR from mouse brain mRNA. Specific primers for ␣1B and ␣1C were designed from the reported cDNA sequences (36,37) (GenBank TM accession numbers U04999 and L01776). The primers for ␣1D were derived from the sequence of the rat homologue (38). The ␣1B (amino acids 713-933), ␣1C (amino acids 754 -899), and ␣1D (amino acids 773-907) fragments generated by RT-PCR were sequenced and then subcloned in myc-pcDNA3, a modified pcDNA3 vector (Invitrogen) designed to insert a myc epitope at the N terminus of the amino acid sequence (21).
DNA Constructs-To generate the corresponding GST fusion proteins, the C 2 domains of RIM were amplified by polymerase chain reaction and subcloned in bacterial expression vector pGEX-KG (39). RIM-C 2 A was amplified from the KIAA0340 clone (residues 744 -839). RIM-C 2 B was amplified from one of the RIM clones obtained by RT-PCR (residues 615-760). The same procedure was used to generate an expression plasmid encoding the C 2 B domain (residues 262-386) of rat Syt-I.
Interaction of RIM C 2 Domains with ␣1 Subunits of Ca 2ϩ Channels and with Syntaxin-1A-The C 2 domains of RIM and the cytosolic portion of Syt-I produced as GST fusion proteins were immobilized on glutathione-agarose beads. The beads were resuspended in 20 mM HEPES, pH 7.5, 150 mM KCl, 1 mM dithiothreitol, 5% glycerol, 0.05% Tween 20, and 1 mg/ml bovine serum albumin. They were then incubated with 35 S-labeled fragments of ␣1 subunits or with the cytoplasmic domain of syntaxin-1A produced by in vitro translation (Promega). The proteins that remained associated with the affinity columns after extensive washing were analyzed by SDS-PAGE and revealed by autoradiography or, for quantification experiments, by Molecular Imager FX (Bio-Rad).
Interaction of the C 2 Domains of RIM with Synaptotagmins and SNAP-25-Rat brain detergent extracts were prepared as described previously (11) in HEPES-buffered saline (50 mM HEPES, pH 7.4, 100 mM NaCl, and 1% Triton X-100). 1-ml aliquots of the brain extract (about 1 mg of protein) were preincubated for 30 min at 4°C in the presence of 2 mM EGTA or 500 M CaCl 2 and then incubated in the same buffer with immobilized GST fusion proteins for 2 h at 4°C. The proteins remaining attached to the beads after several washes were detected by Western blotting using the indicated antibodies. To investigate the effect of different Ca 2ϩ concentrations on the binding of RIM C 2 domains to Syt-I, the preincubations and incubations were performed in Ca 2ϩ /EGTA buffers (40). The final concentration of EGTA in the buffers was 2 mM. The fraction of Syt-I bound to the GST affinity columns was estimated by analyzing the films with Scion Image software. The same quantification method was used for saturation experiments.
Site-directed Mutagenesis-Site-directed mutagenesis was performed using the Quickchange kit (Stratagene) and verified by sequencing the insert of the plasmids.

Cloning of the C Terminus of Six Different Isoforms of RIM
from Human Brain-To shed light on the role of the Rab3interacting molecule RIM, we searched for additional binding partners interacting with the C terminus of the protein. For this purpose, the 3Ј region of RIM encompassing the two C 2 domains was amplified by RT-PCR from human brain mRNA. Experiments performed on first-strand cDNA generated several heterogeneous DNA fragments with a size between 1600 and 2300 base pairs. The sequences of six DNA fragments have been submitted to GenBank TM (GenBank TM accession numbers AF263305-AF263310). The sequences encoded by the DNA fragments display a high degree of homology with the corresponding region of rat RIM. Alignment of the sequences indicates that the RT-PCR fragments result from a combination of different splice variants of human RIM mRNA that diverge in two domains corresponding to the splicing site B and C described in rat RIM (30,32). Site B appears to be an hypervariable region because each of the six clones displays a different sequence. These observations demonstrate that human brain contains several RIM isoforms that differ from each other in the distance between the two C 2 domains.
Direct comparison of the C 2 domains of RIM with the C 2 domains of Syt-I ( Fig. 1) reveals that the overall homology between the sequences is rather low (RIM-C 2 A has 20% and 19% homology with Syt-I C 2 A and C 2 B; RIM-C 2 B has 16% and 22% homology with Syt-I C 2 A and C 2 B). However, both C 2 domains of RIM contain a cluster of four amino acids homologous to the KKKT sequence of Syt-I C 2 B. The KKKT sequence is conserved in RIM-C 2 B, whereas in RIM-C 2 A, two lysines are replaced with two arginines (KRRT) (Fig. 1). The KKKT sequence is known to be required for the interaction with the synprint motif of Ca 2ϩ channels (14,41), for Syt-I oligomerization, and for binding to the adaptor protein AP-2 (41,42). More recently, this particular sequence has been postulated to be the oligomerization interface for Syt-I (43).
Binding of the C 2 Domains of RIM to ␣1 Subunits of Ca 2ϩ Channels-The presence of the KKKT sequence prompted us to investigate whether the C 2 domains of RIM interact with known partners of Syt-I. We first tested the ability of the C 2 domains of RIM to bind to the ␣1 subunits of N-and L-type Ca 2ϩ channels. To address this issue, GST fusion proteins containing the cytoplasmic domain of Syt-I, RIM-C 2 A, or RIM-C 2 B were immobilized on glutathione-agarose beads. They were then incubated with recombinant in vitro-translated proteins representing the cytoplasmic loop connecting domain II and III of ␣1B (N-type), ␣1C (L-type), and ␣1D (L-type) subunits. As shown in Fig. 2A, the in vitro-translated fragment of the neuron-specific ␣1B subunit associated equally well with the GST affinity columns of Syt-I, RIM-C 2 B, and RIM-C 2 A. The binding was not affected by the concentration of Ca 2ϩ in the incubation buffer (data not shown). A weak interaction of the three recombinant proteins with the L-type ␣1C subunit could also be observed, but a much longer exposure of the film was required ( Fig. 2A). In contrast, no specific binding was detectable with the cytoplasmic loop of ␣1D. To determine more precisely the affinity of the pore-forming subunit of N-type Ca 2ϩ channels for Syt-I and for the C 2 domains of RIM, we performed dose-response experiments using a fixed amount of GST fusion proteins and variable amounts of in vitro-translated ␣1B (Fig. 2B). Using this approach, we estimated that half-maximal binding of ␣1B synprint to Syt-I is obtained at about 25 nM. This value is in good agreement with the K D of 25 nM measured previously using a solid phase assay (14). The affinity of ␣1B for RIM-C 2 A and RIM-C 2 B (14 and 18 nM, respectively) was almost identical to that of ␣1B for Syt-I C 2 B. These results suggest that in neurons, the interaction between the C 2 domain of RIM and the N-type calcium channel could contribute to ensure a tight structural and functional association of the exocytotic machinery with the sites of calcium entry.
Syt-I has been reported to bind to syntaxin-1. This interaction is modulated by Ca 2ϩ and is mediated mainly by the C 2 A domain of Syt-I (11,44). As expected, we observed a strong binding of in vitro-translated syntaxin-1 with the cytoplasmic domain of Syt-I in the presence of calcium and a weaker binding in the absence of calcium (Fig. 3). In contrast, under the same conditions, syntaxin-1 was unable to interact efficiently with the C 2 domains of RIM or with GST alone.
Interaction of the C 2 Domains of RIM with Syt-I and SNAP25-We next tested the ability of the C 2 domains of RIM to associate with Syt-I and SNAP-25. In this case, GST fusion proteins containing Syt-I C 2 B, RIM-C 2 A, or RIM-C 2 B were incubated with brain extracts in the presence of either EGTA or 500 M Ca 2ϩ . In the absence of Ca 2ϩ , binding of rat brain Syt-I to the GST affinity columns was barely detectable (Fig. 4). However, in the presence of Ca 2ϩ , the binding to RIM-C 2 A and RIM-C 2 B and the oligomerization with Syt-I were dramatically increased. Similar results were obtained with an antibody directed against Syt-II, indicating that the C 2 domains can interact in a Ca 2ϩ -dependent manner with different Syt isoforms (data not shown). Ca 2ϩ -dependent binding was also observed in An aliquot (10%) of each in vitro translation product used for the reactions is shown for comparison (IVT). The proteins remaining associated with the GST affinity columns were resolved by SDS-PAGE and detected by autoradiography. The autoradiography film shown for ␣1B was exposed for 1 day, whereas the films for ␣1C and ␣1D were exposed for 11 days. B, measurement of the affinity of the different C 2 domains for ␣1B. Increasing concentrations of the radioactively labeled ␣1B synprint fragment were incubated with 1 mg of the indicated GST fusion proteins. The material retained on the affinity columns was analyzed by SDS-PAGE. The radioactive protein band was quantified using Molecular Imager FX (Bio-Rad). The results are expressed as a percentage of ␣1B maximal binding. The calculated EC 50 values were 14 nM for RIM-C 2 A, 18 nM for RIM-C 2 B, and 25 nM for Syt-C 2 B. the presence of the purified cytoplasmic domain of Syt-I (data not shown), showing that RIM-C 2 A and RIM-C 2 B associate directly with synaptotagmins. To confirm the specificity of these interactions, we estimated the binding affinities at 500 M Ca 2ϩ using different amounts of brain extracts. Using a recombinant protein as a standard, in our rat brain extracts Syt-I was found to represent ϳ0.1% of the protein content. By applying this estimate, half-maximal binding of rat brain Syt-I to the GST affinity columns containing Syt-I C 2 B, RIM-C 2 A, and RIM-C 2 B was calculated to be ϳ12, 24, and 16 nM, respectively (Fig. 5A). We also estimated the Ca 2ϩ dependence of the binding of Syt-I to the C 2 domains of RIM (Fig. 6). The halfmaximal effect for hetero-oligomerization with the C 2 domains of RIM and homo-oligomerization with Syt-I was very similar and was observed in the presence of 70 -100 M Ca 2ϩ (Fig. 6). This value is consistent with the Ca 2ϩ dependence for exocytosis (45). Thus, Ca 2ϩ triggers not only the self-association of synaptotagmins but also promotes the hetero-oligomerization of distinct C 2 domains with properties analogous to those of Syt-I C 2 B.
To test whether the C 2 domains of RIM are able to associate with SNAP-25, an aliquot of the proteins remaining attached to the affinity columns was analyzed by Western blotting, using an antibody against SNAP-25. As shown in Fig. 4, in the absence of Ca 2ϩ , RIM-C 2 A and RIM-C 2 B were able to fix SNAP-25. In the presence of Ca 2ϩ (500 M), the amount of SNAP-25 associated with the affinity columns was strongly reduced. To determine the affinity of the different C 2 domains for SNAP-25, we performed dose-response experiments in the absence of Ca 2ϩ . SNAP-25 in our rat brain extract was estimated to represent about 0.35% of the protein content. Using this estimated value, under our experimental conditions, the EC 50 of the interaction of SNAP-25 with Syt-I C 2 B, RIM-C 2 A, and RIM-C 2 B was ϳ253, 203, and 224 nM, respectively (Fig.  5B).
Definition of the Binding Interface of the C 2 Domains of RIM with ␣1B and Syt-I-The two lysines in the middle of the KKKT sequence of Syt-I are known to be required for selfoligomerization and for the binding of the protein to the Ca 2ϩ channel (41). To assess whether the interaction of RIM with its partners occurs by a similar mechanism, the two positively charged amino acids in the middle of the KRRT and KKKT sequence of RIM-C 2 A and RIM-C 2 B, respectively, were mutated to alanine. As shown in Fig. 7, these mutants of RIM-C 2 A and RIM-C 2 B were unable to bind to the ␣1B subunit of the N-type Ca 2ϩ channel and to Syt-I. This confirms that the binding properties of the C 2 domains of RIM are indeed provided by the KK(R)K(R)T motif.
The Binding of the C 2 Domains of RIM to Syt-I Is Sensitive to Inositol Polyphosphates-Synaptotagmins are known to bind inositol polyphosphates (15). The binding site has been shown to overlap the domain required for self-oligomerization of synaptotagmins (46). When the experiment in Fig. 4 was repeated in the presence of 100 M IP6, rat brain Syt-I was not retained by the GST affinity column containing the cytoplasmic domain of Syt-I (Fig. 8). Interestingly, the binding of Syt-I to the C 2 domains of RIM was also abolished by IP6. Similar results were obtained in the presence of 10 M IP6 (data not shown). These findings again substantiate the similarities existing between self-oligomerization of synaptotagmins and hetero-oligomerization of synaptotagmin with the C 2 domains of RIM.

DISCUSSION
Rab3 is thought to control docking and fusion of secretory vesicles at the plasma membrane. In this study, we provide evidence that RIM, a putative effector of Rab3, constitutes a functional link between the GTPase and the machinery for exocytosis. The association of RIM with the components of the secretory apparatus is mediated by two C 2 domains located at the C terminus of the protein. C 2 domains are widespread motifs that provide Ca 2ϩ and phospholipid binding activity to many of their parent molecules (47). The various C 2 domains display distinct features that enable them to participate in multiple biological functions. The two C 2 domains of Syt-I are endowed with unique binding properties that render the protein ideally suited to respond, in a millisecond time scale, to variations in the intracellular Ca 2ϩ concentration (10). Both C 2 domains of Syt-I cooperate to perform Ca 2ϩ -dependent interactions. However, Ca 2ϩ -dependent binding to phospholipids  GST alone or GST fusion proteins including Syt-I C 2 B, RIM-C 2 A, and RIM-C 2 B were immobilized on glutathione-agarose beads. They were then incubated at 100 mM Ca 2ϩ with a rat brain detergent extract in the presence (ϩ) or absence (Ϫ) of 100 mM IP6. The proteins bound to the beads were resolved by SDS-PAGE and analyzed by Western blotting using an antibody against Syt-I. and syntaxin-1 depends more on the C 2 A domain, whereas the C 2 B domain mediates synaptotagmin multimerization (42). C 2 B is also involved in several Ca 2ϩ -independent interactions including binding to Ca 2ϩ channels (14), inositol polyphosphates (48), and SNAP-25 (16). Genetic studies have demonstrated that synaptotagmin-C 2 B plays a critical role in excitation-secretion coupling and that C 2 B-mediated oligomerization of different synaptotagmin isoforms modulates the strength of synaptic transmission (43,49). The overall homology between the C 2 domains of RIM and the C 2 B domain of Syt-I is relatively low. Despite this, RIM-C 2 A and RIM-C 2 B possess properties that are remarkably similar to those of the C 2 B domain of Syt-I. Thus, the C 2 domains of RIM interact with the poreforming subunit of N-type Ca 2ϩ channels and with SNAP-25 and Syt-I with affinities almost identical to the one displayed by the C 2 B domain of Syt-I. The presence of two "C 2 B-like" domains within the same molecule could permit simultaneous interaction with multiple binding partners, placing the Rab3 effector in an ideal position to coordinate docking and fusion of secretory vesicles. In rats (30,32) and humans (the present study), the distance between the two C 2 domains varies considerably between RIM isoforms. This may provide each isoform with specific properties, permitting a selective tuning of the secretory process in different parts of the brain. The "C 2 Blike" properties of the C 2 domains of RIM are provided by the presence of the amino acid motif KK(R)K(R)T. In fact, alterations of the KK(R)K(R)T motif abolished the capacity of RIM C 2 domains to associate with its potential partners.
In neurons, positioning of synaptic vesicles in the immediate proximity of Ca 2ϩ channels is critical for rapid and efficient synaptic transmission. A tight association of RIM with the ␣1 subunits of the N-type Ca 2ϩ channel could contribute to guide the docking of secretory vesicles carrying Rab3 close to Ca 2ϩ entry sites. The C 2 domains of RIM are also able to interact weakly with the ␣1C (L C -subtype) but not the ␣1D (L D -type) subunit of Ca 2ϩ channels. This suggests that close apposition of RIM to the sites of Ca 2ϩ influx may be restricted to secretory systems specialized for fast release.
The binding of RIM to the components of the fusion machinery is differentially modulated by Ca 2ϩ . The association with SNAP-25 does not require the presence of Ca 2ϩ and could therefore pre-exist on the plasma membrane of resting cells. In contrast, the complex with Syt-I is formed at micromolar Ca 2ϩ concentrations and is likely to occur exclusively under stimulatory conditions. The hetero-oligomeric interaction is most probably driven by Ca 2ϩ -triggered changes in the electrostatic potential of the C 2 B domain of Syt-I. In fact, the C 2 domains of RIM are unlikely to bind Ca 2ϩ because they lack the aspartic residues required for the coordination of the cation (see Fig. 1). RIM is therefore probably not a Ca 2ϩ sensor but rather an acceptor for Ca 2ϩ -activated Syts. The C 2 domains of RIM (in particular, RIM-C 2 B) display a very high affinity for Syt-I, and RIM is quite abundant in the brain (we estimated that it represents about 1-2% of the proteins associated with synaptosomal membranes). For these reasons, the interaction of the C 2 domains of RIM with Syt-I is likely to play an important role in the modulation of the exocytotic process. Our findings would be consistent with a model in which the Ca 2ϩ -dependent hetero-oligomerization of the C 2 B domain of Syt-I with the C 2 domains of RIM liberates SNAP-25 and permits the assembly of the SNARE complex.
Inositol polyphosphates, such as IP6, reduce Ca 2ϩ -triggered vesicular fusion by a mechanism that is not yet fully elucidated (12,50). They have been proposed to act by preventing the oligomerization of synaptotagmins, by blocking the binding of Syt-I with phospholipids, or by interfering with the association of Syt-I with plasma membrane proteins (12). Here, we demonstrate that IP6 precludes the formation of the Ca 2ϩ -dependent complex between Syt-I and the C 2 domains of RIM. Thus, in view of these results, it is possible that at least part of the effects of inositol polyphosphates on exocytosis reported in the literature are due to the disruption of the interaction between Syt-I and RIM.
The properties of the C 2 domains of RIM described in this study are linked to the presence of two positively charged amino acids. The amino acid motif KK(R)K(R)T, which plays a critical role in the interaction of RIM with the components of the secretory machinery, is also found in the C 2 domains of other proteins concentrated in active zones such as Piccolo and Aczonin (51,52). Future studies will have to determine whether these proteins are also endowed with properties analogous to those of RIM. If this is the case, the array of proteinprotein interactions taking place in active zones could turn out to be much more complex than previously supposed.
In conclusion, we have found that the C 2 domains of RIM display previously unsuspected binding properties that could favor the targeting and docking of secretory vesicles near Ca 2ϩ channels. In addition, the C 2 domains of RIM are high-affinity binding sites for the Ca 2ϩ -bound form of Syt-I, suggesting that this Rab3-binding protein is also involved in the control of vesicle fusion. Thus, our data indicate that through its multiple binding partners, RIM is able to coordinate different stages of the secretory process.