Rab3-interacting Molecule γ Isoforms Lacking the Rab3-binding Domain Induce Long Lasting Currents but Block Neurotransmitter Vesicle Anchoring in Voltage-dependent P/Q-type Ca2+ Channels*

Assembly of voltage-dependent Ca2+ channels (VDCCs) with their associated proteins regulates the coupling of VDCCs with upstream and downstream cellular events. Among the four isoforms of the Rab3-interacting molecule (RIM1 to -4), we have previously reported that VDCC β-subunits physically interact with the long α isoform of the presynaptic active zone scaffolding protein RIM1 (RIM1α) via its C terminus containing the C2B domain. This interaction cooperates with RIM1α-Rab3 interaction to support neurotransmitter exocytosis by anchoring vesicles in the vicinity of VDCCs and by maintaining depolarization-triggered Ca2+ influx as a result of marked inhibition of voltage-dependent inactivation of VDCCs. However, physiological functions have not yet been elucidated for RIM3 and RIM4, which exist only as short γ isoforms (γ-RIMs), carrying the C-terminal C2B domain common to RIMs but not the Rab3-binding region and other structural motifs present in the α-RIMs, including RIM1α. Here, we demonstrate that γ-RIMs also exert prominent suppression of VDCC inactivation via direct binding to β-subunits. In the pheochromocytoma PC12 cells, this common functional feature allows native RIMs to enhance acetylcholine secretion, whereas γ-RIMs are uniquely different from α-RIMs in blocking localization of neurotransmitter-containing vesicles near the plasma membrane. γ-RIMs as well as α-RIMs show wide distribution in central neurons, but knockdown of γ-RIMs attenuated glutamate release to a lesser extent than that of α-RIMs in cultured cerebellar neurons. The results suggest that sustained Ca2+ influx through suppression of VDCC inactivation by RIMs is a ubiquitous property of neurons, whereas the extent of vesicle anchoring to VDCCs at the plasma membrane may depend on the competition of α-RIMs with γ-RIMs for VDCC β-subunits.

Assembly of voltage-dependent Ca 2؉ channels (VDCCs) with their associated proteins regulates the coupling of VDCCs with upstream and downstream cellular events. Among the four isoforms of the Rab3-interacting molecule (RIM1 to -4), we have previously reported that VDCC ␤-subunits physically interact with the long ␣ isoform of the presynaptic active zone scaffolding protein RIM1 (RIM1␣) via its C terminus containing the C 2 B domain. This interaction cooperates with RIM1␣-Rab3 interaction to support neurotransmitter exocytosis by anchoring vesicles in the vicinity of VDCCs and by maintaining depolarization-triggered Ca 2؉ influx as a result of marked inhibition of voltage-dependent inactivation of VDCCs. However, physiological functions have not yet been elucidated for RIM3 and RIM4, which exist only as short ␥ isoforms (␥-RIMs), carrying the C-terminal C 2 B domain common to RIMs but not the Rab3binding region and other structural motifs present in the ␣-RIMs, including RIM1␣. Here, we demonstrate that ␥-RIMs also exert prominent suppression of VDCC inactivation via direct binding to ␤-subunits. In the pheochromocytoma PC12 cells, this common functional feature allows native RIMs to enhance acetylcholine secretion, whereas ␥-RIMs are uniquely different from ␣-RIMs in blocking localization of neurotransmitter-containing vesicles near the plasma membrane. ␥-RIMs as well as ␣-RIMs show wide distribution in central neurons, but knockdown of ␥-RIMs attenuated glutamate release to a lesser extent than that of ␣-RIMs in cultured cerebellar neurons. The results suggest that sustained Ca 2؉ influx through suppression of VDCC inactivation by RIMs is a ubiquitous property of neurons, whereas the extent of vesicle anchoring to VDCCs at the plasma membrane may depend on the competition of ␣-RIMs with ␥-RIMs for VDCC ␤-subunits.
Protein complexes play essential roles in various cellular responses including neurotransmission via synapses in the nervous system. In central synapses, the postsynaptic density is formed by protein complexes, containing neurotransmitter receptors, signaling, and cytoskeletal proteins, and scaffolding proteins, such as PSD-95 (1), whereas active zones are formed by the cytomatrix and other proteins responsible for neurotransmitter release from presynaptic nerve terminals (2)(3)(4)(5)(6). For Ca 2ϩ influx upon membrane potential depolarization to evoke neurotransmitter release, efficient coupling of VDCCs 2 to protein machineries, such as soluble N-ethylmaleimide-sensitive factor attachment protein receptors (SNAREs), mediating fusion of neurotransmitter-containing vesicles with presynaptic membranes, is critical (7)(8)(9)(10)(11)(12)(13). Thus, it is extremely important to identify protein associations and their functional significance in understanding neurotransmission.
Multiple types of VDCCs are distinguished on the basis of biophysical and pharmacological properties (14). In neurons, high voltage-activated (HVA) VDCC types, such as the N-, P/Q-, R-, and L-types are essential for neurotransmitter release from presynaptic terminals (15)(16)(17). Furthermore, presynaptic Ca 2ϩ channels were considered to serve as the regulatory node in a dynamic, multilayered signaling network that exerts short term control of neurotransmission in response to synaptic activity (13). Biochemically, VDCCs are known as heteromultimeric protein complexes composed of the pore-forming ␣ 1 and auxiliary subunits ␣ 2 /␦, ␤, and ␥ (18). The ␣ 1 -subunit, designated as Ca V , is encoded by 10 distinct genes, whose correspondence with functional subtypes has been largely elucidated (14,18). VDCC complexes are primarily known for association with presynaptic and postsynaptic proteins, including syntaxin, SNAP-25, synaptotagmin, CASK, and Mint via interactions with the ␣ 1 -subunit (8, 9, 19 -24). It was traditionally believed that the presence of anchoring Ca 2ϩ channels close to the Ca 2ϩ microdomain-dependent release machinery was the main reason for the physical interactions between channels and synaptic proteins. However, it is now becoming clear that these proteins additionally regulate channel activity. The ␤-subunit interacts with ␣ 1 from the cytoplasmic side to enhance functional channel trafficking to the plasma membrane (25,26) and to modify multiple kinetic properties (27,28). Association with other proteins has also been revealed for ␤-subunits (12, 29 -31). Considering the cytoplasmic disposition of ␤-subunits, it is intriguing to investigate whether ␤-subunits are involved in targeting specific subcellular machinery to VDCC complexes at presynaptic active zones for neurotransmitter release through as yet unidentified protein interactions.
Originally identified as a putative effector of the synaptic vesicle protein Rab3 (32), RIM1 is part of the RIM superfamily, whose members share a common C 2 B domain at their C termini (33). With regard to RIM1 and RIM2, a long isoform ␣ and short isoforms ␤ and ␥ that lack the Rab3-interacting Zn 2ϩ finger domain are known, whereas only short ␥ forms are known for RIM3 and RIM4 (33,34). A recent paper (35) identified RIM3␥ as a postsynaptic protein. RIM1 has been shown to interact with other presynaptic active zone protein components, including Munc13, ELKS (also known as CAST), RIMbinding protein (RIM-BP), and liprins, to form a protein scaffold in the presynaptic nerve terminal (2, 36 -39). Recently, we reported that the association of RIM1 with ␤-subunits supports neurotransmitter release via two distinct mechanisms: sustaining Ca 2ϩ influx through inhibition of channel inactivation and anchoring neurotransmitter-containing vesicles in the vicinity of VDCCs (12). Importantly, presynaptic VDCC currents manifest resistance to voltage-dependent inactivation and exhibit inactivation largely dependent upon the magnitude of the inward Ca 2ϩ current (40,41), as observed for VDCCs associated with RIM1 (12). Furthermore, we demonstrated that the mouse RIM1 arginine-to-histidine substitution (R655H), which corresponds to the human autosomal dominant conerod dystrophy mutation (42), modifies RIM1 function in regulating VDCC currents elicited by the P/Q-type Ca V 2.1 and L-type Ca V 1.4 channels (43).
Genetic studies using mouse knockouts have shown that RIM1 is essential for different forms of synaptic plasticity in different types of synapses (39,44,45). In the CA1 region Schaffer collateral excitatory synapses and GABAergic synapses, RIM1 maintains normal neurotransmitter release and short term synaptic plasticity. In excitatory CA3-region mossy fiber synapses, cerebellar parallel fiber synapses, and cortico-lateral amygdala synapses, RIM1 is necessary for presynaptic long term plasticity. Using pharmacological and genetic approaches, a presynaptic signaling pathway via cAMP, protein kinase A, and RIM1␣ was elucidated as a general mechanism that underlies the long term modulation of transmitter release at both excitatory and inhibitory synapses (46). In autapses, the RIM1␣ deletion significantly reduces the readily releasable pool of ves-icles, and it alters short term plasticity and the properties of evoked asynchronous release (47). More recently, severe impairment of mouse survival by deletion of both RIM1␣ and RIM1␤ was reported (34). Electrophysiological analyses showed that the deletion of both RIM1␣ and RIM1␤ abolished long term presynaptic plasticity, as does RIM1␣ deletion alone, but aggravated the impairment in synaptic strength and short term synaptic plasticity that is caused by the RIM1␣ deletion. Mice deficient in both RIM1␣ and RIM2␣ showed lethality due to defects in Ca 2ϩ -triggered release, despite normal presynaptic active zone length and normal spontaneous neurotransmitter release (48). In Caenorhabditis elegans, the loss of the single RIM homolog, UNC10, caused a reduction in membrane-contacting synaptic vesicles within 30 nm of the dense projection at neuromuscular junctions (49). In support of our hypothesis regarding VDCC-RIM1␣ association (12), a recent report states that RIM colocalizes with the Ca 2ϩ channels and provides a mechanism to target vesicles to the presynaptic density through direct interaction with Rab3 at C. elegans presynaptic densities (50). However, despite this progress in genetic studies of RIMs, RIM2, RIM3, and RIM4 have yet to be characterized functionally (33,35,36,51), and physiological roles played by RIM1 remain controversial (52).
Here, we analyze a physical and functional interaction between RIM family members and VDCC ␤-subunits. ␥-RIMs as well as ␣-RIMs interact with four VDCC ␤-subunits and markedly suppress the voltage-dependent inactivation of P/Qtype Ca V 2.1 VDCC expressed in BHK cells, and are essential for regulation of Ca 2ϩ -triggered exocytosis in PC12 cells and in cultured cerebellar neurons. Unlike ␣-RIMs, ␥-RIMs suppress anchoring of neurotransmitter-containing vesicles to VDCCs.
Production of Glutathione S-Transferase (GST) Fusion Proteins and Recombinant ␤ 4 -Subunit Proteins-For production of GST fusion proteins for RIMs, each cDNA for RIM constructs and cDNA for GST were subcloned together into the pET23 vector (Novagen). The Rosetta strain (Novagen) of Escherichia coli was transformed by the expression vectors, and protein expression/purification was performed according to the manufacturer's instructions (Novagen). For production of recombinant ␤ 4 -subunits, the gene encoding amino acid residues 47-475 of rat ␤ 4b -subunit and cDNA for GST were subcloned into the pET23 vector (Novagen). The GST-␤ 4 fusion proteins were purified by glutathione-Sepharose beads (GE Healthcare), and the GST tag was cleaved by incubation with thrombin (4 units/ml; Sigma) for 6 h at 4°C. Resultant GST and thrombin were removed by glutathione-Sepharose beads and by benzamidine beads (Sigma) to obtain purified recombinant ␤ 4 -subunits. These proteins were stored at Ϫ80°C.
GST Pull-down Assay and Coimmunoprecipitation in HEK293T Cells-HEK293T cells were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum, 30 units/ml penicillin, and 30 g/ml streptomycin. Forty-eight h after transfection, HEK293T cells were solubilized in Nonidet P-40 buffer (150 mM NaCl, 50 mM Tris, 1% Nonidet P-40, and protease inhibitors) and then centrifuged at 17,400 ϫ g for 20 min. For the pull-down assay, the cell lysate was incubated with glutathione-Sepharose beads bound with purified fusion proteins at 4°C, and then the beads were washed with Nonidet P-40 buffer. The proteins retained on the beads were characterized by Western blot (WB) with anti-GFP antibody (Clontech catalog no. 632460). For coimmunoprecipitation, the cell lysate was incubated with anti-FLAG M2 monoclonal antibody (Sigma catalog no. F3165), and then the immunocomplexes were incubated with protein A-agarose beads (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), and the beads were washed with Nonidet P-40 buffer. Immunoprecipitated proteins were characterized by WB with anti-␤ 4 antibody.
In Vitro Binding of the Purified RIM-GST Fusion Proteins and Recombinant ␤ 4 -Protein-RIM-GST fusion proteins at various concentrations were incubated with 50 pM purified recombinant ␤ 4 -subunits for 3 h at 4°C in phosphate-buffered saline buffer containing 0.1% Nonidet P-40 and 50 g/ml bovine serum albumin and then incubated with glutathione-Sepharose beads for 1 h. The beads were washed twice with the phosphatebuffered saline buffer. The proteins retained on the beads were characterized by WB with the anti-␤ 4 antibody (12) and detected by enhanced chemiluminescence (Thermo Scientific). The densities of protein signals, obtained using NIH Image under a linear relationship with the applied amount of proteins, were normalized to the densities from the maximal binding.
Cell Culture and cDNA Expression in BHK Cells-The baby hamster kidney BHK6-2 cell line stably expressing Ca V 2.1, ␣ 2 /␦, and ␤ 1a was described previously (53). BHK6-2 cells were cultured in DMEM containing 10% fetal bovine serum, 30 units/ml penicillin, and 30 g/ml streptomycin. Transfection of cDNA plasmids was carried out using Effecten transfection reagent (Qiagen). The cells were subjected to electrophysiological measurements 48 h after transfection.
Current Recordings-Whole-cell mode of the patch clamp technique was carried out at 22-25°C with an EPC-9 (HEKA Elektronik) patch clamp amplifier as described previously (54). Patch pipettes were made from borosilicate glass capillaries (1.5-mm outer diameter, 0.87-mm inner diameter; Hilgenberg) using a model P-87 Flaming-Brown micropipette puller (Sutter Instrument Co.). The patch electrodes were fire-polished.
Pipette resistance ranged from 2 to 3.5 megaohms when filled with the pipette solutions described below. The series resistance was electronically compensated to Ͼ60%, and both the leakage and the remaining capacitance were subtracted by the ϪP/4 method. Currents were sampled at 100 kHz after low pass filtering at 8.4 kHz (3 db) in the experiments of activation kinetics, otherwise sampled at 20 kHz after low pass filtering at 3.0 kHz (3 db). Data were collected and analyzed using Pulse version 8.77 (HEKA Elektronik). An external solution contained 3 mM BaCl 2 , 155 mM tetraethylammonium chloride, 10 mM HEPES, and 10 mM glucose (pH 7.4-adjusted with TEA-OH) for BHK cells and 10 mM BaCl 2 , 153 mM tetraethylammonium chloride, 10 mM HEPES, and 10 mM glucose (pH 7.4-adjusted with tetraethylammonium hydroxide) for PC12 cells. The pipette solution contained 95 mM CsOH, 95 mM aspartate, 40 mM CsCl, 4 mM MgCl 2 , 5 mM EGTA, 2 mM disodium ATP, 5 mM HEPES, and 8 mM creatine phosphate (pH 7.2-adjusted with CsOH).
Voltage Dependence of Inactivation-To determine the voltage dependence of inactivation (inactivation curve) of VDCCs, Ba 2ϩ currents were evoked by 20-ms test pulse to 5 mV after the 10-ms repolarization to Ϫ100 mV (Ϫ80 mV for PC12) following 2-s holding potential (V h ) displacement (conditioning pulse) from Ϫ100 to 20 mV (from Ϫ80 to 20 mV for PC12) with 10-mV increments. Amplitudes of currents elicited by the test pulses were normalized to those elicited by the test pulse after a 2-s V h displacement to Ϫ100 mV (Ϫ80 mV for PC12). The mean values were plotted against potentials of the 2-s V h displacement. When the inactivation curve was monophasic, the mean values were fitted to the single Boltzmann equation: where a is the rate of inactivating component, V 0.5 is the potential to give a halfvalue of inactivation, and k is the slope factor. Otherwise, the mean values were fitted to the sum of two Boltzmann equations: High Ϫ V h )/k High )), where a, b, and (1 Ϫ a Ϫ b) are the ratios of a low voltage-induced phase, a high voltage-induced phase, and a non-inactivating phase; V 0.5 Low and V 0.5 High are the potentials that give a half-value of components susceptible to inactivation at low voltages in inactivation curves and at high voltages; and k Low and k High are the slope factors.
Voltage Dependence of Activation-Tail currents were elicited by repolarization to Ϫ60 mV after 5-ms test pulse from Ϫ40 to 30 mV with 5-mV increments. Currents were sampled at 100 kHz after low pass filtering at 8.4 kHz. Amplitudes of tail currents were normalized to the tail current amplitude obtained with a test pulse to 30 mV. The mean values were plotted against test pulse potentials and fitted to the Boltzmann equation: n(V m ) ϭ 1/(1 ϩ exp((V 0.5 Ϫ V m )/k)), where V m is membrane potential, V 0.5 is the potential to give a half-value of conductance, and k is the slope factor.
RNA Preparation, Northern Blot Analysis, and Real-time PCR-Total RNA was prepared from various tissues of 2-month-old C57BL/6 mice with ISOGEN total RNA isolation reagent (Nippon Gene) according to the manufacturer's instructions. For Northern blot analysis, 30 g of total RNA was separated by electrophoresis in a denaturating gel and blotted onto nylon membrane (Roche Applied Science). The probes used to detect RIM1 and RIM2 RNA were cDNA fragments corresponding to nucleotides 1035-1491 of mouse RIM1␣ and nucleotides 471-1372 of mouse RIM2␣, respectively. The probes used to detect RIM3 and RIM4 were cDNA fragments corresponding to coding sequence of mouse RIM3␥ and RIM4␥, respectively. The cDNA probes were labeled with digoxigenin (DIG) using the PCR DIG labeling kit (Roche Applied Science). The hybridization, washing, and detection were performed according to the manufacturer's instructions. Equal loading of total RNA was estimated by ethidium bromide staining of ribosomal RNAs.
For real-time PCR, reverse transcription of RNA to cDNA was performed using an RNA LA-PCR kit (TaKaRa). Quantification was performed by real-time PCR (LightCycler Instrument, Roche Applied Science) using the LightCycler FastStart DNA Master HybProbe Kit (Roche Applied Science). Primer sequences are indicated in supplemental Table S1. Temperature cycles were as follows. Initial 95°C for 10 min was followed by 40 cycles at 95°C for 10 s, 60°C for 10 s, 72°C for 30 s for RIM1; at 95°C for 10 s, 60°C for 10 s, 72°C for 30 s for RIM2; at 95°C for 10 s, 60°C for 10 s, 72°C for 30 s for RIM3; at 95°C for 10 s, 60°C for 10 s, 72°C for 30 s for RIM4; and at 95°C for 10 s, 57°C for 5 s, 72°C for 5 s for 18 S ribosomal RNA (18 S). The results were analyzed with LightCycler software. The identity of the PCR product was confirmed by automated determination of the melting temperature of the PCR products. The results for each gene were normalized relative to 18 S, as described previously (55), and were expressed relative to the brain given the arbitrary value of 1.
In Situ Hybridization Histochemistry-For histological staining of the central nervous system (CNS), adult mice (C57BL/6, body weight 20 -25 g) were deeply anesthetized with an overdose of Nembutal and then transcardially perfused by 0.9% NaCl, followed by 3% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). The brains of the animals were dissected. Cryoprotection of the tissue blocks in 30% sucrose for 24 h at 4°C was followed by histological sectioning on a cryostat (Leica). For details about in situ hybridization histochemistry, see Kagawa et al. (56). Briefly, in vitro transcribed DIG-labeled cRNA probe was generated against template RIM3␥ or RIM4␥ cDNA fragment corresponding to cording sequence of mouse RIM3␥ or RIM4␥ using DIG High Prime kit (Roche Applied Science). The probe (0.3 g/ml) was hybridized overnight to mouse CNS histological 35-m-thick sections at 50°C. Positive signals were detected by alkaline phosphatase-conjugated antidigoxigenin antibody and the nitro blue tetrazolium/5-bromo-4-chloro-3Ј-indolyl phosphate reaction.
Biochemistry of Native Neuronal VDCC Complexes-To obtain the crude synaptic membrane (CSM) fraction, subcellular fractionation was performed based on a previously described method (24), with slight modification. Whole mouse brains (8 g) were homogenized in a homogenization buffer containing 4 mM HEPES, 0.32 M sucrose, 5 mM EDTA, 5 mM EGTA, and protease inhibitors (pH 7.4). Cell debris and nuclei were removed by centrifugation at 800 ϫ g for 10 min. The supernatant (S1) was centrifuged at 9,000 ϫ g for 15 min to obtain crude synaptosomal fraction as pellet (P2). The supernatant was stored as S2. The crude synaptosomes were resuspended in the homogenization buffer and centrifuged at 10,000 ϫ g for 15 min. The washed crude synaptosomes were lysed by hypoosmotic shock in water, rapidly adjusted to 1 mM HEPES/NaOH (pH 7.4), and stirred on ice for 30 min. After centrifugation of the lysate at 25,000 ϫ g for 20 min, the pellet was resuspended in 0.25 M buffered sucrose. The synaptic membranes were then further enriched through a discontinuous sucrose gradient containing 0.8/1.0/1.2 M sucrose. After centrifugation at 65,000 ϫ g for 2 h, the CSM fraction was collected from the 1.0/1.2 M sucrose interface. S1, P2, S2, and CSM fractions were analyzed by WB using rabbit anti-RIM1 polyclonal antibody raised against amino acid residues 11-399 of mouse RIM1, anti-RIM3 polyclonal antibody raised against the peptide containing GSQQAGGGAGTTTAKK, anti-PSD-95 monoclonal antibody (Thermo scientific catalog no. MA1-045), and antisynaptophysin monoclonal antibody (Sigma catalog no. S5768).
Synaptic membrane proteins were extracted from the CSM with solubilization buffer containing 50 mM Tris, 500 mM NaCl, a mixture of protease inhibitors, and 1% digitonin (Biosynth) (pH 7.4). After centrifugation at 147,600 ϫ g for 37 min, supernatant was diluted with a buffer containing 50 mM Tris and 1% digitonin to adjust NaCl concentration to 150 mM and incubated overnight at 4°C. After centrifugation at 6,654 ϫ g for 15 min, the supernatant was incubated with protein A-agarose coupled to anti-RIM1/2 antibody (Synaptic Systems catalog no. 140-203) or anti-RIM3 polyclonal antibody for 6 h at 4°C. Immunoprecipitated proteins were subjected to WB with anti-Ca V 2.1 antibody (Alomone catalog no. ACC-001), anti-Ca V 2.2 antibody (Alomone catalog no. ACC-002), or anti-␤ 4 antibody.
PC12 Cell Culture and siRNA Suppression of Endogenous RIMs in PC12 Cells-PC12 cells were cultured as described previously (57). The siRNA sequences for rat RIM1 and RIM2 were described previously (12). The sense siRNA sequences 5Ј-AAGCTCCGAGGGCACGTTTAT-3Ј and 5Ј-AACTG-GCTGGTACAAACTCTT-3Ј for RIM3 and 5Ј-AACTATG-GAGGAGTTTGTCTA-3Ј and 5Ј-AACTGCCAGCTGC-CTATATCA-3Ј for RIM4 were used. To construct siRNA oligomers, the Silencer siRNA construction kit (Ambion) was used. The GAPDH siRNA (siControl) used was the control provided with the kit. Transfection of siRNAs to PC12 cells was carried out using Lipofectamine TM 2000 (Invitrogen). Cells were transfected with combinations of 0.5 g each of siRNAs in a 35-mm culture dish. Suppression of RNA expression was confirmed by reverse transcription (RT)-PCR analyses (29 cycles) using specific primers listed in supplemental Table S1 and WB analyses. RT-PCR was performed using the LA-PCR kit (TaKaRa), according to the manufacturer's instructions. The cells treated with siRNAs were subjected to WB analyses, patch-clamp measurements, imaging of secretory vesicles, or release assay 48 h after transfection. To generate expression vectors (pCI-neo) for siRNA-resistant RIMs (denoted by an asterisk), silent mutations were introduced in the siRNA-binding regions of RIM. Silent mutations of RIM1 were introduced by substituting the nucleotide sequence 4198 aga atg gac cac aaa tgc 4217 for 4198 agg atg gat cat aag tgt 4217 and 4267 gtg att gga tgg tat aaa 4284 for 4267 gta atc ggc tgg tac aag 4284 . Silent mutations of RIM2 were introduced by substituting the nucleotide sequence 4429 gcc cag ata ctc tta gat 4446 for 4429 gcg caa atc Functional and Structural Impacts of ␥-RIMs on VDCC JULY 9, 2010 • VOLUME 285 • NUMBER 28 ctt ttg gac 4446 and 4453 gaa cta tcc aac 4464 for 4453 gag ctt tcg aat 4464 . Silent mutations of RIM3 were introduced by substituting the nucleotide sequence 346 agc tcc gag ggc acg ttt 363 for 346 agt tcc gaa ggt aca ttc 363 and 811 ggc tgg tac aaa 822 for 811 ggt tgg tat aag 822 . Silent mutations of RIM4 were introduced by substituting the nucleotide sequence 244 tat gga gga gtt tgt 258 for 244 tac ggc ggc gta tgc 258 and 445 ctg cca gct gcc tat atc 462 for 445 cta cct gca gct tac ata 462 . Letters in boldface type indicate mutated nucleotides. In rescue experiments, rescue plasmids were cotransfected with siRNAs using Lipofectamine TM 2000. 1.0 g each of siRNA-resistant RIM1* and RIM2* and/or 0.4 g each of siRNA-resistant RIM3* and RIM4* were cotransfected with 0.5 g each of siRNA in a 35-mm culture dish.
Total Internal Reflection Fluorescence (TIRF) Imaging of Secretory Vesicles-PC12 cells cotransfected with pVenus-N1-NPY carrying a DNA construct for the fusion protein of neuropeptide Y (NPY) and the fluorescent protein Venus (NPY-Venus) and a mixture of RIM siRNAs using OptiFect (Invitrogen) were plated onto poly-L-lysine-coated coverslips. The imaging was performed in modified Ringer's buffer that contained 130 mM NaCl, 3 mM KCl, 5 mM CaCl 2 , 1.5 mM MgCl 2 , 10 mM glucose, and 10 mM HEPES (pH 7.4). Fluorescence images of NPY-Venus in large dense core vesicles were observed as previously reported (12,58). In brief, a high numerical aperture objective lens (Plan Apochromatic, ϫ100, numerical aperture ϭ 1.45, infinity-corrected, Olympus) was mounted on an inverted microscope (IX71, Olympus), and incident light for total internal reflection illumination was introduced from the high numerical objective lens through a single mode optical fiber. A diode-pumped solid state 488-nm laser (kyma488, 20 milliwatts, MELLES GRIOT) was used for total internal fluorescence illumination, and a 510-nm long pass filter was used as an emission filter. Images were captured by a cooled CCD camera (EM-CCD, Hamamatsu Photonics) operated with Metamorph (Molecular Devices). Densities of vesicles were assessed by counting the number of individual fluorescent spots in the area where vesicles show uniform distribution in TIRF images, and obtained numbers were divided by areas. Area calculations were performed using Metamorph software. The cells with distribution of vesicles by a dark spot with area of Ͼ10 m 2 were omitted, to select the cells in which vesicles were uniformly distributed for the analyses. Ten m 2 was adopted because 10 m 2 was the maximal dark circle area that can be located in between vesicles in the images from siRIM1&2-transfected cells with uniform vesicle distribution.
Release Assay in PC12 Cells-Acetylcholine (ACh) secretion experiments were performed as previously reported with slight modifications (57). PC12 cells were plated in poly-D-lysinecoated 35-mm dishes (BD Biosciences) with 5 ϫ 10 5 cells/dish. Cells were cotransfected with 5 g of each RIM plasmid and 1 g of pEFmChAT encoding mouse ChAT cDNA using Lipofectamine TM 2000. Three days after transfection, PC12 cells were washed with a 5.9 mM K ϩ solution that contained 0.01 mM eserine, 140 mM NaCl, 4.7 mM KCl, 1.2 mM KH 2 PO 4 , 2.5 mM CaCl 2 , 1.2 mM MgSO 4 , 11 mM glucose, and 15 mM HEPES (pH 7.4-adjusted with NaOH) and incubated for 30 s with the 5.9 mM K ϩ solution at 37°C. The release of ACh during this period was considered as basal release. To measure K ϩ -stimulated release of ACh, the cells were then incubated for 30 s with a 51. . Supernatant from cells solubilized in Nonidet P-40 buffer and centrifuged at 17,400 ϫ g for 20 min at 4°C was taken as the cellular ACh that was not secreted. ACh was measured using HPLC with electrochemical detection (HTEC-500, EiCOM).
Fluorescent [Ca 2ϩ ] i Measurements-PC12 cells cotransfected with pEGFP-N1 and mixture of RIM siRNAs using Opti-Fect (Invitrogen). Thirty-six h after transfection, PC12 cells were plated onto poly-L-lysine-coated glass coverslips. Fortyeight h after transfection, cells on coverslips were loaded with fura-2 by incubation in DMEM containing 10 M fura-2/AM (Dojindo Laboratories), 0.04% Pluronic F-127 (Biotium), and 10% fetal bovine serum at 37°C for 40 min and washed with the 5.9 mM K ϩ solution. The coverslips were then placed in a perfusion chamber mounted on the stage of the microscope. Fluorescence images of the cells were recorded and analyzed with a video image analysis system (AQUACOSMOS, Hamamatsu Photonics). The fura-2 fluorescence at an emission wavelength of 510 nm was observed at 37 Ϯ 1°C by exciting fura-2 alternately at 340 and 380 nm. Measurement was carried out in 5.9 mM K ϩ solution, 51.1 mM K ϩ solution, and 28.4 mM K ϩ solution. The 340:380 nm ratio images were obtained on a pixel by pixel basis and were converted to Ca 2ϩ concentrations by in vivo calibration using 40 M ionomycin (59).
Construction of Short Hairpin RNA (shRNA) Vectors-shRNA vectors were constructed based upon a pSUPER.neoϩ GFP vector (OligoEngine). To create shRIM2, shRIM3, and shRIM4, two complementary 60-bp oligonucleotides carrying antisense and sense sequence for GGCCCAGATACTCTTA-GAT (19 bp corresponding to nucleotides 4554 -4572 of rat RIM2), CTGGCTGGTACAAATTCTT (nucleotides 1252-1270 of rat RIM3), and CTGCCAGCTGCCTATATCA (nucleotides 445-463 of rat RIM4) were annealed and ligated to pSUPER.neoϩGFP vectors in accordance with the Oligo-Engine instructions. To create shRIM1, 60-bp oligonucleotides carrying GTAATAAATGAAGAGGTTTAAACTTCTTCG-TTTGTTGC and its complementary sequence were annealed and ligated to pSUPER.neoϩGFP vectors. In construction of shRIM1, we inserted a mutation in the stem sequence of shRNA according to the previous report (60) for better silencing efficiency. Control shRNA (shControl) was generated similarly except that an artificial 19-mer sequence (ATCCGCGCGAT-AGTACGTA) was used as a target. This sequence was based upon a commercially available negative control siRNA sequence (B-Bridge International), and we confirmed that it had no significant identity to any known mammalian gene based on a BLAST search.
Rat Cerebellar Neuron Primary Cultures and Transfection-Primary cerebellar neuron cultures were prepared as previously described (61) with minor modifications. Briefly, cerebella from 7-9-day-old Wister rat pups were digested with 1% trypsin (Difco). The solution was removed after a brief centrifugation, and the tissue was mechanically dissociated by repeated pipetting in DMEM containing 10% fetal calf serum (Invitrogen). The cell suspension was centrifuged at 200 ϫ g for 3 min. The pellet was resuspended in DMEM containing 10% fetal calf serum and filtered through a 70-m cell strainer. After centrifugation at 100 ϫ g for 5 min, 5 ϫ 10 6 cells were resuspended in 100 l of rat neuron nucleofector solution (Amaxa) with 3 g of shRNA vector (total amount of DNA was 4 g in all cases). The cell suspension was electroporated using the G-13 program (Amaxa). The cells were diluted with DMEM containing 10% fetal calf serum and seeded onto a polyethyleneimine-coated 15-mm diameter 4-well multidish (Nunc) at a density of 1 ϫ 10 6 cells/well. Two h after initial plating, the medium was removed and gently replaced with Neurobasal medium containing 0.5 mM glutamine, 26 mM KCl, 60 units/ml penicillin, 60 g/ml streptomycin, and 2% B-27 (Invitrogen). The cultures were maintained at 37°C with 5% CO 2 . The medium was very gently replaced with fresh medium every 3 days.
Statistical Analysis-All data accumulated under each condition from at least three independent experiments are expressed as means Ϯ S.E. Student's t test or analysis of variance followed by Fisher's test was employed.
Functional Effects of RIMs on P/Q-type VDCC Currents-To elucidate the functional significance of the direct interaction between RIMs and VDCC ␤-subunits, we examined whole-cell Ba 2ϩ currents through recombinant P/Q-type VDCCs expressed as ␣ 1 ⅐␣ 2 /␦⅐␤ complexes containing the BI-2 variant of Ca V 2.1 (25) and ␤ 1a in BHK cells. Prominent effects of RIMs on P/Q-type VDCC currents were observed for the parameters of voltage-dependent inactivation. Inactivation was markedly decelerated ( Fig. 2A), whereas the voltage dependence of the inactivation was significantly shifted toward depolarizing potentials by RIMs: the inactivation curve showed a component susceptible to inactivation at high voltages (the half-inactivation potentials (V 0.5 ) ranged from Ϫ28 to Ϫ23 mV) and a noninactivating component (Fig. 2B and supplemental Table S2). Furthermore, we coexpressed BADN, a dominant negative suppressor ␤ 4b fusion construct designed to disrupt the association of ␤-binding proteins, such as RIMs, with the functional VDCC complex without interfering with the ␣ 1 -␤ interaction (12). BADN significantly diminished the effect of RIMs on P/Q channel inactivation. The inactivation kinetics were markedly accelerated, and biphasic inactivation curves were elicited (Fig. 2). In the presence of BADN, we observed components susceptible to inactivation at low voltages in inactivation curves (V 0.5 values Functional and Structural Impacts of ␥-RIMs on VDCC ranged from Ϫ72 to Ϫ63 mV) and at high voltages (V 0.5 values ranged from Ϫ27 to Ϫ19 mV) ( Fig. 2B and supplemental Table S2). Thus, the strong suppressive effect on voltagedependent inactivation of P/Q-type VDCCs via interaction with ␤-subunits is a functional feature common to RIMs.
RIM proteins differently modulated the activation kinetics of P/Q-type VDCCs (Fig. 3A). The time constant ( activation ) obtained by fitting the activation time course of inward currents with a single exponential was "bell-shaped" when plotted against different voltages. Activation showed the slowest speed at Ϫ15 mV in vector-, RIM1␣-, and RIM2␣-expressing cells and at Ϫ10 mV in RIM3␥-and RIM4␥-expressing cells. ␥-RIMs significantly decelerated the activation kinetics at membrane potentials over Ϫ10 mV, whereas RIM1␣ only slightly decelerated the activation at membrane potentials over 10 mV, and RIM2␣ had no significant effect ( Fig. 3A and supplemental Table S3). RIMs failed to exert effects on other functional parameters, such as the voltage dependence of activation and the current densities at different voltages (Fig. 3, B and C, and supplemental Table S3). Thus, ␥-RIMs more significantly affect P/Q-type VDCC currents in terms of activation kinetics compared with ␣-RIMs.
Tissue Distribution of RIM3 and RIM4 Expression-RNA preparations from different mouse tissues and brain regions were subjected to Northern blot analyses using cDNA probes specific for RIM1, RIM2, RIM3, or RIM4 (Fig. 4A). RIM1 RNA was detected abundantly in the brain and at low levels in the eye. In the brain, RIM1 RNA was most abundant in the cerebrum and cerebellum. The principal transcripts of RIM2 (ϳ7.0 and ϳ5.5 kb) were detected at a high level in the brain and eye and at lower levels in the kidney and testis. In the brain, RIM2 RNA transcripts were detected at the highest level in the cerebellum. RIM3 RNA was also detected at the highest level in the brain. In the brain, RIM3 RNA was detected at similar levels in the cerebrum, cerebellum, brainstem, and olfactory bulb. RIM4 RNA was also detected at the highest level in the brain. In the brain, RIM4 RNA was detected ubiquitously but at relatively low levels in the brainstem and olfactory bulb. Thus, compared with the relatively selective expression of RIM1 and RIM2, RIM3 and RIM4 are more widely distributed in the brain.
To quantify levels of expression, RNA preparations from different mouse tissues were subjected to real-time PCR analyses using RIM1-, RIM2-, RIM3-, and RIM4-specific primers ( Fig.  4B and supplemental Table S1). The brain was a tissue with high levels of expression common to all RIMs. This is consistent with Northern blot analysis (Fig. 4A). Other tissues of high expression were the testis for RIM2, RIM3, and RIM4 and the eye for RIM2. Moderate levels of expression (Ͼ10% of the level in the brain) were observed in the dorsal root ganglion, eye, stomach, large intestine, and testis for RIM1; in the dorsal root ganglion, stomach, small intestine, large intestine, kidney, and ovary for RIM2; in almost all tested tissues except for the stomach and small intestine for RIM3; and in the eye, heart, thymus, and ovary for RIM4.
To determine exactly which cell types express RIM3 and RIM4 mRNA in the CNS, sections of 8-week-old C57BL/6 mouse brains were subjected to in situ hybridization histochemistry using cRNA probes specific for RIM3 or RIM4. RIM3 and RIM4 RNAs were expressed in neurons throughout the CNS, supporting the widespread distribution of RIM3 and RIM4 (Fig. 5A). For both RIM3 and RIM4, regions rich in glial cells (e.g. the white matter of the cerebral cortex and cerebellum) were not significantly labeled (Fig. 5, A (a) and B (a)). In the hippocampus, pyramidal neurons of the CA1-CA3 region, granule cells of the dentate gyrus, and interneurons exhibited positive labeling for RIM3 and RIM4 (Fig. 5, A (b) and B (b)). In the neocortex, stainings for RIM3 and RIM4 were more intense in layers III and IV of the neocortex in comparison with that in layers II, V, and VI (Fig. 5, A (c) and B (c)). In the thalamus, staining for RIM3 was relatively strong (Fig. 5A (d)), whereas RIM4 hybridization was relatively weak (Fig. 5B (d)). These results support our Northern blotting data indicating a relatively lower expression of RIM4 in the brainstem (Fig. 4A). In the cerebellar cortex, prominent RIM3 and RIM4 hybridization signals were observed in the granule and Purkinje cell layers (Fig. 5, A (e) and B (e)). The RIM3 distribution, which was consistent with a recent report (35), is similar to the expression pattern of RIM4 in the mouse brain except in the brainstem.
␣-RIM and RIM3 Physically Associate with Native VDCCs in the Crude Synaptic Membrane from the Mouse Brain-Subcellular fractionation analysis of the mouse brain showed that RIM1 and RIM3 were highly concentrated in the CSM fraction, which was similar to that for PSD-95 and synaptophysin, established marker proteins of the postsynapse and presynapse, respectively (Fig. 6A). In immunoprecipitation analysis of the CSM fraction solubilized with 1% digitonin-containing buffer, the P/Q-type (Ca V 2.1) and N-type (Ca V 2.2) VDCC ␣ 1 -subunits and the VDCC ␤ 4 -subunit were coimmunoprecipitated with ␣-RIM (RIM1 or RIM2) and RIM3 (Fig. 6B). Thus, the P/Q-type and N-type channels are physically associated with RIM3 as well as with ␣-RIM in the CSM fraction of the mouse brain.
Physiological Relevance of RIM Effects on the Inactivation Properties of VDCCs-To investigate the physiological roles of RIM⅐␤-subunit complexes in native systems, expression of RIM was suppressed by treatment with siRNA (Fig. 7). In rat neuron-like pheochromocytoma PC12 cells, diverse HVA VDCC types have been precisely characterized (66,67). Previously, we identified P/Q-, N-, and L-type Ca 2ϩ currents by pharmacological dissection of HVA Ca 2ϩ current and detected ␣ 1 mRNA species (P/Q-type (Ca V 2.1), N-type (Ca V 2.2), and L-type (Ca V 1.2 and Ca V 1.3)) and ␤-subunits (␤ 1 , ␤ 2 , and ␤ 3 ) in PC12 cells (12). RT-PCR analysis revealed expression of RNA species encoding RIM1, RIM2, RIM3, and RIM4 in PC12 cells (Fig. 7A). RT-PCR analysis also showed that combinations of siRNAs specific for RIM1 and RIM2 (siRIM1&2), RIM3 and RIM4 (siRIM3&4), or all four RIMs (siRIM1&2&3&4) effectively suppressed the expression of target genes, whereas control siRNA (siControl) failed to exert significant effects on RIM RNA expression (Fig. 7A). WB also indicated that application of siRNAs for RIMs effectively suppressed expressions of RIM proteins (Fig. 7B). siRNAs for RIMs accelerated inactivation of whole-cell Ba 2ϩ currents through VDCCs endogenously expressed in PC12 cells (Fig. 7C, top). The voltage dependence of VDCC inactivation in PC12 cells consisted of three phases: a low voltage-induced phase, a high voltage-induced phase, and a non-inactivating phase (Fig. 7C (bottom) and supplemental Table  S4). siRNA application increased the proportion of the low voltageinduced phases (from 0.21 to 0.35 for siRIM1&2, to 0.38 for siRIM3&4, and to 0.46 for siRIM1&2&3&4) and reduced the proportion of the non-inactivating phases (from 0.42 to 0.30 for siRIM1&2, to 0.26 for siRIM3&4, and to 0.17 for siRIM1&2&3&4) (Fig. 7C (bottom) and supplemental Table S4). siRIM1&2&3&4 showed a more marked effect than siRIM1&2 or siRIM3&4. These inactivation properties were rescued by siRNA-resistant RIM constructs. Expression of siRNA-resistant (denoted by an asterisk) RIM To determine the voltage dependence of inactivation, currents were evoked by a 20-ms test pulse to 5 mV after the 10-ms repolarization to Ϫ100 mV following 2-s V h displacements (conditioning pulses) from Ϫ100 to 20 mV with 10-mV increments. See supplemental Table S2 for statistical significance of the differences. Error bars, S.E. cDNAs reversed the accelerated inactivation in RIM knockdown cells (siRIM1&2 ϩ RIM1*&2* and siRIM3&4 ϩ RIM3*&4*) (Fig. 7C, middle). The increase of the ratios of low voltage-induced phases and the reduction of ratios of non-inactivating phases were also rescued by the siRNA-resistant RIMs (0.05 and 0.41 for siRIM1&2 ϩ RIM1*&2*; 0.12 and 0.42 for siRIM3&4 ϩ RIM3*&4*). Other functional current parameters, such as current densities and activation kinetics at different voltages, were unaffected by siRNA for RIMs (supplemental Fig. 2 and Table S5). Thus, RIMs exert suppressive effects on voltage-dependent inactivation of native VDCCs in PC12 cells.
␥-RIMs Inhibit but ␣-RIMs Support Anchoring of Neurotransmitter Vesicles to VDCCs-Dense core vesicles were identified by a fusion protein of NPY and the fluorescent protein Venus (NPY-Venus) in PC12 cells. When cotransfected NPY-Venus was used as an siRNA transfection marker, comparison of fluorescence intensities of immunoreactivities to anti-RIM1 and RIM2 antibodies in NPY-Venus-positive PC12 cells with those in NPY-Venus-negative PC12 cells revealed that combination of specific siRNAs efficiently suppressed RIM1 and RIM2 expression in confocal images (supplemental Fig. 3). We directly observed fluorescent images in the plasma membrane area using evanescent wave microscopy, which illuminates only the subcellular area from the surface to a depth of less than 100 nm by TIRF. When cotransfected Ds-Red monomer was used as an siRNA transfection marker, comparison of fluorescence intensities of immunoreactivities to anti-RIM1 and RIM2 antibodies in Ds-Red-positive PC12 cells with those in Ds-Rednegative PC12 cells also revealed that combination of specific siRNAs efficiently suppressed RIM1 and RIM2 expression in TIRF images (supplemental Fig. 4). The number of vesicles docked to the plasma membrane was decreased significantly by combinatory application of RIM1 and RIM2 siRNAs, whereas it was increased by RIM3 and RIM4 siRNA (Fig. 8). Importantly, the inhibitory effect of siRIM1&2 and the enhancing effect of siRIM3&4 were reversed by expression of siRNA-resistant RIM cDNAs (siRIM1&2 ϩ RIM1*&2* and siRIM3&4 ϩ RIM3*&4*). Effects of siRIM1&2 or siRIM3&4 are unlikely to be due to altered densities of VDCCs at the plasma membrane in PC12 cells because the siRNAs used did not affect VDCC current densities and VDCC expression levels ( Fig. 7B and supplemental Fig. 2A and Table S5). It is important to note that endogenous Rab3 and Munc13 were immunoprecipitated with FLAG-tagged ␣-RIMs but not ␥-RIMs, whereas Liprin1-␣ was immunoprecipitated with ␣-RIMs and ␥-RIMs (supplemental Fig. 6). This is consistent with the previous report that ␣-RIMs are associated with Rab3 and Munc13 via their N termini and with Liprin1-␣ via their C termini (32,37,39). Thus, ␥-RIMs as well as ␣-RIMs play important roles in regulating neurotransmitter vesicle anchoring to VDCCs.

FIGURE 3. Effects of RIMs on the activation properties of P/Q-type Ca V 2.1.
A, activation kinetics of P/Q-type Ca V 2.1 currents in BHK cells expressing ␣ 2 /␦-and ␤ 1a -subunit. Left, families of representative Ba 2ϩ currents. Currents evoked by 5-ms step depolarization from Ϫ20 to 20 mV in 10-mV increments from a V h of Ϫ100 mV are displayed. Right, activation time constants plotted as a function of test potential. The activation phases are well fitted by a single exponential function at all potentials. Activation time constants ( activation ) were obtained from currents elicited by 5-ms step depolarization from Ϫ25 to 30 mV in 5-mV increments from a V h of Ϫ100 mV. B, activation curves of P/Q-type Ca V 2.1 currents. Tail currents elicited by repolarization to Ϫ60 mV after 5-ms test pulses from Ϫ40 to 30 mV were used to determine activation curves. C, effects of RIMs on P/Q-type Ca V 2.1 currents. Left, representative traces for Ba 2ϩ currents on application of test pluses from Ϫ40 to 40 mV with 10-mV increments from a V h of Ϫ100 mV. Right, current density-voltage (I-V) relationships of Ca V 2.1. See supplemental Table S3 for statistical significance of the differences. Error bars, S.E. JULY 9, 2010 • VOLUME 285 • NUMBER 28 ␥-RIMs Enhance Neurotransmitter Secretion Less Potently than ␣-RIMs in PC12 Cells-We studied the physiological relevance of RIM interactions with the VDCC complexes by assessing neurotransmitter release from PC12 cells. PC12 cells were transfected with RIM cDNAs or with siRNAs for RIMs along with Chat, encoding choline acetyltransferase, which synthesizes ACh for synaptic vesicles (57). ACh release, triggered by Ca 2ϩ influx in response to high K ϩ -induced (extracellular K ϩ concentration elevated from 5.9 to 51.1 mM for 30 s) membrane depolarization, was less potently potentiated by recombinant RIM3␥ and RIM4␥ than by recombinant RIM1␣ and RIM2␣ (Fig. 9A). A similar result has been reported previously for RIM3␥ for the secretion of human growth hormone in PC12 cells (36). siRNAs specific to RIMs significantly decreased ACh release. siRIM1&2&3&4 suppressed ACh secretion more markedly than siRIM1&2 or siRIM3&4. Furthermore, ACh release was abolished in the presence of 0.3 mM Cd 2ϩ , a selective blocker of HVA Ca 2ϩ channels, in extracellular solutions (Fig. 9B). Likewise, ACh release triggered by moderate depolarization (extracellular K ϩ concentration elevated from 5.9 to 28.4 mM for 2 min) was also significantly decreased by siRNAs specific to RIMs (Fig. 9C). The inhibitory effects of siRIM1&2 and siRIM3&4 on depolarizationinduced ACh release were rescued by expression of siRNA-resistant RIM cDNAs (Fig. 9, B and C). Notably, ACh release triggered by 28.4 mM K ϩ was more susceptible to suppression by siRIM1&2 than by siRIM3&4. The amount of control ACh secretion by 51.1 mM K ϩ solution nearly doubled that by 28.4 mM K ϩ solution. The [Ca 2ϩ ] i under stimulation with extracellular K ϩ elevation was estimated with fluorescent measurement using fura-2. In 51.1 and 28.4 mM K ϩ -containing extracellular solution, [Ca 2ϩ ] i elevations (⌬[Ca 2ϩ ] i ) were observed and effectively decreased by knockdown of RIMs (Fig. 9, D and E). Control [Ca 2ϩ ] i elevation in 51.1 mM K ϩ solution was approximately 1 order of magnitude higher than those in 28.4 mM K ϩ solution. It is important to note that inhibition of ACh secretion by RIM siRNAs is unlikely to be due to their secondary effects that cause a mislocalization or reduction of the release machinery in PC12 cells because the siRNAs used did not affect the localization of syntaxin (supplemental Fig. 5) or the expression levels of essential components of the release machinery, such as synaptotagmin, SNAP-25, syntaxin, and Rab3 (Fig. 7B). Munc13 was slightly decreased in RIM1 and RIM2 knockdown cells (Fig. 7B), as reported in the RIM1 knock-out mouse, presumably because Munc13 binds to ␣-RIMs (supplemental Fig. 6) (37) and is destabilized in their absence (39). Thus, native ␥-RIMs together with ␣-RIMs play important roles in neurotransmitter release in the PC12 cells. ␥-RIMs Support Neurotransmitter Secretion in Rat Cerebellar Neurons-To investigate physiological roles of RIMs in neurons, we next examined neurotransmitter release from cultured cerebellar neurons. RT-PCR analyses revealed expression of RNA species encoding RIM1, RIM2, RIM3, and RIM4 in cultured rat cerebellar neurons (Fig. 10A). Combinations of shR-NAs specific for RIM1 and RIM2 (shRIM1&2), RIM3 and RIM4 (shRIM3&4), or four RIM isoforms (shRIM1&2&3&4) effectively suppressed the expression of target genes, whereas shControl failed to exert significant effects on RIM RNA expression in rat cerebellar neurons 10 days after transfection (Fig. 10A). Glutamate release from cerebellar neurons, triggered by Ca 2ϩ influx in response to high K ϩ -induced (extracellular K ϩ concentration elevated from 5.9 to 51.1 mM for 60 s) membrane depolarization, was significantly decreased by shRIM1&2, shRIM3&4, and shRIM1&2&3&4 (Fig. 10B). Notably, knockdown of ␣-RIMs (shRIM1&2) more markedly reduced glutamate release than that of ␥-RIMs (shRIM3&4) and that of all RIMs (shRIM1&2&3&4). These results suggest that both ␣-RIMs and ␥-RIMs are important for neurotransmitter release, but their contributions to glutamate release from cultured cerebellar neurons are different.

DISCUSSION
An understanding of the physiological roles played by RIMs other than RIM1 has been elusive, despite evolutionary conservation of the C 2 B domain-containing C terminus among all RIM isoforms. The present investigation revealed physical associations of RIM2␣, RIM3␥, and RIM4␥ with VDCC ␤-subunits. In vitro binding assays and coimmunoprecipitation experiments identified protein complexes formed by direct interactions of VDCC ␤-subunits with ␣ 1 -subunits and with RIMs ( Figs. 1 and 6). The interactions between RIMs and VDCC ␤-subunits decelerated VDCC inactivation to sustain depolarization-induced Ca 2ϩ influx (Figs. 2, 7, and 9, D and E) and enhanced depolarization-induced ACh release in PC12 cells (Fig. 9). TIRF imaging of dense core vesicles indicated that, unlike RIM1␣ and RIM2␣, RIM3␥ and RIM4␥ inhibit the anchoring of neurotransmitter vesicles in the vicinity of VDCCs (Fig. 8).
K d values of the binding of RIM1␣(1079 -1463), RIM2␣(1193-1572), RIM3␥, and RIM4␥ to the ␤ 4 -subunit indicate that the binding affinities of RIM3␥ and RIM4␥ are almost an order of magnitude lower than those of RIM1␣ and RIM2␣ (12) (Fig. 1D). This is consistent with the fact that the amino acid sequence of the primary ␤-subunit interaction region in the RIM1␣ (RIM1␣(1079 -1257)) (12) is better conserved in RIM2␣ (60% identity in RIM2␣(1193-1366)) than in RIM3␥ and RIM4␥ (25 and 12% identities for RIM3␥(1-104) and RIM4␥(1-63), respectively). Despite these differences in affinity among the primary interaction regions for ␤-subunits, kinetic modulation of VDCC inactivation was observed for RIM3␥ and RIM4␥ as well as for RIM2␣. This functional conservation in terms of VDCC regulation may be mainly attributable to the modulatory region (12) well conserved among all RIM proteins;  identities with the RIM1␣ modulatory region (residues 1258 -1463) (Fig. 1A) are 84, 74, and 67% for RIM2␣(1367-1572), RIM3␥(105-307), and RIM4␥(64 -269), respectively. Although expression of RIM3␥ or RIM4␥ significantly inhibits inactivation and slows the activation kinetics of recombinant P/Q-type VDCCs in BHK cells, knockdown of endogenous RIM3 and RIM4 promotes inactivation but fails to affect the activation kinetics of native VDCC currents in PC12 cells. In the recombinant BHK cell system, the P/Q-type formed by the Ca V 2.1 ␣ 1 -and ␤ 1a -subunit was the only VDCC type expressed. However, PC12 cells express various VDCC subunits (the P/Q-, N-, and L-type ␣ 1 -subunits and the ␤ 1 -, ␤ 2 -, and ␤ 3 -subunits) (12,66,67). It is therefore possible that multiple types of VDCCs formed by various ␣ 1 -and ␤-subunits are regulated in a different manner by ␥-RIMs in PC12 cells in terms of parameters such as activation. In fact, we previously reported, using the recombinant system, that RIM1␣ modulates VDCC activation properties, such as current density, voltage dependence of activation, and activation kinetics, depending on the subunit combination of the VDCC complex (12). In contrast to these functional parameters, similar RIM1␣ suppression of inactivation was observed for P/Q-type VDCCs containing a ␤ 2a -, ␤ 3 -, or ␤ 4b -subunit (12). Furthermore, a similar effect on inactivation by RIM1␣ was shown in N-type (Ca V 2.2), R-type (Ca V 2.3), and L-type (Ca V 1.2) VDCCs containing a ␤ 4b -subunit. Thus, considering that the other ␤-subunits interact with RIMs even more strongly than ␤ 1a (Fig. 1, B and C), suppressive effects of RIMs on inactivation are universal among HVA types, whereas effects of RIMs on activation kinetics are variable, depending on the subunit combination of the VDCC complex. Since the submission of this manuscript, similar modulation of Ca V 1.3 VDCC inactivation has been most recently reported for RIM2 (68).
Previous reports have demonstrated functional impacts of syntaxin, SNAP-25, and synaptotagmin on VDCCs through their physical association with the "synprint" region in the II-III linker of ␣ 1 -proteins (9,19,22). It has also been reported that RIM1 and RIM2 associate with the synprint (38,69) directly via the C 2 A domain and with the ␣ 1 C-terminal tail indirectly via the RIM-BP (70) (Fig. 1A). However, inhibition of VDCC inactivation by RIMs may be independent of the synprint-or RIM-BP-mediated association because ␥-RIMs, which lack both the C 2 A domain necessary for synprint binding (38,69) and the PXXP motif for the RIM-BP binding, inhibit VDCC inactivation (Fig. 2). This idea is supported by our observation that BADN was sufficient to inhibit the effects of RIMs on inactivation (Fig. 2). In addition, we have observed that replacement of ␤ with the C-terminal truncation construct ␤ 4 -GK (71), which directly interacts with ␣ 1 but lacks the ability to bind RIM1, was also sufficient to disrupt the RIM1 effects on inactivation (12). In our previous paper (12), we speculated that RIM1␣ may affect a conformational transition to the inactivated state of VDCCs via association with the ␤-subunit.  Table S4 for statistical significance of the differences.
Recently, it has been proposed that the HOOK domain, between the conserved Src homology 3 and guanylate kinase domains of the ␤-subunit, is an important determinant of inactivation (72). Hence, RIMs may act on the HOOK domain to suppress the regulatory function of the ␤-subunit in VDCC inactivation.
In PC12 cells, biphasic voltage dependence of inactivation was observed for native VDCC currents (Fig. 7C, bottom). In the recombinant expression system, a similar biphasic inactivation curve was observed in the presence of BADN, suggesting that a dominant negative effect of BADN partially dissociates the RIMs from the VDCC complexes. Furthermore, V 0.5 values of the low voltage-induced phase and the high voltage-induced phase were similar to those of inactivation curves without RIMs and with RIMs, respectively, whereas the proportions of each phase were affected by RIMs ( Fig. 2 and supplemental Table S2). These findings suggest that at least two types of VDCC complexes, one with RIMs and the other without RIMs, coexist in PC12 cells. VDCC complexes not associated with RIMs may be weakly and indirectly linked to neurotransmitter release in PC12 cells, because knockdown of all four RIMs markedly decreased but did not completely abolish Ca 2ϩ influx-triggered neurotransmitter release (Fig. 9).
When voltage-dependent inactivation of VDCCs is suppressed by RIMs, the responses to depolarizing membrane potentials of Ca 2ϩ sensors, such as synaptotagmins, can be potentiated at presynaptic active zones. RIM-VDCC associa-tion can thus support synaptic plasticity by regulating depolarization-dependent neurotransmitter release. In PC12 cells, expression of both ␣-RIMs and ␥-RIMs enhanced depolarization-dependent neurotransmitter release (Fig. 9A). Thus, not only ␣-RIMs but also ␥-RIMs may play an important role in Ca 2ϩ influx-triggered exocytosis in neuroendocrine cells, such as PC12 cells. In the case of ␣-RIMs, such as RIM1␣, their association with VDCC ␤-subunits supports release via two distinct mechanisms: sustaining Ca 2ϩ influx, as described above, and anchoring neurotransmitter-containing vesicles in the vicinity of VDCCs (12). The former function requires only the RIM1␣ C terminus (RIM1␣(1258 -1463)), which is conserved among RIMs, whereas the latter function requires the "full-length" structure of ␣-RIMs. Because ␥-RIMs exert only the former function, they can act as dominant negative suppressors of vesicle docking. In fact, knockdown of RIM3␥ and RIM4␥ by specific siRNAs increased the number of neurotransmitter vesicles anchored near the plasma membrane in PC12 cells (Fig. 8). In addition, the expression of ␣-RIMs enhanced neurotransmitter release more significantly than did ␥-RIMs upon depolarization elicited by 51.1 mM K ϩ (Fig. 9A), and knockdown of ␣-RIMs decreased neurotransmitter release more prominently than ␥-RIMs upon moderate depolarization elicited by 28.4 mM K ϩ in PC12 cells (Fig. 9C). Upon depolarization elicited by 51.1 mM K ϩ , neurotransmitter release in cells with knockdown of ␣-RIMs and in cells with knockdown of ␥-RIMs was indistinguishable (Fig. 9B). This is presumably because the robustness of global [Ca 2ϩ ] i elevation upon 51.1 mM K ϩ stimulation (Fig. 9, D and E) masks the vesicle-anchoring effect of ␣-RIMs in PC12 cells. Importantly, compared with 51.1 mM K ϩ , 28.4 mM K ϩ induced global [Ca 2ϩ ] i elevation by an order of magnitude less (Fig. 9, D and E) but ACh secretion by only 50% less (Fig. 9, B and C), suggesting that 28.4 mM K ϩ is a depolarizing stimulation almost sufficient for physiological ACh secretion in PC12 cells. Hence, the presence of two functionally different groups of RIMs, ␣-RIMs and ␥-RIMs, becomes physiologically relevant only at moderate stimulations and [Ca 2ϩ ] i elevations in ACh secretion from PC12 cells. By contrast, in cerebellar neurons, contributions of ␣-RIMs and ␥-RIMs to glutamate secretion were differentiated even upon robust depolarization elicited by 51.1 mM K ϩ stimulation (Fig.  10B). We can therefore hypothesize that a transition of the associating partner of presynaptic VDCCs between ␣-RIMs and ␥-RIMs would modify the tightness of coupling of Ca 2ϩ influx with fusion of synaptic vesicles and eventually modulate the efficacy of synaptic transmission.
Relative contributions of ␣-RIMs and ␥-RIMs in neurotransmitter release are very likely different between neurons and PC12 cells. In particular, glutamate release is less reduced by knockdown of all RIMs (shRIM1&2&3&4) than by knockdown of ␣-RIMs (shRIM1&2) in cerebellar neurons in contrast to ACh release in PC12 cells (Fig. 10B), suggesting that contributions to neurotransmitter release of ␥-RIMs through "inhibition" of vesicle anchoring to VDCCs is greater in cerebellar neurons than in neuroendocrine PC12 cells. This is presumably because knockdown of all RIMs abolishes the inhibition of anchoring by ␥-RIMs and, at the same time, the enhancement of anchoring by ␣-RIMs, whereas knockdown of only ␣-RIMs
In central neurons, overlapping distributions of RIM isoforms revealed by in situ hybridization (Fig. 5) (35,48) suggested that RIM isoforms coexist in the same neurons in the brain. Redundant functions of RIM1␣ and RIM2␣ in Ca 2ϩtriggered neurotransmitter release have been reported, but ␣-RIMs were not replaceable by ␥-RIMs with regard to their role in neurotransmitter release (48). This supports the possibility that ␣-RIMs and ␥-RIMs function antagonistically in terms of vesicle anchoring to VDCCs and may also suggest that ␥-RIMs are involved in physiological functions other than neurotransmitter release in neurons. Different subcellular distribu-tions of RIMs support this hypothesis; whereas RIM1␣ is a presynaptic protein (32,39), RIM3␥ is present in many neuronal dendrites and dendritic spines, especially in or near the postsynaptic densities (35). Interestingly, studies on the subcellular localization of the P/Q-and R-type channels have defined both axonal and somatodendritic localizations in contrast to the constrained nerve terminal localization of N-type channels (74 -76). Thus, in regulation of VDCC inactivation, the respective RIM isoforms may interact with particular VDCC types at axonal and somatodendritic regions and at presynaptic regions in neurons.
In terms of Ca V 2.1, which forms P/Q-type VDCCs, the P-type current characterized by very slow inactivation kinetics was first identified in cerebellar Purkinje cells (77,78), whereas the Q-type current characterized by fast inactivation kinetics was later identified in cerebellar granule cells (79,80). Although P-and Q-type VDCCs show distinct properties, knock-out studies have clearly demonstrated that both are encoded by the same Ca V 2.1 gene (80,81). In addition, recombinant human Ca V 2.1 expressed in HEK cells produced the Q-type current but failed to produce the P-type current (82). However, the P-type current was recorded in knock-in mice that express the human Ca V 2.1 instead of mouse Ca V 2.1 (83). These results suggest that some additional factors, such as post-translational modification or interaction with other proteins, may be necessary to produce the P-type current. An interesting candidate that determines the channel types of the Ca V 2.1-forming channel is RIMs. This idea is supported by the following findings. First, our in situ hybridization of cerebellar sections revealed that RIM3␥ and RIM4␥ are more highly expressed in Purkinje cells than in granule cells (Fig. 5, A (e) and B (e)). Second, RIM3␥ has been reported as a postsynaptic protein in the molecular layer of the cerebellar cortex, which contains dendrites of Purkinje cells (35), and P-type has also been located in the dendrites and soma of Purkinje cells (77,84).
In conclusion, we present a common physiological function for RIM proteins, maintaining Ca 2ϩ influx triggered by depolarization through suppression of VDCC inactivation. As observed in PC12 cells and cerebellar neurons, specific physiological roles played by ␥-RIMs can weaken the physical coupling of VDCCs with synaptic vesicles. Further studies using ␥-RIM knock-out mice are necessary to test this hypothesis and reveal its relevance to neural processes.