RIM1 confers sustained activity and neurotransmitter vesicle anchoring to presynaptic Ca2+ channels

Abstract

The molecular organization of presynaptic active zones is important for the neurotransmitter release that is triggered by depolarization-induced Ca2+ influx. Here, we demonstrate a previously unknown interaction between two components of the presynaptic active zone, RIM1 and voltage-dependent Ca2+ channels (VDCCs), that controls neurotransmitter release in mammalian neurons. RIM1 associated with VDCC beta-subunits via its C terminus to markedly suppress voltage-dependent inactivation among different neuronal VDCCs. Consistently, in pheochromocytoma neuroendocrine PC12 cells, acetylcholine release was significantly potentiated by the full-length and C-terminal RIM1 constructs, but membrane docking of vesicles was enhanced only by the full-length RIM1. The beta construct beta-AID dominant negative, which disrupts the RIM1-beta association, accelerated the inactivation of native VDCC currents, suppressed vesicle docking and acetylcholine release in PC12 cells, and inhibited glutamate release in cultured cerebellar neurons. Thus, RIM1 association with beta in the presynaptic active zone supports release via two distinct mechanisms: sustaining Ca2+ influx through inhibition of channel inactivation, and anchoring neurotransmitter-containing vesicles in the vicinity of VDCCs.

Figure 1. Direct interaction of RIM1 with the VDCC β_4b-subunit

(a) Domain structure of mouse RIM1. Arrows indicate molecules interacting with RIM1 at the Zn²⁺-finger like domain (Zn²⁺); PDZ domain (PDZ); first and second C₂ domains (C₂A and C₂B); proline-rich region (PXXP)4,6-10. The protein region encoded by clone #2-5 is also indicated. (b) Mapping of RIM1 binding sites on β_4b by the yeast two-hybrid assay. β-subunit constructs in bait vectors are tested with RIM1 in the prey vector. The interactions are scored by β-galactosidase activity and His⁺ prototrophy. (c) Pulldown assay of β_4b with GST fusion RIM1 mutants. GST fusion proteins immobilized on glutathione-Sepharose beads are incubated with cell lysate obtained from myc-β_4b-transfected HEK293 cells. Bound proteins are analyzed by western blotting (WB) using anti-myc antibody. (d) in vitro association between the purified GST-RIM1 fusion constructs and recombinant β₄-subunit (47–475). GST-RIM1 proteins at various concentrations, incubated with β₄ (50 pM), are captured by glutathione-Sepharose beads. Captured β₄ proteins are examined by WB. The bottom panel shows the quantitative densitometric analysis of bands shown in the upper panel and in Supplementary Fig. 2. The saturation curves are subjected to the nonlinear least-squares curve-fitting method to evaluate the apparent dissociation constant (K_d). (e) Interaction of recombinant β_4b and RIM1 in HEK293 cells. The interaction is evaluated by immunoprecipitation (IP) with anti-FLAG antibody, followed by WB with anti-myc antibody. Top: physical association of myc-β_4b with FLAG-RIM1 in comparison to a positive control FLAG-Ca_v2.1(I–II linker). Bottom: physical association of FLAG-β_4b with myc-RIM1.

Figure 2. Association of RIM1 with native neuronal VDCC complexes

(a) Sucrose gradient fractionation of neuronal VDCC complexes from wild-type (WT) mouse brains and subsequent WB demonstrate cosedimentation of RIM1 with Ca_v2.1 and β₄. Syntaxin showed similar cosedimentation with RIM1. (b) Densitometry of Ca_v2.1, β₄, and RIM1 from western blots of sucrose gradient fractions. The normalized density of each protein is plotted as a function of the sucrose density fraction number. (c) Coimmunoprecipitation of RIM1 with the β₄-subunit. Immunoprecipitation (IP) using an anti-β₄ antibody and subsequent WB for RIM1 is performed on heparin purified samples. As a negative control, a preparation from lethargic mice is used. (d) Sucrose gradient fractionation of neuronal VDCC complexes from lethargic mouse brains. (e) Coimmunoprecipitation of Ca_v2.1 with RIM1. The immunocomplexes are disrupted by GST-BADN or GST-RIM1(1079–1463). IP using anti-RIM1 antibody and subsequent WB for Ca_v2.1 is performed.

Figure 3. RIM1 clusters with the VDCC subunits near presynaptic termini in cultured hippocampal neurons

(a) Ca_v2.1 elicits plasma membrane (PM) colocalization of β_4b with RIM1. BADN disrupts colocalization. Left: confocal imaging of HEK293 cells expressing EGFP-β_4b and RIM1-DsRedmonomer with vector, Ca_v2.1, or Ca_v2.1 plus BADN. Scale bar: 5 μm. Nuclei are stained with Hoechst 33342. Right: subcellular location of EGFP-β_4b or RIM1-DsRedmonomer in 1 μm widths PM region and in the cytosolic area (CYT) (n = 5). ^***P < 0.001 vs vector. ^##P < 0.01 vs Ca_v2.1. (b) β_4b elicits PM colocalization of Ca_v2.1(I–II linker) and RIM1. BADN disrupts colocalization. Left: confocal imaging of HEK293 cells expressing CD8-Ca_v2.1(I–II linker)-EGFP and RIM1-DsRedmonomer with vector, β_4b, or β_4b and BADN. Right: subcellular location of CD8-Ca_v2.1(I–II linker)-EGFP or RIM1-DsRedmonomer (n = 5). ^***P < 0.001 vs vector. ^##P < 0.01 vs β_4b. (c) Immunolocalization of tagged-RIM1 and β_4b in cultured hippocampal neurons. Clustering of RIM1 and β_4b is undetected at 8 DIV, but present at a substantially later stage at 23 DIV. Scale bar: 10 μm. (d) Late clustering of EGFP-Ca_v2.1 (arrowheads) in hippocampal neurons. Synapsin I-positive puncta are already abundant at 8 DIV, while Ca_v2.1 distribution is still diffuse. Like RIM1 and β_4b, Ca_v2.1 clusters much later on. (e, f) Accumulation of EGFP-Ca_v2.1 (arrowheads) in presynaptic varicosities is achieved between 9 and 22 DIV. This maturation process is impaired by RIM1(1079–1463) or BADN, suggesting that the local VDCC concentration at AZs is influenced by RIM1-β-subunit interaction during a postsynaptogenic maturation period.

Figure 4. Effects of RIM1 on the inactivation properties of recombinant neuronal VDCCs

(a)Inactivation of P/Q-type Ca _v2.1 currents in BHK cells. The peak amplitudes before and after coexpression of RIM1 constructs are normalized for Ba²⁺ currents elicited by 2-s pulses to 0 mV from a holding potential (V_h) of −100 mV. (b) Inactivation of N-type Ca_v2.2, R-type Ca_v2.3, or L-type Ca_v1.2 currents (with β_4b). The V_h is −100 mV (Ca_v2.2, Ca_v1.2) or −110 mV (Ca_v2.3). (c) Left: inactivation curves for Ca_v2.1 (with β_1a). Right: inactivation curves for Ca_v2.1 in BHK cells expressing α₂/δ and different β-subunits. See Supplementary Table 2 and 3 for statistical significance of the differences. (d) Inactivation curves for Ca_v2.2, Ca_v2.3 (left), or Ca_v1.2 (right) (with β_4b). See Supplementary Table 1 for statistical significance of the differences. (e) RIM1 prolongs the time between first channel opening and last closing within a single-channel trace of Ca_v2.1 (with β_4b). Seven consecutive unitary traces are shown. The mean values for the time of each trace are 184.2 ± 33.3 ms (n = 117 traces) for vector and 502.8 ± 33.3 ms, (n = 101) for RIM1. The time for traces without opening is counted as 0 ms. (f) Left: Ca_v2.1 currents (with β_1a) induced by 100 Hz AP trains for 1 s. Right: percentage of currents in response to the last stimulus compared to the peak current (n = 6 for vector and n = 4 for RIM1). ^***P < 0.001.

Figure 5. Physiological relevance of effects of RIM1 on inactivation properties of VDCCs

(a) Effects of RIM1 on inactivation time courses of Ca_v2.1 Ba2+ and Ca²⁺ currents (with β₁) in HEK293 cells. The peak amplitudes before and after RIM1 coexpression are normalized for Ba²⁺ and Ca²⁺ currents elicited by 2-s pulses to 0 and 10 mV, respectively, from a V_h of −80 mV. (b) Inactivation curves for Ca_v2.1 Ba2+ and Ca²⁺ currents (with β₁) in HEK293 cells. See Supplementary Table 6 for statistical significance of the differences. (c) Effects of RIM1 and BADN on the inactivation properties of native VDCCs in PC12 cells (7–9 culture passage). Left: normalized current traces. Middle: inactivation curves induced by 2-s holding potential displacement. See Supplementary Table 7 for statistical significance of the differences. Right: comparison of current densities at 10 mV (n = 18, 15, and 9 cells for vector, RIM1, and BADN, respectively). (d) Acceleration of inactivation by coapplication of siRNAs specific for RIM1 and RIM2 (siRIM1 and siRIM2) in VDCC currents recorded from PC12 cells (2–3 culture passages). Left: normalized current traces. Middle: inactivation curves. See Supplementary Table 7 for statistical significance of the differences. Right: comparison of current densities at 10 mV (n = 6 and 8 cells, for siControl and siRIM1 and siRIM2, respectively).

Figure 6. Effects of RIM1 on the activation properties of VDCCs

(a) Effects of RIM1 on activation curves of Ca_v currents (with β_4b) elicited in BHK cells. Tail currents elicited by repolarization to −60 mV after 5-ms test pulse from −50 to 50 mV are used to determine activation curves. See Supplementary Table 1 for statistical significance of the differences. (b) Left: effects on activation speed of Ca_v2.1 channels containing various β-subunits. Time constants are obtained by fitting the activation phase of currents elicited by 5-ms test pulse to 20 mV with a single exponential function. Right: effects of RIM1 on activation speed of Ca_v currents at 20 mV in various VDCC types. ^*P < 0.05 and ^***P < 0.001. (c) Effects of RIM1 proteins on P/Q-type Ca_v2.1. Left: representative traces for Ba²⁺ currents (with β_4b) upon application of test pluses from −40 mV to 50 mV with 10-mV increments. Right: current density-voltage (I–V) relationships of Ca_v2.1. The V_h is −100 mV. See Supplementary Table 2 for statistical significance of the differences. (d) I–V relationships of Ca_v2.2 (left) and Ca_v2.3 or Ca_v1.2 (right) (with β_4b). The V_h is −100 mV (Ca_v2.2, Ca_v1.2) or −110 mV (Ca_v2.3). See Supplementary Table 1 for statistical significance of the differences.

Figure 7. RIM1 and β-subunits associate to anchor neurotransmitter vesicles to VDCCs at the plasma membrane

(a) NPY-containing secretory vesicles are colocalized with RIM1 and VAMP, that is not colocalized with caveolin-1 in PC12 cells. NPY-Venus and RIM1-DsRedmonomer, VAMP-Venus and RIM1-DsRedmonomer, NPY-Venus and VAMP-DsRedmonomer, or caveolin-1-EGFP and VAMP-DsRedmonomer are coexpressed in PC12 cells, and live images of the cells are obtained by confocal microscopy. Scale bar: 10 μm. (b,c) Effects of RIM1 constructs and BADN on the density of docked vesicles. In b, typical TIRF images of plasma membrane-docked vesicles containing NPY-Venus. Left: BADN-co-transfected PC12 cell. Middle: vector-co-transfected PC12 cell. Right: RIM1-co-transfected PC12 cell. Scale bar: 10 μm. In c, the vesicle density is determined by counting the vesicles in each image (n = 20 cells in each). ^*P < 0.05, ^***P < 0.001 vs vector. ^###P < 0.001 vs RIM1.

Figure 8. The RIM1-β association enhances neurotransmitter release in PC12 cells and cultured cerebellar neurons

(a) Effects of RIM1 constructs and BADN on depolarization-dependent release of ACh from ChAT-co-transfected PC12 cells. Three days after transfection, PC12 cells were incubated for 30 s with the low-K⁺ solution (5.9 mM K⁺) at 37 °C. The release of ACh during this period was considered as basal release. To measure ACh release, the cells were then incubated for 30 s with a high-K⁺ solution (51.1 mM K⁺). The amount of secreted ACh is determined as a percentage of the cellular content for each dish. ^***P < 0.001 vs vector. ^##P < 0.01, ^###P < 0.001 vs RIM1. (b) Effects of RIM1 and BADN on depolarization-dependent release of glutamate from cultured cerebellar neurons. 24 h after the introduction of cDNAs, cerebellar neurons (9–11 DIV) were incubated for 1 min with the low-K⁺ solution (5.9 mM K⁺) at 37 °C. The release of glutamate during this period was considered as basal release. To measure glutamate release, the cells were then incubated for 1 min with a high-K⁺ solution (51.1 mM K⁺). ^*P < 0.05.