Introduction

At chemical synapses, an action potential arriving at the presynaptic terminal opens voltage-gated Ca²⁺ channels, and Ca²⁺ influx through voltage-gated Ca²⁺ channels triggers neurotransmitter release from synaptic vesicles within ms. Synaptic vesicle fusion occurs at a specialized region of the presynaptic membrane called the active zone (AZ), where Ca²⁺ channel clusters and vesicle fusion sites are accommodated via AZ scaffold proteins (Südhof, 2012). For rapid transmission, accumulation of Ca²⁺ channels, co-localization of Ca²⁺ channels and fusion sites, and molecular priming of synaptic vesicles for fusion are required, and the differences between these factors might provide a basis for synaptic diversity. However, it remains unknown how these factors are regulated by AZ-scaffold proteins.

Rab3-interacting molecule-binding proteins (RIM-BPs) represent one principal, conserved family of AZ proteins and bind to RIMs and Ca²⁺ channels (Wang et al., 2000; Hibino et al., 2002). RIM-BP family includes RIM-BP1, RIM-BP2, and RIM-BP3 in mammals, but RIM-BP3 expression is low in the nervous system (Mittelstaedt and Schoch, 2007). In Drosophila neuromuscular junctions (NMJs), loss of RIM-BPs decreases Ca²⁺ channel density and reduces release probability (Liu et al., 2011). Moreover, RIM-BP here is necessary for tight coupling of synaptic vesicles to Ca²⁺ channels and replenishment of high release probability vesicles at Drosophila NMJs (Liu et al., 2011; Müller et al., 2015, Petzoldt et al., 2020). In mammalian synapses, the observations from KO mice are diverse. In RIM-BP1,2 DKO mice, the coupling between Ca²⁺ channels and synaptic vesicles became loose, and action potential-evoked neurotransmitter release was reduced at the calyx of Held synapse (Acuna et al., 2015). At hippocampal CA3-CA1 synapses, RIM-BP2 deletion alters Ca²⁺ channel localization at the AZs without altering total Ca²⁺ influx. Besides, RIM-BP1,2 DKO has no additional effect, indicating that RIM-BP2 dominates the function of RIM-BP isoforms (Grauel et al., 2016). Tight coupling between Ca²⁺ channels and vesicles is crucial for rapid and efficient transmitter release (Wadel et al., 2007).

In contrast, hippocampal mossy fiber-CA3 synapses are characterized by low release probability (Vyleta and Jonas, 2014). At hippocampal mossy fiber synapses, gSTED imaging suggested that RIM-BP2 KO altered the Munc13-1 cluster number and distribution. In addition, RIM-BP2 deletion decreased the number of docked vesicles (Brockmann et al., 2019). However, because of technically difficulty, direct and quantitative measurements of exocytosis in RIM-BP2 KO terminals have not been performed so far.

To quantify the RIM-BP2 function in mossy fiber-CA3 synapses, we performed whole-cell patch-clamp recordings from WT and RIM-BP2 KO hippocampal mossy fiber boutons. We measured the size of the readily releasable pool (RRP) of synaptic vesicles, Ca²⁺ influx, Ca²⁺ sensitivity of vesicle fusion, and the sensitivity of release to an intracellular Ca²⁺ buffer. From electrophysiological recordings, we show that Ca²⁺ current amplitudes are reduced in RIM-BP2 KO terminals, and that the reduction of Ca²⁺ currents is the dominant factor for that of transmitter release. Using STED imaging, we find the densities of P/Q-type Ca²⁺ channels in terminals to be reduced in RIM-BP2 KO. We suggest that RIM-BP2 regulates the number of P/Q-type Ca²⁺ channels at the AZ and is critical for rapid transmitter release at hippocampal mossy fiber terminals.

Results

RIM-BP2 KO reduces synaptic vesicle release and calcium currents at hippocampal mossy fiber boutons

To investigate the roles of RIM-BP2 in synaptic transmission, we examined the kinetics of exocytosis and Ca²⁺ influx in WT and RIM-BP2 KO synapses. We applied whole-cell patch-clamp recordings and capacitance measurements to hippocampal mossy fiber terminals in acute slices from WT and RIM-BP2 KO mice, in order to measure Ca²⁺ currents and exocytosis directly. The patch pipette contained Cs⁺-based internal solution and 0.5 mM EGTA to isolate Ca²⁺ currents (Midorikawa and Sakaba, 2017). The external solution contained 1 μM TTX to block Na⁺ currents. Membrane capacitance was measured by applying a sinusoidal wave (1000 Hz, ±30 mV in amplitudes, Lindau and Neher, 1988). When the terminal was depolarized from −80 mV to +10 mV for 10 ms, Ca²⁺ currents and membrane capacitance were recorded as shown in Fig. 1A. During the depolarizing pulse, capacitance was not measured due to large conductance changes, and the increase was used for measuring synaptic vesicle exocytosis. To determine the size of the RRP and the time course of exocytosis, we measured capacitance changes (ΔC_m) in response to various durations of depolarization. Here, the length of the depolarizing pulse was varied between 5 ms and 100 ms, and ΔC_m plotted against pulse duration (Fig. 1B, bottom). ΔC_m in response to a 10 ms pulse was reduced in RIM-BP2 KOs (p = 0.0499). ΔC_m became larger as the duration of pulse was prolonged, but the increase started to be saturated at a 100 ms pulse in WT terminals, suggesting depletion of the RRP. ΔC_m in response to a 100 ms pulse was 52 ± 9.1 fF (n = 8). The amplitude corresponds to the fusion of about 500-600 synaptic vesicles (Hallermann et al., 2003; Midorikawa and Sakaba, 2017). The time course of ΔC_m could be fitted by a single exponential with a time constant of 58 ± 11 ms. In RIM-BP2 KO terminals, ΔC_m in response to a 100 ms pulse was smaller (41 ± 5.9 fF, n = 8) than that of WT terminals, and the release time constant was slower (64 ± 14 ms) than in WT terminals. We should note that the time constant is not necessarily reliable because the ΔC_m values do not reach a plateau value at 100 ms pulse in some terminals. The amplitudes of Ca²⁺ currents were smaller in RIM-BP2 KO terminals as compared to WT terminals (Fig. 1B, top). The Ca²⁺ current in response to a 100 ms pulse was reduced by ∼ 30% in RIM-BP2 KO terminals (27 ± 3.9 pA, n = 8) compared to WT terminals (41 ± 4.3 pA, n = 8) (p = 0.0379). Thus, the results show that RIM-BP2 deletion reduces both neurotransmitter release and Ca²⁺ influx.

Figure 1.

To investigate the Ca²⁺ influx in more detail, we examined the voltage dependence of Ca²⁺ currents. Ca²⁺ currents were recorded in the presence of 1 μM TTX, a Na⁺ channel blocker, and 10 mM TEA, a K⁺ channel blocker. Fig. 1C shows voltage step protocols and representative current traces. Ca²⁺ currents were elicited by 5 ms depolarizing pulses from a holding potential of −80 mV to various potentials to obtain I-V relationships. In both genotypes, Ca²⁺ currents started to be activated at around −20 mV, and maximal amplitude was observed at ∼ +10 mV (Fig. 1D). There was no major difference in the voltage dependence of Ca²⁺ currents between WT and RIM-BP2 KO mice. Consistently, activation time constants of Ca²⁺ currents were similar between WT and RIM-BP2 KO, when the terminal was depolarized to +10 mV (τ = 1.4 ± 0.2 ms in WT, n = 6; τ = 1.3 ± 0.2 ms in KO, n = 4). These results suggest a reduction in the number of Ca²⁺ channels rather than changes in activation kinetics is responsible for the current decrease observed.

High extracellular calcium concentration rescues the decrease of synaptic vesicle release in RIM-BP2 KO

Ca²⁺ influx controls exocytosis by regulating vesicular release probability and/or the RRP size (Katz, 1969; Thanawala and Regehr, 2013). Because Ca²⁺ currents were reduced in RIM-BP2 KO mice, we tested whether the reduction of release in RIM-BP2 KO mice could be explained by that of Ca²⁺ influx or impaired synaptic vesicle docking / priming (Brockmann et al., 2017). To increase Ca²⁺ influx, we raised the extracellular Ca²⁺ concentration ([Ca²⁺]_ext) from 2 mM to 4 mM. Ca²⁺ currents and membrane capacitances were recorded at 4 mM [Ca²⁺]_ext, and amplitudes then plotted against pulse duration (Fig. 2A). Although Ca²⁺ current elicited by a 100 ms pulse in RIM-BP2 KO was still smaller (33 ± 2.7 pA, n = 3) than that in WT (47 ± 6.6 pA, n = 4) (Fig. 2B, top), elevated [Ca²⁺]_ext increased Ca²⁺ currents in both WT and RIM-BP2 KO mice. At 4 mM [Ca²⁺]_ext, ΔC_m in response to a 100 ms pulse in RIM-BP2 KO was larger (51 ± 11 fF, n = 3) than at 2 mM [Ca²⁺]_ext, and indeed showed a similar value in WT at 2 mM [Ca²⁺]_ext (Fig. 2B, bottom). At the same time, ΔC_m in WT was not altered (53 ± 3.2 fF, n = 4) when higher [Ca²⁺]_ext, suggesting that the RRP in WT was already depleted at 2 mM [Ca²⁺]_ext. Also, the time course of exocytosis at 4 mM [Ca²⁺]_ext was similar to that of WT at 2 mM [Ca²⁺]_ext, as the time courses were superimposable in Fig 2B. These results suggest that a decrease in neurotransmitter release in RIM-BP2 KO could be rescued by an increase in Ca²⁺ influx, easiest explained by a reduction of Ca²⁺ channel number.

Figure 2.

Decreased Calcium currents are mainly responsible for reduced release in RIM-BP2 KO

In order to further address whether the reduction of presynaptic Ca²⁺ influx observed was responsible for the compromised release of RIM-BP2 KO mice, we adjusted Ca²⁺ current amplitudes of WT to that of RIM-BP2 KO by lowering [Ca²⁺]_ext. The amplitudes of Ca²⁺ currents and capacitance jumps were plotted against pulse duration (Fig. 3A). At 1 mM [Ca²⁺]_ext, Ca²⁺ currents and ΔC_m were smaller than those of RIM-BP2 KO at 2 mM [Ca²⁺]_ext. At 1.5 mM [Ca²⁺]_ext, Ca²⁺ currents became comparable to those of RIM-BP2 KO at 2 mM [Ca²⁺]_ext. Here, the ΔC_m evoked by a 100 ms pulse (43 ± 6.7 fF, n = 5) was comparable between WT and RIM-BP2 KO (Fig 3A). The average time course of capacitance increase of WT at 1.5 mM [Ca²⁺]_ext was similar to that of RIM-BP2 KO at 2 mM [Ca²⁺]_ext. Therefore, by reducing Ca²⁺ currents, WT data could reproduce the release time course of RIM-BP2 KO. In Fig. 3B, ΔC_m in response to a 100 ms pulse at various [Ca²⁺]_exts were plotted against the peak Ca²⁺ current amplitudes. The relationship could be fitted by a Hill plot with n = 3 in WT (Schneggenburger et al., 1999). Consistently, when capacitance jumps elicited by various pulse durations are plotted against total Ca²⁺ influx (Fig. S1), the RIM-BP2 KO data were superimposed on the WT data, indicating that Ca²⁺-sensitivity for release was not altered and that the reduction of Ca²⁺ currents determine the reduced release in RIM-BP2 KO. In Fig. 3C, relative amplitudes of Ca²⁺ currents and capacitance in response to a 100 ms pulse are plotted. The RIM-BP2 KO data were superimposed on the WT data. These results confirmed that synaptic vesicles have similar sensitivity to Ca²⁺ influx in both genotypes.

Figure 3.

High EGTA experiments suggest unaltered coupling between calcium channels and synaptic vesicles

Previous studies have used Ca²⁺ chelator such as EGTA to examine the sensitivity of release to intracellular Ca²⁺ buffers (Adler et al., 1991; Borst & Sakmann, 1996; Vyleta & Jonas, 2014). When the physical coupling between Ca²⁺ channels and synaptic vesicles is tight, EGTA (slow Ca²⁺ chelator) has less effect on release probability, but EGTA is effective when the coupling is loose. Previous experiments have suggested that RIM-BPs regulate the channel-vesicle coupling at the calyx of Held (Acuna et al., 2015). If this observation also applies to the mossy fiber synapse, EGTA sensitivity of release should be higher in RIM-BP2 KO terminals. To investigate the coupling at mossy fiber boutons in RIM-BP2 KO, we changed the concentration of EGTA in the patch pipette. In the presence of 5 mM EGTA, we depolarized the terminal from −80 mV to +10 mV for 5 ms and recorded Ca²⁺ currents and membrane capacitance (Fig. 4A). Ca²⁺ currents and ΔC_m were plotted against pulse duration (Fig. 4B). In both WT and RIM-BP2 KO mice, ΔC_m were reduced under 5 mM EGTA compared with the control condition of 0.5 mM EGTA. We compared the effect of EGTA on average amplitudes of peak Ca²⁺ currents in response to a 20 ms pulse (Fig. 4D). Interestingly, Ca²⁺ current amplitudes were ∼ 1.5 times larger at 5 mM EGTA (53 ± 12 pA, n = 5) than at 0.5 mM EGTA (39 ± 3.7 pA, n = 8) in WT, though the data scattered under 5 mM EGTA condition and there was no statistically significant difference (p = 0.524). In RIM-BP2 KO, we did not observe such an increase. EGTA might inhibit Ca²⁺ channel inactivation in WT (von Gersdorff and Matthews, 1996). In Fig. 4C, we compared the effect of EGTA on the ΔC_m in response to a 20 ms pulse. In both genotypes, ΔC_m values were inhibited by 5 mM EGTA, and the inhibition was larger in RIM-BP2 KO.

Figure 4.

Ca²⁺ current amplitudes in RIM-B2 KO are smaller than in WT (Fig. 4B). It is possible that strong effect of high EGTA on release maybe due to reduced Ca²⁺ currents in RIM-BP2 KO. Therefore, we changed [Ca²⁺]_ext from 2 mM to 1 mM and adjusted Ca²⁺ current amplitudes of WT to the RIM-BP2 KO level. The capacitance increase and the amplitudes of Ca²⁺ currents became similar to those of RIM-BP2 KO (Fig. 4C and Fig. 4D). These results indicate that WT and RIM-BP2 KO terminals have similar sensitivity of neurotransmitter release to EGTA. Thus, it is unlikely that changes in the coupling distance are responsible for the reduced exocytosis of the mutants.

STED microscopy reveals that RIM-BP2 KO reduces the density of P/Q-type Ca²⁺ channels

Our electrophysiological data suggest that RIM-BP2 may regulate the density of Ca²⁺ channels, or possibly Ca²⁺ channel properties. In order to directly study the Ca²⁺ channel density at the AZ, we performed STED microscopy analysis of Ca²⁺ channels in WT and RIM-BP2 KO hippocampal mossy fiber terminals. We here focused on P/Q-type and N-type Ca²⁺ channels because both Ca²⁺ channel types are relevant for transmitter release at this synapse (Castillo et al., 1994; Pelkey et al., 2006; Li et al., 2007). We performed immunohistochemistry on thin brain cryosections from WT and RIM-BP2 KO mice. STED microscopy confirmed complete loss of RIM-BP2 proteins in KO terminals (Fig. S2). We then immunohistochemically labeled the α-subunit of either P/Q-type Ca²⁺ channel (Cav2.1) or N-type Ca²⁺ channel (Cav2.2) and Munc13-1 (AZ/release site marker, Böhme et al., 2016; Sakamoto et al., 2018) with VGLUT1 (presynaptic marker) and PSD-95 (postsynaptic marker) (Fig. 5A, Fig. 5B and Fig. S3). In the stratum lucidum of the hippocampal CA3 region, mossy fibers form large synapses containing multiple and clustered AZs on CA3 pyramidal cells, and also small synapses containing a single AZ on interneurons (Acsády et al., 1998; Rollenhagen et al., 2007). Thus, we identified AZs using Munc13-1 STED images and quantified the signal intensity of Cav2.1 or Cav2.2 only at AZs clustered in large VGLUT1 positive terminals in the CA3 stratum lucidum, allowing us to confine the analysis to large mossy fiber terminals innervating onto CA3 pyramidal cells. The total signal intensity of Cav2.1 at the AZ was 22% lower in RIM-BP2 KO mice (n = 4 animals) than in WT mice (n = 4 animals) (p = 0.0286) (Fig. 5C, left). These data are consistent with the reduction of Ca²⁺ currents in RIM-BP2 KO mice. In contrast to Cav2.1, the total signal intensity of Cav2.2 and Munc13-1 did not significantly differ between WT and RIM-BP2 KOs (p = 0.3143 and p = 0.0831) (Fig. 5C, right). Furthermore, we found that the number of Cav2.1 clusters within the AZ identified by STED deconvolution analysis was not altered (2.5 ± 0.11 in WT, n = 4 animals; 2.3 ± 0.02 in KO, n = 4 animals) (p = 0.114). These data suggest that RIM-BP2 KO reduces the number of P/Q-type Ca²⁺ channels per cluster. To optically estimate physical distances between Ca²⁺ channels and release sites, we next analyzed the nearest neighboring distance between Cav2.1 and Munc13-1 clusters. These distances were unchanged (61 ± 1.4 nm in WT, n = 4 animals; 64 ± 1.3 nm in KO, n = 4 animals) (p = 0.200) (Fig. 5D), consistent with the similar sensitivity of release to EGTA between WT and RIM-BP2 KO mice. From these results, we conclude that decreased Ca²⁺ influx in RIM-BP2 KO hippocampal mossy fiber terminals is caused by reduction in the density of P/Q-type Ca²⁺ channels at the AZ.

Figure 5.

Discussion

Accumulation of Ca²⁺ channels at the AZ, and efficient vesicle docking and priming are essential factors for fast synaptic vesicle exocytosis. AZ proteins are critical for fast synchronous release and synaptic diversity. RIM-BPs have been implicated in recruitment of Ca²⁺ channels to the AZ (Kaeser et al., 2011; Liu et al., 2011) by interacting with RIMs and Ca²⁺ channels (Wang et al., 2000; Hibino et al., 2002). In addition, RIM-BPs, like RIM1/2, recruit Munc-13 and accelerate synaptic vesicle priming (Brockmann et al., 2020). Concerning the relative importance of RIM-BPs on priming vs Ca²⁺ channel recruitment, we still do know little.

At murine central synapses, RIM-BPs deficient synapses demonstrate overall a rather mild decrease in the amounts of transmitter release (Acuna et al., 2015; Grauel et al., 2016; Luo et al., 2017; Krinner et al., 2017; Brockmann et al., 2019; Butola et al., 2021). Most studies so far have been focused on synapses with high release probability (phasic synapses) such as the calyx of Held and CA3-CA1 synapses. This said, at mossy fiber-CA3 synapses, KO of RIM-BPs has a strong influence on synaptic transmission (Brockmann et al., 2019). Notably, at mossy fiber-CA3 synapses, release probability is relatively low (tonic synapses). It was unknown whether the reduction of transmitter release at RIM-BP2 KO mossy fiber synapses is due to impaired Ca²⁺ channel recruitment, compromised vesicle docking or priming. Using STED microscopy, Brockmann et al. (2019) suggested that the physical recruitment of Munc13-1 might be an important function of RIM-BP2, though functional demonstration had been lacking.

By using direct presynaptic patch clamp recordings, we here observed a decrease of Ca²⁺ current amplitudes (∼ 30%) in RIM-BP2 KO mice (Fig. 1). As the rates of exocytosis steeply depend on Ca²⁺ influx (Dodge and Rahamimoff, 1967), the reduced Ca2+ influx is likely sufficient to explain the slowed release time course at RIM-BP2 KO mossy fiber synapses. Moreover, STED microscopy supported decreased density of P/Q-type Ca²⁺ channels (Cav2.1) in the mutant mossy fiber terminal. Brockmann et al. (2019) observed a reduction of Munc13-1 cluster number and docked vesicles in RIM-BP2 deficient synapses at hippocampal mossy fiber terminals. They identified the terminals by using ZnT3, enriched at mossy fiber (Brockmann et al., 2019; Wenzel et al., 1997). Our STED imaging surprisingly did not detect differences in Munc13-1 cluster number at the AZ between WT and RIM-BP2 KO mice. Notably, we analyzed the number of Cav2.1, Cav2.2 and Munc13-1 clusters only within AZs showing direct co-localization with VGLUT1 and PSD-95, which mainly reflects the AZs facing CA3 pyramidal cells rather than the synapses onto interneurons (Acsády et al., 1998; Rollenhagen et al., 2007). Such a difference in areas analyze between previous and our study might explain the difference. Still, our results do not exclude roles of RIM-BP2 in synaptic vesicle priming. We however suggest that the reduced Ca²⁺ influx might be more critical here when compared to effects of impaired vesicle priming.

Recent studies have reported that RIM-BPs deletion altered Ca²⁺ channel localization and loosened the coupling between Ca²⁺ channels and synaptic vesicles at phasic synapses (Acuna et al., 2015; Grauel et al., 2016; Butola et al., 2021). We investigated the Ca²⁺ channel-vesicle coupling at hippocampal mossy fiber terminals. The sensitivity of release to Ca²⁺ chelator EGTA was not changed between WT and RIM-BP2 KO mice. We also indicated RIM-BP2 KO does not alter the coupling of Ca²⁺ channels to release sites by using STED microscopy.

At phasic synapses, deletion of RIM-BP2 affects physical distance between Ca²⁺ channels and synaptic vesicles. In contrast, at hippocampal mossy fiber terminals having low release probability (Vyleta and Jonas, 2014), RIM-BP2 KO decreased the number of P/Q-type Ca²⁺ channels, thereby reducing Ca²⁺ influx and neurotransmitter release. Thus, we hypothesize the specific molecular-architectural and biochemical contributions by RIM-BPs might be of different relevance between phasic and tonic synapses. Brockmann et al. (2020) proposed that RIM-BP2 controls Ca²⁺ channel recruitment and vesicle priming by profound interaction with RIMs and Munc13s. These results have been obtained through studying hippocampal cultures dominated by phasic synapses. It seems possible that the interaction among three protein types is tight at phasic synapses, simply because of high density of these proteins at AZs. Alternatively, one might speculate about additional proteins, different interaction surfaces taken, transsynaptic columns (Tang et al., 2016). Deletion of RIM-BP2 does not lead to loss of Ca²⁺ channels themselves and/or vesicle priming due to remaining RIMs, Munc13, and other proteins, which can regulate recruitment of Ca²⁺ channels and synaptic vesicle priming without RIM-BP2. Instead, fine-tuning of release such as synchronization is compromised. We hypothesize that AZ scaffold might be differently built, leading to different consequences when eliminating one component. Hence deletion of RIM-BP2 leads to immediate loss of Ca²⁺ channels and/or vesicle priming. This is in line with the proposal that at tonic synapses, synaptic vesicles are only loosely coupled with Ca²⁺ channels (Vyleta and Jonas, 2014) and vesicles are de-primed under resting conditions (Miki et al., 2016; Neher and Brose, 2018). Future research along this line may lead to an understanding of how fine molecular differences in the AZ scaffolds might orchestrate synaptic diversity.

Materials and Methods

Slice preparation

All animal experiments were conducted in accordance with the guidelines of the Physiological Society of Japan, and were approved by Doshisha University Animal Experiment Committee (A22063, D22063).

We used male and female C57BL/6 mice (postnatal days 35-40). WT and RIM-BP2 KO mice (Grauel et al., 2016) were deeply anesthetized with isoflurane and decapitated. Their brains were quickly removed and chilled in sherbet-like cutting solution containing (in mM): 87 NaCl, 75 sucrose, 25 NaHCO₃, 1.25 NaH₂PO₄, 2.5 KCl, 10 glucose, 0.5 CaCl₂ and 7 MgCl₂ bubbled with 95% O₂ and 5% CO₂ (Hallermann et al., 2003). Hippocampal slices (300 μm thick) were prepared from brains using a vibratome (VT1200S, Leica) in ice-cold cutting solution. After slicing, slices were incubated in cutting solution at 37°C for 30 min, and subsequently kept at room temperature (22-25°C) up to 4 h.

Whole-cell recordings

Electrophysiological recordings were performed in a recording chamber filled with the extracellular solution containing (in mM): 125 NaCl, 2.5 KCl, 25 glucose, 25 NaHCO₃, 1.25 NaH₂PO₄, 0.4 ascorbic acid, 3 myo-inositol, 2 Na-pyruvate, 2 CaCl₂ and 1 MgCl₂ saturated with 95% O₂ and 5% CO₂. In some experiments, the concentration of CaCl₂ was changed to 1 mM, 1.5 mM and 4 mM. For membrane capacitance measurements, 1 μM tetrodotoxin (TTX, Wako) was added to block Na⁺ channels. Moreover, for recording Ca²⁺ currents, 10 mM TEA-Cl was added to block K⁺ channels. Slices were visualized by an upright microscope (BX-51, Olympus). Whole-cell patch-clamp recordings were applied to hippocampal mossy fiber terminals at room temperature. The patch pipettes were filled with the intracellular solution containing (in mM): 135 Cs-gluconate, 20 TEA-Cl, 10 Hepes, 5 Na₂-phosphocreatine, 4 MgATP, 0.3 GTP and 0.5 EGTA (pH = 7.2 adjusted with CsOH). In some experiments, the concentration of EGTA was changed to 5 mM. Presynaptic patch pipettes (BF150-86-10, Sutter Instrument) had a resistance of 15-20 MΩ and series resistance (R_s) was 30-70 MΩ. R_s was compensated so that residual resistance was 30-35 MΩ. Patch-clamp recordings were performed using an EPC10/3 amplifier (HEKA) in voltage-clamp mode, controlled by Patchmaster software (HEKA). Membrane currents were low-pass filtered at 2.9 kHz and sampled at 20 kHz. Membrane capacitance measurements were performed using an EPC10/3 amplifier in the sine + DC configuration (Gillis, 2000) using Patchmaster software (HEKA). A sine wave (30 mV in amplitude, 1000 Hz in frequency) was superimposed on the holding potential of −80 mV. Data were obtained from at least three different WT and RIM-BP2 KO terminals in each set of experiments (biological replicate).

Immunostaining

Brains of WT and RIM-BP2 KO mice were quickly removed and transferred to cryomold (Sakura Finetek) filled with tissue freezing medium (Leica). Brains were instantaneously frozen on aluminum block in liquid nitrogen and stored at −80°C. For cryosectioning, frozen brains were kept in cryostat (CM1860, Leica) at − 18°C for 30 min. Then, 10 μm-thick sections were sliced and collected on cover glasses (Matsunami glass, 25×25 No.1). Sections were quickly fixed by dehydration with a heat blower at 50°C for 1 min, and further dehydrated in ethanol for 30 min at −25°C and in acetone for 10 min on ice. After blocking with PBS containing 0.3% BSA at room temperature, sections were incubated with primary antibodies diluted in PBS containing 0.3% BSA: anti-VGLUT1 (1:2000, Synaptic Systems, guinea pig polyclonal, RRID:AB_887878), anti-PSD-95 (1:50, NeuroMab, Mouse monoclonal IgG2a, clone K28/43, TC Supernatant, RRID:AB_2877189), anti-Munc13-1 (1:1000, Mouse monoclonal IgG1, clone 11B-10G, Sakamoto et al., 2018), and anti-Cav2.1 (1:400, Synaptic Systems, rabbit polyclonal, RRID:AB_2619841) or anti-Cav2.2 (1:400, Synaptic Systems, Mouse monoclonal IgG2b, clone 163E3, RRID:AB_2619843) or anti-RIM-BP2 (1:200, Mouse monoclonal IgG2b, clone 8-4G, Sakamoto et al., 2022) for 3 hours at room temperature. Afterwards, sections were washed and incubated with fluorescence dye labeled secondary antibodies: DyLight 405 Anti-Guinea Pig IgG (Jackson ImmunoResearch, RRID:AB_2340470), Alexa Fluor 488 (Thermo Fisher Scientific) labeled Anti-Mouse IgG2a (Jackson ImmunoResearch, RRID:AB_2338462), STAR635P (Abberior) labeled Anti-Mouse IgG1 (Jackson ImmunoResearch, RRID:AB_2338461), and Alexa Fluor 594 (Thermo Fisher Scientific) labeled Anti-Rabbit IgG (Jackson ImmunoResearch, RRID:AB_2340585) or Anti-Mouse IgG2b (Jackson ImmunoResearch, RRID:AB_2338463) for 1 hour at room temperature. Sections were post-fixed with 4% PFA in PBS and washed with PBS. Sections were mounted on coverslips using Prolong Glass Antifade Mountant (Thermo Fisher Scientific). The same number of sections were obtained from WT and RIM-BP2 KO brains and processed in parallel at the same experimental day (biological replicate).

STED imaging

STED imaging was performed using TCS SP8 STED 3x microscope (Leica) equipped with a 405 nm diode laser, a pulsed white light laser (WLL), a continuous 592 nm STED laser for alignment, a pulsed 775 nm STED laser, HyD detectors, and a 100× oil-immersion objective lens (NA = 1.4). STED images were acquired using the Leica LAS-X software with an image format of 1024×1024 pixels, 16-bit sampling, 8-line accumulations, and 11.36-zoom factor, yielding a pixel size of 10 nm. HyD detectors were configured to counting mode with a gating from 0.5 to 6.5 nanosecond. A 405 nm diode laser was used to excite DyLight405. Alexa488, Alexa594, and STAR635P were excited using WLL at 488 nm, 561 nm, and 633 nm, respectively. The 775 nm STED laser power was set to 75% and 100% of maximum power for depletion of STAR635P and Alexa594, respectively, and delay time was set to 300 picoseconds. STED imaging was performed on thin sections obtained from four different WT and RIM-BP2 KO mice (biological replicate). Images were acquired three times from each section (technical replicate).

Analysis

For electrophysiological experiments, the data were analyzed by Igor Pro (WaveMetrics) or Excel (Microsoft Corp). To determine the time constant of release, an exponential fit was applied. Values are given as mean ± SEM, and n indicates the number of recorded terminals. Statistical method of sample size determination was not done, but our sample sizes are similar to those of previous studies (Hallermann et al., 2003, Midorikawa and Sakaba, 2017).

For STED imaging, data were analyzed using custom-designed programs in Mathematica (Wolfram), values were processed by Igor Pro (WaveMetrics), and pictures were created with ImageJ (NIH). To determine the area of AZs, image masks were generated from STED images of Munc13-1 by unsharp masking and image binarization. Only AZs that show co-localization with VGLUT1 and PSD-95 immunofluorescence signals were included in the analysis. To confine the analysis to the large mossy fiber boutons, AZs in presynaptic terminals of small VGLUT1-positive area were excluded. Then, the background subtracted integral of signal intensity of Cav2.1 or Cav2.2 was quantified using the image masks of AZs. The background signal intensity was estimated from non-AZ area in the same STED images. For quantification of Cav2.1 clusters in the AZs, STED images of Cav2.1 were deconvolved using a gaussian kernel with a radius of 40 nm. Image masks of each Ca²⁺ channel were generated from deconvolved STED images by unsharp masking and image binarization. Then, the total signal intensity, the size, and the number of Cav2.1 clusters at the AZs were quantified using the image masks of Cav2.1 clusters. Custom-designed programs in Mathematica (Wolfram) are available from Source code 1. Values are given as mean ± SEM, and n indicates the number of animals. Statistical method of sample size determination was not done, but our sample sizes are similar to those of previous studies (Brockmann et al., 2019).

Experiments were not fully randomized and blinded.

Statistical analysis was done in MATLAB (The MathWorks). To consider the data variability among terminals, Wilcoxon rank sum test was used for significance test of two groups. Significance level was set at α = 0.05, and p values were described in the main text or figure legends.

Data availability

All data generated in this study are included in the main text or shown as scatter plots in each graph. Numerical values of graphs are provided in Source data files. The custom-written code files in Mathematica are uploaded as Source code file.

Acknowledgements

We thank Stephan Sigrist, Dietmar Schmitz and Christian Rosenmund for kindly providing us RIM-BP2 KO mice. We also thank Stephan Sigrist for comments on the manuscript and The IRCN Imaging Core, The University of Tokyo Institutes for Advanced Studies, for the use of STED microscopy and for assistance. This works has been supported by JSPS KAKENHI (20J20550 to RM, 20KK0171, 21K15183 to HS, 20H03427 to KH, 20KK0171, 21H02598 to TS), JSPS core-to-core program A, Advanced Research Networks (JPJSCCA2022007 to TS) and JST, PRESTO (JPMJPR21E7 to HS).

Competing interests

None