Superpriming of synaptic vesicles after their recruitment to the readily releasable pool

Significance

During sustained nerve activity, synapses must continuously recycle vesicles. We used the unique opportunities for quantitative analysis offered by the calyx of Held synapse to study late stages in the process that renders vesicles release-ready. We dissect two sequential steps with distinct pharmacology and kinetics, the characterization of which is essential for an understanding of molecular mechanisms of transmitter release and short-term plasticity.

Abstract

Recruitment of release-competent vesicles during sustained synaptic activity is one of the major factors governing short-term plasticity. During bursts of synaptic activity, vesicles are recruited to a fast-releasing pool from a reluctant vesicle pool through an actin-dependent mechanism. We now show that newly recruited vesicles in the fast-releasing pool do not respond at full speed to a strong Ca²⁺ stimulus, but require approximately 4 s to mature to a “superprimed” state. Superpriming was found to be altered by agents that modulate the function of unc13 homolog proteins (Munc13s), but not by calmodulin inhibitors or actin-disrupting agents. These findings indicate that recruitment and superpriming of vesicles are regulated by separate mechanisms, which require integrity of the cytoskeleton and activation of Munc13s, respectively. We propose that refilling of the fast-releasing vesicle pool proceeds in two steps, rapid actin-dependent “positional priming,” which brings vesicles closer to Ca²⁺ sources, followed by slower superpriming, which enhances the Ca²⁺ sensitivity of primed vesicles.

The release rate of a synaptic vesicle (SV) is governed by two factors, the intrinsic Ca²⁺ sensitivity of the vesicle fusion machinery and the distance of the SV to Ca²⁺ channels. As Munc13s and Munc18s confer fusion competence on a docked SV, the regulation of release rate by Munc13s and Munc18s is called “molecular priming” (1). It is distinguished from “positional priming,” a process that is thought to regulate the proximity of an SV to the calcium source (2, 3). However, it is not known how these two priming mechanisms are manifested in the kinetics of quantal release. Deconvolution analyses of excitatory postsynaptic currents (EPSCs) evoked by long presynaptic depolarizations at the calyx of Held (a giant nerve terminal in the auditory pathway) showed that releasable SVs can be separated into fast-releasing pools (FRPs) and slowly releasing pools (SRPs) (4). The differences in SV priming that underlie the differences in release kinetics between SVs in the FRP and the SRP are currently unclear (3, 5). Wadel et al. (3) found that SVs in the SRP can be released by homogenous Ca²⁺ elevation only 1.5 to 2 times slower than SVs in the FRP, even though they are released 10 times slower by depolarization-induced Ca²⁺ influx. This was interpreted as evidence that the differences in their release kinetics arise from differences primarily in positional priming. In contrast, Wölfel et al. (5) showed that release with two kinetic components is even observed if the intracellular Ca²⁺ concentration is homogenously elevated throughout the calyx terminal, indicating that SVs in the FRP and the SRP differ with regard to their molecular priming.

We found recently that SVs in the SRP rapidly convert into the FRP after specific FRP depletion by a short depolarizing pulse (6). Such rapid refilling of the FRP with SRP vesicles, which is referred to as SRP-dependent recovery (SDR), was suppressed by actin depolymerization or inhibition of myosin, implying that SDR involves a transport process, steering docked and partially primed vesicle toward Ca²⁺ channels. In the same study, we noted that the time constant of release τ from newly primed FRP SVs after FRP depletion is initially slower than the time constant of FRP release τ under resting conditions. This finding is in agreement with the previously published notion that the Ca²⁺-sensitivity of SVs after a specific depletion of the FRP is 1.5 to 2 times lower than that of SVs under control conditions (3, 7). Thus, an additional SV maturation process, which is closely related to the Ca²⁺-sensitivity of vesicle fusion, appears to be required for newly primed FRP SVs to acquire full release competence.

In the present study, we characterize this maturation step, which we refer to as “superpriming“ (see also ref. 8). We show that the mechanism regulating recovery of Ca²⁺ sensitivity is distinct from that regulating recovery of the FRP size, in that the former and the latter require activation of Munc13s and the integrity of the cytoskeleton, respectively. The Ca²⁺ sensitivity is known to be profoundly affected by phorbol esters, which lower the energy barrier for vesicle fusion (9, 10). Munc13 has been identified as a presynaptic receptor of phorbol esters together with PKC (11–13). We therefore propose that the recovery of Ca²⁺ sensitivity represents a final step in the maturation of the intrinsic properties of newly recruited SVs involving Munc13 proteins, whereas the FRP size represents the number of release-competent SVs close to Ca²⁺ sources.

Results

By using dual whole-cell patch-clamp recordings on the pre- and postsynaptic compartments of calyx of Held synapses, we studied EPSCs induced by applying long depolarizing pulses to calyx terminals. The quantal release rate was estimated from EPSCs by using the deconvolution method (14). For better separation of the FRP and SRP, 0.5 mM EGTA was included in the presynaptic pipette solution (4). To prevent saturation and desensitization of AMPA-receptor currents, cyclothiazide, and γ-D-glutamylglycine were included in the bath solution.

The FRP Size and Its Release Rate Are Regulated by Distinct Mechanisms.

We studied the recovery time courses of the FRP size and the rate at which it is rereleased after various degrees of depletion induced by depolarizing pulses. The depleting stimulus was composed of two steps (Fig. 1A). A first depolarization of 2 ms length (to fully open Ca²⁺ channels) was followed by episodes lasting 3 ms or 10 ms or 30 ms [denoted as predepleting pulses (preDPs) preDP3, preDP10, and preDP30, respectively, and shown as broken lines in Fig. 1A; see also Table S1 ]. We showed previously that the preDP3 completely depletes the FRP while releasing very few SRP SVs (6). The preDP10 depletes the SRP and the FRP, and the preDP30 induces Ca²⁺-dependent pool recovery, as shown previously. To study the size and Ca²⁺ sensitivity of the recovered FRP, a second depleting pulse (0 mV for 30 ms) was applied at a fixed interstimulus interval (ISI) of 750 ms. Fig. 1C shows the averaged traces of second EPSCs (EPSC₂s) superimposed on the corresponding first EPSCs (EPSC₁s) for the three cases, preDP3, preDP10, and preDP30 (Fig. 1C, Left, Center, and Right, respectively). In agreement with Lee et al. (6), the amplitude of the recovered response (solid trace, Fig. 1C) is smallest for the preDP10 and larger for the preDP3 and preDP30. A dotted horizontal line in each of the panels of Fig. 1C indicates the level of the preDP3 response. The extra recovery relative to the preDP10 case was called SDR for the preDP3 case and Ca²⁺-dependent recovery (CDR) for the preDP30 case by Lee et al. (6) because these components depend on an intact SRP and on a Ca²⁺/calmodulin (CaM)-dependent mechanism, respectively.

Fig. 1.

FRP size and its release synchronicity are regulated by distinct mechanisms. (A) Paired pulse protocol. The first pulse (broken line) and the second pulse (solid line) are superimposed. Interstimulus interval (ISI) = 750 ms. (B) Corresponding average traces of Ca²⁺ currents. (C–E) Averaged EPSC₁ (broken line) and EPSC₂ (solid line) evoked by a paired-pulse protocol with different lengths of preDPL (columns) and under different presynaptic conditions [C, in the presence of 1/1,000 DMSO as a control; D, 20 µM CaMip (red); E, 20 µM latrunculin B (LatB, blue)]. A green dotted horizontal line in each panel indicates the mean amplitude of EPSC₂ after a preDP3. (Insets) EPSC₁ and EPSC₂ scaled to the same peak for comparison of their time courses. The SE range of averaged traces is depicted by shading of the traces with a light color.

In addition, we illustrate (Fig. 1C, Insets) that the recovered EPSCs of the three cases not only differ in their amplitude but also in their time course. Throughout this study, we will compare the responses after depletion by prepulses of different lengths (preDP3, preDP10, and preDP30), as they report on distinct properties of SDR and CDR. To compare time courses, the paired EPSCs were scaled to the same peak (Fig. 1C, Insets). As evident from Fig. 1C, there are marked differences in the times to peak of the EPSC₂s. They are prolonged relative to those of EPSC₁ for preDP3 and preDP10, whereas they are quite similar for preDP30. This indicates that prolonged depolarization during pool depletion accelerates subsequent maturation of recovered SVs. The time to peak of the EPSC reflects the synchronicity of FRP release. For quantitative analysis, we deconvolved EPSC traces such as those in Fig. 1C and integrated the resulting time courses of quantal release to calculate cumulative release (Fig. S1). We then fitted double exponentials to the cumulative release plots, which, in agreement with previous work (15), were interpreted as release from two pools (the SRP and the FRP). Here, we use the parameters of such fits to describe time courses of pool recovery, namely the ratio of the amplitudes of the fast component of preDP and test pulses (denoted as FRP₂/FRP₁) as a measure for the relative amount of recovered FRP size and the ratio of fast time constants (denoted as τ_fast,2/τ_fast,1 or τ-ratio) as a measure of the Ca²⁺ sensitivity of the recovered FRP. Absolute values of parameters are given in Fig. S2. After a preDP3, the τ_fast of EPSC₂ (τ_fast,2) was slower than that of EPSC₁ (τ_fast,1; τ_fast,2/τ_fast,1, 1.69 ± 0.06; n = 16). As the length of the preDP (preDPL) increased, the fast time constant of EPSC₂ was accelerated despite the finding that the amplitude of Ca²⁺ currents induced by a DP30 was slightly reduced (Fig. 1B). The time constant almost caught up with that of EPSC₁ (τ_fast,1) when the preDPL was increased to 30 ms (τ-ratios, 1.54 ± 0.07 after preDP10; 1.16 ± 0.02 after a preDP30; n = 10; Fig. 1C).

Fig. 1 D and E show the effects of a CaM inhibitory peptide (CaMip) and of latrunculin B, a cytoskeleton disruptor. Each panel in Fig. 1 D and E shows averaged EPSC₁ (broken line) and EPSC₂ (solid line) evoked by a dual pulse protocol with different preDPLs (columns) and under different presynaptic conditions (rows). Control traces without drugs are shown in black. In agreement with previous reports (6, 16), latrunculin B (15 μM; n = 7) inhibited CDR and SDR, and CaMip (20 μM; n = 7) abolished CDR (Fig. 1D). Considering times to peak, however, a very different pattern was observed. Neither drug changed the rise times in any major way at the chosen ISI of 750 ms. This indicates that the mechanism regulating the τ_fast recovery (i.e., superpriming) is distinct from that of recruiting vesicles via SDR or CDR.

Distinct Recovery Time Courses of the Size and Release Time Constant of FRP.

Fig. 1 shows SV pool recoveries after a fixed time interval (ISI, 750 ms). We used a paired-pulse protocol with various ISIs (0.2, 0.5, 1, 2, 5, and 10 s) to explore in detail the recovery time courses of the FRP size and τ_fast after a preDP3 or a preDP30 (Fig. 2, Left, shows protocols used). The τ-ratio at the shortest ISI (200 ms) after a preDP3 was 1.8 ± 0.17 (n = 7), reminiscent of the previous result that SRP vesicles have 1.5 to twofold lower Ca²⁺ sensitivity (3). Consistent with Fig. 1, latrunculin B had no effect on the recovery of τ_fast, whereas it retarded the recovery of the FRP size after depletion by a preDP3 (Fig. 2A). Similarly, after a preDP30, latrunculin B and calmidazolium (CMZ), a CaM inhibitor, had no effect on the τ_fast recovery, whereas they slowed down the recovery of the FRP size (Fig. 2B). Blebbistatin, a myosin II inhibitor that abolishes CDR and SDR like latrunculin B (6), retarded the FRP size recovery after a preDP30, but had no significant effect on the recovery of τ_fast. In Fig. 2C, we compare the recovery time courses of the FRP size and τ_fast after a preDP3 with those after a preDP30 under control conditions. Recovery time courses of τ_fast were significantly faster after a preDP30 than after a preDP3 (Fig. 2C, Right), although the recovery time courses of FRP sizes were rather comparable between the two cases (Fig. 2C, Left). The different recovery time courses further support the notion that the recovery of τ_fast and FRP size are regulated by distinct mechanisms. In summary, 30 ms predepolarization accelerates superpriming, which is not affected by drugs that retard the recovery of SV pool sizes.

Fig. 2.

Recovery time courses of the FRP size and its release time constant (τ) after a preDP3 or preDP30. (A) Recovery time courses of the FRP size (Center) and release τ of the FRP (τ_fast; Right) after a preDP3 in the presence of 1/1,000 DMSO (control, open triangles) and latrunculin B (filled circles). (B) Recovery time course of the FRP size and τ_fast after a preDP30. (C) Recovery time courses after a preDP3 (brown open triangles) and preDP30 (black, open circles) under control conditions are compared. The recovery time courses of τ_fast were fitted with monoexponential functions (dotted lines; recovery time constants, 0.52 s after a preDP30 and 2.74 s after a preDP3). Note that both τ_fast recovery time courses display very slow components, which were not taken into account by the monoexponential fit.

The Acceleration of τ_fast Recovery May Be Mediated by Activation of Phospholipase C.

The black symbols in Fig. 3B summarize the aforementioned findings that longer prepulse durations are associated with faster recovery of τ_fast, resulting in a monotonous dependence of τ_fast recovery on the prepulse duration. Such dependence indicates that Ca²⁺-dependent mechanisms may facilitate the recovery of τ_fast. Thus, we tested the possibility that acceleration of τ_fast recovery is mediated by Ca²⁺-induced activation of phospholipase C (PLC), which activates Munc13s, which are essential mediators of molecular priming (10, 12, 17). Inclusion of U73122 (10 µM), a PLC inhibitor, in the presynaptic pipette had no effect on the recovery of FRP size after preDP3 (P = 0.48) and preDP10 (P = 0.27; n = 12; Table S1), and partially suppressed it after a preDP30 (42.1 ± 1.9%; n = 12; P < 0.01; Fig. 3 A and B, red symbols). However, U73122 had rather pronounced inhibitory effects on the recovery of τ_fast at longer preDPLs, resulting in weaker dependence of τ_fast recovery on the preDPL (Fig. 3 A and B, 3, red symbols). Similar to U73122, edelfosine, a phosphoinositide-specific PLC inhibitor, significantly retarded the τ_fast recovery at the preDP30 with smaller effects at shorter preDPs (τ-ratio, 1.42 ± 0.07 at preDP30; n = 6; P < 0.01; Fig. 3 B, 3, and Fig. S3), and inhibited the FRP size recovery only after a preDP30 (41.6 ± 3.0%; n = 6; P < 0.01; Fig. 3 B, 2). Neither the recovery of τ_fast nor the recovery of the FRP size were affected by presynaptic application of U73343 (10 µM), an inactive analogue of U73122 (Fig. S3). The ratio of Ca²⁺ current amplitudes (I_Ca,2/I_Ca,1) was not significantly altered by these drugs (Fig. 3 B, 1). These results indicate that activation of PLC contributes to recovery time courses of τ_fast and FRP size after a preDP30.

Fig. 3.

Inhibition of PLC retards superpriming of newly recruited FRP-SVs after a strong prepulse. (A) Averaged traces of EPSC₁ (broken line) and EPSC₂ (solid line) evoked by a dual pulse protocol (as shown in Fig. 1) with different preDPLs (Left, 3 ms; Center, 10 ms; Right, 30 ms) in the presence of U73122 (red). EPSCs were normalized to the peak amplitude of the EPSC₁. EPSC₁ and EPSC₂ are superimposed. The SE range of averaged traces is depicted by shading of traces with a light color. (B) The ratio of the second to the first presynaptic Ca²⁺ current amplitude (I_Ca,2/I_Ca,1, 1), the fraction of the FRP size (FRP₂/FRP₁, 2), and the release time constants (τ_fast) of FRPs (τ_fast,2/τ_fast,1, 3) as a function of preDPL (1 and 3) or the fraction of SRP released by the first pulse (2). (C) The second-to-first ratio of the presynaptic Ca²⁺ current amplitude (1), the FRP size (2), and release time constant (τ_fast) of FRP (3) as a function of ISI (0.2, 0.5, 1, 2, 5, or 10 s) after a preDP30. (B and C) Black, red, and green symbols represent values under control conditions and in the presence of U73122 and edelfosine, respectively. Values in the presence of CMZ (light blue symbols) are shown for comparison in C, 2 and 3. Broken lines in C, 3, show the recovery time courses of τ_fast after preDP3 (open circles) and preDP10 (open squares). (Significant at *P < 0.05 and **P < 0.01, control vs. U73122 conditions.)

The data in Fig. 3C extend the analysis of the effects of U73122 on the recovery time courses of the FRP size and τ_fast after depletion of SVs by a preDP30 using a protocol similar to that shown in Fig. 2. We found that U73122 significantly retarded the FRP size recovery and the τ_fast recovery. In Fig. 3C, we compare the effects of CMZ and U73122 on the time courses of the FRP size and τ_fast recovery. Unlike CMZ, U73122 significantly retarded the τ_fast recovery (recovery time constants, 0.52 s for control and 2.0 s for U73122), and somewhat retarded the FRP size recovery. It should be noted, however, that the τ_fast recovery time course after a preDP30 was still faster than recovery time courses after a preDP3 or a preDP10 even under conditions of PLC inhibition (Fig. 3C, 3), indicating that high [Ca²⁺] elevation alone without activation of PLC can make a partial but significant contribution to the acceleration of superpriming.

1-Oleoyl-2-Acetyl-sn-Glycerol Accelerates the Recovery of τ_fast After a preDP3 but Not After a preDP10.

The results described here earlier indicate that a strong depolarization of the calyx of Held activates PLC, and that subsequent production of diacylglycerol (DAG) may accelerate the recovery of τ_fast after a preDP30. Bath-applied 1-oleoyl-2-acetyl-sn-glycerol (OAG), a DAG variant, enhanced both the baseline FRP size and its release rate, with no significant effect on the SRP (Fig. S4). Applying OAG (20 μM) via the presynaptic pipette, we tested whether OAG can accelerate the recovery of τ_fast after a preDP3 or a preDP10, and found that OAG had little effect on the recovered FRP size at 750 ms for all preDPLs (Fig. 4 A and C, 2). In contrast, OAG significantly accelerated τ_fast of the recovered FRP after a preDP3 [τ-ratio, 1.27 ± 0.03 (n = 6) vs. 1.69 ± 0.06 (n = 16); P < 0.01; Fig. 4 A and C, 3]. Intriguingly, however, OAG had little effect on τ_fast after a preDP10 and a preDP30 (Fig. 4 A and C, 3, and Table S1). Although the effect of OAG may be occluded by Ca²⁺-dependent PLC activation at the preDP30, the near-absence of an OAG effect on τ_fast after a preDP10 was surprising. Because SDR contributes to the FRP size recovery after a preDP3 but not after a preDP10 (6), this result indicates that OAG can facilitate the superpriming of FRP vesicles recruited from the SRP, but not those newly recruited from an “unprimed” recycling pool at this short ISI (750 ms). To confirm this idea, we examined whether the effect of OAG on τ_fast after a preDP3 depends on SDR. As expected, latrunculin B, which blocks SDR, abolished the effect of OAG on τ_fast after a preDP3 (Fig. 4B). These results indicate that the effect of OAG on the τ_fast recovery at an ISI of 750 ms is selective for SVs recruited from the SRP and that OAG can superprime SVs of the SRP, at least partially.

Fig. 4.

OAG accelerates release τ of recovered FRP after a preDP3. (A) Averaged traces of the EPSC₁ (broken line) and EPSC₂ (solid line) evoked by a dual-pulse protocol (as shown in Fig. 1) with different preDPLs (Left, 3 ms; Center, 10 ms; Right, 30 ms) in the presence of OAG (20 µM; red). EPSCs were normalized to the peak amplitude of the EPSC₁. EPSC₁ and EPSC₂ are superimposed. The SE range of averaged traces is depicted by shading of the traces with a light color. (B) Same as in A except that OAG and latrunculin B were added to the presynaptic patch pipette (OAG + LatB; blue). (C) Summary of ratios (2nd over 1st) of presynaptic Ca²⁺ current amplitude (C1), FRP size (C2), and FRP release time constant (τ_fast, C3) as functions of preDPLs (C1 and C3) or the SRP fraction released by the 1st pulse (C2) (black, control; red, OAG; blue, OAG + latrunculin B).

Next, we tested whether OAG has any effect on the τ_fast recovery after a preDP10 at longer ISIs. OAG accelerated the τ_fast recovery after a preDP10 at ISIs longer than 1 s (Fig. 5B). This finding is in contrast to the effect of OAG on the τ_fast recovery after a preDP3. For a preDP3, OAG accelerated τ_fast at the very first ISI (200 ms; Fig. 5A). These results indicate that the effect of OAG on τ_fast requires a longer time for SVs that are not recruited from the SRP via SDR but rather from a recycling pool (SI Discussion). This idea may explain the reason for the differential effects of OAG on τ_fast after a preDP3 and a preDP10 at a short ISI (750 ms).

Fig. 5.

The second-to-first ratio of the presynaptic Ca²⁺ current amplitude (Top), FRP size (Middle), and τ_fast (Bottom) as a function of ISI (0.2, 0.5, 1, 2, 5, or 10 s) after a preDP3 (A) or a preDP10 (B). Recovery time courses under control (black) and in the presence of OAG (blue) are superimposed. The broken line in the A (Bottom) shows the τ_fast recovery after a preDP30 (from Fig. 2B). The control recovery time courses after a preDP3 are reproduced from Fig. 2A.

OAG had little effect on the FRP size recovery after a preDP3 (Fig. 5A), whereas it enhanced the recovery of the FRP size and τ_fast after a preDP10 (Fig. 5B). This effect of OAG on the recovery after a preDP10 is in line with the finding that U73122 affected the recovery of both parameters after a preDP30 (Fig. 3), and indicates that the τ_fast recovery may be partially linked to the FRP size recovery after full depletion of the SRP (Discussion).

In the presence of OAG, recovery of τ_fast was enhanced after a preDP3 but still slower than that after a preDP30 (Fig. 5A). This indicates that OAG alone may not be sufficient to accelerate recovery to the same degree as a preDP30, which leads to higher [Ca²⁺] levels during the recovery period. This finding is consistent with Fig. 3C, in which we show that the recovery time course of τ_fast after a preDP30 in the presence of U73122 is not as slow as that after a preDP3. These results imply that high [Ca²⁺] elevation induced by a preDP30 activates a PLC-independent mechanism, which accelerates superpriming together with a PLC-dependent pathway.

Contributions of PLC-Dependent and -Independent Mechanisms to Superpriming Are Mutually Occlusive.

The incomplete effects of U73122 and OAG on the τ_fast recovery after preDP30 and preDP3, respectively, indicate that Ca²⁺ has dual roles in superpriming. To explore whether the PLC-dependent and -independent components display different Ca²⁺-sensitivities, we tested the effect of U73122 under conditions of reduced strength of the Ca²⁺ stimulus during the prepulse. To do so at a fixed duration of 30 ms, we changed the level of depolarization from 0 mV to +30 mV (denoted as “preDP30/30mV”). The Ca²⁺ influx induced by such a pulse was one third (Fig. 6A) of that induced by a 0-mV step pulse (“preDP30/0mV”). It was rather similar to that elicited by a preDP10 (Fig. S5 B, 1), implying that global [Ca²⁺] elevation is similar between preDP30/30mV and preDP10. Nevertheless, the τ_fast recovery at 750 ms after a preDP30/30mV under control conditions was more advanced than after a preDP10, and rather similar to that after preDP30/0mV (n = 6; Fig. 6B). In the presence of U73122, however, the τ-ratio after a preDP30/30mV reported significantly slower recovery than that after a preDP30/0mV (1.78 ± 0.12; n = 7; P = 0.027) and was similar to the τ-ratio estimates after a preDP3 (P = 0.52; Fig. 6C). In summary (Fig. 6C and Table S1), the effect of U73122 on the τ-ratio after a preDP30/30mV (Fig. 6C) is much stronger than that following a preDP30/0mV. These results indicate that the τ_fast recovery after a weak Ca²⁺ stimulus (preDP30/30mV) can mostly be ascribed to the activation of PLC, whereas that after a strong one (preDP30/0mV) depends on cooperative but partially mutually occlusive actions of PLC-dependent and PLC-independent mechanisms.

Fig. 6.

(A) 1, Paired-pulse protocol for estimation of τ_fast recovery at 750 ms after 30-ms depolarizing voltage steps to 0 mV (first row, preDP30/0mV), the resultant presynaptic Ca²⁺ currents (second row, averaged) and EPSCs under control condition (black, third row, averaged) and in the presence of U73122 (red, fourth row, averaged). EPSC₁ (Left, dotted line) and EPSC₂ (Right, solid line) were normalized to the peak amplitude of EPSC₁. (Right, Bottom) Averaged traces of EPSC₁ and EPSC₂ scaled to the same peak for comparison of time courses. 2, Paired pulse protocol to estimate recovery of τ_fast at 750 ms after a 30-ms depolarizing voltage step to +30 mV instead of 0 mV (preDP30/30mV); same cell pair as in 1. (Right, Bottom) Comparison of times to peak of averaged traces of EPSC₁ in 1 and EPSC₂ in 2. For comparison, a normalized EPSC₁–EPSC₂ pair under control conditions after a preDP3 is shown in the bottom of 2 (black; reproduced from Fig. 1A). (B) Ratios of the τ_fast,2 over τ_fast,1 under the different prepulse conditions of A. (C) Summary of τ_fast recovery at 750 ms after a preDP3 or preDP30 (depolarizing step to 0 mV or 30 mV) under different conditions. The mean values for τ_fast under two conditions (ctrl/30mV and OAG/0mV) were not significantly different from control (Ctrl) values [ctrl/0mV, P value not significant (n.s.)]. Paired observations are connected by dotted lines. Asterisks indicate significant differences.

Discussion

The present study provides evidence for differential regulation of the number of fast releasing vesicles (FRP size) and their release rate by showing that the recovery time courses of the two parameters after depletion of the pool of fast releasing vesicles are distinct and differentially affected by the duration of the predepolarization, latrunculin B, CaM inhibitors, PLC inhibitors, and OAG (Figs. 2 and 5). The recovery of release rate (expressed as τ_fast) is primarily regulated by PLC-dependent mechanisms, whereas the FRP size recovery depends on actin- and CaM-mediated mechanisms. τ_fast, which characterizes the release rate of release-competent SVs, probably represents the last step in the stimulus-release chain, whereby a primed SV attains high Ca²⁺ sensitivity for fusion (superpriming). Therefore, recovery time courses of the FRP size and its τ_fast may represent two distinct processes that occur in sequence. Given that the proximity of SVs to the calcium source and the intrinsic Ca²⁺ sensitivity of SVs govern their release rate, our results imply that the recovered FRP size represents the number of recruited release-competent SVs close to calcium sources, whereas the τ_fast recovery represents a final step of superpriming whereby these SVs obtain the capability to be released at a full speed. Furthermore, our results imply that slowly releasing SVs, which are approximately as abundant at the calyx of Held as fast-releasing SVs, are not only remote from Ca²⁺ sources but also less advanced in superpriming.

The Recovery of τ_fast Has PLC-Dependent and PLC-Independent Components and May Involve Munc13s.

Three lines of evidence support the notion that Ca²⁺ has dual effects on the superpriming of FRP-SVs that are mediated by PLC-dependent and PLC-independent pathways. First, after inhibition of PLC (10 µM U73112), higher Ca²⁺ elevation (preDP30/0mV) still enhanced τ_fast recovery more than a smaller Ca²⁺ stimulus (preDP3; Fig. 6C). Second, after pharmacological activation of PLC (OAG, 20 μM), the same two Ca²⁺ stimuli also caused τ_fast recovery to different degrees (Figs. 4 C, 3, 5A, and 6C). Third, in the presence of U73122 or OAG, the τ_fast recovery after a preDP30/30mV, which induces milder [Ca²⁺] elevation, was not different from that after a preDP3 (Fig. 6C). All inhibitor drugs tested in the present study were included in the presynaptic patch pipette at a supramaximal dose. However, the dose of OAG required to elicit maximal effects on PLCs in cells is not known. Therefore, the dose of OAG we used (Figs. 4, 5, and 6C) may have been submaximal, which may have contributed to the different effects of preDP30/0mV and preDP3 in the presence of OAG.

It should be noted that the difference in τ-ratio between control and U73122 conditions after a preDP30/30mV is significantly higher than that after a preDP30/0mV, indicating that the activation of PLC makes a bigger contribution to the τ_fast recovery when the [Ca²⁺] elevation is less pronounced (Fig. 6C). Given that the contributions of PLC-dependent and -independent mechanisms to superpriming are partially mutually occlusive, we propose that these two mechanisms converge on the same regulatory protein or process. Munc13s are the only priming proteins with regulatory domains that sense Ca²⁺ and DAG (11, 12, 18, 19). Therefore, our results indicate that the recovery of τ_fast is controlled by the activity of Munc13s, and support the notion that molecular priming mechanisms (i.e., superpriming) are responsible for the recovery of τ_fast.

Munc13 is thought to act by converting closed syntaxin into an open form of a Munc18/syntaxin complex, thus promoting subsequent SNARE complex formation (20). Binding of DAG to the C1 domain and of Ca²⁺ and phospholipids to the C2B domain of Munc13s mediate membrane binding of Munc13s and/or their activation (11, 18). Recruitment of more Munc13 molecules to the membrane may accelerate the time required to saturate the number of SNARE complexes that can assemble around a single SV. Because the τ_fast recovery depends on the activation of PLC and is accelerated by OAG, we propose that an increase in the number of SNARE complexes assembled per SV, which might be increased upon higher Munc13 activity, may become functionally manifest as an accelerated recovery of τ_fast, which we refer to as superpriming. Alternatively, a conformational change within Munc13s, induced by the modulators, may underlie superpriming. This possibility is supported by recent studies, which show that mutations in the regulatory domains of Munc13-1 enhance the baseline release probability of SVs (9, 21).

CaM-Dependent and PLC-Dependent Roles of Munc13.

CaM inhibitors specifically affect CDR (6, 16) and have little effect on SDR and the recovery of τ_fast (Fig. 2B). Similar to CaM inhibitors, perturbations of proteins involved in endocytosis have a specific effect on CDR, implying that CaM-dependent CDR is closely related to clearing refractory release sites (22). Recently, a knock-in mouse line was established that harbors a CaM-insensitive mutant of Munc13-1 (21). It was shown that recovery of the FRP after prolonged depolarization is slowed down in calyces of such mice, mimicking block of CDR. In contrast, a gain-of-function mutation of the C2B domain of ubMunc13-2 increases vesicular release probability (18). These reports imply that the interaction of DAG and Ca²⁺ with the C1 and C2B domains of Munc13s may have preferential effects on superpriming, whereas the Munc13–CaM interaction is one of the prerequisites for CDR.

Dependence of Superpriming on the SV Positions.

The present study and previous reports by Wadel et al. (3) and Müller et al. (7) show that primed SVs just recruited from SRP after a predepolarization are somewhat less Ca²⁺-sensitive than FRP SVs at steady state. Recently, it has been shown that activation of Munc13 requires its interaction with RIM, which renders the MUN domain of Munc13 to be exposed (23, 24). Rab3-interacting molecule (RIM) interacts with Ca²⁺ channels, and thus may be closely associated with them in the active zone. Given that activation of Munc13 requires its interaction with RIM, available Munc13s may be more concentrated in the vicinity of the calcium source than at the periphery. Our finding supports the notion that full maturation of FRP-SVs with respect to their Ca²⁺ sensitivity requires interaction of Munc13s with RIM (which is associated with Ca²⁺ channels), and may then be taken as an indication that positional priming is a prerequisite for the full maturation of intrinsic Ca²⁺ sensitivity (or superpriming) of a SV. This hypothesis may reconcile the dispute regarding the primary factor that determines the FRP: The proximity to the calcium source or the intrinsic Ca²⁺ sensitivity (3, 5). Our finding that SVs newly recruited from the SRP are more mature in the presence OAG (Fig. 5) may then indicate that OAG binding to Munc13s partially substitutes for the interaction with RIM.

Discrete Pools or a Continuum of States?

So far, we have discussed our results in terms of two discrete SV pools: FRP and SRP. The basis for that is the relative ease of fitting cumulative release with two exponentials. We are aware, however, that a variety of assumptions about SV populations may result in satisfactory fits by two exponentials. In particular, SRP SVs, which we assume to be more remote from Ca²⁺ channels, may be located at variable distances, some of them contributing to the slow and the fast components of the fit. Under these assumptions, it may be understood why OAG and U73122 have differential effects on the FRP size recovery depending on the prepulse duration. If the Ca²⁺ sensitivity of vesicle fusion is increased by superpriming, SVs that reside at the borderline between pools will be released with a faster release time constant, and thus may be counted as FRP SVs. Such “spillover” may happen in cases when SRP vesicles are partially superprimed by OAG and may explain the small effects of OAG and U73122 on the recovery of the FRP size (Figs. 3 C, 2, and 5B). This idea is in line with the enhancing effect of OAG on the baseline FRP size (Fig. S4).

General Implications for Short-Term Plasticity.

Short-term plasticity is essential for understanding the computation in a defined neural network (25). Analysis of the priming steps associated with refilling of the FRP at mammalian glutamatergic synapses has not been trivial because release-competent SVs are heterogeneous in release probability and their recovery kinetics (26, 27). The present study indicates that such SVs are fully matured only when they are positioned close to the Ca²⁺ source. We demonstrate that the time course for such full maturation or superpriming of newcomer SVs is slower (τ = 3.6 s) than that of cytoskeleton-dependent conversion of reluctant SVs into FRP SVs (τ = 60 ms) (6). Therefore, we propose a two-step model for refilling of the FRP: rapid “positional priming,” which brings vesicles closer to Ca²⁺ sources, followed by slower superpriming, which enhances the Ca²⁺ sensitivity of vesicles. Given that the presence of reluctant SVs is a common property of small glutamatergic synapses and calyx of Held synapses, our two-step model for refilling of the FRP may provide a general scheme for characterizing a variety of short-term plasticity features that have been experimentally observed in such synapses.

Materials and Methods

SI Materials and Methods provides further details of experimental procedures. Transverse brainstem slices containing the medial nucleus of trapezoid body were prepared from 7- to 9-d-old Sprague–Dawley rats. Pre- and postsynaptic compartments of a calyx of Held synapse were simultaneously whole-cell patch-clamped at −80 mV and −70 mV, respectively, at room temperature. EPSCs were recorded in the artificial cerebrospinal fluid, to which 1 µM tetrodotoxin, 50 µM D(-)-2-amino-5-phosphonovalerate, 10 mM tetraethylammonium-Cl, 100 µM cyclothiazide and 2 mM γ-D-glutamylglycine were added. To induce square-like presynaptic calcium currents, a presynaptic depolarizing pulse was comprised of depolarization to 0 mV preceded by predepolarizations to +70 mV for 2 ms. The duration of a presynaptic depolarizing pulse is defined by the duration of the 0-mV step. Quantal release rates were estimated by using a deconvolution method developed by Neher and Sakaba (14). Statistical data are expressed as mean ± SEM, with statistical significance determined at a threshold P value of 0.05 or 0.01.