Dynamic Modulation of Phasic and Asynchronous Glutamate Release in Hippocampal Synapses

Abstract

Although frequency-dependent short-term presynaptic plasticity has been of long-standing interest, most studies have emphasized modulation of the synchronous, phasic component of transmitter release, most evident with a single or a few presynaptic stimuli. Asynchronous transmitter release, vesicle fusion not closely time locked to presynaptic action potentials, can also be prominent under certain conditions, including repetitive stimulation. Asynchrony has often been attributed to residual Ca²⁺ buildup in the presynaptic terminal. We verified that a number of manipulations of Ca²⁺ handling and influx selectively alter asynchronous release relative to phasic transmitter release during action potential trains in cultured excitatory autaptic hippocampal neurons. To determine whether other manipulations of vesicle release probability also selectively modulate asynchrony, we probed the actions of one thoroughly studied modulator class whose actions on phasic versus asynchronous release have not been investigated. We examined the effects of the phorbol ester PDBu, which has protein kinase C (PKC) dependent and independent actions on presynaptic transmitter release. PDBu increased phasic and asynchronous release in parallel. However, while PKC inhibition had relatively minor inhibitory effects on PDBu potentiation of phasic and total release during action potential trains, PKC inhibition strongly reduced phorbol-potentiated asynchrony, through actions most evident late during stimulus trains. These results lend new insight into PKC-dependent and -independent effects on transmitter release and suggest the possibility of differential control of synchronous versus asynchronous vesicle release.

INTRODUCTION

Short-term, frequency-dependent modulation of transmitter release has been of interest for many decades. Primarily, studies have focused on modulation of phasic, synchronous synaptic transmitter release. However, asynchronous release can complement phasic release and contributes strongly to postsynaptic responses under certain conditions (Best and Regehr 2009; Hefft and Jonas 2005; Iremonger and Bains 2007; Lu and Trussell 2000; Taschenberger et al. 2005). Less is known about this form of release. Asynchrony is characterized by temporal dispersion of vesicle release after presynaptic action potential arrival and Ca²⁺ influx. The slow buildup and removal of Ca²⁺ in the presynaptic terminal by buffering and clearance mechanisms likely participate in asynchrony (Atluri and Regehr 1998; Barrett and Stevens 1972; Cummings et al. 1996; Goda and Stevens 1994). In addition, separate Ca²⁺ sensors for phasic and asynchronous release are possible (Geppert et al. 1994; Goda and Stevens 1994; Sun et al. 2007). Mutations, genetic deletions, and nonphysiological divalent ion substitution alter the relative proportion of phasic to asynchronous release (Calakos et al. 2004; Geppert et al. 1994; Pan et al. 2009; Rahamimoff and Yaari 1973; Tang et al. 2006), However, whether second messengers and synaptic modulators can alter the proportion of phasic to asynchronous release is less clear.

Hippocampal principal neurons in vivo experience wide range of firing frequencies partly dependent on the behavioral task (Czurkó et al. 1999; Hirase et al. 1999). During high-frequency activity, phasic transmitter release gradually depresses, partly resulting from vesicle depletion (Zucker and Regehr 2002). During the same repetitive stimulation, asynchronous vesicle release in hippocampal neurons becomes more prominent in the postysynaptic response and can carry most of the postsynaptic charge late in excitatory postsynaptic current (EPSC) trains (Cummings et al. 1996; Hagler and Goda 2001; Otsu et al. 2004). Modulation of the phasic to asynchrony ratio could have an important influence on the temporal relationship between the arrival of a presynaptic action potential and the corresponding postsynaptic spike (Iremonger and Bains 2007; Jones et al. 2007; Wyart et al. 2005). Strong phasic release will promote a postsynaptic spike soon after the presynaptic action potential; strong asynchronous release will introduce a delay and more temporal jitter between presynaptic and postsynaptic firing.

In this study, we examined the correlation between the depression of phasic release and the increase of asynchrony in evoked release during action potential trains delivered to autaptic synapses from dissociated hippocampal excitatory neurons. We tested the possibility of vesicle recycling in promoting asynchronous release and compared the effect of elevating release probability (P_r) by different manipulations, including elevated Ca²⁺ and phorbol ester stimulation, on phasic versus asynchronous transmitter release. Although a phorbol ester strongly promoted increased phasic release and asynchrony in parallel, protein kinase C (PKC) inhibition more strongly compromised the phorbol-potentiated asynchronous component. However, we found no evidence that PKC activation is involved in the asynchrony generated by augmented Ca²⁺ influx during action potential trains, suggesting that PKC activity is not needed for increases in asynchrony during strong Ca²⁺ influx.

METHODS

Materials

Unless otherwise specified, reagents were purchased from Sigma-Aldrich (St. Louis, MO).

Cultures

Microisland cultures were prepared as previously described (Mennerick et al. 1995; Moulder et al. 2007). Briefly, hippocampal neurons from Sprague-Dawley rats at postnatal days 1–3 were dissociated with 1 mg/ml papain. The dissociated neurons were seeded at ∼100 cells/mm² in 35-mm culture dishes that were precoated with 0.15% agarose and type I collagen (0.5 mg/ml) as the substrate. Plating media was composed of Eagle's minimal essential medium (Invitrogen, Carlsbad, CA) supplemented with 5% heat-inactivated horse serum, 5% FBS, 17 mM glucose, 400 μM glutamine, 50 U/ml penicillin, and 50 μg/ml streptomycin. Glial proliferation was inhibited by 6.7 μM cytosine arabinoside 3–4 days after plating. One half the culture media was removed and replaced with Neurobasal medium plus B27 supplement 4–5 days after plating. Cells were recorded 9−15 days after plating.

Solutions

Whole cell recordings were conducted in extracellular solution (bath) consisting of (in mM) 138 NaCl, 4 KCl, 2 CaCl₂, 1 MgCl₂, 10 glucose, and 10 HEPES (Invitrogen; pH 7.25). α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid (AMPA) receptor–mediated current was isolated with 25–50 μM d-amino-5-phosphonovaleric acid (d-APV, Tocris, Ellisville, MO) in bath solutions, and 400 μM kynurenate was added to minimize AMPA receptor saturation and to reduce access resistance errors caused by the large autaptic currents. We and others have previously shown that kynurenate shields receptors from saturation and desensitization, thus promoting a more accurate assessment of presynaptic release (Moulder and Mennerick 2005; Neher and Sakaba 2001; Otsu et al. 2004). ESPCs were quantified by subtracting traces obtained in the presence of 1–2 μM 2,3 dioxo-6-nitro-1,2,3,4 tetrahydrobenzo [f] quinoxaline-7-sulfonamide (NBQX, Tocris). Solution was perfused by a gravity-based multibarrel perfusion system at the rate of 0.2 ml/min. For application of hyperkalemic (45 mM KCl, equimolar substitution for NaCl) or hyperosmolaric (0.5–0.75 M sucrose) solution, the solution was perfused at 1.6 ml/min with <50 ms complete solution switch. The internal pipette solution contained (in mM) 140 K-gluconate, 4 NaCl, 0.5 CaCl₂, 1 EGTA, and 10 HEPES (pH 7.25). Stock solutions of ω-agatoxin IVA (1 mM) and ω-conotoxin GVIA (1 mM, Tocris) were dissolved in distilled water and diluted to the indicated concentrations. Stock solutions of EGTA-AM (100 mM, Invitrogen), folimycin (67 μM, Calbiochem, Gibbstown, NJ), Gö6983 (2 or 20 mM), and phorbol 12,13-dibutyrate (PDBu, 1 or 5 mM), phorbol 12-myristate 13-acetate (PMA, 5 mM) were made in DMSO and diluted as indicated.

Electrophysiology and data analysis

Whole cell recordings were performed with a MultiClamp 700B amplifier and Digidata 1440A acquisition system (Axon Instruments, Sunnyvale, CA); data were acquired in Clampex10 (Molecular Devices, Axon instruments). Electrode pipettes were pulled from borosilicate glass (World Precision Instruments, Sarasota, FL) with pipette resistance 3–6 MΩ. After whole cell mode was established, only the cells with access resistance <15 MΩ, membrane resistance >150 MΩ, and leak current <200 pA were accepted for analysis. Series resistance was compensated at 80%. Cells were voltage clamped at −70 mV, and EPSCs were evoked by a brief (1 ms) depolarization to 0 mV. Signals were sampled at 5 or 10 kHz and filtered at 2 kHz. All recordings were performed at room temperature. Leak current was subtracted off-line; data were analyzed by Clampfit10 (Molecular Devices, Axon Instruments) or by customized Igor Pro procedures (WaveMetrics. Lake Oswego, OR). Data fitting was performed by commercially available fitting routines (Clampfit and Igor Pro). Results are presented as means ± SE. Paired or unpaired t-test was used for statistic analysis; Pearson's test was used to evaluate the significance of correlation.

RESULTS

Asynchronous charge transfer in train-evoked release

We examined excitatory autaptic currents evoked by single or repetitive stimulation from island cultures of hippocampal neurons (Bekkers and Stevens 1991). To improve the accuracy of quantifying transmitter release, we recorded in the presence of a rapidly dissociating glutamate receptor antagonist (400 μM kynurenate) to reduce voltage-clamp errors associated with large currents and to shield postsynaptic receptors from saturation and desensitization (Jones and Westbrook 1996; Neher and Sakaba 2001). Under standard conditions, a single evoked AMPA receptor–mediated EPSC, caused by receptor deactivation kinetics and the brief transmitter concentration profile (Clements et al. 1992), had an average decay time constant (τ) of 2.83 ± 0.3 ms (mean amplitude = 3.5 ± 0.6 nA, n = 7). Repetitive high-frequency stimulation at 20 Hz for 2 s generated an EPSC waveform that exhibited progressive alteration during stimulation. As observed by others (Cummings et al. 1996; Hagler and Goda 2001; Otsu et al. 2004), the amplitude of each sequential EPSC early in the stimulus train showed strong depression (Fig. 1A1), which has been previously attributed to a combination of vesicle depletion (Zucker and Regehr 2002) and to incompletely defined mechanisms not directly related to depletion (Brody and Yue 2000; He et al. 2002; Moulder and Mennerick 2005). By contrast, during the same stimulus train, a late component of EPSCs emerged (Fig. 1A2), which has previously been attributed to asynchronous vesicle release at hippocampal and other synapses (Best and Regehr 2009; Cummings et al. 1996; Hagler and Goda 2001; Hefft and Jonas 2005; Hjelmstad 2006; Iremonger and Bains 2007; Lu and Trussell 2000; Otsu et al. 2004; Taschenberger et al. 2005). This delayed charge transfer resulted from a gradual EPSC waveform widening associated with increased synaptic noise (Fig. 1A2), and a failure of the postsynaptic current to decay completely to baseline between stimuli (Fig. 1A2, B). When fitted with a single exponential function, the final EPSC of trains exhibited an apparent decay τ of 5.03 ± 0.68 ms (n = 7, P < 0.012 compared with the initial EPSC of the train).

Fig. 1.

Repetitive stimulation at 20 Hz resulted in complementary alterations in phasic and asynchronous release. A1 and A2: an excitatory postsynaptic current (EPSC) waveform evoked by a train of action potentials at 20 Hz for 2 s showed depression in the phasic component and increases in asynchronous component. The stimulus artifact in the sample trace (and hereafter) is blanked for clarity. A2: the magnified trace from the boxed region in A1. The first EPSC from the same EPSC train in A1 was scaled to the amplitude of the 38th to 40th EPSCs. The charge transfer of the scaled EPSC (open area) was defined as the phasic component (phasic Q), whereas the residual area after subtracting the phasic component from the entire EPSCs was denoted as the asynchronous component (asynchronous Q, shaded area). Note that the block shading indicates the residual, accumulated asynchronous current on which the subsequent EPSC was superimposed. B: superimposed 1st (black thick) and the 40th EPSCs (gray) averaged from 3 independent recordings. The 40th EPSC was scaled to the peak amplitude of the 1st EPSC, and the baseline current before the stimulation was aligned to the same level as the baseline of the 1st EPSC. The decay of both EPSCs was fitted to a single exponential function (thin black lines) with τ_1st = 2.6 ms and τ_40th = 4.9 ms. C: summary of the gradually depressed phasic component and progressively increased asynchronous component in response to 20-Hz, 2-s stimulation. The charge transfer of phasic and asynchronous components from each EPSC was normalized to the initial EPSC (n = 8). D1: no significant correlation between the size of the initial EPSC and the proportion of asynchrony (asyn, asynchronous release is abbreviated as “asyn” in the figures hereafter) in the same EPSC train. The amplitude of the initial EPSC was plotted against the cumulative proportion of asynchrony to total release (asynchronous plus phasic). The correlation of the initial EPSC amplitude with the proportion of asynchronous release was fitted by least square linear regression with r² = 0.13 (P > 0.05). D2: by contrast, the depression of the phasic component (expressed as the 40th EPSC/1st EPSC) was correlated with the proportion of asynchronous release. The response ratio of the phasic component from the same set of cells in D1 was plotted against the proportion of asynchronous release. Linear regression of the plot yields r² = 0.49 (P < 0.05).

We quantified the phasic and the late components of EPSCs by scaling and superimposing the initial EPSC to the peak of each subsequent EPSC in the train, after accounting for residual asynchrony contributed by preceding EPSCs (Fig. 1A2). The area under the scaled EPSC was designated phasic charge. The charge contributed by the residual current after the phasic event (the shading immediately after the phasic EPSCs in Fig. 1A2) plus the steady-state component remaining from the preceding EPSC (the horizontal shaded area in Fig. 1A2) were denoted asynchronous charge. The total charge transfer therefore was the cumulative integral of the entire EPSC waveform (the sum of phasic and asynchronous postsynaptic charge). Figure 1C shows a summary from eight cells in which phasic and asynchronous charge were quantified in this way. The measured opposing effects of stimulation on phasic and asynchronous charge are similar to previously published work using alternate analysis methods (Hagler and Goda 2001; Otsu et al. 2004). We found that there was considerable variability among cells in the contribution of asynchrony to total release during train stimulation, but this variation did not correlate with the initial EPSC size (Fig. 1D1). Therefore initial transmitter output did not predict the relative contribution of the late EPSC components during trains. However, we found that the synapses with stronger depression of phasic EPSCs resulted in a greater contribution of asynchrony to total release (Fig. 1D2). These results are consistent with the idea that the two release processes work in opposition by competing for a common pool of vesicles (Otsu et al. 2004), but the results do not exclude the possibility of differential regulation of the two components.

Selective sensitivity of asynchrony to Ca²⁺ manipulations

Previous work has suggested that asynchronous release at hippocampal synapses is sensitive to presynaptic Ca²⁺ level (Cummings et al. 1996; Hagler and Goda 2001; Otsu et al. 2004). We verified this by examining the sensitivity of phasic versus asynchronous EPSC components to the slow Ca²⁺ buffer EGTA-AM. In several studies, EGTA has been shown to block the late, presumed asynchronous phase of release selectively. However, under the conditions of most studies, EGTA-AM also changes the dynamics of phasic release during a train (Cummings et al. 1996; Hagler and Goda 2001; Otsu et al. 2004). We found that >10-min exposure to 100 μM EGTA-AM resulted in 47 ± 7% reduction in total charge transfer over the course of a 40-pulse train (n = 7). Phasic release was depressed by 22 ± 14%, even though early pulses in the train usually exhibited facilitation relative to baseline (pre-EGTA). In this condition, asynchrony was reduced by 73 ± 3%.

To determine whether putative asynchrony can be selectively depressed without affecting total release, we titrated EGTA-AM concentration and exposure time to 20 μM for 5 min, which produced minimal effects on initial release (Fig. 2A, initial EPSC was reduced by 15 ± 5%, P = 0.07, n = 7) or total release over a 40-pulse, 20-Hz train (Fig. 2C). Under these conditions, the asynchronous charge was reduced to 57 ± 6% of control, whereas total release was maintained at 94 ± 8% of control (n = 7; Fig. 2C). Phasic release during the 40-pulse train under this condition was slightly but not significantly increased by 20 ± 14% relative to baseline. The effect on asynchrony was strongest toward the middle of the pulse train (Fig. 2B, left), with a gradual re-emergence of asynchrony late in the train (Fig. 2B, right). This late re-emergence likely resulted from buffer saturation, because it was less prominent with the longer incubations in 100 μM EGTA-AM (data not shown). These results suggest that asynchrony can be selectively depressed by Ca²⁺ buffering and is very sensitive to presynaptic Ca²⁺ concentration, even when overall phasic release is nearly intact.

Fig. 2.

EGTA-AM preferentially eliminated the asynchronous component. A and B: 20 μM EGTA-AM had little effect on the initial EPSC (inset in A), but significantly reduced the asynchronous component. Superimposed EPSC waveforms were recorded before (black) and after (gray) EGTA-AM treatment from the same cell. The boxed areas are magnified in B. C: summary of the EGTA-AM effect on total, phasic, and asynchronous charge transfer. The charge transfer of each component after EGTA-AM treatment was compared with the charge transfer measured before EGTA-AM application (n = 7, *P < 0.002).

The sensitivity of asynchronous release to a Ca²⁺ chelator and the prerequisite of repetitive presynaptic activity for promoting asynchronous release suggest that changing Ca²⁺ influx may differentially influence phasic and asynchronous release. Work from others has suggested a positive correlation between asynchrony during stimulus trains and the concentration of extracellular Ca²⁺ (Hagler and Goda 2001; Hjelmstad 2006; Otsu et al. 2004). In agreement with these previous reports, we saw an increase in both phasic and asynchronous EPSC components as the extracellular calcium concentration ([Ca²⁺]_o) was increased (Fig. 3, A–C). The preferential enhancement of the asynchronous component was revealed when the asynchronous charge was normalized to the total synaptic charge (Fig. 3B1). Increasing [Ca²⁺]_o from 1 to 1.6 to 4 mM shifted the release mode from phasic-dominant release at low [Ca²⁺]_o to stronger asynchronous release in 1.6 and 4 mM [Ca²⁺]_o (25 ± 4% in 1 mM Ca²⁺, 37 ± 5% in 1.6 mM Ca²⁺, and 49 ± 3% in 4 mM Ca²⁺, n = 6; Fig. 3, A and B1). Total charge transfer also increased modestly with increasing [Ca²⁺]_o (Fig. 3B2). This could be consistent with incomplete depletion at lower [Ca²⁺]_o (Moulder and Mennerick 2005) or with Ca²⁺-dependent vesicle replenishment during stimulus trains (Stevens and Wesseling 1998). These results emphasize that manipulating Ca²⁺ influx differentially modulates asynchrony versus phasic release during stimulus trains.

Fig. 3.

Asynchronous release was sensitive to [Ca²⁺]_o. A: the proportion of asynchrony to total release over the train was increased as [Ca²⁺]_o was raised. EPSCs recorded from 1 (black), 1.6 (gray), and 4 mM Ca²⁺ (light gray) in the same cell were superimposed. Boxed areas are magnified in the inset. B: summarized effect of altering [Ca²⁺]_o on the proportion of asynchronous (top) and total release (bottom) (top: n = 6, *P < 0.03 compared with 1 Ca²⁺, **P < 0.003 compared with 1.6 Ca²⁺; bottom: n = 6, *P < 0.04 compared with 1 Ca²⁺, **P < 0.05 compared with 1.6 Ca²⁺). C: higher [Ca²⁺]_o increased phasic depression (top) and enhanced asynchronous release (bottom). The phasic and asynchronous components of each EPSC were normalized to the 1st EPSC in 1 mM Ca²⁺ (n = 6).

A closer examination of the release dynamics during the train showed that phasic release was more strongly depressed in high [Ca²⁺]_o, despite greater potentiation in the early part of the stimulus train. The coincidence of phasic depression and increased asynchrony late in the train is particularly prominent when the postsynaptic charge transfer during the last half of the train is quantified. When assessed only for the last 20 stimuli, asynchrony represented 55 ± 7% of total release in 1.6 mM Ca²⁺ and 80 ± 5% of total release in 4 mM Ca²⁺. These results, again, likely reflect activity-dependent increases in the asynchronous component by gradual calcium accumulation in the later part of the train. The increased depression of phasic release is usually interpreted as a higher rate of vesicle depletion in higher calcium compared with low calcium (Zucker and Regehr 2002).

The sensitivity of asynchronous release to [Ca²⁺]_o and to the slow chelator EGTA suggests that asynchrony may be driven by sensors relatively distant from the sites of Ca²⁺ influx, although we cannot exclude the possibility of a separate Ca²⁺ sensor (Geppert et al. 1994; Goda and Stevens 1994; Sun et al. 2007). Ca²⁺ influx at hippocampal synapses is driven by two classes of voltage-gated Ca²⁺ channels, N and P/Q classes of high-voltage activated channels (Reid et al. 1997; Takahashi and Momiyama 1993; Wu and Saggau 1994), with the possibility of some contribution of R-type channels (Gasparini et al. 2001). At some synapses, there may be a looser association of N-type channels with vesicles (Mintz et al. 1995). It is unknown whether differential gating, kinetics, or location of N and P/Q channels relative to sites of release might favor one mode of release over the other. Therefore we tested the possibility of channel-specific influence on release mode using selective blockers of N or P/Q type Ca²⁺ channels. We used ω-conotoxin GVIA (ConoTX) to block N type Ca²⁺ channels and ω-agatoxin IVA (AgaTX) to block P/Q Ca²⁺ channels, and we compared asynchronous and total release under conditions of altered extracellular [Ca²⁺] designed to match the initial output of phasic release.

Figure 4 A shows the effect of ConoTX on initial phasic release under the various conditions used to evaluate asynchrony. On average, ConoTX (0.5–1 μM) reduced the initial EPSC to 50 ± 14% of baseline (n = 5), similar to the effect of lowering [Ca²⁺]_o to 1 mM in the absence of blocker (Fig. 4A; single EPSC in 1 mM Ca²⁺ was 63 ± 10% of baseline). In the presence of the Ca²⁺ channel blocker, we found that the initial phasic release was restored to baseline levels (2 mM Ca²⁺ without channel blocker) by increasing [Ca²⁺]_o to 5 mM (Fig. 4A, right). The initial EPSC amplitude in 5 mM [Ca²⁺]_o plus ConoTX was 124 ± 23% of baseline (n = 6). With train stimulation during N-type Ca²⁺ channel inhibition, asynchrony relative to total release decreased to a level similar to that observed in the absence of blocker with low [Ca²⁺]_o (21 ± 5% in 1 mM Ca²⁺, 18 ± 5% in ConoTX /2 mM Ca²⁺, n = 5, Fig. 4B, left bars). Likewise, when [Ca²⁺]_o was elevated in the presence of ConoTX to match initial phasic release to the original levels, the degree of asynchronous release during trains was indistinguishable from that under original baseline conditions (n = 6; Fig. 4B, middle bars). These results were mirrored by experiments using the selective P/Q channel blocker AgaTX (0.5–1 μM). P/Q channel blockade depressed single EPSC amplitude by 58 ± 8%, and the EPSC recovered close to the original EPSC amplitude with 6 mM [Ca²⁺]_o (100 ± 18% of baseline response, n = 5). Again, matching initial phasic transmitter output in the presence of toxin produced a similar ratio of asynchrony to total release during 20-Hz trains (Fig. 4B, right bars).

Fig. 4.

Asynchronous release was not dependent on specific presynaptic Ca²⁺ channels. A: by altering [Ca²⁺]_o, the EPSC measured in the presence of N-type channel blocker ω-conotoxin GVII (0.5–1 μM; ConoTX) can be matched to the EPSC in the absence of channel blocker. The evoked EPSC in the presence of channel blocker in 2 mM Ca²⁺ (ConoTX/2 Ca²⁺) was similar to the EPSC measured in 1 mM Ca²⁺ without the blocker (1 Ca²⁺; left). Similarly, the EPSC recorded in the presence of channel blocker can be matched to the control EPSC (2 mM Ca²⁺) by increasing [Ca²⁺]_o to 5 mM Ca²⁺(ConoTx/5 Ca²⁺; right). B: blocking either N or P/Q type channels did not alter the proportion of asynchronous release compared with matched EPSCs. Summary of channel blockers' effects on proportion of asynchronous release (1 Ca²⁺-ConoTX/2 Ca²⁺, n = 5; 2 Ca²⁺-ConoTX/5 Ca²⁺, n = 6; 2 Ca²⁺-AgaTX/6 Ca²⁺, n = 5).

These findings suggest that neither distinct functional properties of the Ca²⁺ channels nor relative location of Ca²⁺ channels participate strongly in shaping the relative dynamics of asynchrony to phasic release. Furthermore, it is interesting that manipulation of Ca²⁺ influx by either altering driving force (manipulation of [Ca²⁺]_o) or altering conductance (channel blockade) produces similar effects on the asynchrony ratio. Taken together, it seems likely that bulk intraterminal Ca²⁺ concentration dictates overall asynchrony. The results again highlight the ability to differentially modulate asynchrony versus phasic release by manipulation of presynaptic Ca²⁺ influx.

Asynchronous release and newly retrieved vesicles

To help explain the slow development of asynchrony during stimulus trains, it has been proposed that asynchronous release may draw from the vesicle pool that has been replenished during ongoing stimulation (Hagler and Goda 2001; Hjelmstad 2006; Otsu et al. 2004). Replenishment can occur through endocytosis or through recruitment of a pool of reserve vesicles (Kavalali 2007). Because endocytosis may support transmitter release during prolonged activity (Ertunc et al. 2007), we asked if vesicles endocytosed during stimulation participate in asynchronous release. We tested this idea using folimycin, an inhibitor of the vacuolar H⁺-ATPase (Ertunc et al. 2007; Sara et al. 2005). With complete folimycin poisoning, neurotransmitter loading into vesicles is blocked, and subsequent vesicle fusion will be postsynaptically silent.

We first assessed the efficacy of folimycin by measuring the recovery of evoked EPSCs in the presence of folimycin after vesicle depletion (Fig. 5 A). Folimycin exposure (67 nM for 20 min) slightly reduced the evoked EPSC, although this effect was not statistically significant (Fig. 5B1, left; 27 ± 17% depression, n = 10, P = 0.31), suggesting that folimycin did not cause strong transmitter leak from vesicles within this time window.

Fig. 5.

Asynchronous release was insensitive to blocking vesicle supply from newly recycled vesicles. A: the availability of newly endocytosed vesicles was functionally inhibited by folimycin (67 nM for 20 min). After incubation, folimycin had little effect on the average amplitude of EPSCs (dotted line, n = 10, also see B1, left) compared with the control (solid line, n = 10). Folimycin effectively attenuated vesicle recovery after extensive vesicle depletion. Top: experimental protocol and the averaged EPSCs corresponding to the stimulation. The ratio of EPSC recovery was assessed by comparing the recovered EPSC 40 s after vesicle depletion (45 mM K⁺ for 90 s) to the initial EPSC before depletion. Bottom: summary of folimycin's effect on EPSC recovery (n = 5, *P < 0.007). B: despite the effectiveness at inhibiting transmitter refilling, folimycin did not alter the level of phasic depression during a 2-s, 20-Hz train (B1, right). B2: summary of folimycin's effect on isolated evoked EPSCs (top) and on the proportion of asynchronous release (bottom; n = 10).

Folimycin, however, significantly retarded the recovery of evoked EPSCs after extensive vesicle depletion by application of 45 mM K⁺ for 90 s, designed to deplete the readily releasable and recycling vesicle pools (Harata et al. 2001). EPSC recovery 40 s after the K⁺ stimulus was 68.7 ± 7.2% of the initial EPSC in the control condition (n = 4) and 25 ± 5.6% in folimycin (n = 4; Fig. 5A). The strong attenuation of subsequent EPSC recovery after extensive vesicle depletion suggests that inhibition of transmitter refilling into vesicles was effective. Note that we did observe eventual recovery of EPSCs (data not shown), indicating that folimycin slowed but did not completely prevent vesicle refilling. Because this EPSC recovery was evaluated after washout of the folimycin, it could represent slow folimycin reversibility (Ertunc et al. 2007). Alternatively, it might suggest that there is slow mobilization of reserve pool vesicles after extensive vesicle depletion (Ikeda and Bekkers 2009). Regardless of the reason for the slow recovery, the results show that, over short intervals, such as during action potential trains, folimycin inhibition was effective.

During the brief, 40-pulse train, folimycin did not increase stimulus-dependent depression of phasic release (Fig. 5B1, right). Previous work shows that folimycin aggravates depression during substantially longer stimulus trains by interfering with the contribution of recycled vesicles to sustained transmitter release (Ertunc et al. 2007). The lack of folimycin's effect on release in our results likely results from the short stimulation time period, in which release depends more heavily on the internal supply of preexisting vesicles (Granseth et al. 2006; Ryan et al. 1996; Sankaranarayanan and Ryan 2000).

Despite this confirmed inhibitory effect of folimycin on newly endocytosed vesicles, we observed no change in the degree of asynchronous release during action potential trains in folimycin-poisoned cultures (percentage of asynchronous release: 43 ± 4% in control, n = 10; 41 ± 4% in folimycin, n = 10; Fig. 5B). We conclude that vesicle retrieval is not an important contributor to asynchronous release over short time periods.

Asynchronous release and phorbol ester modulation

The results thus far suggest that bulk [Ca²⁺] in the presynaptic terminal is prerequisite for asynchronous release. An obvious effect of increased [Ca²⁺]_o is an elevation of P_r. We tested the possibility that raising P_r by other methods may similarly increase relative asynchrony. We studied whether second messenger–activated P_r increases, like the Ca²⁺-induced P_r increase, enhance phasic release depression while increasing asynchrony. Among modulators of transmission, phorbol esters have received intensive interest (Brose and Rosenmund 2002; Malenka et al. 1986). Phorbol esters apparently activate PKC-dependent and independent mechanisms to potentiate transmitter release. We focused here on phorbol-mediated increases in P_r (Oleskevich et al. 2000; Rhee et al. 2002; Wu and Wu 2001; Yawo 1999). PDBu treatment (1 μM, 2–3 min) potentiated isolated EPSCs as expected, and the potentiation, although somewhat larger, was compatible with that caused by increasing [Ca²⁺]_o from 2 to 4 mM (Fig. 6, A1, B1, inset, and C).

Fig. 6.

Pharmacological manipulation of P_r potentiated EPSCs but did not affect the proportion of asynchronous release. 4 mM Ca²⁺ (A1), and 1 μM PDBu (B1) had superficially similar effects on isolated EPSCs, paired-pulse modulation and phasic depression during train-evoked EPSCs. A1: comparison of release dynamics during stimulation stimulus train between control (2 mM Ca²⁺) and 4 mM Ca²⁺. The phasic component of each EPSC was normalized to the initial EPSC and plotted against the stimulus episode (n = 7). Inset: sample paired EPSCs (50-ms interstimulus interval) recorded in control (2 mM Ca²⁺, solid line) and in 4 mM Ca²⁺ (dotted line). A2: the late onset EPSCs (36th- 40th) from 20-Hz train stimulation in control (black) were superimposed on the EPSCs from the same stimulus episodes in 4 mM Ca²⁺ (gray). B1 and B2 are similar to A1 and A2, using PDBu as modulator. C: summarized effects of 4 mM Ca²⁺ and PDBu on synaptic potentiation. EPSCs measured in 4 mM Ca²⁺ or PDBu were normalized to the control (*P < 0.02, n = 7; **P < 0.004, n = 9). D: summary of the effect of 4 mM Ca²⁺ and PDBu on paired-pulse ratio (PPR). Normalized PPR was expressed as the ratio of the 2nd EPSC to the 1st EPSC, relative to the baseline paired-pulse ratio in 2 mM Ca²⁺, which was taken as 100% (*P < 0.02, n = 7; **P < 0.03, n = 9). The paired-pulse interval was 50 ms. E: summary of the effect of Ca²⁺ and PDBu on the proportion of asynchrony to total release during the train (*P < 0.05, n = 7).

Because the strong phorbol potentiation may involve presynaptic potentiation mechanisms in addition to increased vesicle release probability (Stevens and Sullivan 1998; Waters and Smith 2000), we examined whether Ca²⁺ increase and PDBu produce similar effects on P_r. A relatively straightforward estimate of P_r effects can be obtained from paired-pulse ratios (Fig. 6, A1, B1, inset, and D). PDBu's effect was consistent with the effect of increasing [Ca²⁺]_o to 4 mM. Both treatments significantly and similarly decreased the paired-pulse ratio (ratio of 2nd to 1st EPSC, 50 ms apart), suggesting an increase in P_r (Fig. 6D). Moreover, during train stimulation, PDBu increased phasic depression similar to Ca²⁺ elevation (Fig. 6, A1 and B1). The decays of the phasic peak response were fitted to a double exponential function for each cell, and the weighted decay time constant was similar for both 4 Ca²⁺ and PDBu (4.0 ± 1.1 vs. 4.7 ± 0.8 stimuli, respectively; P = 0.61). These results confirmed that PDBu and raising [Ca²⁺]_o to 4 mM similarly increased P_r and depressed phasic release.

Both PDBu and high [Ca²⁺]_o enhanced initial phasic release (Fig. 6C) and enhanced asynchronous release (Fig. 6, A2 and B2). However, when we compared the ratio of asynchronous to total release, elevated [Ca²⁺]_o, but not PDBu treatment, changed the asynchrony ratio (Fig. 6E). This result indicates that PDBu potentiated both release modes in parallel; unlike increased Ca²⁺, P_r elevation by PDBu did not preferentially encourage either asynchrony or phasic release during repetitive stimulation. The differential effect of raising [Ca²⁺]_o and PDBu on release mode is particularly significant when only the charge transfer during the steady state of release was quantified. During the steady-state release (the last 20 stimuli), asynchrony to total release increased from 56 ± 5% in baseline to 76 ± 2% in 4 mM Ca²⁺ but was almost unchanged in PDBu (55 ± 7% in control; 59 ± 6% in PDBu). Note that, although PDBu robustly potentiated single evoked EPSCs (280 ± 50% of control), a stronger depression during the train resulted in differential potentiation, in which the early EPSCs were more strongly potentiated then the steady-state EPSCs. Such differential potentiation led to a less than twofold (182 ± 30% of control) increase in overall phasic charge transfer during the train.

Because presynaptic phorbol ester effects have been attributed to a combination of PKC-dependent and PKC-independent mechanisms (Betz et al. 1998; Wierda et al. 2007), we wondered whether both pathways contributed to the overall parallel modulation of both phasic and asynchronous release or whether dissection of the two pathways might show differential regulation. We used the broad-spectrum PKC inhibitor Gö6983 to block PKC activation by PDBu. To evaluate the efficacy of our Gö6983 treatment, we tested two previously documented presynaptic effects of PKC. First, we examined whether Gö6983 could depress the ability of phorbol ester to speed replenishment after Ca²⁺-independent depletion of the readily releasable vesicle pool (Stevens and Sullivan 1998) with a 3-s application of hypertonic sucrose (Rosenmund and Stevens 1996; Stevens and Sullivan 1998). Figure 7 A shows that phorbol-stimulated speeding of EPSC recovery after sucrose challenge, as observed by Stevens and Sullivan (1998). Gö6983 completely prevented phorbol ester's ability to hasten EPSC recovery (Fig. 7A).

Fig. 7.

Protein kinase C (PKC) inhibition prevented increases in asynchronous release in PDBu. A: PKC inhibition by 2–3 μM Gö6983 (Gö) effectively attenuated PDBu (1 μM) stimulated recovery from Ca²⁺-independent vesicle depletion. EPSC recovery assay and the corresponding EPSC are shown in the top panel. The EPSC recovery was calculated by normalizing the EPSC recovered 3 s after hyperosmotic-induced vesicle depletion to the initial EPSC before depletion; 0.5–0.75 M sucrose was applied for 3 s to induce depletion. The raw traces represent examples from separate cells. Note especially the similar depression induced by depletion in the control (black trace) and PDBu plus Gö6983 (PDBu/Gö dashed trace) conditions. By contrast, PDBu alone (gray trace) promoted substantial recovery. The average ratios of EPSC recovery in control, PDBu, or PDBu with Gö6983 (PDBu/Gö) are summarized in the bottom panel (*P < 0.004, n = 7). B1: the effect of PDBu (PDBu, top) or PDBu plus Gö6383 (PDBu/Gö, bottom) on a single EPSC (inset) and on asynchronous release during train stimulation. The asynchronous component in each stimulus episode was plotted as a function of time (stimulus number; n = 8 in control-PDBu, n = 9 in Gö-PDBu/Gö; see C for comparative statistics). B2: effect of Gö6383 on increased asynchrony in 4 mM Ca²⁺ (n = 10). C: summarized effects (from B1) of PDBu alone or PDBu/Gö on cumulative phasic (left) or asynchronous (right, *P < 0.04) charge transfer. D: summaries of asynchrony to total release for the conditions shown in B1 and B2: PDBu (top left, n = 8), Gö alone (top right, n = 10; P < 0.05), PDBu/Gö (bottom left, n = 10, *P < 0.007), and 4 mM Ca²⁺/Gö (bottom right, n = 10) on the proportion of asynchronous release during the train.

As a second test of Gö6983 effectiveness, we examined the documented ability of phorbol esters to occlude effects of presynaptic inhibitory G protein–coupled receptor stimulation. This effect has been shown to be PKC dependent (Zamponi et al. 1997). As previously reported, we found that A1 adenosine receptor stimulation with 2-chloroadenosine (1 μM) depressed isolated EPSCs by 64.5 ± 10.3% (n = 4) (Mennerick and Zorumski 1995; Scholz and Miller 1991; Swartz 1993). Exposure to phorbol ester before 2-chloroadenosine application diminished the inhibition by 2-chloroadenosine (level of depression: 32 ± 9%, n = 4). Inhibiting PKC activity by Gö6983 reversed phorbol ester's occlusion (level of depression: 60 ± 8%, n = 4). Thus by two measures, Gö6983 effectively prevented presynaptic PKC activation by PDBu under our experimental conditions.

In experiments designed to probe the effect of PDBu and Gö6983 on train-evoked EPSCs, we again found that PDBu increased phasic and asynchronous release in parallel (Fig. 7, B and C, 1st panel). When the effect of Gö6983 alone on synaptic transmission was examined, we found that Gö6983 did not affect EPSC amplitude (108 ± 10%, n = 10) or P_r estimated by paired-pulse ratio (2nd EPSC to 1st EPSC: 0.73 ± 0.07 in control; 0.71 ± 0.08 in Gö6983, n = 10). We did find that incubation in Gö6983 alone slightly reduced the overall asynchrony contribution to total release (42 ± 4% asynchrony to total release ratio before Gö6983, 38 ± 4% ratio after Gö6983, n = 10, P < 0.004; Fig. 7D, top right), suggesting a possible small contribution of PKC-dependent mechanisms to basal asynchrony. Pre-exposure to Gö6983 (>3 min) before PDBu application did not significantly inhibit PDBu potentiation of isolated EPSCs (ratio of EPSC potentiation: 2.4 ± 0.2 by PDBu alone, n = 8; 2.0 ± 0.3 in PDBu/Gö6983, n = 10, P = 0.49; Fig. 7B1, right), suggesting the importance of PKC-independent pathways to the potentiation of phasic release (Rhee et al. 2002). PKC inhibition also slightly but nonsignificantly reduced PDBu-mediated potentiation of total charge transfer in train-evoked release (ratio of total charge transfer potentiation 1.6 ± 0.2 in PDBu, n = 8; 1.3 ± 0.1 in PDBu/Gö6983, n = 10, P = 0.08). In contrast with the trend-level effect on total release, PKC inhibition completely prevented the increase of asynchronous charge transfer in the presence of PDBu (Fig. 7, B1 and C). PKC inhibition therefore reduced the ratio of asynchronous to total charge transfer (Fig. 7D, bottom left). It should be noted that these effects were compared with the effect of Gö6983 alone; the results therefore represent the true effect of Gö6983 on PDBu rather than the small effect of Gö6983 on baseline asynchrony. The results suggest that PKC-dependent pathways selectively enhance asynchronous release, whereas PKC-independent pathways sustain the potentiation of phasic release, leading to the net parallel increase in both components when PKC-dependent and -independent pathways are intact.

Although PKC inhibition only subtly decreased PDBu-induced potentiation of isolated EPSCs, a closer examination of phasic depression showed that PKC inhibition also dampened potentiation during steady-state phasic release. This effect led to an unchanged asynchronous to total release when only the steady-state release (the last 20 stimuli) was quantified (47 ± 4% in Gö; 45 ± 3% in PDBu/Gö). Therefore the PKC-dependent component of potentiation was most prominent during the steady-state phase of release in 20-Hz trains, where asynchrony was strongest. This led to the relatively selective inhibition of asynchrony over the course of the train.

To verify that the effects of PDBu on asynchrony were not specific to one particular phorbol ester, we also evaluated the effect of PMA (1 μM) on release during trains. Similar to PDBu, PMA increased individual EPSCs to 212 ± 20% of baseline (n = 9). Synchronous and asynchronous release increased largely in parallel over the course of the train, so that asynchrony contributed 39 ± 5% of total charge transfer at baseline before PMA treatment and 38 ± 4% after treatment. At steady state (stimuli 21–40), asynchrony contributed 57 ± 7% of postsynaptic current at baseline and 59 ± 6% after PMA treatment. Finally, Gö6983 selectively compromised the increased in asynchrony produced by PMA. After Gö6983 incubation, asynchrony's contribution to total release was reduced from 40 ± 4% to 33 ± 4% (P < 0.03 compared with Gö6983 alone, n = 9).

Presynaptic activity and Ca²⁺ influx could result in PKC activation directly or indirectly (Steinberg 2008). We wondered if the Ca²⁺-dependent increases in asynchronous release observed with the manipulations of [Ca²⁺]_o shown in Figs. 2–4 could result from PKC activation. We therefore tested whether the preferential increase of asynchronous release in high [Ca²⁺]_o was sensitive to PKC inhibition. We assayed the asynchrony ratio in high Ca²⁺ in the presence of the PKC inhibitor. In 4 mM Ca²⁺, asynchronous to total charge transfer was 47 ± 3% (n = 10). Inhibiting PKC activity during train stimulation did not significantly change the ratio of asynchronous to total charge transfer (44 ± 3%, n = 10; Fig. 7, B2 and D, bottom right). The insensitivity of asynchronous release to PKC inhibition in high [Ca²⁺]_o suggests that, unlike PDBu potentiation of asynchronous release, PKC activity likely does not contribute to increased asynchrony with Ca²⁺ elevation.

DISCUSSION

In this study, we observed a disproportionate shift between phasic and asynchronous release with several manipulations of Ca²⁺ influx or buffering. Phasic depression was correlated with increased asynchrony. In contrast to the effects of manipulating P_r with [Ca²⁺]_o alterations, P_r modulation with phorbol esters increased phasic and asynchronous release in parallel. However, we found that PKC-dependent mechanisms explained most of the effect on asynchrony and a small component of the potentiation of phasic release. Our results are important because they are a proof of principle that modulators can influence asynchrony and phasic release through separate mechanisms. Furthermore, our results add to a list of PKC-dependent and PKC-independent presynaptic effects of diacylglycerol analogs.

An alternative interpretation of the late EPSC components that accumulate during repetitive stimulation is transmitter pooling. We have previously shown that transmitter pooling is evident in island cultures primarily with strong, synchronous synaptic output (Mennerick and Zorumski 1995). Here, we failed to find any evidence of a positive correlation between initial EPSC amplitude and the degree of late (asynchronous) charge transfer (Fig. 1D). Initial EPSCs decayed with time constants very similar to those of mEPSCs in this preparation (Diamond and Jahr 1995; Zorumski et al. 1996). This confirms that asynchrony and spillover are both negligible after a single EPSC. Because phasic release strongly depresses during train stimulation, it is unlikely that pooling becomes prominent during the train. Finally, the sensitivity of late EPSC components to EGTA, even when phasic release was intact, strongly suggests that release asynchrony is the major contributor to late EPSC components. We cannot completely exclude a role for transmitter spillover/pooling in these results; however, in the context of previous work that has explored conditions promoting pooling, our results strongly support the idea that release asynchrony underlies the majority of late EPSC components during train stimulation in these experiments (Cummings et al. 1996; Hagler and Goda 2001; Otsu et al. 2004).

The mechanisms by which phorbol esters potentiate transmitter release have been debated (Lou et al. 2008; Rhee et al. 2002; Wierda et al. 2007). Some studies have indicated that potentiation of Ca²⁺ influx may be important to the effects of phorbol esters (Bartschat and Rhodes 1995; Honda et al. 2000; Swartz et al. 1993) (but see Hori et al. 1999; Redman et al. 1997; Waters and Smith 2000; Yawo 1999). Our results and others show that alterations of Ca²⁺ buffering or influx (by manipulating either Ca²⁺ driving force or conductance) selectively alter the asynchrony contribution to total release. In addition, our results suggest that PKC-dependent mechanisms of phorbol ester stimulation selectively alter asynchrony. One parsimonious explanation for the effect of phorbol esters is that PKC-dependent mechanisms may alter presynaptic Ca²⁺ dynamics, such as Ca²⁺ influx or handling, which alter the temporal profile of bulk Ca²⁺ in the presynaptic terminal. In other words, changes in presynaptic intracellular [Ca²⁺] may be downstream of PKC activation. On the other hand, PKC-dependent increases in mEPSC frequency may be independent of Ca²⁺ influx (Capogna et al. 1995; but see Waters and Smith 2000).

An alternative, nonmutually exclusive explanation for the selective effect of PKC-dependent mechanisms on asynchrony is that asynchronous release draws from replenished vesicles (Otsu et al. 2004), and PKC-dependent effects may involve speeding vesicle replenishment after initial release (Minami et al. 1998; Waters and Smith 2002; Wierda et al. 2007). Our results suggest that rapid retrieval of vesicles is not important for supporting asynchronous release. However, this result leaves open the possibility that recruitment from the reserve pool is important for fueling the development of asynchronous release during a train (Hjelmstad 2006; Otsu et al. 2004). Past experiments (Stevens and Sullivan 1998) and this work (Fig. 7A) suggest that replenishment of Ca²⁺-independent sucrose-evoked release can be accelerated by PKC-dependent mechanisms, which could lead to increased asynchrony. Furthermore, PKC-dependent phorbol effects were more evident on phasic and asynchronous release late during 2-s, 20-Hz stimulus trains, possibly suggesting acceleration of replenishment. However, in previous studies of hippocampal neurons, phorbol stimulation did not speed replenishment during electrical stimulation; Ca²⁺-dependent acceleration of vesicle replenishment occluded phorbol-induced replenishment (Stevens and Sullivan 1998). Therefore PKC-dependent mechanisms could selectively increase overall asynchrony by speeding replenishment. However, this explanation is difficult to test directly during stimulus trains. Other studies have assumed that most asynchrony during brief trains is part of the initial readily releasable vesicle pool (Moulder and Mennerick 2005; Stevens and Williams 2007). Accurate quantification of the readily releasable vesicle pool size accessible to action potentials will require further clarification of the contribution of asynchrony to this vesicle pool.

In contrast to PKC-dependent asynchrony increases, PKC-independent effects apparently underlie selective potentiation of phasic release. This is consistent with the idea that PKC-independent mechanisms (such as binding of Munc13 by phorbol esters) enhances fusion efficiency of vesicles (Basu et al. 2007; Rhee et al. 2002), rendering vesicles more easily released during the synchronous phase of transmission.

In summary, our work shows that second messenger pathways that potentiate vesicle release probably increase phasic and asynchronous release in parallel. However, in the case of phorbol ester modulation, this parallel increase can be subdivided into a component that is PKC-independent that mainly affects phasic release and a PKC-dependent component that disproportionately affects asynchronous release. This PKC-dependent effect on asynchrony could be explained by Ca²⁺-dependent increases in asynchronous vesicle release and/or by PKC-dependent effects on vesicle replenishment. The results suggest that asynchronous and phasic transmitter release may be controlled relatively independently. However, because physiological PKC activation is usually associated with rises in presynaptic [Ca²⁺] through inositol triphosphate receptor activation and DAG increases, which can produce the PKC-independent effects studied here and elsewhere, the precise physiological conditions under which differential regulation may occur remain unclear. Nevertheless, this modulation has the potential to fine tune the temporal relationship between incoming presynaptic firing and postsynaptic output.

GRANTS

This work was supported by National Institute of Mental Health Grant MH-78823 to S. Mennerick.