Synchronisation ofneurotransmitter release during postnatal development in a calyceal presynaptic terminal of rat

Abstract

1. Mechanisms contributing to the synchronisation of transmitter release during development were studied in synapses of the medial nucleus of the trapezoid body (MNTB) using patch recording and Ca²⁺ imaging techniques in a rat brainstem slice preparation.

2. Excitatory postsynaptic currents (EPSCs) were generated in an all-or-none manner at immature synapses (postnatal days earlier than P6). Many delayed miniature EPSC (mEPSC)-like currents followed EPSCs at immature synapses, while observations of delayed mEPSC-like currents were rare at mature synapses (later than P9).

3. At immature synapses bath application of either ω-conotoxin GVIA or ω-agatoxin-IVA reduced EPSCs (both to 40% of control), and Ca²⁺ currents in the presynaptic terminal (both to 70% of control). The frequency of delayed mEPSC-like currents was reduced by ω-conotoxin GVIA, but not by ω-agatoxin IVA.

4. At immature synapses delayed mEPSC-like currents were rare after incubation of the slice with extrinsic Ca²⁺ buffers (EGTA AM).

5. At mature synapses many mEPSC-like currents followed evoked EPSCs after partial block of Ca²⁺ channels by bath application of a low concentration of Cd²⁺ (3 μm) or ω-agatoxin IVA (50 nm) but not by low [Ca²⁺]_o (0.5-1 mm).

6. Ca²⁺ transients induced by action potentials in presynaptic terminals were monitored by adding a high concentration of fura-2 (200 μm) to the pipette. Their decay time course was slower at immature presynaptic terminals than at mature terminals. Both the Ca²⁺ extrusion rate and the endogenous Ca²⁺ binding capacity were estimated to be smaller at immature terminals than at mature terminals.

7. These results suggest that the maturation of synaptic transmission in MNTB progresses with the capacity for Ca²⁺ clearance from the presynaptic terminal. The possible importance of developmental increases in both Ca²⁺ clearance capacity and Ca²⁺ currents is discussed in relation to the synchronisation of transmitter release.

The medial nucleus of the trapezoid body (MNTB) is an auditory relay nucleus. The MNTB’s principal neurone receives a large single glutamatergic terminal called the calyx of Held from the contralateral antero-ventral cochlear nucleus (AVCN; Held, 1893; Wu & Kelly, 1992; Suneja et al. 1995). The MNTB’s principal neurone projects to the ipsilateral lateral superior olive (LSO; Morest et al. 1968), and plays a role in sound localization. Localizing a sound source is only possible when precisely timed transmission is made in all the relay nuclei including MNTB. The mature MNTB synapse is known to mediate highly reliable, phase-locked transmission (Goldberg & Brown, 1968; Aitkin, 1986). The EPSC recorded in a mature MNTB neurone is large in amplitude and is generated in an all-or-none manner when the stimulus intensity to the presynaptic fibre is varied, indicating innervation by a single fibre from AVCN (Forsythe & Barnes-Davies, 1993; Borst et al. 1995).

In immature synapses both the timing and amplitude of EPSCs fluctuate, and many delayed mEPSC-like currents follow the EPSC. During the early days of postnatal development, the amplitude of presynaptic Ca²⁺ currents increases more than twofold between postnatal days 5-6 (P5-6) and P10-11. The amplitude of EPSCs increases and the coefficient of variation of EPSC amplitudes decreases over the same period attaining a steady level at P9-11. We therefore compared the nature of synaptic transmission in animals older than P9 with that in immature animals (Chuhma & Ohmori, 1998). During this period, the immunoreactivity to calbindin-D_28K appears around P8 in MNTB neurones and increases during the next 10 days (Friauf, 1993). The outer ear canals open after P12, and auditory brainstem responses are first recorded at P12-14 (Jewett & Romano, 1972; Blatchley et al. 1987). Therefore, the maturation of synaptic transmission precedes and progresses along with the development of hearing. We are interested in the maturation of synaptic transmission, and in the mechanisms by which fast and reliable transmission, including synchronisation of transmitter release, are attained at MNTB synapses. We have therefore focused our experiments on the regulation of presynaptic Ca²⁺, and demonstrated that maturation of synaptic transmission involves increases of both Ca²⁺ influx and clearance capacity in MNTB synapses.

METHODS

Preparation of brain slices

Brain slices containing MNTB were prepared from Wister rats of postnatal days 4-6 (P4-6) and P9-17. Animals were decapitated after deep anaesthesia with ether. Procedures conformed to the guiding principles for the care and use of animals in the field of physiological sciences set by the Japanese Physiological Society. Detailed procedures for slice preparation were described previously (Chuhma & Ohmori, 1998). Briefly, transverse brain slices (200-300 μm sections) were cut by a vibratome (DTK-2000; Dosaka, Kyoto, Japan), and were incubated for 1 h at 36°C in high-glucose artificial cerebrospinal fluid (high-glucose ACSF; composition, mm: 75 NaCl, 2.5 KCl, 26 NaHCO₃, 1.25 NaH₂PO₄, 0.7 CaCl₂, 2 MgCl₂ and 100 glucose, pH 7.4) saturated with 95% O₂-5% CO₂. Experiments were performed at room temperature (20-25°C) unless otherwise noted.

Recording of EPSCs

Details of the electrophysiological experiments have been described previously (Chuhma & Ohmori, 1998). EPSCs were recorded in continuously circulated ACSF (mm: 125 NaCl, 2.5 KCl, 26 NaHCO₃, 1.25 NaH₂PO₄, 2 CaCl₂, 1 MgCl₂ and 17 glucose, pH 7.4) saturated with 95% O₂-5% CO₂. ACSF was supplemented with 20 μm strychnine (Sigma), 10 μm bicuculline (Sigma) and 50 μmd-2-amino-5-phosphonovalerate (APV; Tocris) to block glycinergic and GABAergic inputs, and to block NMDA receptors. In some experiments (Fig. 6), extracellular CaCl₂ was varied from 0.25 to 10 mm: at [Ca²⁺]_o lower than 2 mm, total divalent cation concentration ([Ca²⁺]_o+[Mg²⁺]_o) was 3 mm. At higher [Ca²⁺]_o, Mg²⁺ was omitted. When using 10 mm CaCl₂, 10 mm Hepes was used in place of NaHCO₃ and NaH₂PO₄. The composition of the pipette solution was as follows (mm): 136 caesium glucuronate, 14 CsCl, 10 Hepes and 5 EGTA, and the pH was adjusted to 7.2. To block Na⁺ currents, 5 mmN-(2,6-dimethylphenylcarbamoylmethyl) triethylammonium bromide (QX314, Alamone Labs, Jerusalem, Israel) was added to the pipette solution.

Figure 6. Cooperativity to Ca²⁺ of mature EPSCs after partial block of Ca²⁺ channels by Cd²⁺ with immature and mature controls.

Relative amplitudes (I/I_max) of EPSCs (I) to the maximum EPSC (I_max) obtained at 5-10 mm[Ca²⁺]_o were transformed as (I/I_max)/(1 - I/I_max) and plotted against [Ca²⁺]_o on log-log coordinates. Mature synapses after application of Cd²⁺ (3 μm) (P9-11, •), immature controls (P4-5, ^) and mature controls (P10-11, □) are shown. The slope of the fitted line gives the Hill coefficient (n). K_EPSC (eqn (4)) was estimated from the concentration corresponding to 1 (indicated by the horizontal dotted line). Hill coefficients and K_EPSC values and number of cells tested are tabulated in Table 3.

Neurones were voltage clamped at -70 mV (Axopatch 200A; Axon Instruments). The liquid junction potential (approximately -10 mV) was corrected. Patch pipette resistances were ≈2-5 MΩ. Series resistances were 9-36 MΩ and compensated 70-85%. Presynaptic nerve fibres were electrically stimulated (0.5-7 V, 100 μs duration) every 5 s as described previously (Chuhma & Ohmori, 1998).

Ca²⁺ channel block with neurotoxins and Cd²⁺

The effects of ω-conotoxin GVIA (Peptide Institute, Osaka, Japan) and ω-agatoxin IVA (Alamone labs) were investigated at immature synapses. The presence of both N-type and P/Q-type Ca²⁺ channels is known for the immature presynaptic terminals (Iwasaki & Takahashi, 1998). Stock solutions of these neurotoxins were prepared using water with cytochrome c (Sigma). Stock solutions were directly added to the external medium in the recording chamber (0.5 ml) after stopping the circulation; the final concentration of ω-conotoxin was 1-2 μm, ω-agatoxin was 200 nm and cytochrome c was 0.1 mg ml⁻¹. During these experiments, the chamber solution was maintained saturated with 95% O₂-5% CO₂ by directly blowing the mixed gas over the surface. When toxin sensitivities of presynaptic Ca²⁺ channels were investigated (Table 2), amplitudes of Ca²⁺ currents were repeatedly measured every 30 s by applying 100 ms depolarization from -70 to -10 mV. Other details were the same as described (Chuhma & Ohmori, 1998).

Table 2.

Effects of neurotoxins on EPSC and presynaptic Ca²⁺ currents in P4-5 synapses

	EPSC
	Size (% of control)	Asynchronicity indices Test/control	Ca2+ current size (% of control)
ω-Conotoxin GVIA	42.9 ± 4.4 (5)	0.90 ± 0.15 (5)/1.60 ± 0.26 (5)*	67 ± 6 (3)
ω-Agatoxin IVA	40.7 ± 12.3 (5)	1.45 ± 0.11 (5)/1.32 ± 0.26 (5)**	69 ± 2 (4)

ω-Conotoxin GVIA was at 1–2 μm and ω-agatoxin IVA was at 200 nm.

^*P < 0.01

^**P > 0.2. Test and control of asynchronicity indices were records from the same cells. Number of cells examined is indicated in parentheses.

In experiments where Ca²⁺ channels were partially blocked at the mature synapse, 50 nmω-agatoxin was dissolved in ACSF with cytochrome c, and the external medium was circulated. This concentration of ω-agatoxin reduced the EPSC amplitude to 20-30% of control but did not block the EPSC completely. It took more than 60 min before a stable, partially blocked condition was attained.

EGTA AM loading

Brain slices prepared from the same P4-5 animals were divided into two groups and one was incubated with the acetoxymethyl ester (AM) form of EGTA. EGTA AM was mixed with Pluronic F-127 to facilitate cell loading. Both EGTA AM (Molecular Probes) and Pluronic F-127 (Molecular Probes) were dissolved separately in dimethyl sulphoxide (DMSO) as stock solutions. After mixing by sonication, the mixture was added to the incubation medium (high-glucose ACSF) saturated with 95% O₂-5% CO₂. The final concentration of EGTA AM was 60 μm, Pluronic F-127 was 0.05% and DMSO was 0.6%. Slices were incubated for 1 h at 36°C. Control slices were incubated similarly with Pluronic F-127 and DMSO. After incubation, both groups of slices were washed and maintained in high-glucose ACSF, as in other experiments.

Measurements of presynaptic Ca²⁺ transients

When Ca²⁺ flux into the presynaptic terminal was monitored, whole-cell recordings of presynaptic terminals were made with a K⁺-based pipette solution containing 200 μm fura-2 pentapotassium salt (fura-2-5K, Molecular Probes). The composition of the pipette solution was (mm): 20 KCl, 120 potassium gluconate, 10 Hepes, 5 Mg-ATP and 5 creatine phosphate, and pH was adjusted to 7.4. Pipette resistances were 5-10 MΩ. Membrane potential was maintained near -75 mV by current injection. Fura-2 was alternately excited using a wavelength pair of 360 and 380 nm, and imaging was made through a 500 nm long-pass filter by a cooled CCD camera system (HiSCa, Hamamatsu Photonics). Images were acquired at 136-490 ms intervals. Fluorescence ratios (R =f₃₆₀/f₃₈₀) of the presynaptic terminal area were calculated off-line, after subtraction of the background fluorescence intensity taken from a nearby area. Subtraction of the background decreased decay time constants of Ca²⁺ transients slightly, and affected the estimation of [Ca²⁺]_i. Therefore, windows for the synaptic area and the background area were fixed for a single series of measurements. Ratios were converted to [Ca²⁺]_i by an equation determined through in situ calibration (Neher, 1989):

(1)

In these calibrations, fluorescence ratios from MNTB neurones corresponding to 0, 0.13 μm, 1.15 μm, 24.5 μm and 10 mm[Ca²⁺]_i were measured with 100 μm fura-2-5K at pH 7.4 in the K⁺-based pipette solution.

Action potentials in the presynaptic terminal were generated by current injection through the patch pipette (5 ms duration, 20 ms interval), and a train of five action potentials was used in most experiments to improve the signal-to-noise ratio. The amplitudes of Ca²⁺ transients were dependent on the number of spikes, increased linearly up to 12 action potentials, and failed to increase further probably due to saturation of Ca²⁺-bound fura-2. Decay time constants (τ_decay) of Ca²⁺ transients were measured by fitting a single exponential function.

The Ca²⁺ binding capacity of exogenous buffer (k_B), namely fura-2, was calculated as in Neher & Augustine (1992):

(2)

where [CaB] was the concentration of Ca²⁺-bound fura-2, [B] the total fura-2 concentration in the terminal, estimated from the relative intensity of f₃₆₀ at each trial to f₃₆₀ at saturation (200 μm) towards the end of each experiment (see Fig. 7A), K_D the fura-2 Ca²⁺ binding constant (0.27 μm), [Ca²⁺]_b the resting Ca²⁺ concentration, and [Ca²⁺]_p the Ca²⁺ concentration at the peak of the Ca²⁺ transient. The endogenous Ca²⁺ binding capacity (k_S), and the Ca²⁺ extrusion rate (γ) were estimated from the plot of τ_decayvs.k_B (Helmchen et al. 1997 and see Fig. 7C) following the equation:

(3)

Figure 7. Estimation of endogenous Ca²⁺ binding capacity and Ca²⁺ extrusion rate.

A, time course of increase in fura-2 concentration at a P6 presynaptic terminal. Fura-2 concentration within the presynaptic terminal (right ordinate) was estimated linearly from f₃₆₀ intensities (arbitrary units (a.u.), left ordinate). Ca²⁺ transients marked with filled circles are illustrated in insets. The right inset is the ensemble average of the last 3 Ca²⁺ transients. B, Ca²⁺ transients measured at P6 (dotted line) and P11 (continuous line) presynaptic terminals recorded with 200 μm fura-2, amplitude scaled and superimposed to facilitate comparison of decay time courses. The decay time constant was 2.9 s at P6 and 1.0 s at P11. C, exogenous Ca²⁺ binding capacity (k_B) was calculated (eqn (2)) and decay time constants of Ca²⁺ transients are plotted against k_B for P6 (filled symbols, 4 cells) and for P10-11 (open symbols, 3 cells). Different symbols indicate different cells. Lines are linear regression fits and are τ_decay= 0.33 +k_B/185 for P10-11, and τ_decay= 0.60 +k_B/29 for P6.

Counting mEPSC-like currents

In order to evaluate mEPSC-like currents quantitatively, we registered the time of occurrences of both EPSCs and mEPSC-like currents in each trace by visually identifying peaks using commercially available software (Axograph; Axon Instruments). Occurrence times were assembled into a histogram (event time histogram, bin width 0.5 ms). We scored small current transients as mEPSC-like currents when they showed a clear transient peak with steep rising and slower decay, and a total time course of 5-10 ms. When these mEPSC-like currents occurred in the decay phase of the evoked EPSC, identification of their peaks was difficult, and they were not counted. Consequently, this produced a gap in the event time histogram immediately after the major peak (see Fig. 4Ac). Each event time histogram was made from 30-50 consecutive records of 100 ms duration obtained from a single neurone. The frequency of mEPSC-like currents was evaluated as an asynchronicity index by dividing the total counts during a time window of 60 ms (from 20 to 80 ms) after the EPSC by the number of traces.

Figure 4. EGTA AM incubation of immature synapses.

Aa and b, control EPSCs and EPSCs after EGTA-AM loading recorded from the sister slice prepared from a single animal (P5). Seven traces are superimposed. c and d, event time histograms of control EPSCs shown in a and for the EGTA-incubated synapse shown in b, each made from 50 consecutive traces. B, asynchronicity indices for control (open bar, 5 cells) and for EGTA-incubated synapses (hatched bar, 5 cells).

Since background spontaneous mEPSCs were rare (see Fig. 1C and 2A), the asynchronicity index was calculated without subtracting the rate of spontaneous mEPSCs. Data are given as means ±s.e.m. (number of cells), unless otherwise noted.

Figure 1. EPSCs and mEPSC-like currents at immature synapses.

A, sample traces of EPSCs recorded for 100 ms before and after electrical stimulation (P4). Five sequential traces are superimposed. B, event time histogram of 50 consecutively recorded EPSCs, some of which are shown in A. C, asynchronicity index, computed for mEPSC-like currents recorded from 20 to 80 ms before (open bar) and after stimulation (hatched bar, 3 cells). Means ±s.e.m. are shown in this and in all subsequent figures. D and E, all-or-none nature of EPSC and mEPSC-like currents at immature synapses (P5). D, sample traces of EPSCs generated by stimuli of 2, 3 and 8 V. E, mean peak amplitudes of EPSCs plotted against stimulus intensity. Each point is the average of 5 records.

Figure 2. Evoked EPSCs at mature synapses and their asynchronicity indices.

Aa, EPSCs from a mature synapse (P10). Five records are superimposed. b, late phases of EPSCs shown in a at magnified amplitude. A mEPSC-like current is indicated by an asterisk. c, event time histogram of EPSCs measured from 50 consecutively recorded traces. B, asynchronicity indices of immature (open bar, 11 cells) and mature EPSCs (hatched bar, 7 cells).

RESULTS

We have classified animals older than P9 as mature and younger than P6 as immature in respect to the nature of synaptic transmission of the MNTB synapse (see Introduction). Since our previous work was limited to synapses younger than P13 (Chuhma & Ohmori, 1998), EPSCs of much older animals (P15-17) were compared with those of P9-11 animals first. We found no differences in several EPSC parameters between these two mature groups (Table 1). The EPSC amplitude showed a small difference between P9-11 and P15-17, but this difference was not statistically significant (P = 0.32, Student’s paired t test). Very large EPSCs were occasionally observed by us (see Fig. 2A, 7.2 nA at -70 mV, P10) and were also reported by others (8-9 nA) in P8-11 synapses (Borst & Sakmann, 1996; von Gersdorff et al. 1997).

Table 1.

EPSCs in mature animals

	Rise time (20–80%) (ms)	Decay time constant (ms)	Coefficient of variation*	Amplitude at −70 mV (nA)	Asynchronicity indices	Number of cells tested
P9–11	0.26 ± 0.01	1.29 ± 0.10	0.07 ± 0.01	4.1 ± 0.8	0.10 ± 0.02	7
P15–17	0.27 ± 0.04	1.07 ± 0.07	0.09 ± 0.01	5.2 ± 0.2	0.10 ± 0.02	3
P	0.88	0.16	0.19	0.32	0.85

^*Coefficient of variation of the EPSC amplitude. Statistical significance was determined by Student’s t test.

Asynchronous nature of synaptic transmission in the immature synapse

Many mEPSC-like currents followed EPSCs at immature synapses but the occurrence of delayed mEPSC-like currents was rare at mature synapses (see Fig. 2-3 of Chuhma & Ohmori, 1998). It was also rare to observe mEPSC-like currents before electrical stimulation was applied to the presynaptic fibre at immature synapses (Fig. 1A). In the event time histogram made from 50 consecutive records, only one mEPSC-like current was seen before electrical stimulation, but many were observed after the stimulation (Fig. 1B). The mean number of occurrences during the 60 ms window period (asynchronicity index, see Methods) was 0.16 ± 0.10 before stimulation and 1.25 ± 0.28 after stimulation (n = 3 cells, Fig. 1C). It is not likely that these delayed mEPSC-like currents were due to spontaneous release.

Figure 3. Effects of Ca²⁺ channel blockers at immature synapses (P4-5).

A, effects of ω-agatoxin IVA (200 nm) at a P4 synapse. a and c, event time histogram of control EPSCs and in ω-agatoxin. All histograms in this figure were measured from 30 consecutive traces. b and d, EPSCs of control and after application of ω-agatoxin recorded from the same cell. Six traces are superimposed in all sample records of this figure. B, effects of ω-conotoxin GVIA (2 μm) at a P5 synapse. a and c, event time histogram of control and after application of ω-conotoxin. b and d, EPSCs of control and in ω-conotoxin recorded from the same cell. C, effects of toxins on asynchronicity indices (hatched bar) with corresponding controls (open bar) recorded in the same cells (n = 5 cells).

When the intensity of presynaptic stimulation at immature synapses was increased, EPSCs were generated in an all-or-none manner (P5, Fig. 1D and E), as was observed at mature synapses (Barnes-Davies & Forsythe, 1995). Both the large first EPSC and the mEPSC-like currents in the later phase emerged only when the stimulus intensity to the presynaptic fibre exceeded threshold (3 V, in Fig. 1D). When the stimulus was of subthreshold intensity, there were no mEPSC-like currents. At suprathreshold intensities, both the amplitudes of the first EPSCs and the frequencies of delayed mEPSC-like currents were not affected by the stimulus intensity (Fig. 1D and E). The asynchronicity index was almost independent of the stimulus intensity, and was 1.8 at 3.0 V, 2.3 at 6 V and 1.8 at 8 V in this synapse. Similar results were observed in three other experiments. This all-or-none appearance of the EPSC and mEPSC-like currents strongly indicates that the two have a common origin, and reflect activity of a single presynaptic fibre.

The average amplitude of these mEPSC-like currents was 36.9 ± 3.1 pA (n = 5 cells) at -70 mV and their decay time constant was 2.1 ± 0.2 ms (n = 5 cells). These values were not significantly different from those of mEPSCs induced by Ba²⁺ at P4-13 synapses: 30.2 ± 0.3 pA for the amplitude and 2.1 ± 0.1 ms for the decay time constant (Chuhma & Ohmori, 1998).

Synchronisation of synaptic transmission in MNTB synapses during development

At mature MNTB synapses, electrical stimulation applied to the presynaptic fibre generated a single large EPSC with little amplitude or timing fluctuation (Fig. 2Aa). In these mature synapses, the occurrence of mEPSC-like currents was rare (Fig. 2Ac). The event time histogram shows only two mEPSC-like currents in 50 consecutive records (one of them is indicated by an asterisk in Fig. 2Ab).

The asynchronicity index was thus close to 0 for mature synapses (Fig. 2B, 0.10 ± 0.02, n = 7 cells, P9-11), and was 1.34 ± 0.21 (n = 11 cells) for immature synapses. The asynchronicity index of much older animals (P15-17) was the same as for P9-11 (Table 1). The decreasing trend of asynchronicity indices during development was not affected by higher experimental temperature (35°C). The index was 2.8 ± 0.5 at P6 (n = 4 cells) and 0.1 ± 0.02 at P11 (n = 3 cells).

Subtypes of presynaptic Ca²⁺ channels partly modify the asynchronous nature of release at immature synapses

The P/Q-type is reported to be the major subtype of Ca²⁺ channels in the mature presynaptic terminal of MNTB, but both P/Q-type and N-type Ca²⁺ channels coexist in the immature terminal (Wu et al. 1997; Iwasaki & Takahashi, 1998). We have investigated the possibility that particular subtypes of Ca²⁺ channels might contribute to the generation of delayed mEPSC-like currents in the immature synapse.

Both ω-conotoxin GVIA (1-2 μm) and ω-agatoxin IVA (200 nm) reduced the size of EPSCs at P4-5 synapses to a similar extent (Fig. 3A and B and Table 2). However, delayed mEPSC-like currents were differentially affected, and event time histograms clearly indicate a reduced incidence of delayed mEPSC-like currents when ω-conotoxin was applied (Fig. 3Bc). The asynchronicity index was significantly reduced by ω-conotoxin (Table 2, P < 0.01, paired t test), but was slightly increased by ω-agatoxin (Fig. 3C and Table 2, P > 0.2). These results suggest that these two subtypes of Ca²⁺ channels are present in the presynaptic terminal and contribute equally to the synchronized transmitter release at immature synapses, but that their contributions to delayed asynchronous release are different.

Asynchronous transmitter release was suppressed by EGTA loading into the immature presynaptic terminal

Increased residual [Ca²⁺]_i in the presynaptic terminal is known to increase the frequency of mEPSCs (Hubbard et al. 1968; Blioch et al. 1968). We assumed that the occurrence of mEPSC-like currents in the immature synapse might be due to a higher level of [Ca²⁺]_i after each action potential. To test this hypothesis we loaded extrinsic Ca²⁺ buffer into the presynaptic terminal by incubating the immature slices with EGTA AM.

An initial large EPSC was observed as in the control slice, but delayed mEPSC-like currents were absent in slices incubated with EGTA AM (Fig. 4Ab and d). The asynchronicity index was close to 0 (0.03 ± 0.01, n = 5 cells, Fig. 4B). In control sister slices (see Methods), robust mEPSC-like currents followed the first large EPSC, and the asynchronicity index was 1.39 ± 0.28 (n = 5 cells, Fig. 4Aa and c and B). This difference was statistically significant (P < 0.01, t test).

Peak EPSC amplitudes in control and EGTA-loaded slices were not different: 0.57 ± 0.24 nA in control slices (n = 5 cells) and 0.76 ± 0.15 nA in EGTA-loaded slices (n = 5 cells). At mature synapses, as at immature synapses, incubation with EGTA AM did not change the amplitude of EPSCs. The concentration of EGTA loaded presynaptically might have accelerated the decay of [Ca²⁺]_i (Chen & Regehr, 1999), but this was apparently not enough to affect the transient rise of [Ca²⁺]_i induced by action potentials. Borst & Sakmann (1996) reported that 1 mm EGTA decreased EPSC amplitude but that 200 μm EGTA had no effect on the EPSC when loaded into the presynaptic terminal through the patch electrode.

Resting [Ca²⁺]_i of presynaptic terminals, estimated by fura-2 fluorometry, was lower at mature terminals (181 ± 55 nm, n = 21 cells) than at immature terminals (337 ± 63 nm, n = 9 cells). These estimates are slightly higher than that (46 ± 7 nm) of Helmchen et al. (1997) at mature MNTB terminals. Although these estimates of resting [Ca²⁺]_i may not be exact (because of the factor of background subtraction, see Methods), mature terminals may well have lower resting [Ca²⁺]_i than immature terminals.

Effects of reduced Ca²⁺ influx in the mature synapses

The development of endogenous Ca²⁺ buffers may be correlated with an increase in Ca²⁺ influx into presynaptic terminals. Ca²⁺ current amplitudes of immature presynaptic terminals were 43% of those of mature terminals when compared at the potential giving maximum current amplitude (Fig. 7 of Chuhma & Ohmori, 1998). If this low Ca²⁺ influx was an essential factor for induction of the delayed mEPSC-like currents of immature synapses, delayed mEPSC-like currents might appear at mature synapses if the Ca²⁺ influx per action potential was reduced. We therefore reduced the total Ca²⁺ influx into mature presynaptic terminals by two methods: either by decreasing [Ca²⁺]_o to reduce Ca²⁺ influx through each Ca²⁺ channel evenly, or by blocking a fraction of Ca²⁺ channels while leaving others normal.

When [Ca²⁺]_o was reduced from 2 to 1 mm at the mature synapse (Fig. 5A), EPSC amplitudes were reduced to the level observed at immature synapses (0.59 ± 0.14 nA, n = 4 cells). However, delayed mEPSC-like currents were not induced. The event time histogram in 1 mm[Ca²⁺]_o demonstrated only a single large peak (Fig. 5Ab), and the asynchronicity index remained close to 0 (0.18 ± 0.06, n = 4 cells, Fig. 5D). Essentially similar results were obtained when [Ca²⁺]_o was reduced to 0.5 mm, where the asynchronicity index was 0.11 ± 0.06 (n = 3 cells) and the EPSC amplitude was reduced to 16.9 ± 3.2% of the control.

Figure 5. Effects of reduced Ca²⁺ influx at mature synapses (P10).

Aa, EPSCs after reduction of [Ca²⁺]_o to 1 mm. Six to seven traces are superimposed in the sample records shown in this figure. b, event time histogram. All histograms in this figure were made from 50 consecutive records. Ba, EPSCs in 3 μm Cd²⁺. b, event time histogram. Ca, EPSCs in ω-agatoxin (50 nm). b, event time histogram. D, asynchronicity indices in 1 mm[Ca²⁺]_o(n = 4 cells), after fractional block of Ca²⁺ channels by Cd²⁺ (3 μm, n = 6 cells) and by ω-agatoxin (50 nm, n = 3 cells) indicated with corresponding controls (open bars).

Reduced EPSC size at this low [Ca²⁺]_o did not affect the decay phase of EPSCs (1.26 ± 0.10 ms for control and 1.10 ± 0.10 ms for 1 mm[Ca²⁺]_o, P = 0.37), and probably did not affect desensitization of postsynaptic glutamate receptors either. While cyclothiazide (CTZ; 20 μm; Tocris), which inhibits or largely slows down the AMPA-receptor desensitization (Patneau et al. 1993), prolonged the decay time constant of EPSCs from 1.36 ± 0.25 to 2.34 ± 0.27 ms (n = 4 cells), delayed release was not observed at mature synapses (P9-13). The decay time course of mEPSCs was similarly prolonged (from 1.6 ms to 2.8 ms). It is unlikely that delayed release at mature synapses was masked by desensitization of postsynaptic glutamate receptors.

Partial block of presynaptic Ca²⁺ channels in the mature synapse

We then reduced total influx of Ca²⁺ to the presynaptic terminal by blocking some fraction of Ca²⁺ channels using low concentrations of either Cd²⁺ or ω-agatoxin. Cd²⁺ reduced EPSC amplitudes with a K_D of 1.77 μm, and cooperativity of 1.76 when added in ACSF. Cd²⁺ is known to induce flickering block of Ca²⁺ channels with a block duration of about 1 ms (L-type Ca²⁺ channel; Nilius et al. 1985). These intervals are equivalent to the duration of Ca²⁺ influx (0.8 ms) during an action potential (Borst & Sakmann, 1996), and a proportion of Ca²⁺ channels are likely to be completely blocked during a spike.

Cd²⁺ (3 μm) reduced the EPSC size to about 20% of the control, and induced mEPSC-like currents in the late phase of traces (Fig. 5B). In the event time histogram, a substantial number of events were scored in the late phase following the first dominant peak (Fig. 5Bb). The asynchronicity index was 0.45 ± 0.11 (n = 6 cells) and was larger than that in the control (0.11 ± 0.02, P < 0.05, paired t test) measured in the same synapse (Fig. 5D). However, the frequency of mEPSC-like currents was still lower than that at immature synapses for which the asynchronicity index was 1.34 ± 0.21 (n = 11 cells). When Ca²⁺ transients from the mature presynaptic terminals were monitored, this concentration of Cd²⁺ (3 μm) reduced the size of the Ca²⁺ transient to 55 ± 20% of control (n = 3 cell) but did not affect the decay time constant (τ_decay was 1.76 ± 0.24 s in control and 1.60 ± 0.33 s in Cd²⁺, n = 3 cells). We have tried higher concentrations of Cd²⁺ (5-10 μm), but these concentrations blocked EPSCs completely (10 μm).

ω-Agatoxin induced delayed mEPSC-like currents when applied at a concentration (50 nm) lower than that (200 nm) used to completely block P/Q-type Ca²⁺ channels (Fig. 5C). EPSC amplitudes were reduced to 20-30% of control, and mEPSC-like currents followed the EPSC. The asynchronicity index was 0.50 ± 0.18 (n = 3 cells), larger than the control (0.05 ± 0.02) recorded from the same synapses (Fig. 5D). Recordings of partially blocked EPSCs were difficult with ω-agatoxin, and we were only successful in three cases out of 13 trials (see also Methods). In other cases EPSCs were either blocked completely, or reduced only slightly from the control size. Since Cd²⁺ reproduced delayed mEPSC-like currents more consistently than did ω-agatoxin, we used 3 μm Cd²⁺ in the following experiments.

Partial block of Ca²⁺ channels with Cd²⁺ reduces the Ca²⁺ dependence of transmitter release

The amplitude of synaptic currents is dependent on [Ca²⁺]_o, and increases as a power function of [Ca²⁺]_o (Dodge & Rahamimoff, 1967; Barnes-Davies & Forsythe, 1995; Takahashi et al. 1996; Wu et al. 1999). We measured EPSC amplitudes after partial block by Cd²⁺ at 0.5-5 mm[Ca²⁺]_o, and compared the sensitivity and cooperativity in relation to [Ca²⁺]_o between immature and mature EPSCs. The amplitude of EPSCs (I) relative to the one recorded at 5-10 mm[Ca²⁺]_o (I_max) was transformed as (I/I_max)/(1 - I/I_max) and plotted against [Ca²⁺]_o on log-log coordinates (Fig. 6). The slope of the plot gives the cooperativity (n) to Ca²⁺. These plots were evaluated by the Hill equation (eqn (4)) and parameters obtained are listed in Table 3:

(4)

where K_EPSC is the dissociation constant.

Table 3.

[Ca²⁺]_o sensitivity of EPSC amplitude

Postnatal days	Hill coefficient	KEPSC (mm)	No. of cells tested
P4–5	2.0	1.27	2–10
P10–11	3.7	1.05	2–9
P9–10 Cd2+*	2.2	1.43	3–12

^*Cd²⁺ (3 μm) was added in the ACSF. The Hill coefficient and K_EPSC were calculated from Fig. 6 by eqn (4). Number of cells is the number of experiments in each plot of Fig. 6.

n was larger and K_EPSC was smaller at mature synapses than at immature synapses. When mature EPSCs were partially blocked by Cd²⁺, n became smaller and K_EPSC larger, and the overall features were closer to those of immature EPSCs (Table 3). This suggests that partial block of presynaptic Ca²⁺ channels at mature synapses changes aspects of Ca²⁺ dependency to be similar to those observed at immature synapses.

Estimation of Ca²⁺ clearance capacity in the presynaptic terminals

We have shown (Fig. 4) that the occurrence of delayed mEPSC-like currents was reduced after introduction of an extrinsic Ca²⁺ buffer into immature presynaptic terminals. This result suggests that the increase in [Ca²⁺]_i after an action potential might be prolonged in immature presynaptic terminals possibly because of a smaller capacity for Ca²⁺ clearance.

The decay time course of the intraterminal Ca²⁺ transient is affected by several factors including the endogenous Ca²⁺ binding capacity, the extrusion rate for Ca²⁺, and particularly the concentration of exogenous Ca²⁺ buffers (fura-2) during fluorometric experiments (Neher & Augustine, 1992; Helmchen et al. 1997). When the presynaptic terminal Ca²⁺ transient was monitored repeatedly at an immature synapse (P6), its time course of decay was prolonged as the concentration of fura-2 within the terminal increased (two insets in Fig. 7A). Fura-2 concentration was monitored as the fluorescence intensity at 360 nm excitation (circles in Fig. 7A). Figure 7B illustrates Ca²⁺ transients from immature (P6, dotted line) and mature (P11, continuous line) presynaptic terminals when saturated with 200 μm fura-2, superimposed after scaling in order to facilitate comparison. The time constant of decay was 2.5 ± 0.52 s (P6, n = 8 cells) for immature terminals and was 0.9 ± 0.3 s (P10-11, n = 6 cells) for mature terminals.

The Ca²⁺ binding capacity of exogenous buffer (k_B) was calculated (eqn (2)) in each trial and the Ca²⁺ transient decay time constant (τ_decay) was plotted against k_B (Fig. 7C). A near linear relationship was observed for both mature (open symbols, n = 3 cells) and immature (filled symbols, n = 4 cells) terminals. However, there are marked differences in slope and x-intercept. The slope represents the reciprocal of the Ca²⁺ extrusion rate (γ) and the x-intercept (k_S+ 1) gives an estimate of the endogenous Ca²⁺ binding capacity (Helmchen et al. 1997 and see eqn (3)). The endogenous Ca²⁺ binding capacity (k_S) thus estimated for mature synapses was 60 and the extrusion rate was 185 s⁻¹. These estimates are similar to the ones reported for mature MNTB synapses by Helmchen et al. (1997). k_S for immature synapses was 16 and γ was 29 s⁻¹. These results indicate that presynaptic terminals at immature MNTB synapses have smaller Ca²⁺ extrusion rates and smaller endogenous Ca²⁺ binding capacities than terminals at mature synapses.

DISCUSSION

The immature presynaptic terminal of MNTB is not of calyceal shape but is a plate-like structure with many filopodial-like processes (Fig. 1B of Chuhma & Ohmori, 1998) and a large surface area (300-600 μm² was estimated for a P5 synapse, Fig. 13 of Kandler & Friauf, 1993). This large immature presynaptic terminal, combined with the relatively small capacity for Ca²⁺ clearance (Fig. 7), might facilitate generation of delayed mEPSC-like currents due to the longer lifetime of the Ca²⁺ transient. A lower Ca²⁺ clearance capacity is matched by smaller Ca²⁺ currents at immature terminals (Chuhma & Ohmori, 1998). Mature presynaptic terminals have large Ca²⁺ currents (Chuhma & Ohmori, 1998), together with a well-developed capacity for Ca²⁺ clearance (Fig. 7C). A large fraction of the readily releasable synaptic vesicles seem to be released by a single action potential at mature synapses (Chuhma & Ohmori, 1998). Increases in both Ca²⁺ currents and Ca²⁺ clearance seem essential to achieving synchronization of transmitter release, because (1) delayed release at immature synapses was eliminated by loading with extrinsic Ca²⁺ buffer (Fig. 4) and (2) delayed release emerged after fractional block of Ca²⁺ channels at mature terminals, either by ω-agatoxin or by Cd²⁺ (Fig. 5). After fractional block of Ca²⁺ channels at mature synapses, both the sensitivity and the cooperativity of synaptic transmission in relation to [Ca²⁺]_o were reduced, and the overall features of synaptic function became similar to those of immature synapses (Fig. 6). These observations may indicate that delayed asynchronous release of transmitter at immature synapses emerges as a consequence of relatively loose linkage between Ca²⁺ channels and synaptic vesicles, with the looseness of this linkage exaggerated by the small capacity for Ca²⁺ clearance. However, this hypothesis is still preliminary and does not exclude other additional possibilities, one of which would be a developmental increase of the affinity of Ca²⁺ sensor molecules. Synaptotagmin is thought to be the most likely candidate for the Ca²⁺ sensor (Geppert et al. 1994) and several subtypes have been identified (Li et al. 1995).

The immature synapse of MNTB may be reminiscent of synapses during path finding, where the leading edge of an axon terminal extends many filopodia and releases transmitter before contacting the target neurone (Evers et al. 1989). Growth cones are reported to increase [Ca²⁺]_i by generating repetitive action potentials after contacting the target (Funte & Haydon 1993), and the direction of growth may be determined by local concentrations of intracellular Ca²⁺ (Zheng, 2000). The shapes of presynaptic terminals, their Ca²⁺ dynamics, and the asynchronous nature of neurotransmitter release give immature terminals many similarities to developing synapses, and these properties may be essential in forming the permanent mature synapse in MNTB.

The probability of release at mature MNTB synapses estimated by assuming binomial statistics was very large (0.87; Chuhma & Ohmori, 1998). This estimate was even larger than that estimated for P8-10 MNTB synapses of rats of 0.2, determined from the ratio of relative EPSC amplitude to those corresponding to the maximum size of the synaptic vesicle pool by Schneggenburger et al. (1999). They further demonstrated the [Ca²⁺]_o dependence of EPSC amplitude. When estimated from their Fig. 1C, the Hill coefficient n was 2.4 and K_EPSC was 1.5 mm (eqn (4)). These values are closer to those of immature EPSCs and those of mature EPSCs after partial block with Cd²⁺ (Fig. 6 and Table 3). The differences between our observations and those of Schneggenburger et al. (1999) may be due to the different method of estimation, as well as the different stage of maturation of animals.

Borst & Sakmann (1996) observed asynchronous EPSCs at very low [Ca²⁺]_o (0.25 mm) at P8-10 rat MNTB synapses, and Isaacson & Walmsley (1995) made similar observations for bushy cell synapses of P12-19 rat AVCN after applying a relatively high concentration of Cd²⁺ (100 μm). Both experiments substantially reduced the release probability, and EPSCs of a single quantum or a few quanta were evoked with many failures. These EPSCs demonstrated extensive fluctuations in amplitude and in delay after stimulation. The delay fluctuation was extended to 5 ms or more. These EPSCs were not accompanied by delayed mEPSC-like currents which were observed for nearly 100 ms as was reported here (see Fig. 1). When we reduced [Ca²⁺]_o to 0.5-1 mm, EPSC amplitude was reduced but delayed mEPSC-like currents were not observed (Fig. 5A). Delayed mEPSC-like currents were only observed at mature synapses when the EPSC amplitude was reduced by conditions where only a fraction of Ca²⁺ channels were blocked either by Cd²⁺ (3 μm) or by ω-agatoxin (50 nm) (Fig. 5). We do not believe that delayed release arose due to flux of Cd²⁺ into the presynaptic terminal. Cd²⁺ reduced the Ca²⁺ transient but did not affect its decay time course (Results, ‘Partial block of presynaptic Ca²⁺ channels in the mature synapse’), and Cd²⁺ blocked EPSCs at slightly higher concentrations (5-10 μm). Moreover, delayed mEPSC-like currents were observed after block of Ca²⁺ channels by the low concentration of ω-agatoxin (Fig. 5C). We have not examined EPSCs at much higher levels of Cd²⁺ (100 μm). At a low Cd²⁺ concentration (5 μm), Otis & Trussell (1996) reported a slight acceleration of the EPSC decay time course (to 88% of control) in the nucleus magnocellularis of the chick cochlear nucleus (chick embryo E16-21). After ensemble averaging, the possible presence of delayed release could not be assessed from their records. The decay phase of ensemble averaged EPSCs was not significantly affected in our experiments in 3 μm Cd²⁺, although it accelerated slightly to 95% of the control.

Subtypes of Ca²⁺ channels

Blocking N-type Ca²⁺ channels with ω-conotoxin reduced the emergence of asynchronous mEPSC-like currents in the immature terminal, while block of P/Q-type Ca²⁺ channels only slightly enhanced their occurrence (Fig. 3). Wu et al. (1999) demonstrated closer placement of P/Q-type Ca²⁺ channels than N-type channels to the active zone in P8-10 MNTB synapses, and that P/Q-type Ca²⁺ channels trigger the release of a larger fraction of neurotransmitter than do N- or R-type Ca²⁺ channels. However, both N-type and P/Q-type Ca²⁺ channels contribute equally to the synchronous release at immature synapses (Fig. 3, Table 2). These observations may suggest that the two subtypes of Ca²⁺ channels at immature presynaptic terminals are not segregated to the same extent as they are at mature terminals.

Asynchronous transmitter release and Ca²⁺ channel blockers

Asynchronous release of neurotransmitter was reported when Sr²⁺ was added to the extracellular medium (Goda & Stevens, 1994). At mature MNTB synapses, Sr²⁺ (2 mm) in ACSF induced delayed mEPSC-like currents, and the asynchronicity index was 0.64 ± 0.17 (n = 4 cells, N. Chuma, K. Koyano & H. Ohmori, unpublished observation). Affinities of Sr²⁺ for all subtypes of synaptotagmin are lower than those of Ca²⁺ (Li et al. 1995), and the lifetime of Sr²⁺ in the presynaptic terminals was reported to be longer than that reported for Ca²⁺ (Xu-Friedman & Regehr 1999). When a peptide having the synprint sequence was introduced into presynaptic terminals of cultured superior cervical sympathetic ganglion neurones of rats, the size of EPSCs was reduced and the occurrence of EPSPs became asynchronous (Mochida et al. 1996). When a similar experiment was performed on a co-cultured neurone-muscle preparation of Xenopus the sensitivity of synaptic transmission to Ca²⁺ was reduced (Retting et al. 1997). This group have further demonstrated that when Ca²⁺ channels were partially blocked by ω-conotoxin GVIA, the Ca²⁺ sensitivity of synaptic transmission was reduced to that seen in the presence of synprint in the presynaptic terminal. Synprint is the synaptic protein interaction site of the N-type Ca²⁺ channel, and these phenomena could be interpreted as being due to the obstruction of linkages between Ca²⁺ channels and synaptic vesicles by synprint.

We have reproduced similar phenomena, delayed asynchronous release (Fig. 5), reduced cooperativity and reduced affinity of EPSC amplitude to [Ca²⁺]_o (Table 3), by reducing Ca²⁺ influx into presynaptic terminals by fractional block of Ca²⁺ channels either by Cd²⁺ or by ω-agatoxin at mature synapses. Since the simple reduction of [Ca²⁺]_o did not reproduce these phenomena (Fig.5), we have assumed that these Ca²⁺ channel blockers produced a patchy arrangement of operating Ca²⁺ channels and non-operating Ca²⁺ channels. If most of the docked readily releasable synaptic vesicles were linked with Ca²⁺ channels, then some vesicles would be sited at a distance from active Ca²⁺ channels after channel block, and consequently these vesicles would be exposed to reduced [Ca²⁺]_i after a delay due to diffusion from nearby non-blocked Ca²⁺ channels. The low Ca²⁺ cooperativity and reduced Ca²⁺ sensitivity of the immature synapse (Table 3) suggest that the linkage between Ca²⁺ channels and transmitter vesicles might not be as tight in immature presynaptic terminals as it is in mature terminals.