Changes in synaptic transmission properties due to the expression of N-type calcium channels at the calyx of Held synapse of mice lacking P/Q-type calcium channels

Abstract

P/Q-type and N-type calcium channels mediate transmitter release at rapidly transmitting central synapses, but the reasons for the specific expression of one or the other in each particular synapse are not known. Using whole-cell patch clamping from in vitro slices of the auditory brainstem we have examined presynaptic calcium currents (I_pCa) and glutamatergic excitatory postsynaptic currents (EPSCs) at the calyx of Held synapse from transgenic mice in which the α_1A pore-forming subunit of the P/Q-type Ca²⁺ channels is ablated (KO). The power relationship between Ca²⁺ influx and quantal output was studied by varying the number of Ca²⁺ channels engaged in triggering release. Our results have shown that more overlapping Ca²⁺ channel domains are required to trigger exocytosis when N-type replace P/Q-type calcium channels suggesting that P/Q type Ca²⁺ channels are more tightly coupled to synaptic vesicles than N-type channels, a hypothesis that is verified by the decrease in EPSC amplitudes in KO synapses when the slow Ca²⁺ buffer EGTA-AM was introduced into presynaptic calyces. Significant alterations in short-term synaptic plasticity were observed. Repetitive stimulation at high frequency generates short-term depression (STD) of EPSCs, which is not caused by presynaptic Ca²⁺ current inactivation neither in WT or KO synapses. Recovery after STD is much slower in the KO than in the WT mice. Synapses from KO mice exhibit reduced or no EPSC paired-pulse facilitation and absence of facilitation in their presynaptic N-type Ca²⁺ currents. Simultaneous pre- and postsynaptic double patch recordings indicate that presynaptic Ca²⁺ current facilitation is the main determinant of facilitation of transmitter release. Finally, KO synapses reveal a stronger modulation of transmitter release by presynaptic GTP-binding protein-coupled receptors (γ-aminobutyric acid type B receptors, GABA_B, and adenosine). In contrast, metabotropic glutamate receptors (mGluRs) are not functional at the synapses of these mice. These experiments reinforce the idea that presynaptic Ca²⁺ channels expression may be tuned for speed and modulatory control through differential subtype expression.

Transmitter release at central synapses is triggered by calcium influx through voltage-gated calcium channels (VGCCs). At many synapses, transmitter release is mediated by multiple calcium channel subtypes early during development, but it increasingly relies on P/Q-type Ca²⁺ channels with maturation, as has been shown at the neuromuscular junction (Rosato Siri & Uchitel, 1999, Rosato-Siri et al. 2002) and at the calyx of Held synapse (Iwasaki & Takahashi, 1998, Fedchyshyn & Wang, 2005). In mice, deletion of the α_1A pore-forming subunit of Ca_v2.1 (P/Q-type) calcium channels induces a progressive syndrome characterized by ataxia, dystonia and early death around 20 days after birth. These progressive neurological alterations underscore the importance of α_1A for normal function of the central nervous system. In humans, diseases arising from defects in this subunit include certain forms of migraine, epilepsy and ataxia (Pietrobon, 2002). Thus, genetic ablation of the α_1A subunit (KO) allows a critical examination of those features of neurotransmission which depend on this particular channel type (Jun et al. 1999; Fletcher et al. 2001) and provides an excellent model for studying fast synaptic transmission.

Transmission at the neuromuscular junction of the KO mouse is codependent on both N- and R-type Ca²⁺ channels (Urbano et al. 2003). R-type channels are positioned close to the Ca²⁺ sensors for exocytosis while the N-type channels are less well localized, but nevertheless influence secretion strongly, particularly when action potentials are prolonged. Transmission also displays lower quantal content but greater ability to withstand reductions in the Ca²⁺/Mg²⁺ ratio, and little or no paired-pulse facilitation. Further studies at the neuromuscular junction are technically limited by the difficulty in measuring Ca²⁺ currents under voltage clamp conditions at this site.

The large synapse formed by the calyx of Held presynaptic terminals onto principal cells of the medial nucleus of the trapezoid body (MNTB) offers the means to make direct electrophysiological analysis of presynaptic Ca²⁺ currents (I_pCa) and induced excitatory postsynaptic currents (EPSCs). This synapse is an axosomatic glutamatergic synapse critically involved in sound localization (Forsythe, 1994). At the calyx of Held synapse of young animals, cooperative actions of multiple Ca²⁺ channels are required for the release of single synaptic vesicles but developmental changes reduce the number of recruited Ca²⁺ channels by the presynaptic action potential and at the same time increase their coupling efficacy (Yang & Wang, 2006). Using the calyx of Held synapse preparation from KO mice, we and others have shown that N-type channels functionally compensate for the absence of P/Q subunits although EPSC facilitation is greatly diminished and direct recording of presynaptic calcium currents revealed the absence of calcium-dependent facilitation (Inchauspe et al. 2004, 2005).

Our results demonstrate that replacing presynaptic P/Q calcium channels by N-type calcium channel results in a functional synapse with only a slight decrease in EPSC amplitudes, but that presents several changes in synaptic plasticity and in its regulation by presynaptic receptors.

In KO synapses, N-type channels seem to be located at greater distances from actives sites of the release machinery, and consequently Ca²⁺ influx through these channels is less effective in triggering transmitter release. The relation between EPSCs and I_pCa indicates that in KO synapses an increased number of overlapping Ca²⁺ channel domains are required to trigger exocytosis.

Changes in synaptic plasticity properties include no pair pulse facilitation (of both presynaptic and postsynaptic currents) and a slower rate of replenishment of the releasable pool, which slows down the recovery of EPSCs after the short-term depression induced by repetitive stimulation at high frequencies.

We also show that KO synapses are more susceptible to inhibitory modulation by G protein-coupled receptors (GPCRs), like GABA_B and adenosine receptors, than are their wild-type counterparts.

Methods

Preparation of brainstem slices

Mice of 11–15 days old were killed by decapitation, and the brain was removed rapidly and placed into an ice-cold low-sodium artificial cerebrospinal fluid (aCSF). The brainstem was mounted in the Peltier chamber of an Integraslice 7550PSDS (Campden Instruments Ltd, UK) vibrating microslicer. Transverse slices containing MNTB were cut sequentially and transferred to an incubation chamber containing normal aCSF with low calcium (0.1 mm CaCl₂ and 2.9 mm MgCl₂) at 37°C for 1 h. After incubation the chamber was allowed to return to room temperature. Slices of 200 μm thickness were used for presynaptic Ca²⁺ current recordings and 300 μm for EPSC recordings. Normal aCSF contained (mm): NaCl 125, KCl 2.5, NaHCO₃ 26, NaH₂PO₄ 1.25, glucose 10, ascorbic acid 0.5, myo-inositol 3, sodium pyruvate 2, MgCl₂ 1 and CaCl₂ 2. Low sodium aCSF was as above but NaCl was replaced by 250 mm sucrose and MgCl₂ and CaCl₂ concentrations were 2.9 mm and 0.1 mm, respectively. The pH was 7.4 when gassed with 95% O₂–5% CO₂.

Electrophysiology

Slices were transferred to an experimental chamber perfused with normal aCSF at 25°C. MNTB neurons were visualized using Nomarski optics on a BX50WI (Olympus, Japan) microscope, and a 60×, 0.90 NA water immersion objective lens (LUMPlane FI, Olympus). Whole-cell voltage clamp recordings were made with patch pipettes pulled from thin-walled borosilicate glass (Harvard Apparatus, GC150F-15, UK). Electrodes had resistances of 3.6–4.2 MΩ for presynaptic recordings and 3.0–3.5 MΩ for postsynaptic recordings, when filled with internal solution. Patch solutions for voltage clamp recordings contained (mm): CsCl 110, Hepes 40, TEA-Cl 10, Na₂phosphocreatine 12, EGTA 1, MgATP 2, LiGTP 0.5 and MgCl₂ 1. The pH was adjusted to 7.3 with CsOH. To block Na⁺ currents and avoid postsynaptic action potentials, 10 mmN-(2,6-diethylphenylcarbamoy-lmethyl)-triethyl-ammonium chloride (QX-314) was added to the pipette solution when recording EPSCs. Lucifer Yellow was also included for visualizing presynaptic terminals.

Patch clamp recordings were made using Multiclamp 700A amplifier (Axon Instruments, Union City, CA, USA), a Digidata 1322A (Axon Instruments) and pCLAMP 9.0 software (Axon Instruments). Data were sampled at 50 kHz and filtered at 4–6 kHz (Low pass Bessel). Series resistances ranged from 6 to 15 MΩ. Whole-cell membrane capacitances (15–25 pF) were registered from the amplifier after compensation of the transient generated by a 10 ms voltage step. Leak currents were subtracted on-line with a P/4 protocol. Ca²⁺ currents were recorded in the presence of TTX (1 μm) and TEA-Cl (10 mm). EPSCs were evoked by stimulating the globular bushy cell axons in the trapezoid body at the midline using a bipolar platinum stimulating electrode and an isolated stimulator (stimuli of 0.1 ms duration and voltage amplitude between 4 and 7 V), or by evoking a presynaptic Ca²⁺ current. Strychnine (1 μm) was added to the external solution to block inhibitory glycinergic synaptic responses. When high frequency stimuli were applied, kynurenic acid (1 mm, Tocris) was used to minimize postsynaptic AMPA receptor saturation.

l-(1)-2-Amino-4-phosphonobutyric acid (l-AP4), 1-aminocyclopentane-1,3-dicarboxylic acid (ACPD) and baclofen were purchased from Tocris.

Presynaptic action potentials (APs) were measured in whole-cell configuration under current clamp mode. Patch solutions for current clamp recordings contained (mm): potassium gluconate 110, KCl 30, Hepes 10, Na₂phosphocreatine 10, EGTA 0.2, MgATP 2, LiGTP 0.5 and MgCl₂ 1. Only cells that have a membrane potential between −68 mV and −78 mV without current injection were selected for recording. Electrode series resistance and electrode capacitance were compensated electronically. APs were elicited by injecting step current pulses between 0.1 nA and 0.2 nA for 0.25 ms.

Average data are expressed and plotted as means ±s.e.m. Statistical significance was determined using Student's t test.

Animals were kept and used in accordance with the UK Animals (Scientific Procedures) Act 1986.

Results

Calcium dependence of transmitter release

Neurotransmitter release depends strongly on the extracellular Ca²⁺ concentration and calcium influx into the presynaptic terminal. The relation is nonlinear and can be approximated by the following power relationship:

(1)

where r is an estimative measure of the cooperative binding of the intracellular Ca²⁺ to the sensor at a release site.

The involvement of N-type instead of the P/Q-type channels in KO mice provides the opportunity to test whether the different pathways for Ca²⁺ entry are coupled with equal effectiveness to neurotransmitter release by determining if they affect the measured Ca²⁺ cooperativity. For this purpose we studied the calcium dependence of nerve-evoked transmitter release by changing the extracellular calcium concentration, [Ca²⁺]_o, while stimulating the calyx of Held axons from WT or KO mice, where transmitter release is controlled almost exclusively by P/Q type and N-type Ca²⁺ channels respectively (Inchauspe et al. 2004).

At both WT and KO calyces, EPSCs show synchronous release, display all or nothing behaviour and have amplitudes independent of stimulus intensity when above threshold. We recorded EPSCs at different [Ca²⁺]_o ranging from 0.25 mm to 2 mm. Plotted on logarithmic scales, we found that the power relation was linear for [Ca²⁺]_o up to 1 mm, as seen in Fig. 1A. The slope of the linear regression was similar for both WT (r = 2.14 ± 0.10, n = 14, R = 0.998) and KO mice (r = 2.14 ± 0.14, n = 6, R = 0.996). EPSC amplitudes were 15% lower in KO (6.3 ± 0.50 nA) than in WT mice (7.36 ± 0.55 nA) at 2 mm[Ca²⁺]_o (P = 0.014, n = 10 for both WT and KO) and this reduction was greater at lower [Ca²⁺]_o.

Figure 1.

Calcium dependence of transmitter release at the calyx of Held synapse A, relationship between EPSC amplitude and external calcium concentration for WT and KO mice (note double logarithmic axes). EPSCs were evoked by trapezoid fibre stimulation and recorded under voltage clamp conditions at a holding potential of −70 mV. The line is the fit to a linear function showing that the amplitude of EPSC is proportional to the calcium concentration raised to a power of 2.14 ± 0.10 for WT (n = 14, R = 0.998) and 2.14 ± 0.14 for KO (n = 6, R = 0.996). B, double patch recordings from the calyx of Held presynaptic terminal and the principal MNTB neuron to assess the relationship between EPSC and I_pCa upon variation of the number of Ca²⁺ channels engaged in triggering release. Recordings of presynaptic Ca²⁺ currents (PRE) and EPSCs (POST) in WT and P/Q KO mice, in response to presynaptic voltage protocols (lower traces) with action potential-like waveforms (ramps from −70 to +60 mV; rise and fall time 0.2 ms, plateau duration increasing from 0.1 to 0.7 ms in 0.1 ms increments). External solution contains 1 mm Ca²⁺ and 2 mm Mg²⁺. C, area integral (charge) of presynaptic Ca²⁺ currents and EPSCs are normalized and plotted on a log–log scale. According to eqn (2), Ca²⁺ domain cooperativity values (m) are obtained from the slope of the linear regressions: 3.0 ± 0.1 (n = 6) for WT and 4.1 ± 0.3 for KO (n = 6), P = 0.008.

Calcium channel domain cooperativity

It has been shown that during development of the calyx of Held, VGCCs become spatially closer to synaptic vesicles (SVs), thereby reducing the number of Ca²⁺ domains required to elevate the intracellular Ca²⁺ to the fusion threshold (Fedchyshyn & Wang, 2005; Yang & Wang, 2006). To further explore the presynaptic coupling between VGCCs and vesicular release on compensation of P/Q- by N-type calcium channels, we made simultaneous pre- and postsynaptic voltage clamp recordings from the calyx of Held presynaptic terminal and its postsynaptic target, the MNTB neuron. We studied the relationship between calcium influx and transmitter release under conditions where the calcium influx is mediated by either PQ-type channels (WT) or N-type channels (KO). This relationship is nonlinear and can be described by the power function:

(2)

The m value is an indirect readout of the spatial interaction between SVs and Ca²⁺ ions near the inner mouth of open channels (Borst & Sakmann, 1999; Gentile & Stanley, 2004) and hence may be referred as the ‘Ca²⁺ channel/domain’ cooperativity (Fedchyshyn & Wang, 2005; Yang & Wang, 2006). The value of m depends on the external Ca²⁺ concentration, temperature and calcium channel types involved in synaptic transmission, changing also with development (Fedchyshyn & Wang, 2005), but it is always greater than 2. This nonlinearity implies that slight variations in Ca²⁺ entry into the terminal can profoundly affect synaptic strength.

Presynaptic calcium currents were elicited by pseudo action potential-like voltage ramps from a holding potential of −70 mV to +60 mV, with a rise and fall time of 0.2 ms. By increasing the plateau duration at +60 mV from 0.1 to 0.7 ms in 0.1 ms steps, we recruited an increasing number of VGCCs engaged in triggering release, while maintaining the driving force for Ca²⁺ as well as the on and off kinetics of the command voltage steps. External solution contained 1 mm of Ca²⁺ and 2 mm of Mg²⁺. Simultaneous recordings of I_pCa (PRE) and EPSCs (POST) in WT and P/Q KO mice are shown in Fig. 1B. Total EPSC charge (normalized integral) was plotted against normalized I_pCa charge on a double logarithmic scale. When this input–output relationship was fitted with a linear function, we found that the slope value was higher for KO synapses (Fig. 1C). The Ca²⁺ channel/domain cooperativity coefficient m was 3.0 ± 0.1 for WT (n = 6) and 4.1 ± 0.3 for KO (n = 6), P = 0.008, indicating that more overlapping Ca²⁺ channel domains are necessary to trigger vesicle release at the active sites of the KO calyx of Held synapse, compared to WT. The presynaptic calcium current required to evoke an EPSC with one-tenth of the maximal amplitude should be 46% and 57% of the maximum I_pCa in synapses from WT and KO mice, respectively.

VGCCs in both groups of animals have similar kinetic properties, since no significant differences are observed in the rise or decay time of Ca²⁺ currents or in the rise and decay time of EPSCs evoked by the action potential protocol, indicating that the difference in m values is independent of the gating kinetics of VGCCs (see Table 1).

Table 1.

Rise time (10–90%) and decay times (10–90%) of I_pCa and EPSC from WT and KO synapses

	Rise time of IpCa	Decay time of IpCa	Rise time of EPSC	Decay time of EPSC
WT	0.20 ± 0.04 ms	0.43 ± 0.06 ms	0.36 ± 0.01 ms	1.49 ± 0.05 ms
KO	0.27 ± 0.03 ms	0.48 ± 0.04 ms	0.47 ± 0.03 ms	1.57 ± 0.04 ms

As inferred from developing synapses, the lower release efficiency could arise from a differential location of the channels at the release sites, being P/Q-type Ca²⁺ channels located closer to the release site than N-type Ca²⁺ channels. To test this hypothesis we studied synaptic transmission after introducing a slow Ca²⁺ buffer into presynaptic calyces. Slices were preincubated for 30 min with the tetra-acetoxymethyl ester of EGTA (EGTA-AM, 0.2 mm). Under this conditions, the mean amplitude of the EPSCs measured at KO mice synapses was reduced from 6.3 ± 0.5 nA (n = 7) to 4.7 ± 0.8 nA (n = 20), while it remained unchanged in synapses from WT mice (8.0 ± 0.6 nA, n = 7 with EGTA versus 7.36 ± 0.55 nA, n = 24 in control conditions). These data support the hypothesis that P/Q-type Ca²⁺ channels are located closer than N-type Ca²⁺ channels to the release site, and are therefore most favourably suited to mediate exocytosis, confirming results obtained by Wu et al. (1999), Rosato-Siri et al. (2002) and Urbano et al. (2003).

Altered mechanism of short-term plasticity in synaptic transmission of KO mice

Probability of vesicle release

The probability of release can be evaluated by estimating the fraction of the ready releasable pool released by a single action potential. We measured the cumulative amplitudes of EPSCs during a 100 Hz train in WT and KO mice to estimate the size of the ready releasable pool assuming that depression is largely caused by a transient decrease in the number of readily releasable quanta (Schneggenburger et al. 1999). To minimize saturation of glutamate receptors, experiments were performed in the presence of the rapidly dissociating AMPA receptor competitive antagonist kynurenic acid (1 mm). In Fig. 2A, peak amplitudes of EPSC are summed to give a plot of cumulative EPSC amplitudes during 100 Hz trains. Data points in a range of 140–200 ms are fitted by linear regression and back-extrapolated to time 0. This estimate effectively takes into account the cumulative EPSC amplitudes reached within the first five to six stimuli, corresponding to a time interval of about 50 ms, before depression reaches its steady state level. It assumes that recovery from depression is negligible for the time interval of about 50 ms used for the analysis. The zero time intersect gives an estimate of the size of the readily releasable pool of SVs (N) multiplied by the mean quantal amplitude (q). Since this is dependent on the initial amplitude of the EPSCs and the extent of the effect of kynurenic acid added to minimize postsynaptic AMPA receptor saturation, each EPSC amplitude during stimulation was normalized to that of the first EPSC in the train. Nq values obtained from the extrapolated zero time intercept of normalized data (2.2 ± 0.3, n = 22 for WT and 2.0 ± 0.3, n = 11 for KO) multiplied by the mean amplitude of EPSCs at 2 mm Ca²⁺ concentration (7.36 ± 0.55 nA, n = 24 for WT and 6.3 ± 0.5 nA, n = 20 for KO), gave the real Nq values: 16.2 ± 3.4 nA and 12.6 ± 2.9 nA for WT and KO mice, respectively. The release probability (p_r), shown in Fig. 2B, was estimated by dividing the mean amplitude of the first EPSC by Nq. Values obtained by this method may be overestimates because Ca²⁺ influx during the first stimuli is lower than during the last pulses, and Ca²⁺ concentration rises during stimulation. Nevertheless, we conclude that there are no significant differences in the release probability between WT (p_r= 0.45 ± 0.06) and KO (p_r= 0.50 ± 0.07) synapses (P = 0.2), in accordance with results published by Ishikawa et al. (2005).

Figure 2.

Release probability in WT and P/Q-KO synapses A, cumulative amplitude of EPSCs during 100 Hz stimulation in WT and KO mice. EPSC amplitudes are normalized to that of the first stimulus. Data are the average of n = 22 cells for WT and n = 11 cells in KO. The last six data points of the curve are fitted by linear regression and extrapolated to time zero to estimate the readily releasable pool size (N) multiplied by the mean quantal amplitude (q). The normalized Nq values obtained for WT and KO are 2.2 ± 0.3 and 2.0 ± 0.3, respectively. B, the release probability (p_r) is estimated by dividing the first EPSC amplitude by Nq. There are no significant differences in the release probability between WT (p_r= 0.45 ± 0.06, n = 22) and KO (p_r= 0.50 ± 0.07, n = 11) synapses (P = 0.2). C, histogram showing the probability distribution of mEPSC amplitudes. Mean mEPSC amplitudes are 39 ± 2 pA in WT and 38 ± 2 pA in KO mice, while frequencies are 1.6 ± 0.4 Hz and 1.4 ± 0.5 Hz, respectively.

An estimate of the quantal size can be obtained from measurement of miniature EPSCs (mEPSCs). We found no differences in the mean amplitude and frequency probability distribution of mEPSCs between WT and KO mice, as shown in the histograms of Fig. 2C. The mean mEPSC amplitude was 39 ± 2 pA in WT (n = 11) and 38 ± 2 pA in KO mice (n = 10). The frequency of spontaneous events was also similar between WT and KO mice: 1.6 ± 0.4 Hz and 1.4 ± 0.5 Hz, respectively.

EPSCs at low and high frequency stimulation

In order to compare WT and KO synaptic performance, evoked release was triggered by low (10 Hz) or high (100 Hz) frequency stimulation. To minimize saturation of glutamate receptors during stimulation, kynurenic acid (1 mm) was added to the external solution. Stimulation at 10 Hz or 100 Hz generates short-term depression (STD) of EPSCs, mainly caused by depletion of synaptic vesicles rather than postsynaptic AMPA receptor desensitization (Wong et al. 2003). The amplitude of the 10 Hz evoked EPSCs after 3 s stimulation depressed up to 22.9 ± 0.2% and 18.9 ± 0.4% of the maximum current for WT and KO, respectively. Decay time constants were 93 ± 3 ms for WT and 128 ± 6 ms for KO (Fig. 3A). Thus, a similar magnitude of depression was achieved with similar time courses at the calyx of Held synapse of both WT and KO mice, in agreement with Ishikawa et al. (2005).

Figure 3.

Recovery of EPSCs from short-term depression at 10 Hz frequency A, depression of EPSC amplitudes during 3 s stimulation at 10 Hz (conditioning train, recorded at a holding potential of −70 mV). The amplitudes are normalized to the first EPSC in the train. Data are fitted to a single exponential decay function, with a decay time constant τ= 93 ± 3 ms for WT (n = 8) and τ= 128 ± 6 ms for KO (n = 6), P = 0.07. Magnitude of depression is 22.9 ± 0.2% and 18.9 ± 0.4% of the first pulse for WT and KO, respectively. Kynurenic acid (1 mm) was added to the extracellular solution to reduce postsynaptic AMPA receptor saturation. B, time course of recovery from synaptic depression, measured by eliciting a single test EPSC at increasing time intervals following the conditioning train. The fraction of recovery is calculated as (I_test−I_ss)/(I₁−I_ss), where I₁ and I_ss are the amplitudes of the first and last EPSCs in the train and I_test is the amplitude of the test EPSC. Data are fitted to an exponential decay function of first order, with time constants τ= 4.20 ± 0.15 s (n = 8) and τ= 8.0 ± 0.4 s (n = 6) for WT and KO, respectively (P = 0.005). C, upper panel: representative traces of I_pCa during application of action potential-like waveforms at the calyx of Held presynaptic terminal, at a frequency of 10 Hz, in WT and KO mice. Presynaptic calcium currents were elicited by pseudo action potential-like voltage ramps from a holding potential of −70 mV to +60 mV, rise time and plateau duration of 0.2 ms and decay time of 0.6 ms. Lower panel: normalized average peak calcium currents recorded during 10 Hz depolarization (filled symbols). There is no decrease of calcium current in synapses from either WT or KO mice, so inhibition of Ca²⁺ currents cannot be the cause of EPSC depression at this frequency. Open symbols are the theoretical calculation of I_pCa depression needed to account for the experimentally measured EPSC depression, according to the relationship described by eqn (2) and data presented in Fig. 1C.

Presynaptic Ca²⁺ currents amplitudes evoked at similar rates (either by depolarizing voltage steps of 1 ms or by pseudo action potential-like voltage ramps) were recorded at the WT and KO calyx of Held presynaptic terminals. Superimposed traces of typical recordings and mean normalized peak amplitudes are plotted as a function of time during the train in Fig. 3C. Neither in the WT nor in the KO do Ca²⁺ currents showed depression during 3 s stimulation at 10 Hz, suggesting that depression is mainly due to presynaptic vesicle depletion (Takahashi et al. 2000; Wang & Kaczmarek, 1998) Based on the relationship between EPSCs and I_pCa (eqn (2)), we superimposed in Fig. 3C the theoretical depression of I_pCa that would be necessary to achieve the experimentally measured EPSC depression. Recovery of EPSC during stimulation was not taken into account since recovery time constants (in the order of a few seconds) are much greater than the time course of depression (in the order of ms).

During 0.2 s stimulation at 100 Hz, EPSCs in WT depressed with a slower decay time (τ= 16.4 ± 0.4 ms) than in KO mice (τ= 12.7 ± 0.7 ms, P = 0.03). This is consistent with EPSC facilitation at early time points (10 ms inter stimulus interval) in WT synapses. EPSCs depressed to a similar extent by the end of the stimulus train: 9.7 ± 0.3% in WT (n = 22) and 11.2 ± 0.5% in KO (n = 11) mice (Fig. 4A).

Figure 4.

Recovery of EPSCs from short-term depression at 100 Hz frequency A, depression of EPSC amplitude during 100 Hz stimulation. Summary plot of average EPSC amplitudes, normalized to that of the first pulse in the train. The amplitudes of the EPSCs at the end of the stimuli are 9.7 ± 0.3% and 11.2 ± 0.5% of the first pulse for WT (n = 22) and KO (n = 11), respectively. The time course of the current decay during the train is fitted with a single exponential decay function, with a time constant τ= 16.4 ± 0.4 ms for WT and τ= 12.7 ± 0.7 ms for KO. B, recovery time course from synaptic depression was studied by eliciting a single test EPSC at varying time intervals following the conditioning train. Data are fitted to an exponential decay function of first order, with recovery time constants τ= 1.95 ± 0.05 s (n = 7) and τ= 2.62 ± 0.02 s (n = 7) for WT and KO, respectively (P = 0.043). C, upper panel: representative traces of I_pCa elicited by brief depolarizations (1.2 ms at a potencial of −10 mV) at the calyx of Held presynaptic terminal, at a frequency of 100 Hz, in WT and KO mice. Lower panel: normalized average peak calcium currents recorded during 100 Hz depolarization (filled symbols). Calcium currents do not inactivate in synapses from either WT or KO mice, but there is activity-dependent facilitation in synapses from WT mice. Open symbols are the theoretical calculation of I_pCa depression needed to account for the experimentally measured EPSC depression, according to the relationship described by eqn (2) and data presented in Fig. 1C.

As in the 10 Hz trains, I_pCa amplitudes elicited by brief depolarizing voltage steps (1 ms) applied at 100 Hz during 0.2 s did not show depression. Even more, in synapses from WT mice, I_pCa amplitudes increased for the first 10 pulses (activity dependent facilitation, see Fig. 4C) in accordance with previous results of Cuttle et al. (1998), Forsythe et al. (1998) and Ishikawa et al. (2005). Figure 4C shows the mean peak amplitude of calcium currents evoked at 100 Hz frequency (normalized to the amplitude of the first I_pCa in the train), together with the theoretical calculation (based on eqn (2)) of the magnitude of calcium current depression that would be necessary to achieve the EPSC depression experimentally measured. The theoretical calculation (as well as the equivalent for 10 Hz shown in Fig. 3C) is an underestimate because the m values used were obtained in conditions of 1 mm of Ca²⁺ and 2 mm of Mg²⁺ and not in 2 mm of Ca²⁺ and 1 mm of Mg²⁺ as with the experimental data in Figs 3 and 4. Even underestimated, the theoretical calculation shows that the I_pCa depression should be quite large to account for the EPSC STD, in contrast to the experimental results.

The efficacy of synaptic transmission during repetitive stimulation is also determined by the rate of recovery from synaptic depression, which depends on the refilling rate of the readily releasable pool of synaptic vesicles. The time course of recovery from synaptic depression was studied by eliciting a single test EPSC at varying time intervals (from 0.5 s to 12 s) following the conditioning train. The fraction of recovery was calculated a follows:

where I₁ and I_ss are the amplitudes of the first and last EPSCs in the train and I_test is the amplitude of the test EPSC. The fraction of recovery as a function of time after the stimuli train could be fitted by a single exponential for the WT and KO mice. The recovery from 10 Hz stimulation was slower in KO calyx of Held synapses, with a time constant of 8.0 ± 0.4 s (n = 6) compared to 4.20 ± 0.15 s (n = 8, P = 0.005) in WT (see Fig. 3B). After 100 Hz stimulation, the recovery time constants were τ= 1.95 ± 0.05 s (n = 7) for WT and τ= 2.62 ± 0.02 s (n = 7) for KO (Fig. 4B).

In order to investigate the mechanism underlying this difference in the recovery from STD, we compared the action potential waveforms (measured in current clamp mode) between WT and KO synapses as well as the Ca²⁺ currents elicited by similar pseudo action potential waveforms. We found no differences in AP waveforms between WT and KO synapses. The mean of n = 3 and n = 7 cells for WT and KO, respectively, can be seen in Fig. 5A. The parameters that characterize the APs are depicted in Table 2. Action potentials did not change their shape during repetitive stimulation at 10 Hz or 100 Hz. A train of APs elicited at the calyx of Held presynaptic terminal of both WT and KO mice at 100 Hz frequency is shown in Fig. 5B. Nevertheless, calcium currents evoked by action potential waveforms (Fig. 5C) were significantly smaller (26%) at the KO calyx of Held presynaptic terminals (mean peak amplitudes: 780 ± 60 pA, n = 12) than at WT ones (1060 ± 70 pA, n = 9, P = 0.004). The parameters characterizing these calcium currents are shown in Table 3. The reduction in Ca²⁺ influx may cause a reduced rate of replenishment, which contributes to the slower kinetics of recovery in KO synapses compared to WT.

Figure 5.

KO mice generate similar action potential (AP) waveforms but smaller Ca²⁺ currents A, action potential waveforms measured at the calyx of Held presynaptic terminals of WT and KO mice, in whole cell current clamp mode. Data are the means of n = 3 and n = 7 cells for WT and KO, respectively. B, train of action potentials evoked at 100 Hz. At the right, the first (black line) and last (grey line) AP in the train are superimposed, showing no change in waveform during high frequency stimulation. C, Ca²⁺ currents elicited by pseudo action potential voltage ramps (−70 to + 45 mV; rise time and plateau duration of 0.2 ms, decay time of 0.6 ms) are significantly smaller in the calyx of Held presynaptic terminal of KO mice (mean peak amplitudes: 780 ± 60 pA, n = 12) than in WT ones (1060 ± 70 pA, n = 9, P = 0.004).

Table 2.

Parameters of action potential (AP) waveforms from WT (n = 3) and KO (n = 7) synapses

	Membrane potential at rest (mV)	Peak potential (mV)	Half-width (ms)	Rise time (10–90%) (ms)	Decay time (90–10%) (ms)
WT	−72 ± 1	44 ± 2	0.39 ± 0.01	0.25 ± 0.01	0.36 ± 0.02
KO	−73 ± 1	45 ± 2	0.42 ± 0.03	0.30 ± 0.02	0.39 ± 0.03
		P = 0.12	P = 0.17	P = 0.21	P = 0.19

Table 3.

Parameters of Ca²⁺ currents generated by AP waveforms, from WT (n = 9) and KO (n = 12) synapses

	Peak amplitude (pA)	Half-width (ms)	Rise time (10–90%) (ms)	Decay time (90–10%) (ms)
WT	1060 ± 70	0.38 ± 0.02	0.31 ± 0.06	0.34 ± 0.06
KO	780 ± 60	0.43 ± 0.06	0.40 ± 0.08	0.48 ± 0.09
	P = 0.004	P = 0.22	P = 0.09	P = 0.09

Facilitation of presynaptic calcium currents and transmitter release

Previous experiments have shown that presynaptic calcium currents undergo paired pulse facilitation (PPF) in WT but not in KO mice. In a different series of experiments it was also shown that PPF of EPSCs was absent in the KO mice (Inchauspe et al. 2004). In order to confirm this correlation and to determine whether the increased transmitter release is due to facilitation of the presynaptic calcium current, we made simultaneous pre- and postsynaptic patch recordings at the calyx of Held presynaptic terminal and its postsynaptic neurons in the MNTB. EPSCs were evoked by applying two depolarizing voltage steps to the presynaptic terminal (2 ms from −70 mV to −10 mV) at inter pulse intervals of 5–10 ms. Alternatively, two pseudo action potential-like voltage ramps were applied from a holding potential of −70 mV to +60 mV (0.2 ms of rise time, 0.2 ms plateau duration and 0.6 ms of repolarizing time) with the same interval of 5–10 ms (Fig. 6A). In normal aCSF (2 mm Ca²⁺ and 1 mm Mg²⁺), the release probability and hence EPSC magnitude were too large. Synaptic vesicle depletion and postsynaptic AMPA receptor saturation generated postsynaptic depression of the second EPSC. These experiments were therefore conducted in aCSF containing low Ca²⁺ concentration (1 mm) and high Mg²⁺ (2 mm) to unmask the EPSC facilitation effect.

Figure 6.

EPSC facilitation arises from facilitation of presynaptic calcium currents in WT mice, while no facilitation of I_pCa and EPSCs is seen in KO A, simultaneous recordings of presynaptic Ca²⁺ currents (top) in the calyx of Held presynaptic terminal and EPSCs (bottom) in postsynaptic neurons, in WT and KO mouse, induced by two action potential-like voltage ramps (−70 to +60 mV; rise time and plateau duration of 0.2 ms, decay time of 0.6 ms), separated by 5 ms time interval. External solution contains low Ca²⁺ (1 mm) and high Mg²⁺ (2 mm), to avoid postsynaptic AMPA receptor saturation which unmasks facilitation of EPSC. B, mean magnitude of facilitation of the second I_pCa with respect to the first one is 12.8 ± 0.8% in WT and −2.2 ± 2.5% in KO. The average induced facilitation of EPSCs is 46.5 ± 6.6% for WT and 5.2 ± 3.9% for KO. The arrow indicates the expected EPSC facilitation caused by a 12.8% facilitation in I_pCa, estimated to be 43% according to the power relationship expressed by eqn (2), with m = 3.0.

The facilitation of the second I_pCa with respect to the first was 12.8 ± 0.8% in WT and −2.2 ± 2.5% in KO, while EPSC-induced facilitation was 46.5 ± 6.6% for WT and 5.2 ± 3.9 for KO (Fig. 6B). According to the power relationship expressed by eqn (2), with m = 3.0, the enhancement of the EPSC amplitude caused by a 12.8% facilitation in I_pCa was estimated to be 43%, a value close to the 46% experimentally measured in synapses of WT mice. This confirms that the lack of transmitter release facilitation in KO synapses is directly related to the absence of facilitation in the presynaptic N-type calcium currents.

Modulation of EPSCs by presynaptic receptors

Presynaptic inhibition of EPSCs by GABA_B receptors

In a previous work (Inchauspe et al. 2004), we have shown that at the calyx of Held synapses transmitter release is supported by presynaptic P/Q-type calcium channels in WT mice and N-type channels in KO mice (80 and 90% of the total current, respectively). It has been suggested that G-protein coupled receptors preferentially inhibit N-type Ca²⁺ channels and have much less effect on P/Q-type Ca²⁺ channels (Currie & Fox, 1997; Zhang et al. 1996). It is also known that at this nerve terminal baclofen activates G-protein G_o, and its β/γ subunits inhibit Ca²⁺ channels via a membrane-delimited pathway (Kajikawa et al. 2001). This intracellular signalling pathway is likely to be shared by adenosine receptors, metabotropic glutamate receptors and GABA_B receptors.

To investigate the effect of presynaptic receptor activation on transmitter release, EPSCs were evoked in both WT and KO mice. The selective GABA_B agonist baclofen added to the bath solution caused a potent inhibition of synaptic transmission in a dose-dependent manner. The inhibitory effect was reversible after washout. Figure 7A shows EPSCs recorded from the soma of an MNTB neuron under voltage clamp at a holding potential of −70 mV for different concentrations of the agonist in WT and KO mice. The time course of the inhibitory effect of baclofen on WT and KO cells can be seen in figures of the online supplemental material. On average, cells from KO mice showed a higher sensitivity to the agonist. The concentration–response relationship for baclofen is shown in Fig. 7B, for pooled data from six cells for WT and four cells for KO. Both curves were well fitted by the Hill equation:

Figure 7.

Synaptic transmission at KO synapses is more susceptible to presynaptic inhibition by GABA_B receptors A, EPSCs were evoked in MNTB neurons by stimulation with a bipolar electrode placed at the midline of the brainstem slice (stimuli of 0.1 ms duration and 3–10 V amplitude), and recorded under whole-cell voltage clamp conditions at a holding potential of −70 mV, during application of the GABA_B receptor agonist baclofen in the bath solution. Each trace is the average of 5–8 EPSCs recorded in control conditions and after application of successive baclofen doses: (0.02, 0.05, 0.1, 0.2, 0.3, 0.5, 0.7, 1.5 μm for WT and 0.02, 0.04, 0.06, 0.08, 0.1, 0.2 μm for KO). B, concentration–response relationship for baclofen induced inhibition of the calyx of Held–MNTB transmission (presented as percentage of inhibition relative to the EPSC recorded in control conditions). Curves are well fitted by a Hill equation. Maximum EPSC inhibition is 65.6 ± 2.3% and 87.8 ± 1.1% (P = 0.028) and IC₅₀ values are 0.16 ± 0.01 μm and 0.051 ± 0.001 μm (P = 0.018) for WT (n = 6) and KO (n = 5), respectively. C, presynaptic calcium current inhibition by GABA_B receptors activation. Presynaptic calcium currents elicited by stepping the voltage to −10 mV for 10 ms from a holding potential of −70 mV, before and after (bold trace) application of 2 μm baclofen, in WT and KO mice. Activation rate of I_pCa becomes slower and its amplitudes smaller. D, plot summarizing the calcium current inhibition by 2 μm of baclofen in WT (n = 7) and KO mice (n = 5). The current change is presented as percentage inhibition relative to the calcium current recorded in control conditions (20 ± 3% in WT and 52 ± 6% in KO, P = 0.003). Ca²⁺ currents were measured 3 ms after the depolarizing voltage step. E, depolarization voltage steps to +100 mV for 20 ms relieve baclofen mediated inhibition of Ca²⁺ currents. Traces show the Ca²⁺ current before (in bold) and after the depolarizing voltage step to +100 mV in WT and KO mice. Note the stronger inhibition relief in KO mice.

where I_M is the maximum current inhibition, C is the baclofen concentration, IC₅₀ the concentration for half-current inhibition, and n_H the Hill coefficient. The fitted curves indicated a maximum inhibition of 65.6 ± 2.3% and 87.8 ± 1.1%, and an IC₅₀ of 0.16 ± 0.01 μm and 0.051 ± 0.001 μm for WT and KO, respectively.

To clarify the target of GABA_B-induced presynaptic inhibition, we examined the effect of baclofen on I_pCa (Takahashi et al. 1996, 1998). Under whole-cell voltage-clamp, I_pCa were evoked by a depolarizing voltage step from a holding potential of −70 mV to −10 mV. Saturated concentrations of baclofen (1–2 μm) applied via perfusion reduced the amplitude of I_pCa (measured 3 ms after the onset of the pulse) by 20 ± 3% in WT and 52 ± 6% in KO (Fig. 7C and D). The effect was reversible after washout and by 1–2 μm of the selective GABA_B antagonist CGP55845. Given the highly nonlinear dependence of transmitter release on presynaptic Ca²⁺ influx (eqn (2)), the reduction in presynaptic Ca²⁺ current is likely to account for a great proportion of the reduction in EPSC amplitude, suggesting that presynaptic Ca²⁺ channels are the main target of GABA_B receptors, although other postsynaptic effects of baclofen cannot be discarded.

The inhibition of presynaptic Ca²⁺ currents by baclofen could be relieved by strong depolarization of the presynaptic terminal to +100 mV during 20 ms, as shown in Fig. 7E.

Presynaptic inhibition mediated by adenosine receptors and metabotropic glutamate receptors (mGluRs)

Adenosine receptors are also present at the calyx of Held synapse (Kimura et al. 2003; Barnes-Davies & Forsythe, 1995) and its activation suppressed transmitter release through inhibition of Ca²⁺ currents. Bath application of adenosine attenuated the amplitude of EPSCs in a dose-dependent manner. Figure 8A shows EPSCs recorded from the soma of an MNTB neuron under voltage clamp at a holding potential of −70 mV for different concentrations of adenosine in WT and KO mice. In Fig. 8B, mean experimental data showing the concentration–response relationship for adenosine have been fitted with the Hill function. Maximum inhibition was 36.10 ± 0.34% and 46.1 ± 0.2% and concentration for half current inhibition was 4.14 ± 0.06 μm and 2.91 ± 0.02 μm in neurons from WT (n = 6) and KO (n = 5), respectively.

Figure 8.

A, presynaptic inhibition of EPSCs mediated by adenosine. Representative traces of EPSCs evoked in MNTB neurons under whole-cell voltage clamp conditions at a holding potential of −70 mV during bath application of different concentrations of adenosine. Each trace is the average of 5–8 EPSC recordings in control conditions and after application of 4, 8, 12 and 20 μm of adenosine. B, concentration–response relationship for adenosine induced inhibition of the calyx of Held–MNTB transmission. Experimental data have been fitted by a Hill equation. Maximum inhibition is 36.10 ± 0.34% and 46.1 ± 0.2%, and concentration for half current inhibition is 4.14 ± 0.06 μm and 2.91 ± 0.02 μm in neurons from WT (n = 6) and KO (n = 5), respectively. Hill coefficient is 1.79 ± 0.04 and 1.78 ± 0.02 for WT and KO. C, presynaptic calcium current inhibition by adenosine. Presynaptic calcium currents elicited by stepping the voltage to −10 mV for 10 ms from a holding potential of −70 mV, before and after (bold trace) application of 20–50 μm of adenosine, in WT and KO mice. Activation rate of I_pCa becomes slower and its amplitude smaller. D, plot summarizing the calcium current inhibition by adenosine in WT (n = 4) and KO mice (n = 5). Saturated concentrations of adenosine (20–50 μm) applied via perfusion reduce the amplitude of I_pCa (measured 3 ms after the onset of the pulse) by 13 ± 1% in WT (n = 4) and 19 ± 5% in KO (n = 5), relative to the calcium current recorded in control conditions. E, metabotropic glutamate receptors have no effect on WT and KO synapses. Representative traces of EPSCs evoked in two different MNTB neurons under whole-cell voltage clamp conditions at a holding potential of −70 mV, before (a) and after (b) application of the selective agonist of group III mGluR, l-AP4, at a saturated concentration of 100 μm. Each trace is the average of 4–6 EPSCs responses. F, mean inhibition of EPSC by 100 μm of l-AP4 is 11.2 ± 2.6% relative to the current recorded in control conditions, for WT mice (n = 9) and 5.0 ± 3.5% (n = 4), for KO mice.

We investigated the effect of adenosine on I_pCa. Under whole-cell voltage-clamp, I_pCa were evoked by depolarizing voltage steps from a holding potential of −70 mV to −10 mV. Saturated concentrations of adenosine (20–50 μm) applied via perfusion reduced the amplitude of I_pCa (measured 3 ms after the onset of the pulse) by 13 ± 1% in WT (n = 4) and 19 ± 5% in KO (n = 5), as shown in Fig. 8C and D. This inhibition could be relieved by strong depolarization of the presynaptic terminal to +100 mV during 20 ms.

In rats, the calyx of Held synapse expresses presynaptic group II and III mGluRs that are negatively coupled to neurotransmitter release (Barnes-Davies & Forsythe, 1995; Takahashi et al. 1996; Von Gersdorff et al. 1997). The selective agonist of group III mGluRs (mGlu4, mGlu6, mGlu7, and mGlu8), l-AP4, at a saturated concentration of 100 μm had little or no effect on the mouse EPSC amplitudes (traces of representative EPSCs recordings are shown in Fig. 8E), as well as the non-selective agonist of group I and II mGluRs, ACPD. On average, activation of group III mGluRs by l-AP4 attenuated EPSC amplitudes by 11.2 ± 2.6% in WT (n = 9) and 5.0 ± 3.5% in KO (n = 4), as shown in Fig. 8F. These results are similar to those found by Renden et al. (2005).

Discussion

In this study we analysed the effects of compensatory expression of N-type calcium channels following genetic ablation of P/Q-type Ca²⁺ channel at an identified synapse. The similar amplitudes and time courses of EPSCs demonstrate that P/Q-type channels are not essential for the establishment of functional pre- and postsynaptic elements at the calyx of Held synapse and prompted us to focus our study on possible alterations in the excitation–secretion coupling.

We studied the calcium dependence of transmitter release at the calyx of Held synapse of WT and KO mice, where transmitter release is controlled almost exclusively by P/Q-type and N-type Ca²⁺ channels, respectively. Deletion of the P/Q channels causes a small reduction of EPSC amplitudes (15%) at a physiological [Ca²⁺]_o of 2 mm, in contrast to the strong reduction observed at the neuromuscular junction of the same animal model (Urbano et al. 2003). At [Ca²⁺]_o between 0.25 mm and 1 mm, the reduction is higher, but the power relation between EPSCs and Ca²⁺ concentration is similar for both WT and KO synapses. Therefore replacement of P/Q-type by N-type Ca²⁺ channels does not affect the dependence of transmission on [Ca²⁺]_o

The relation between I_pCa and the evoked EPSCs was studied by varying the number of Ca²⁺ channels domains engaged in triggering release. Similar studies performed during development have shown a reduced slope of the input–output relationship in mature calyx of Held synapses, consistent with a lower number of overlapping Ca²⁺ channel domains required to trigger fusion of synaptic vesicles (Fedchyshyn & Wang, 2005; Yang & Wang, 2006). The power relationship between EPSCs and I_pCa amplitudes at the calyx of Held–MNTB synapse shows a higher coefficient in KO than in WT mice, suggesting that more overlapping N-type Ca²⁺ channel domains contribute to the Ca²⁺ influx required to trigger exocytosis at the active site of the KO calyx of Held presynaptic terminals. As inferred from developing synapses, the higher Ca²⁺ channel/domain cooperativity could arise from a differential location of the channels at the release sites, being P/Q-type Ca²⁺ channels located closer to the release site than N-type Ca²⁺ channels. This hypothesis was confirmed by the reduction in EPSC amplitudes in KO synapses when the slow Ca²⁺ buffer EGTA-AM was introduced into presynaptic calyces. Thus, Ca²⁺ influx through N-type channels seems to be less effective in triggering transmitter release, whereas P/Q-type Ca²⁺ channels may be most favourably suited to mediating exocytosis (Wu et al. 1999; Rosato-Siri et al. 2002; Urbano et al. 2003).

Significant alterations in short-term plasticity phenomena and G-protein coupled receptor mediated modulation of transmitter release are originated as a consequence of the deletion of P/Q-type Ca²⁺ channels. Lack of pair pulse facilitation of presynaptic and postsynaptic currents was an important feature found at the calyx of Held synapse and neuromuscular junction of KO animals (Inchauspe et al. 2004; Urbano et al. 2003). This new series of experiments, where simultaneous pre- and postsynaptic double patch recordings were performed, strengthens the hypothesis that short-term facilitation of transmitter release is directly related to facilitation of the presynaptic calcium currents.

EPSCs recorded at the WT and KO MNTB neurons displayed similar strong STD in response to high frequency stimulation. STD is mainly caused by depletion of synaptic vesicles (Wang & Kaczmarek, 1998) although other factors such us desensitization of postsynaptic AMPA receptors (Wong et al. 2003) and inhibition of transmitter release mediated by presynaptic G-protein-coupled autoreceptors (von Gersdorff et al. 1997; Kimura et al. 2003) are also involved. Presynaptic calcium currents evoked at the same frequency do not show depression or inactivation and so they are not involved in the STD of EPSCs. The absence of depression of I_pCa during a train of 30 stimuli at 10 Hz was also reported by Takahashi et al. (2000), Cuttle et al. (1998), Forsythe et al. (1998) and Ishikawa et al. (2005).

A remarkable difference is observed in the kinetics of recovery from synaptic depression. This is an important feature in determining synaptic efficacy. Recovery from high frequency depression in immature synapses is substantially slower than in more mature synapses (Joshi & Wang, 2002). It is mainly dependent on the recycling rate of vesicles into the readily releasable pool (Wang & Kaczmarek, 1998) which is strongly influenced by preceding synaptic activity. It has been well documented that the increase in Ca²⁺ influx during repetitive firing of action potentials is the key element that enhances the replenishment. As shown in Fig. 5C, Ca²⁺ currents elicited by similar pseudo action potential waveforms are significantly smaller in the calyx of Held presynaptic terminal of KO mice. Therefore it seems likely that the reduction in Ca²⁺ influx is accompanied by a reduced rate of replenishment.

At the calyx of Held synapse, the metabotropic glutamate receptor agonist l-AP4, the GABA_B receptor agonist baclofen, and adenosine can presynaptically attenuate synaptic transmission (Barnes-Davies & Forsythe, 1995). This effect is explained by the reduction in Ca²⁺ influx due to G-protein activation which inhibits Ca²⁺ channels via a membrane-delimited pathway involving the βγ G-protein subunits (Kajikawa et al. 2001), suggesting that adenosine, GABA_B and mGlu receptors may share a common mechanism of action (Takahashi et al. 1996, 1998; Kimura et al. 2003).

Different Ca²⁺ channels expressed in native cells are not equally affected by G-protein inhibition. For example, N-type channels undergo greater inhibition compared with P/Q type (Currie & Fox 1997; Bourinet et al. 1996; Zhang et al. 1996; Zhou et al. 2003). In the present study, we examined the presynaptic inhibitory effect of GABA_B, adenosine and mGlu receptors activation on EPSCs mediated by P/Q-type and N-type Ca²⁺ channels in WT and KO mice, respectively. Our results indicate that in the KO calyx of Held synapse, transmitter release is reduced by the GABA_B receptor agonist baclofen to a greater extend and with an enhanced sensitivity, as demonstrated by the shift in the IC₅₀ of the inhibition–concentration relationship. This increase in the sensitivity, also reported by Iwasaki et al. (2005), correlates with the high expression of the N-type calcium current in the KO calyx of Held presynaptic terminal compared to the P/Q channels expressed in the WT. Thus, under identical conditions and at the same synapse N-type Ca²⁺ channels are more tightly coupled with GABA_B receptors than P/Q-type Ca²⁺ channels. This reinforces observations reported in other systems (Currie & Fox, 1996, 2002, 1997; Bourinet et al. 1996; Zhang et al. 1996) such as parallel fibre terminals of the tottering mouse, where N-type channels are dominant and GABA_B receptor activation produces a 3- to 5-fold more potent inhibition of transmission (Zhou et al. 2003).

The inhibitory effect of adenosine on EPSCs is still high in the P11–15 WT mice, in contrast to the diminished developmental response observed at the rat calyx of Held synapse (Wong et al. 2006). Inhibition by adenosine is more effective and sensitive at the KO calyx of Held synapse. Concerning mGluRs, they are poorly expressed or coupled at P11–15 WT and KO mice, in agreement with Renden et al. (2005).

The tight coupling of GABA_B and adenosine with N-type presynaptic Ca²⁺ channels present at the KO calyx of Held synapse correlates with the enhancement of GPCR-mediated presynaptic inhibition at various central nervous system (CNS) synapses where N-type channels dominate. Mean amplitudes of EPSCs are only 15% lower in KO with respect to WT mice. The discrepancy between our present data and our previous data (Inchauspe et al. 2004) regarding the extent of the decrease of EPSC amplitudes in KO mice is probably due to the variability of EPSC amplitudes and to a lower access resistance in the patch clamp recordings in this new set of data. Even more, Ishikawa et al. (2005) found no difference at all, which is not in contradiction with our results taking into account the variability of EPSC amplitudes. The fact that compensation of P/Q-type Ca²⁺ channels by N-type channels results in a functional synapse with similar EPSC amplitudes and release probability in KO mice, in spite of the decreased presynaptic Ca²⁺ currents (Ishikawa et al. 2005; Inchauspe et al. 2004; and Fig. 5 of this paper) and regardless of the larger distance of N-type Ca²⁺ channels from the release sites, implies that there must be one or more compensatory mechanisms. A possible one is an enhanced Ca²⁺ sensitivity of the release machinery. This could be achieved by compensatory changes in second messengers known to facilitate transmitter release like cAMP (Sakaba & Neher, 2001; Kaneko & Takahashi, 2004) or by changes in the intrinsic Ca²⁺ sensitivity of vesicle fusion (Wölfel et al. 2007) due to altered vesicle proteins like presynaptic munc-13 (Lou et al. 2005; Basu et al. 2007).

The P/Q-type calcium channel KO mouse provides a model system in which the compensatory changes of expression of N-type calcium channels at the calyx of Held synapse can be exploited to assess the functional significance of expression of N- versus P/Q-type calcium channels at the same synapse. We and Ishikawa et al. (2005) have shown that the KO calyx of Held synapse presents different properties in synaptic plasticity and in transmitter release modulation by presynaptic receptors. Our results coincide in that synapses of KO mice depict similar EPSC amplitudes and release probabilities, and undergo similar STD after repetitive high frequency stimulation, but lack activity-dependent facilitation and are more strongly regulated by GABA_B presynaptic receptors compared to synapses of WT mice. In addition, with other GPCRs like adenosine and mGlu receptors, we have demonstrated that STD of EPSCs is not due to inactivation of presynaptic Ca²⁺ currents; we have found that EPSCs from KO mice recover with a slower kinetics after STD at distinct frequencies and that N-type channels present a different Ca²⁺ channel/domain cooperativity in triggering vesicle fusion, which affects the EPSC–I_pCa relationship. It is likely that these alterations are present in many other synapses, causing disruption of well balanced neuronal networks and leading to neurological syndromes like those found in the P/Q deficient mouse (epilepsy ataxia, and dystonia).

Supplementary material

Online supplemental material for this paper can be accessed at:

http://jp.physoc.org/cgi/content/full/jphysiol.2007.139683/DC1

and

http://www.blackwell-synergy.com/doi/suppl/10.1113/jphysiol.2007.139683