Similar Intracellular Ca²⁺ Requirements for Inactivation and Facilitation of Voltage-Gated Ca²⁺ Channels in a Glutamatergic Mammalian Nerve Terminal

Abstract

Voltage-gated Ca²⁺ channels (VGCCs) of the P/Q-type, which are expressed at a majority of mammalian nerve terminals, show two types of Ca²⁺-dependent feedback regulation—inactivation (CDI) and facilitation (CDF). Because of the nonlinear relationship between Ca²⁺ influx and transmitter release, CDI and CDF are powerful regulators of synaptic strength. To what extent VGCCs inactivate or facilitate during spike trains depends on the dynamics of free Ca²⁺ ([Ca²⁺]_i) and the Ca²⁺ sensitivity of CDI and CDF, which has not been determined in nerve terminals. In this report, we took advantage of the large size of a rat auditory glutamatergic synapse—the calyx of Held—and combined voltage-clamp recordings of presynaptic Ca²⁺ currents (I_Ca(V)) with UV-light flash-induced Ca²⁺ uncaging and presynaptic Ca²⁺ imaging to study the Ca²⁺ requirements for CDI and CDF.

We find that nearly half of the presynaptic VGCCs inactivate during 100 ms voltage steps and require several seconds to recover. This inactivation is caused neither by depletion of Ca²⁺ ions from the synaptic cleft nor by metabotropic feedback inhibition, because it is resistant to blockade of metabotropic and ionotropic glutamate receptors. Facilitation of I_Ca(V) induced by repetitive depolarizations or preconditioning voltage steps decays within tens of milliseconds. Since Ca²⁺ buffers only weakly affect CDI and CDF, we conclude that the Ca²⁺ sensors are closely associated with the channel. CDI and CDF can be induced by intracellular photo release of Ca²⁺ resulting in [Ca²⁺]_i elevations in the low micromolar range, implying a surprisingly high affinity of the Ca²⁺ sensors.

Introduction

Beyond being charge carriers, Ca²⁺ ions serve many additional functions in neuronal tissue, the most prominent being the rapid triggering of neurotransmitter secretion from nerve terminals. P/Q-type Ca²⁺ channels (Ca_V2.1) are responsible for Ca²⁺-dependent vesicle exocytosis from many nerve terminals of the mammalian CNS. Presynaptic Ca_V2.1 channels undergo a dual feedback regulation by incoming Ca²⁺ ions, Ca²⁺-dependent facilitation, and inactivation (Lee et al., 1999; Liang et al., 2003; Inchauspe et al., 2004; Ishikawa et al., 2005). Ca²⁺-dependent facilitation (CDF) develops rapidly with the first tens of milliseconds of repeated action potential (AP)-like depolarizations. An increased channel open probability (Chaudhuri et al., 2007) and/or a negative shift in the voltage-dependent activation (Borst and Sakmann, 1998a; Cuttle et al., 1998) may underlie the measured increase in presynaptic Ca²⁺ currents (I_Ca(V)). Among presynaptic voltage-gated Ca²⁺ channels (VGCCs), CDF is unique to Ca_V2.1 channels, whereas Ca²⁺-dependent inactivation (CDI) is a common property of several VGCC subtypes whereby the entry of Ca²⁺ enhances channel closure during a maintained depolarization (DeMaria et al., 2001; Liang et al., 2003).

Because transmitter release is highly nonlinearly related to calcium influx, even minute changes in presynaptic Ca²⁺ flux can strongly impact release (Neher, 1998). Thus, Ca²⁺-dependent regulation of VGCCs represents an important and powerful mechanism to regulate synaptic strength (Borst and Sakmann, 1998b; Xu and Wu, 2005). Indeed, calyx of Held synapses lacking Ca_V2.1-dependent CDF also show reduced EPSC facilitation during repetitive synapse activation (Inchauspe et al., 2004; Ishikawa et al., 2005). Likewise, in cultured superior cervical ganglion neurons, Ca²⁺ channel mutations that block CDI or CDF of Ca_V2.1 channels substantially affect short-term plasticity of synaptic transmission (Mochida et al., 2008), and a gain-of-function missense mutation of the Ca_V2.1 channel that occludes CDF correlates with a reduction in short-term synaptic facilitation at the parallel fiber→Purkinje cell synapse (Adams et al., 2010).

Most mammalian nerve terminals are too small to be accessible for electrophysiological recordings. Thus, our knowledge about Ca²⁺-dependent regulation of presynaptic VGCCs is still limited. Here we took advantage of the large size of a glutamatergic terminal in the mammalian auditory brainstem—the calyx of Held—to study Ca²⁺-dependent regulation of presynaptic Ca_V2.1 channels. In particular, we were interested in determining the intracellular Ca²⁺ requirements for CDI and CDF. To this end, we combined presynaptic whole-cell recordings of I_Ca(V) with Ca²⁺ uncaging via flash photolysis while monitoring [Ca²⁺]_i with indicator dyes. Our results indicate that both regulatory pathways, CDI as well as CDF, can be activated by [Ca²⁺]_i elevations of only a few micromoles. Such elevations of the global volume average [Ca²⁺]_i are easily achieved during repetitive AP firing, especially at higher frequencies when individual AP-evoked [Ca²⁺]_i transients start to summate. Since auditory circuits discharge at high rates during sound stimulation and show varying levels of spontaneous AP firing even in the complete absence of sound (Friauf and Ostwald, 1988; Taberner and Liberman, 2005; Tolnai et al., 2008; Sonntag et al., 2009), presynaptic VGCCs at the calyx of Held and presumably also in other auditory terminals predominantly expressing Ca_V2.1 channels will be subject to strong Ca²⁺-dependent regulation.

Materials and Methods

Slice preparation.

Juvenile (P8–10) Wistar rats of either sex were used for most experiments. For comparison, some recordings were also obtained from postnatal day (P)9–11 C57BL/6N or α_1A-deficient (α_1A^−/−) mice (Jun et al., 1999), and in some cases from P14–16 rats of either sex (see Figs. 3, 7). Brainstem slices were prepared as described previously (Taschenberger and von Gersdorff, 2000; Erazo-Fischer et al., 2007). After decapitation, the whole brain was immediately immersed into ice-cold low Ca²⁺ artificial CSF (aCSF) containing the following (in mm): 125 NaCl, 2.5 KCl, 3 MgCl₂, 0.1 CaCl₂, 10 glucose, 25 NaHCO₃, 1.25 NaH₂PO₄, 0.4 ascorbic acid, 3 myo-inositol, and 2 Na-pyruvate at pH 7.3 when bubbled with carbogen (95% O₂, 5% CO₂). The brainstem was glued onto the stage of a VT1000S vibratome (Leica), and 200-μm-thick coronal slices containing the medial nucleus of the trapezoid body (MNTB) were cut. Slices were incubated for 30–40 min at 35°C in a chamber containing normal aCSF and kept at room temperature (22–24°C) for up to 4 h thereafter. The composition of normal aCSF was identical to that of low Ca²⁺ aCSF, except that 1.0 mm MgCl₂ and 2.0 mm CaCl₂ were used.

Figure 3.

Clustering of presynaptic VGCCs may account for weak sensitivity of Ca²⁺-dependent inactivation to intracellular Ca²⁺ chelators. A1, B1, Time course of peak amplitudes and fractional inactivation of I_Ca(V) during gradual current reduction by either bath application of the Ca_V2.1 blocker ω-AgaTX (A) or slow bath perfusion with a low Ca²⁺ solution (B) (left panels). Samples traces representing I_Ca(V) recorded before (ctrl, control) and after reducing its amplitude by ∼50 and ∼70% are shown superimposed in the right panels. The colored traces represents peak scaled versions of I_Ca(V) after ∼50% amplitude reduction. A2, B2, Comparison of I_Ca(V) inactivation before and after reducing I_Ca(V) by ∼50% by application of either ω-AgaTX (A2) or low Ca²⁺ solution (B2). C, Line fits to scatter plots of inactivation versus normalized peak amplitudes of I_Ca(V) at selected degrees of current reduction (0, ∼33, ∼50, and ∼66%) by application of low Ca²⁺ solution (blue) or ω-AgaTX (red). Dotted lines represent 95% confidence limits. When ω-AgaTX was applied to gradually reduce I_Ca(V), fractional inactivation and normalized amplitudes were highly correlated (Pearson's r = 0.83, p < 0.001). In contrast, the slope of the regression line was not significantly different from zero (p = 0.79) for the data obtained with low Ca²⁺ bath perfusion. All data shown in A–C were obtained with 10 mm BAPTA in the pipette solution. D, I_Ca(V) inactivation is unaltered in calyces of mice lacking Ca_V2.1 (p = 0.93) (0.5 mm EGTA in the pipette solution). w.t., Wild type; k.o., knockout. Inactivation of I_Ca(V) is strongly reduced in more mature calyces. Recordings from two P14 terminals with either 0.5 mm EGTA (left) or 10 mm BAPTA (right) in the patch pipette. The gray trace represents a peak scaled version of I_Ca(V) recorded with 0.5 mm EGTA. E2, Fractional inactivation of I_Ca(V) in P14–16 calyces is significantly attenuated when recorded with 10 mm BAPTA in the patch-pipette (p < 0.001). The dotted lines in A2, B2, D, and E2 indicate the mean fractional inactivation of I_Ca(V) in calyces of P8–10 rats with 0.5 mm EGTA in the pipette (see Fig. 1D).

Figure 7.

Presynaptic I_Ca(V) shows robust Ca²⁺-dependent facilitation during prepulse protocols. A1, I_Ca(V) elicited by a 100 Hz train of 5 ms steps from V_h = −80 mV to −20 mV in two P8 calyces (top and middle) and in a P14 calyx (bottom). With Ca²⁺ as the charge carrier (top and bottom), the activation of I_Ca(V) strongly accelerated from the first to the fifth voltage step and remained fast during later steps. No change in activation kinetics was observed with external Ba²⁺ (middle). A2, First, fifth, and tenth I_Ca(V) shown superimposed for comparison after normalizing to the same peak amplitude. A3, I_Ca(V) facilitation was similar in P8–10 and P14 calyces. Average half-rise times of the first and tenth I_Ca(V) in the train. B–C, Facilitation of I_Ca(V) during paired-pulse protocols. B1, Sample traces of I_Ca(V) elicited by stepping to −15 mV with (red trace) or without (black trace) a preceding prepulse to +20 mV (20 ms interpulse interval). Facilitation was quantified as the ratio of charges transferred during the initial 3 ms of I_Ca(V) with or without prepulse. Note that a hyperpolarization to −140 between prepulse and test pulse (blue trace) did not alter the amount of current facilitation. B2, Increasing intracellular Ca²⁺ buffer strength only slightly suppressed facilitation of I_Ca(V). C1, Sample traces of I_Ca(V) without (black trace) or with prepulse at 10 ms (red) or 70 ms interval (blue). C2, The relaxation of I_Ca(V) facilitation is an order of magnitude faster than recovery from inactivation and insensitive to changes in [Ca²⁺]_i buffering strength. Solid lines represent single exponential fits with the decay time constants as indicated.

Electrophysiology and data analysis.

Patch-clamp recordings were made from calyx of Held terminals using an EPC-10 amplifier controlled by Pulse software (HEKA Elektronik). Sampling intervals and filter settings were 20 μs and 4.5 kHz, respectively. Cells were visualized by differential interference contrast microscopy through a 60× water-immersion objective (numerical aperture 1.0; Olympus) using an upright Axioskop FS microscope (Zeiss). All experiments were performed at room temperature.

Patch pipettes were pulled from borosilicate glass (Science Products) on a P-97 micropipette puller (Sutter Instrument). Pipettes were coated with dental wax to reduce stray capacitance. Open tip resistance was 4–5 MΩ. Access resistance (R_s) was ≤20 MΩ and routinely compensated by 50–60% during presynaptic voltage-clamp experiments. The holding potential (V_h) was −80 mV. For measuring presynaptic I_Ca(V) and changes in membrane capacitance (ΔC_m), pipettes were filled with a solution containing the following (in mm): 100 Cs-gluconate, 30 TEA-Cl, 30 CsCl, 10 HEPES, 5 Na₂-phosphocreatine, 4 ATP-Mg, 0.3 GTP, pH 7.3, with CsOH. The pipette solution was supplemented with varying concentrations of EGTA or BAPTA as indicated. The bath solution was supplemented with 1 μm TTX, 1 mm 4-AP, and 40 mm TEA-Cl to suppress voltage-gated Na⁺ and K⁺ currents. Junction potentials were not corrected. In some experiments a mixture of antagonists (50 μm d-AP5, 5 μm NBQX, 300 μm CPPG, and 5 μm AM251) was applied to study a possible inactivation of presynaptic I_Ca(V) by mechanisms that involve depletion of Ca²⁺ ions from the synaptic cleft and/or activation of metabotropic glutamate or cannabinoid receptors (see Fig. 2).

Figure 2.

Neither depletion of Ca²⁺ from the synaptic cleft nor feedback inhibition of VGCCs via metabotropic receptors account for I_Ca(V) inactivation during sustained 100 ms depolarizations. A, Paired recordings from calyx terminals and synaptically coupled (A1) or neighboring (A2) MNTB principal cell somata. Vesicle exocytosis was elicited by applying step depolarizations from V_h = −80 to 0 mV of various durations (5–30 ms). Dotted lines mark the onset of presynaptic depolarizations. The corresponding I_Ca(V) and ΔC_m responses are shown in the middle and bottom rows, respectively. Postsynaptic neurons were voltage clamped at V_h = +40 mV. Presynaptic depolarizations induced large NMDA EPSCs with short latencies in postsynaptic neurons (A1) and induced spillover NMDA EPSCs in neighboring cells (A2), indicating that synaptically released glutamate can escape the synaptic cleft and activate extrasynaptic receptors. B, Blocking ionotropic and metabotropic receptors pharmacologically does not affect I_Ca(V) inactivation. Paired recordings of I_Ca(V) (left) and EPSCs (right) in control solution (black) and after adding a mixture of glutamate and cannabinoid receptor antagonists (red; see Materials and Methods) to suppress Ca²⁺ influx through GluR channels and feedback inhibition via mGluRs and/or CBRs. C, Scatter plot of fractional inactivation versus peak amplitudes of I_Ca(V). Pooled results obtained with different intracellular Ca²⁺ chelator species and concentrations. Data obtained from recordings in the presence of the antagonist mixture are plotted using red symbols. Solid and broken lines represent line fit and 95% confidence limits to the entire dataset, indicating a weak positive correlation (Pearson's r = −0.30, p = 0.001, n = 115) between I_Ca(V) peak amplitude and degree of inactivation with a slope of ∼5.4% × nA⁻¹.

Changes in membrane capacitance were monitored using the Sine+DC technique (Lindau and Neher, 1988) with a software lock-in amplifier (HEKA Pulse) by adding a 1 kHz sine–wave voltage command (amplitude ±35 mV) to V_h = −80 mV. To block Ca_V2.1 channels, ω-agatoxin IVA (ω-AgaTX; 200 nm) (Peptide Institute) was dissolved in aCSF containing 0.1 mg/ml cytochrome c and applied via bath perfusion. Off-line analysis was done with Igor Pro (WaveMetrics). All values are given as mean ± SEM. Unless indicated otherwise, significance of difference was evaluated with the two-tailed Student's unpaired t test and p < 0.05 was taken as the level of statistical significance.

Flash photolysis and presynaptic Ca²⁺ imaging.

A UV flash lamp (Rapp Optoelectronic) was used to uncage Ca²⁺ from DM-Nitrophen and elicit step-like elevations of the intracellular free Ca²⁺ concentration ([Ca²⁺]_i) in presynaptic terminals (Schneggenburger and Neher, 2000). Different post-flash amplitudes of [Ca²⁺]_i were obtained by controlling the flash light intensity using neutral density filters and/or by varying the concentration of DM-nitrophen in the pipette solution (0.6–9 mm).

Presynaptic [Ca²⁺]_i was monitored using the ratiometric Ca²⁺ indicator dyes Fura 4F (K_d = 1 μm) or Fura 2FF (K_d = 10 μm), which were excited at 350 and 380 nm by a monochromator (Polychrome 5; TILL Photonics). Fluorescent images were collected with an interline-transfer 640 × 480 pixel CCD (VGA, TILL Photonics). To allow for brief exposure times, on-chip pixel binning (8 × 15) was used. Monochromator and CCD camera were controlled by the TILLvisION software (TILL Photonics). Presynaptic patch pipettes contained the following (in mm): 100–125 Cs-gluconate, 20 TEA-Cl, 20 HEPES, 0.5 MgCl₂, 5 Na₂-ATP, 0.3 GTP, 0.5–9 DM-Nitrophen, 0.3–7.8 CaCl₂, and 0.2 Fura 2FF or 0.2 Fura 4F, pH 7.3, with CsOH. Time series images were analyzed off-line using Igor Pro software. The background fluorescence of a small area adjacent to the region of interest was subtracted to obtain the background-corrected fluorescence ratio r = F₃₅₀/F₃₈₀. That ratio was converted into the [Ca²⁺]_i using the following equation (Grynkiewicz et al., 1985): [Ca²⁺]_i = K_eff(R − R_min)/(R_max − R). The calibration constants (K_eff, R_max, and R_min) for each dye species were obtained from in vitro measurements.

Results

Inactivation of presynaptic voltage-gated Ca²⁺ channels

During long-lasting membrane depolarizations of 1 s duration, the inactivation of I_Ca(V) in calyx terminals followed a biphasic time course (Fig. 1A) (Forsythe et al., 1998; Ishikawa et al., 2005). To quantify the contribution of the two kinetically distinct components of I_Ca(V) inactivation during long depolarizations (1 s), we calculated the fractional inactivation during the first 100 ms [(I_peak − I_{100 ms})/I_peak] in comparison to that during the remaining 900 ms [(I_{100 ms} − I_{1 s})/I_peak] (Fig. 1A,B). When external Ca²⁺ was replaced by Ba²⁺ as a charge carrier, I_Ca(V) peak amplitudes decreased slightly from −1.61 ± 0.14 nA (n = 10) to −1.33 ± 0.10 nA (n = 10, p = 0.11; Fig. 1B, left). More importantly, the fast component of I_Ca(V) inactivation was strongly reduced in external Ba²⁺ (45.6 ± 2.0% vs 24.5 ± 1.9%, p < 0.001; Fig. 1A,B), whereas the slowly inactivating component remained nearly unaltered (24.6 ± 1.1% vs 24.3 ± 1.9%, p = 0.90; Fig. 1B). This observation suggests that the fast but not (or to a lesser extent) the slow component of I_Ca(V) inactivating is Ca²⁺ mediated. Because calyces are unlikely to experience long-lasting depolarizations comparable to those shown in Figure 1A, we reasoned that the contribution of the slowly inactivating component to the inactivation of I_Ca(V) under in vivo conditions is very limited and we did not further characterize it in this study.

Figure 1.

Two kinetically distinct components of I_Ca(V) inactivation during sustained depolarizations: fast and Ca²⁺ dependent and slow and Ca²⁺ independent. A, Presynaptic I_Ca(V) elicited by 1 s depolarizations from V_h = −80 mV and recorded with either Ca²⁺ (voltage steps to 0 mV, left) or Ba²⁺ (voltage steps to –10 mV, right) as the charge carrier. Inactivation of I_Ca(V) was biphasic under both conditions. However, the rapidly inactivating component was strongly reduced in amplitude when recording in Ba²⁺. B, Pooled results showing average values of peak amplitudes (left) and normalized fractions of fast (middle) and slowly (right) inactivating components of I_Ca(V). Average amplitudes of I_Ca(V) decreased slightly when substituting extracellular Ca²⁺ with Ba²⁺ (p = 0.11). External Ba²⁺ profoundly reduced the amplitude of the fast inactivating component of I_Ca(V) (p < 0.001) while leaving the slowly inactivating component unaltered (p = 0.9). C, Supplementing the pipette solution with the fast Ca²⁺ chelator BAPTA at ≥10 mm slowed the inactivation time course but had little effect on the amount of I_Ca(V) inactivation during 100 ms steps to 0 mV. C1, Presynaptic I_Ca(V) recorded with either Ca²⁺ or Na⁺ as the charge carrier (left) and different concentrations of Ca²⁺ chelators in the patch pipette (right). For comparison, I_Ca(V) recorded in divalent free external solution is shown (second trace). C2, Corresponding ΔC_m traces. Filled and empty symbols denote the same recordings as in C1. As expected, ΔC_m responses were strongly attenuated in the presence of ≥10 mm intracellular Ca²⁺ chelator. D, Pooled results showing average values for the amount of inactivation, inactivation time constant and vesicle exocytosis. The amount of inactivation during 100 ms depolarizations was similar for all four recording conditions (p = 0.58, ANOVA), while the rate of inactivation of I_Ca(V) decreased with high concentrations of BAPTA in the pipette solution.

To corroborate that a Ca²⁺-dependent mechanism is responsible for the fast component of I_Ca(V) inactivation, we compared I_Ca(V) recorded in nominally Ca²⁺-free solution with that obtained under control conditions. Figure 1C1 illustrates that I_Ca(V) inactivation was nearly absent in Ca²⁺-free bath solution with Na⁺ as the charge carrier (Ca²⁺-free: 4.6 ± 1.1%, n = 7; 2 mm external Ca²⁺: 45.9 ± 1.1%, n = 12, p < 0.001) (Forsythe et al., 1998). We next tested whether increasing the intracellular Ca²⁺ buffering strength affects I_Ca(V) inactivation during 100 ms steps. Figure 1C1 exemplifies I_Ca(V) recorded with pipette solutions containing variable chelator species and concentrations. Supplementing the pipette solution with high concentrations of the fast Ca²⁺ buffer BAPTA slowed the time course of current inactivation. When fit with a single exponential function, the time constants of I_Ca(V) inactivation were 33.3 ± 1.3 (n = 12), 34.4 ± 3.1 (p = 0.99, n = 9), 44.7 ± 3.6 (p = 0.03, n = 19), and 53.0 ± 2.7 (p < 0.001, n = 15) ms for 0. 5 mm EGTA, 10 mm EGTA, 10 mm BAPTA, and 30 mm BAPTA, respectively (Dunnett test) (Fig. 1C1,D). When external Ca²⁺ was replaced by Ba²⁺, I_Ca(V) decayed much slower (time constant 101.2 ± 25.5 ms, n = 13) (compare Fig. 1A). The level of I_Ca(V) inactivation during 100 ms steps was, however, insensitive to changes in the Ca²⁺ buffering strength (45.9 ± 1.1% [0.5 mm EGTA, n = 12] vs 43.6 ± 1.9% [10 mm EGTA, n = 9] vs 42.3 ± 2.2% [10 mm BAPTA, n = 19] versus 44.6 ± 1.6% [30 mm BAPTA, n = 15], p = 0.58, ANOVA) (Fig. 1D). Expectedly, ΔC_m responses were nearly completely abolished with high concentrations of EGTA or BAPTA in the pipette solution (Fig. 1C2,D) (Borst et al., 1995).

Because even a high concentration (30 mm) of the fast Ca²⁺ buffer BAPTA was unable to effectively lower the amount of I_Ca(V) inactivation during 100 ms depolarizations (Fig. 1D), mechanisms other than Ca²⁺-dependent inactivation of I_Ca(V) may alternatively or additionally account for the Ca²⁺ current decline. We therefore considered the following two possibilities: (1) activation of pre- and/or postsynaptic Ca²⁺-permeable conductances may cause a depletion of Ca²⁺ ions from the synaptic cleft (Borst and Sakmann, 1999); or (2) metabotropic receptors could be directly [metabotropic glutamate receptors (mGluRs)] or indirectly [cannabinoid receptors (CBRs)] activated by released glutamate and thereby attenuate I_Ca(V) (Takahashi et al., 1996; Kushmerick et al., 2004). Metabotropic GluRs have been localized synaptically and extrasynaptically in the calyx terminal (Elezgarai et al., 1999, 2001). To test whether released glutamate is able to escape the synaptic cleft and activate extrasynaptic GluRs, we recorded from pairs of calyx terminals and neighboring principal MNTB neurons (Fig. 2A). Upon presynaptic depolarization, spillover NMDA EPSCs with relatively long latency were elicited in neighboring neurons. Thus, synaptically released glutamate can also activate extrasynaptic GluRs beyond the glial sheath, at least those of relatively high affinity. The smaller amplitude, slower rise, and longer latency of these spillover EPSCs (Fig. 2A2) compared to NMDA EPSCs in postsynaptic neurons (Fig. 2A1) is consistent with a longer diffusional distance.

To test for a contribution of metabotropic feedback inhibition and/or Ca²⁺ depletion from the synaptic cleft to presynaptic I_Ca(V) inactivation, we applied a mixture of antagonists with the aim of blocking all Ca²⁺-permeable ionotropic GluRs as well as mGluRs and CBRs (see Materials and Methods for composition). Figure 2B illustrates a paired recording of presynaptic I_Ca(V) and EPSCs before and after application of the antagonist mixture. While the EPSC is nearly completely abolished, I_Ca(V) remained virtually unchanged in the presence of antagonists. For recordings with 0.5 mm EGTA in the pipette solution, the fractional inactivation of I_Ca(V) during 100 ms depolarizations in the presence of antagonists in the bath (49.2 ± 2.6%, n = 8) was statistically not different (p = 0.26) from that in control solution (45.9 ± 1.1%, n = 12). Similar results were obtained with higher EGTA concentration in the pipette (43.6 ± 1.1% [n = 9] vs 44.8 ± 2.0% [n = 8], p = 0.66, 10 mm EGTA). When plotting I_Ca(V) inactivation as a function of peak amplitude for the entire dataset, a weak but statistically significant correlation was observed (Fig. 2C). The fractional inactivation of I_Ca(V) during 100 ms steps increased by ∼5.4% per nA current amplitude.

Reducing the number of functional VGCCs attenuates inactivation

In contrast to somatic VGCCs, those expressed in nerve terminals are known to cluster near release sites (Heuser and Reese, 1981; Pawson et al., 1998; Harlow et al., 2001; Nagwaney et al., 2009). Within overlapping [Ca²⁺]_i nanodomains that build up during channel opening, intracellular chelators cannot effectively buffer [Ca²⁺]_i. Even though individual channels transition stochastically between open and closed states during prolonged depolarizations, there may be always some open channels within a cluster. Assuming that the Ca²⁺ sensor mediating CDI is preassociated with the channel protein (Lee et al., 1999), it may thus be constantly exposed to elevated [Ca²⁺]_i during depolarizations if a channel is part of a larger cluster.

We therefore asked whether a reduction in the number of functional channels would reduce the rate of I_Ca(V) inactivation (Soong et al., 2002; Kreiner et al., 2010). Figure 3A shows that this is indeed the case. When the irreversible Ca_V2.1 channel blocker ω-AgaTX was applied to the bath solution, not only did the amplitude of I_Ca(V) become smaller but its inactivation time constant more than doubled from 43.5 ± 3.5 ms (control) to 92.7 ± 17.7 ms (at ∼50% current block) (p = 0.03, paired t test) and the fractional inactivation decreased notably from 47.8 ± 2.8% (control) to 34.7 ± 1.5% (at ∼50% current block) (p = 0.02, paired t test) (Fig. 3A1,A2). In contrast, gradually reducing I_Ca(V) by slow bath perfusion with a low Ca²⁺ solution neither affected fractional inactivation (46.1 ± 2.5% vs 45.6 ± 2.4% for control and 50% current reduction, respectively; p = 0.72, paired t test) nor inactivation rate (49.2 ± 7.3 ms vs 49.9 ± 8.6 ms for control and 50% current reduction, respectively; p = 0.87, paired t test) (Fig. 3B1,B2). Both sets of experiments were performed under conditions that nearly completely prevent a rise in global [Ca²⁺]_i (10 mm BAPTA in the pipette solution; see also Fig. 4B1 below) and thus isolate the effect of the local [Ca²⁺]_i domains on I_Ca(V) inactivation. Linear regression analysis of scatter plots of I_Ca(V) inactivation versus normalized peak amplitudes measured after selected degrees of current reduction (0, ∼33, ∼50, and ∼66%) is shown in Figure 3C. When I_Ca(V) was gradually reduced by application of ω-AgaTX, fractional inactivation and normalized amplitudes were highly correlated, whereas the slope of the regression line was not statistically significantly different from zero for the data obtained with low Ca²⁺ bath perfusion (Fig. 3C).

Figure 4.

Recovery of I_Ca(V) from inactivation is intrinsically slow, regardless of the Ca²⁺ buffering conditions A, Recovery of I_Ca(V) from inactivation was tested at variable intervals using a paired-pulse protocol consisting of a 100 ms depolarization to 0 mV followed by a 20 ms depolarization to 0 mV at variable interstimulus interval. Pipette solution contained 0.5 mm EGTA. Traces for three different recovery intervals are shown superimposed. B, The time course of recovery from inactivation was biphasic and insensitive to changes in Ca²⁺ chelator species and/or concentrations in the pipette solution. B1, [Ca²⁺]_i transients evoked by 100 ms step depolarizations. The rise of global [Ca²⁺]_i was nearly completely suppressed when adding 10 mm BAPTA to the pipette solution (right panel).B2, Average time course of recovery from inactivation. Solid lines represent double exponential fits. Fast and slow time constants were similar for the three [Ca²⁺]_i buffering conditions (p ≥ 0.56, ANOVA). Relative amplitudes of fast and slow time constants are given in parentheses. The recovery time course following a shorter (50 ms) conditioning depolarization was virtually unchanged (filled symbols in the left panel).

Since only decreasing the density of functional channels reduced inactivation of I_Ca(V) notably, we tentatively postulate that presynaptic VGCCs must be arranged in such a geometrical configuration that blocking individual channels reduces the time integral of [Ca²⁺]_i experienced by the Ca²⁺ sensor for CDI more effectively than decreasing Ca²⁺ flux by reducing external Ca²⁺. Alternatively, I_Ca(V) inactivation may slow down because of a larger relative contribution of ω-AgaTX-resistant channels with possibly different inactivation properties to the whole-cell current. However, this seems unlikely because the fractional inactivation of presynaptic I_Ca(V) in Ca_V2.1 deficient mice was essentially equal to that measured in wild-type terminals (44.5 ± 1.8% [n = 18] vs 44.8 ± 2.4% [n = 12], p = 0.93; Fig. 3D).

During postnatal maturation of the calyx synapse, larger active zones seem to disappear in favor of smaller ones (Taschenberger et al., 2002), and the number of VGCCs associated with individual docked vesicles seems to decrease (Fedchyshyn and Wang, 2005; Wang et al., 2008). These developmental changes may result in smaller VGCC clusters in more mature calyces compared to P8–10 calyces. We therefore asked whether I_Ca(V) inactivation may decline with age. Indeed, the fractional inactivation of I_Ca(V) was strongly reduced in P14–16 calyces (31.0 ± 1.7%, τ = 49.0 ± 5.3 ms, n = 11) compared to younger terminals, and it was further attenuated when recording with a pipette solution supplemented with high concentration of BAPTA (12.8 ± 2.4%, τ = 125.7 ± 51.8 ms, n = 6, p < 0.001) (Fig. 3E1,E2).Together, these results support the notion that a tight clustering of presynaptic VGCCs in the calyx terminal may promote the relative insensitivity of I_Ca(V) inactivation to Ca²⁺ chelators. The developmental or experimental decrease in the number of channels per cluster may allow fast chelators such as BAPTA to more effectively antagonize inactivation.

Recovery of I_Ca(V) from inactivation

During repetitive AP firing, the modulation of AP-evoked presynaptic Ca²⁺ influx by VGCC inactivation critically depends on the rate of recovery of I_Ca(V) from inactivation. To measure the recovery time course of calyceal VGCCs, we used a paired-pulse protocol consisting of a 100 ms depolarization to 0 mV followed by 20 ms steps at various recovery time intervals between 25 ms and 15 s (Fig. 4). The recovery of I_Ca(V) from inactivation was surprisingly slow and followed a biphasic time course with a fast recovery time constant of ∼0.2 s and a slow recovery time constant of ∼5 s (Fig. 4B). The time course of the fractional recovery was fit with a double exponential function 1 − a × e^−t/τ_f − (1 − a) × e^−t/τ_s, where τ_f and τ_s represent the fast and slow time constants, and a is the fraction of the fast recovery (Fig. 4B2). To exclude the possibility that this slow recovery was caused by a delayed clearance of Ca²⁺ entering the terminals during the 100 ms depolarization, we repeated the experiments with elevated intracellular Ca²⁺ chelator concentrations. As expected, the rise of global [Ca²⁺]_i was strongly attenuated with 10 mm EGTA and nearly completely abolished with 10 mm BAPTA in the pipette solution (Fig. 4B1). In contrast, the time course of recovery from inactivation was nearly unaffected (Fig. 4B2). It is unlikely that the slow component of recovery can be attributed to the slow component of I_Ca(V) inactivation (see Fig. 1A), because the time course of recovery was unchanged when the duration of the conditioning depolarization was shortened from 100 to 50 ms (Fig. 4B2, left). The slow ‘intrinsic’ recovery of calyceal I_Ca(V) from inactivation leads to cumulative inactivation of VGCCs during repetitive stimulation and may account for the strong reduction of presynaptic Ca²⁺ influx observed during high-frequency stimulation (Forsythe et al., 1998; Nakamura et al., 2008).

Probing the [Ca²⁺]_i sensitivity of I_Ca(V) inactivation by flash photolysis of caged Ca²⁺

To directly examine the Ca²⁺ requirements for inducing I_Ca(V) inactivation, we used Ca²⁺ uncaging while monitoring [Ca²⁺]_i. In a first set of experiments, a UV-light flash was delivered during a 100 ms depolarization to 0 mV to produce a rapid, step-like elevation of the global, volume-averaged [Ca²⁺]_i. (Fig. 5). To monitor [Ca²⁺]_i, either Fura 4F (low-intensity flashes) or Fura 2FF (high-intensity flashes) were included in the pipette solution. In the absence of the UV-light flash, 100 ms depolarizations resulted in an maximum elevation of [Ca²⁺]_i to ∼15 ± 2 μm (n = 26). As illustrated in Figure 5A, increasing [Ca²⁺]_i via Ca²⁺ uncaging by a few micromoles did not result in a noticeable acceleration of I_Ca(V) inactivation (Fig. 5A). This suggests that the Ca²⁺ sensor is either already saturated or ‘sees’ a significantly higher, possibly local [Ca²⁺]_i signal such that a small increase in global [Ca²⁺]_i has little further effect on the inactivation rate. To differentiate between these two possibilities, we applied a strong UV-light flash that induced an elevation of global [Ca²⁺]_i to ∼100 μm, which resulted in a substantial acceleration of the inactivation (Fig. 5B). Three conclusions may be drawn from these experiments: (1) the accelerated decay of I_Ca(V) after Ca²⁺ uncaging directly demonstrates that I_Ca(V) inactivation is Ca²⁺-dependent; (2) the immediate change in inactivation time course following high post-flash [Ca²⁺]_i without any noticeable delay is incompatible with a mechanism that involves a diffusible molecule and rather suggests that the Ca²⁺ sensor for CDI must be located on a molecule that is preassociated with VGCCs; and (3) because only large elevations in post-flash [Ca²⁺]_i accelerated the inactivation kinetics, we postulate that during depolarizations the Ca²⁺ sensor experiences high [Ca²⁺]_i, further supporting the notion that the Ca²⁺ sensor responsible for triggering CDI must be located within molecular distance to the channel itself.

Figure 5.

The sensor mediating Ca²⁺-dependent inactivation ‘sees’ high [Ca²⁺]_i and therefore must be located within molecular distance from the channel mouth. A, I_Ca(V) (bottom) elicited by step depolarizations to 0 mV with (red) and without (black) a UV-light flash (dotted line) delivered ∼25 ms after current onset. Changes in [Ca²⁺]_i are illustrated in the top panel. The rate of I_Ca(V) inactivation was unaltered after UV-flash uncaging despite a ∼3-fold elevation of volume-averaged global [Ca²⁺]_i relative to control conditions (without flash). B, Similar experiment as illustrated in A except that the post-flash [Ca²⁺]_i increased to ∼100 μm. Note the immediate acceleration of the inactivation time course after Ca²⁺ uncaging. ctrl, Control. C, Scatter plot of flash-evoked I_Ca(V) inactivation as a function of global [Ca²⁺]_i. Flash-evoked I_Ca(V) inactivation was quantified as the ratio of the final I_Ca(V) amplitude measured after a flash compared to that of a control current. Black symbols represent mean ± SEM. for post-flash [Ca²⁺]_i of ≤15 μm and ≥40 μm. Note that for [Ca²⁺]_i. of ≤15 μm, inactivation of I_Ca(V) after Ca²⁺ uncaging was nearly indistinguishable from that under control conditions.

In a second set of experiments, we quantified the fractional inactivation induced by Ca²⁺ uncaging 56 ms before measuring I_Ca(V) with the principal aim of establishing the apparent dose–response relationship between I_Ca(V) inactivation and [Ca²⁺]_i (Fig. 6). We elicited elevations of [Ca²⁺]_i between 0.6 and 100 μm by adjusting flash intensity and/or DM-Nitrophen concentration in the patch pipette while monitoring [Ca²⁺]_i. Inactivation was measured as the ratio of peak amplitudes of post-flash I_Ca(V) relative to a control current measured ∼20 s before the flash. The 20 s interval was long enough to allow full recovery from inactivation induced during the control depolarization (Fig. 6A1). Inactivation of I_Ca(V) was noticeable for post-flash [Ca²⁺]_i ≥1 μm and severe for post-flash [Ca²⁺]_i ≥10 μm (Fig. 6A2). The amount of inactivation did not increase when probed with a second test depolarization delivered 112 ms after a flash (twice the interval between flash and the first depolarization) (Fig. 6B). To exclude the possibilities that (1) the measured I_Ca(V) inactivation merely resulted from a changed driving force due to elevated cytosolic Ca²⁺ or (2) that the release of intracellular Ca²⁺ activated an outward current, we compared the I–V relationship after uncaging with that of a control current recorded before the flash by applying voltage ramps (Fig. 6C) (Morad et al., 1988). As with step depolarizations, the post-flash I_Ca(V) was strongly reduced in amplitude. However, its I–V relationship was virtually unaltered (Fig. 6C, inset) which is the expected behavior for an ion conductance for highly asymmetrically distributed intracellular and extracellular charge carriers. Figure 6D shows pooled data from 75 flashes in 40 terminals. The relationship between [Ca²⁺]_i and the fractional inactivation of I_Ca(V) was fitted by a Hill function. The apparent half maximal inhibitory concentration (IC₅₀) and maximal inactivation were 6.1 ± 0.8 μm and 38 ± 2%, respectively. The amount of inactivation induced by strong flashes was thus comparable to the amplitude of the fast inactivating component during sustained step depolarizations (compare Fig. 1).

Figure 6.

[Ca²⁺]_i sensitivity of I_Ca(V) inactivation assayed by Ca²⁺ uncaging via UV-flash photolysis. A, Comparison of I_Ca(V) elicited by two step depolarizations (20 ms to 0 mV), separated by a 20 s interval, without (A1) or with (A2) a UV-light flash delivered 50 ms before the second I_Ca(V) (red traces). The time course of [Ca²⁺]_i shortly before and during the second depolarization (red dotted line) is shown in the top panel. In the absence of the UV-light flash the two currents were nearly identical, indicating that a 20 s interval was sufficient to allow full recovery from the small amount of inactivation induced during the first depolarization (A1, bottom). In contrast, a global elevation of [Ca²⁺]_i by UV-flash photolysis induced a pronounced reduction of the second current in the same terminal (A2). B, No further reduction of I_Ca(V) was observed when probed with an additional depolarization 112 ms after the UV-light flash. C, Similar experiment as illustrated in A except that I_Ca(V) was elicited by ramping V_m from −80 to +40 mV. Note the reduced current amplitude (bottom panel) but similar I–V relationship (bottom panel inset) of the second I_Ca(V) (red trace) following the UV-light flash. D, Dose–response relationship of Ca²⁺-dependent inactivation versus global [Ca²⁺]_i. Pooled data obtained from 75 flashes in 40 terminals. Solid line represents a Hill function fitted to the data.

Characterization of I_Ca(V) facilitation

Among presynaptically expressed VGCCs, Ca_V2.1 channels are unique in that they can exhibit strong Ca²⁺-dependent facilitation during repetitive depolarizations (Borst and Sakmann, 1998b; Cuttle et al., 1998). We therefore set out to compare the [Ca²⁺]_i requirements for CDF to those for inducing CDI. During repetitive, mild depolarizations, presynaptic I_Ca(V) showed robust facilitation. It manifested itself as a gradually accelerated current activation during the first five voltage steps. The half-rise time of I_Ca(V) decreased by >50% from 991 ± 88 μs to 446 ± 47 μs from the first to the 10th voltage step (Fig. 7A, top). CDF was similar in P14 when compared to P8–10 calyces, and it was virtually absent when external Ca²⁺ was replaced by Ba²⁺ (Fig. 7A, middle and bottom). To quantify CDF, we compared the charge transfer during the initial 3 ms of I_Ca(V) (Q_{3 ms(test)}/Q_{3 ms(ctrl)}) elicited by a 10 ms test pulse with and without a preceding 10 ms conditioning pulse to +20 mV (Fig. 7B1). A hyperpolarization of the terminal to −140 mV during the interpulse interval (Fig. 7B1, blue trace) did not affect the amount of facilitation, ruling out the possibility that the latter was voltage dependent rather than Ca²⁺ dependent (n = 5). The ratio Q_{3 ms(test)}/Q_{3 ms(ctrl)} showed little sensitivity to changes in the intracellular Ca²⁺ buffering strength (Fig. 7B2; 0.5 mm EGTA: 1.82 ± 0.06 [n = 16]; 10 mm EGTA: 1.65 ± 0.05 [n = 15]; 10 mm BAPTA: 1.65 ± 0.06 [n = 15], p = 0.6, ANOVA) and, unlike CDI (compare Fig. 4), CDF decayed very quickly with a mean time constant of ∼30 ms regardless of the type and concentration of Ca²⁺ chelator (Fig. 7C).

To assay the [Ca²⁺]_i sensitivity of CDF, we elicited step-like elevations of [Ca²⁺]_i by flash photolysis. The degree of CDF of I_Ca(V) elicited by Ca²⁺ uncaging was comparable to that elicited by conditioning prepulses (Fig. 8A). The scatter plot in Figure 8B illustrates the relationship between CDF and post-flash [Ca²⁺]_i in the concentration range between ∼1 μm to ∼100 μm. CDF was first observed for [Ca²⁺]_i values ≥2 μm and peaked between ∼5 μm to 10 μm. With higher [Ca²⁺]_i, CDF was difficult to elicit because I_Ca(V) tended to decrease relative to control. Because we were not able to completely separate CDF from CDI, estimating its [Ca²⁺]_i sensitivity is more difficult. To cope with these confounding effects of CDI, we fitted the data with the product of two Hill functions with the parameters for CDI fixed to the values obtained earlier (Fig. 6D) and obtained an apparent half maximum [Ca²⁺]_i for CDF of 2.4 ± 0.7 μm (maximum CDF = 1.6 ± 0.3). The [Ca²⁺]_i sensitivity for CDF was thus surprisingly similar to that of CDI.

Figure 8.

[Ca²⁺]_i sensitivity of I_Ca(V) facilitation assayed by Ca²⁺ uncaging via UV-flash photolysis. A, Comparison of I_Ca(V) facilitation elicited by either prepulse (A1) or Ca²⁺ uncaging (A2) (red traces). The interval between the UV-light flash and the test depolarization was 20 ms. Control current traces without preceding prepulse or without Ca²⁺ uncaging are shown for comparison (black traces). Recordings were obtained from the same terminal. The time course of [Ca²⁺]_i shortly before and during the second depolarization is shown in the top panel. B, Dose–response relationship of I_Ca(V) facilitation versus [Ca²⁺]_i. Maximum facilitation was observed for [Ca²⁺]_i elevations between 5 and 15 μm. At [Ca²⁺]_i >20 μm, Ca²⁺-dependent inactivation of I_Ca(V) was observed. The solid red line represents a product of two Hill functions fitted to the data. The solid blue line represent the simulated dose–response relationship of Ca²⁺-dependent facilitation in the absence of I_Ca(V) inactivation.

Discussion

In this study, we combined whole-cell recordings of presynaptic I_Ca(V) with Ca²⁺ uncaging via flash photolysis and ratiometric Ca²⁺ imaging to determine the intracellular Ca²⁺ requirements for the induction of Ca²⁺-dependent inactivation and facilitation of presynaptic VGCCs. Our results indicate that (1) the sensors for CDI and CDF must be closely associated with the channel protein, (2) relative low elevations in [Ca²⁺]_i by only a few micromoles are sufficient to induce CDI and CDF, and (3) the recovery from CDI is an order of magnitude slower than the relaxation of CDF suggesting different Ca²⁺ binding/unbinding kinetics. There are a number of functional implications of these findings. During sustained channel opening, we expect the Ca²⁺ sensors for CDI and CDF to be close to saturation. Single APs increase the global, volume averaged [Ca²⁺]_i by ∼0.3–0.4 μm (Helmchen et al., 1997; Müller et al., 2007) at the calyx of Held, but several micromoles free Ca²⁺ were measured during high-frequency stimulation (Korogod et al., 2005). Thus, presynaptic Ca²⁺ influx can be expected to be subject to regulation by CDF and CDI during physiological spike activity. Because of its slow recovery, CDI is favored during low-frequency AP firing whereas net facilitation of I_Ca(V) can only be achieved by high-frequency AP bursts.

Two kinetically distinct components of inactivation

During long-lasting depolarizations the amplitude of the calyceal I_Ca(V) declines biexponentially (Fig. 1; Forsythe et al., 1998). Inactivation is absent in Ca²⁺-free bath solution and intracellular Ca²⁺ uncaging causes a similar inactivation of I_Ca(V) in the absence of channel opening. The maximum inactivation observed after flash photolysis was comparable to that seen during 100 ms depolarizations. These results directly demonstrate that elevated [Ca²⁺]_i causes I_Ca(V) inactivation and suggest that CDI accounts for the majority of the fast component of inactivation. Because substituting external Ca²⁺ by Ba²⁺ had little effect on the slow inactivation, we assume that voltage-dependent inactivation (Cox and Dunlap, 1994; Patil et al., 1998) accounts for a majority of the slow inactivation.

Glutamate release-dependent decrease of I_Ca(V)?

Since presynaptic depolarization can trigger the release of large amounts of glutamate from the calyx, we had to consider a negative regulation of I_Ca(V) by activation of mGluRs (Takahashi et al., 1996) or via retroinhibition by cannabinoids acting on CB1Rs (Kushmerick et al., 2004). Additionally or alternatively, a decrease in I_Ca(V) amplitude could be caused by depletion of Ca²⁺ ions from the synaptic cleft because of concomitant activation of postsynaptic Ca²⁺-permeable GluR channels (Borst and Sakmann, 1999). All of the aforementioned mechanisms of I_Ca(V) inactivation are ‘release-dependent’, i.e., they require activation of pre- and/or postsynaptic GluRs by synaptically released glutamate. Given that high concentrations of intracellular Ca²⁺ chelators which strongly attenuate transmitter release in the young calyx of Held (Borst et al., 1995) did not notably change the amount of CDI, we conclude that glutamate release-dependent mechanisms do not prominently contribute to the fast inactivation of I_Ca(V). This conclusion was further substantiated by paired recordings of I_Ca(V) and EPSCs which revealed no change in the inactivation kinetics of I_Ca(V) in the presence of a mixture of antagonists blocking both Ca²⁺ entry through ionotropic GluRs and metabotropic feedback inhibition of I_Ca(V).

Little sensitivity of I_Ca(V) inactivation to intracellular chelators

In some neurons (Kasai and Aosaki, 1988; Cox and Dunlap, 1994; Grant and Fuchs, 2008; Benton and Raman, 2009) and for certain subunit combinations of heterologously expressed VGCCs (Lee et al., 2000; Kreiner and Lee, 2006), CDI can be effectively antagonized by high concentrations of intracellular Ca²⁺ buffers. In the calyx of Held, a high intracellular Ca²⁺ buffering strength slows its time course but has little effect on the amount of CDI (Fig. 1, see also Forsythe et al., 1998; Borst and Sakmann, 1999). We speculate that a tight clustering of presynaptic VGCCs may account for this result. This conclusion was based on the following two observations: First, inactivation gradually decreases when reducing the number of functional VGCCs by means of application of the irreversible antagonists ω-AgaTX (Fig. 3A,B). Second, the rate of inactivation is substantially reduced in more mature calyces (Fig. 3C,D) whose active zones tend to have a smaller average diameter (Taschenberger et al., 2002) and possibly harbor fewer VGCCs (Fedchyshyn and Wang, 2005; Wang et al., 2008). Interestingly, somatic Ca_V2.1 channels in cerebellar Purkinje neurons which presumably exhibit a more homogeneous surface distribution compared to that of presynaptic VGCCs inactivate less in the presence of a high concentration of EGTA (Benton and Raman, 2009).

Because of the rapid onset kinetics of CDI after Ca²⁺ uncaging (Fig. 5; see also Morad et al., 1988), it is likely that the Ca²⁺ sensor triggering CDI is located within molecular distance to and presumably preassociated with the Ca²⁺ channel. If this holds true, the Ca²⁺ sensor will be exposed to the micro- or nanodomain Ca²⁺ which builds up during channel opening and which is only weakly affected by buffers. In addition, clustering of presynaptic VGCCs at the active zone may lead to a scenario in which the Ca²⁺ sensor for inactivation is constantly exposed to high [Ca²⁺]_i during membrane depolarizations regardless of the stochastic opening and closing of individual channels. If the average number of VGCCs in a channel cluster is reduced, either experimentally by applying an open channel blocker or during synapse maturation, Ca²⁺ chelators with a rapid binding kinetics may become more efficient in attenuating [Ca²⁺]_i at the sensor. This may explain why reducing the number of available VGCCs by applying ω-AgaTX also led to a concomitant decrease of I_Ca(V) inactivation and also the significantly reduced I_Ca(V) inactivation in more mature terminals. Additionally, alternative splicing of the pore-forming α₁2.1 subunit (Soong et al., 2002) as well as changes in expression of auxiliary VGCC subunits may contribute to the developmental decline in CDI at the calyx of Held. For example, expression of the β_2a subunit which produces slower inactivation of I_Ca(V) than other β subunits (Lee et al., 2000) may increase postnatally in the developing calyx.

Similar intracellular Ca²⁺ requirements for CDI and CDF

When probing their [Ca²⁺]_i sensitivity by Ca²⁺ uncaging, we found that both regulatory pathways, CDI as well as CDF, could be activated by relatively low elevations of [Ca²⁺]_i (<10 μm) implying a high affinity of the corresponding Ca²⁺ sensors. The high Ca²⁺ sensitivity of CDF is consistent with the finding that the calyceal I_Ca(V) facilitates strongly after conditioning sustained presynaptic depolarizations of small amplitude that do not trigger appreciable whole-cell Ca²⁺ currents (Hori and Takahashi, 2009). Assuming that the effector molecules inducing CDI and CDF are closely associated with VGCCs and thus are exposed to the Ca²⁺ micro- or nanodomains, the sensors may be nearly saturated during sustained channel opening. The similar affinity of their Ca²⁺ sensors on the one hand and >10-fold different recovery time constants (compare Figs. 4 and 7C) on the other hand suggest very different Ca²⁺ binding and unbinding kinetic for CDI and CDF. If the Ca²⁺ sensor for CDI binds Ca²⁺ much more slowly than that for CDF, its occupancy will remain low during very brief, AP-like Ca²⁺ transients. Short high-frequency AP trains therefore favor CDF that may be followed by CDI during prolonged AP trains. In contrast, its rapid relaxation prevents accumulation of CDF during low-frequency trains.

Physiological implications of CDI and CDF at the calyx of Held

Auditory brainstem circuits discharge at high rates during sound stimulation and show varying levels of spontaneous AP firing even in the complete absence of sound (Friauf and Ostwald, 1988; Taberner and Liberman, 2005; Tolnai et al., 2008; Sonntag et al., 2009). Within the MNTB of adult rats, spontaneous discharge rates vary between zero and 133 Hz (mean 29 Hz, Tolnai et al., 2008). Similar values have been reported for gerbils (mean 25 Hz, Hermann et al., 2007). Even higher spontaneous rates were measured in adult mice (mean 71 Hz, range 0.4–174 Hz, Lorteije et al., 2009). Such presynaptic firing rates will lead to a significant increase in presynaptic resting [Ca²⁺]_i levels because individual AP-evoked [Ca²⁺]_i transients summate (Korogod et al., 2005; Habets and Borst, 2006; Müller et al., 2007) and residual [Ca²⁺]_i will equilibrate with endogenous buffers and prevent them from binding incoming Ca²⁺ (Felmy et al., 2003). Therefore, presynaptic VGCCs in some calyces may be permanently in a partially inactivated state which may contribute to the lower release probability under in vivo compared to slice conditions (Lorteije et al., 2009). Surprisingly though, in recordings of trains of AP-like presynaptic I_Ca(V) in mouse endbulb of Held terminals which like calyces of Held predominantly express Ca_V2.1 VGCCs, inactivation was completely absent (Lin et al., 2011, their Fig. 9). Facilitation of presynaptic Ca²⁺ influx during brief high-frequency AP bursts may counteract synaptic depression caused by vesicle depletion and/or postsynaptic desensitization. On the other hand, clearance of Ca²⁺ accelerates with development (Chuhma and Ohmori, 2001) and the expression of calcium-binding proteins strongly increases (Lohmann and Friauf, 1996; Felmy and Schneggenburger, 2004) which may significantly limit the impact of CDI in the mature calyx of Held.