Developmental changes in potassium currents at the rat calyx of Held presynaptic terminal

Abstract

During early postnatal development, the calyx of Held synapse in the auditory brainstem of rodents undergoes a variety of morphological and functional changes. Among ionic channels expressed in the calyx, voltage-dependent K⁺ channels regulate transmitter release by repolarizing the nerve terminal. Here we asked whether voltage-dependent K⁺ channels in calyceal terminals undergo developmental changes, and whether they contribute to functional maturation of this auditory synapse. From postnatal day (P) 7 to P14, K⁺ currents became larger and faster in activation kinetics, but did not change any further to P21. Likewise, presynaptic action potentials became shorter in duration from P7 to P14 and remained stable thereafter. The density of presynaptic K⁺ currents, assessed from excised patch recording and whole-cell recordings with reduced [K⁺]_i, increased by 2–3-fold during the second postnatal week. Pharmacological isolation of K⁺ current subtypes using tetraethylammonium (1 mm) and margatoxin (10 nm) revealed that the density of Kv3 and Kv1 currents underwent a parallel increase, and their activation kinetics became accelerated by 2–3-fold. In contrast, BK currents, isolated using iberiotoxin (100 nm), showed no significant change during the second postnatal week. Pharmacological block of Kv3 or Kv1 channels at P7 and P14 calyceal terminals indicated that the developmental changes of Kv3 channels contribute to the establishment of reliable action potential generation at high frequency, whereas those of Kv1 channels contribute to stabilizing the nerve terminal. We conclude that developmental changes in K⁺ currents in the nerve terminal contribute to maturation of high-fidelity fast synaptic transmission at this auditory relay synapse.

Action potentials are generated by sequential gating of voltage-dependent Na⁺ channels and K⁺ channels (Hodgkin & Huxley, 1952). The properties of voltage-dependent K⁺ channels, as well as those of Na⁺ channels, critically determine the amplitude, duration and firing frequency of action potentials. At fast synapses, fidelity of high-frequency synaptic transmission is of crucial importance for accurate informational flow through neuronal circuits (Riehle et al. 1997; Diesmann et al. 1999). At an auditory relay synapse, the calyx of Held, in the medial nucleus of the trapezoid body (MNTB) of the brainstem, high-fidelity transmission is secured up to 1000 Hz (Wu & Kelly, 1993) for sound source localization (Oertel, 1997). This high-fidelity transmission is acquired during postnatal development in rodents, particularly during the second postnatal week when they start to hear sound (Blatchley et al. 1987).

An important factor determining the maximal frequency of synaptic transmission is the presynaptic action potential duration, which becomes shorter during the second postnatal week (Taschenberger & von Gersdorff, 2000; Fedchyshyn & Wang, 2005). Developmental changes in presynaptic Na⁺ currents (Leão et al. 2005) can contribute to shortening of presynaptic action potential duration. However, presynaptic K⁺ currents (I_pK) might also undergo developmental change and contribute to the action potential shortening. Developmental changes in K⁺ currents have been reported in postsynaptic neurons (Harris et al. 1988; O'Dowd et al. 1988; Gurantz et al. 1996; Gurantz et al. 2000; Riazanski et al. 2001; Hattori et al. 2003). However, little is known about developmental changes in K⁺ currents at the nerve terminal. In the present study, we addressed this issue by directly recording K⁺ currents from the calyx of Held presynaptic terminals of developing rats.

Methods

Preparations and solutions

All experiments were performed in accordance with the guideline of the Physiological Society of Japan. Brainstem slices were prepared from P7–21 Wistar rats as previously described (Forsythe & Barnes-Davies, 1993). Briefly, the rat was decapitated under halothane anaesthesia and the brain was quickly removed. Transverse slices (150–200 μm thick) containing MNTB were cut using a tissue slicer (ZERO-1; Dosaka, Kyoto, Japan). Slices were maintained in artificial cerebrospinal fluid (aCSF) at 37°C for 30–45 min and subsequently maintained at room temperature. MNTB principal cells and calyces were visually identified using a 60× water immersion objective lens (Olympus, Tokyo, Japan) attached to an upright microscope (Axioskop; Zeiss, Oberkochen, Germany).

The standard aCSF contained (mm): 125 NaCl, 2.5 KCl, 26 NaHCO₃, 1.25 NaH₂PO₄, 2 CaCl₂, 1 MgCl₂, 10 glucose, 3 myo-inositol, 2 sodium pyruvate and 0.5 ascorbic acid, pH 7.4 when bubbled with 95% O₂ and 5% CO₂. For recording K⁺ currents, the aCSF contained tetrodotoxin (TTX; 1 μm) (Wako, Osaka, Japan), and the pipette solution contained (mm): 97.5 potassium gluconate, 32.5 KCl, 10 Hepes, 5 EGTA, 1 MgCl₂, 12 Na₂ phosphocreatine, 2 ATP-Mg and 0.5 GTP-Na (295–305 mosmol l⁻¹, pH 7.3 adjusted with KOH; final K⁺ concentration, 143.5 mm). Presynaptic current-clamp recordings were made using the same pipette solution. In the majority of whole-cell K⁺ current recordings (Figs 2, 3, 4 and 5), to reduce driving force of K⁺, we replaced potassium gluconate (97.5 mm) in the pipette solution by equimolar N-methyl-d-glucamine gluconate (pH adjusted to 7.3 with gluconate; final internal K⁺ concentration, 32.5 mm); otherwise the amplitude of I_pK often exceeded the range of amplifier (±20 nA) when a command potential above +20 mV was given. In this low-K⁺ pipette solution, the reversal potential of K⁺ currents was −66 mV. For recording BK currents (Fig. 3) and action potentials (Fig. 7), EGTA concentration in the pipette solution was reduced to 0.2 mm. Presynaptic Na⁺ currents (I_pNa; Supplemental Fig. 1) were recorded with a low-Na⁺ aCSF containing (mm): 20 NaCl, 110 N-methyl-d-glucamine Cl, 20 tetraethylammonium (TEA) Cl, 2.5 KCl, 10 Hepes, 10 glucose, 2 BaCl₂, 0.1 CdCl₂ (pH 7.3 adjusted with HCl) and with a pipette solution containing (mm): 110 CsCl, 20 TEA, 10 Hepes, 10 EGTA, 1 MgCl₂, 5 tris-pohsphocreatine, 2 ATP-Mg, 0.5 GTP-Na (295–305 mosmol l⁻¹, pH adjusted to 7.3 with CsOH). The liquid junction potentials between the pipette solutions and aCSF were +10 mV for potassium gluconate solution, +7 mV for N-methyl-d-glucamine gluconate solution and −3 mV for CsCl solution, which were not corrected for unless otherwise noted. Drugs were applied to aCSF perfusing slices at the rate of 1.0–1.5 ml min⁻¹. TEA and 4-aminopyridine (4-AP) were purchased from Tokyo Kasei Kogyo (Tokyo, Japan) and Wako, respectively. The synthetic peptides margatoxin (MgTX) and iberiotoxin (IbTX) were from Peptide Institute (Osaka, Japan). All other chemicals and salts were from Nacalai (Kyoto, Japan), Sigma or Wako.

Figure 2.

Whole-terminal I_pK recorded with low-K⁺ (32.5 mM) pipette solution A, sample traces at various command voltages (left, superimposed) and current–voltage (I–V) relationships of I_pK (right) at P7–8 (•) and P13–15 (○). B, developmental increase in K⁺ current density. Left, membrane capacitance of calyceal terminals at P7–8 (filled bar) and P13–15 (open bar). *Significant difference (P < 0.05). Right, current density–voltage relationships at P7–8 and P13–15. The current density was calculated from the scaled I_pK amplitude divided by the membrane capacitance (left). The liquid junction potential was corrected for this I–V relationship.

Figure 3.

Developmental changes in 4-AP-sensitive and 4-AP-insensitive K⁺ currents including BK currents A, the current–voltage relationships of I_pK in the absence (circles) and presence (triangles) of 4-AP at P7–8 (filled symbols) and P13–15 (open symbols) calyces. B, K⁺ currents remaining in the presence of 4-AP (5 mm, a), 4-AP and iberiotoxin (IbTX, 100 nm, b), and BK currents isolated as IbTX-sensitive difference currents (a −b) at P7 and P14. C, density (left) and 10–90% rise time (right) of BK currents recorded at +20 mV at P7–8 (filled bars) and at P13–15 (open bars).

Figure 4.

Relative proportions of Kv1 and Kv3 components in I_pK at P7–8 and P13–15 I_pK was recorded in Cd²⁺ (100 μm)-containing aCSF using 32.5 mM K⁺ pipette solution. A, left panel, Kv3 current components obtained as TEA (1 mm)-sensitive difference currents (a −b). Sample traces show records at P13. Bottom, the current density–voltage relationships of Kv3 current components at P7–8 (•) and P13–15 (○). Right panel, Kv1 current component obtained as MgTX-sensitive difference currents (b −c) in the presence of TEA (1 mm). B, activation curves of Kv3 and Kv1 components at P7–8 (left, n = 8) and P13–15 (right, n = 6), with G/G_total in ordinates. The curves were fitted with the Boltzmann function, G = G_max/[1 +exp{−(V −V½)/k}], where V½ and k denote half-activation voltage and slope factor, respectively. Values of these parameters in Kv3 and Kv1 components are listed in Table 1. For the current density–voltage relationships and the activation curves, the liquid junction potential and series resistance remaining after compensation were corrected (see Methods).

Figure 5.

Developmental acceleration in the activation kinetics of Kv3 and Kv1 currents A, activation kinetics of Kv3 currents at P7 and P14. Top, sample traces at +40 mV are normalized with peak amplitudes and superimposed. Bottom, the 10–90% rise time at different voltages at P7–8 (•) and P13–15 (○). B, activation kinetics of Kv1 currents at P7 and P14. Sample traces at 0 mV are normalized with peak amplitudes and superimposed. *Significant difference for groups of data points between P7–8 and P13–15 (two-way ANOVA, P < 0.001 for Kv3; P < 0.05 for Kv1).

Figure 7.

Developmental increase in the fidelity of presynaptic spiking Action potentials were evoked by afferent fibre stimulation at room temperature. A, sample traces show action potentials evoked at 100 Hz and 400 Hz recorded from P8 and P14 calyceal terminals in the presence of IbTX (100 nm, centre column), IbTX + 1-mM TEA (right column) and in their absence (left column). Crosses indicate failures of action potential generation. Resting membrane potential was –68.3 ± 1.4 mV at P7–8 (n = 9) and –65.9 ± 1.9 mV at P13–15 (n = 12), which did not change after application of IbTX and TEA. B, graphs summarizing the fidelity of presynaptic spiking, expressed as a number of action potentials generated by a train of 30 stimuli (ordinates) at 100–400 Hz (abscissa), in the presence of IbTX (▵), TEA + IbTX (•) and in their absence (○), at P7–8 and P13–15. *Significant difference between control and TEA (two-way ANOVA, P < 0.01). C, spiking fidelity, represented as a success rate of action potential generation during a train of 30 stimuli at 400 Hz, in the presence or absence of TEA at P7–8 (filled bars) and P13–15 (open bars). *Significant difference (one-way ANOVA post hoc Duncan's MRT, P < 0.05).

Recording and data analysis

Whole-cell patch-clamp recordings were made from presynaptic calyceal terminals. The pipette resistance was 4.5–7.0 MΩ. Immediately after making a whole-cell patch, the access resistance was 6–15 MΩ, which was routinely compensated by 80%. We monitored access resistance throughout the recording and excluded the data from analysis if measured access resistance changed by ±20%. Voltage-clamp recordings were made using a patch-clamp amplifier (Axopatch-1D or -200B; Axon Instruments, Union City, CA, USA). Presynaptic currents were elicited by depolarizing command pulses in a stepwise manner from a holding potential of −80 mV. Leak currents in whole-cell recordings were subtracted by the scaled pulse (P/8) protocol. The current amplitude was measured at 10 ms after the command pulse onset for I_pK, and at the peak value for I_pNa. To estimate the I_pK density (Figs 2B and 4), the I_pK amplitude was corrected for the error caused by the series resistance remaining after compensation, by scaling a factor

where V_c is the command voltage (after liquid junction potential correction), V_rev is reversal potential for K⁺, I is recorded I_pK amplitude and R_s is remaining series resistance after compensation (Traynelis, 1998). This correction made a significant difference in the estimation of current amplitude, particularly at more mature calyces. The current density (nA pF⁻¹ or pA pF⁻¹) was then estimated from the corrected current amplitude divided by the membrane capacitance of a nerve terminal. The conductance (G) of K⁺ currents was estimated using following equation considering GHK rectification (Clay, 2000):

where R is the gas constant, F is Faraday constant and T is absolute temperature. Current-clamp recordings were made using a MultiClamp 700A (Axon Instruments) equipped with a high-input-impedance (10¹¹ Ω) voltage follower. Presynaptic action potentials were evoked by a depolarizing current injection via recording pipettes or afferent fibre stimulation using a monopolar glass pipette. Records were low-pass-filtered at 5 kHz and digitized at 20–50 kHz using an analog to digital converter (Digidata 1320A) with pCLAMP8.2/9.1 software (Axon Instruments). All experiments were carried out at room temperature (25–27°C).

Values in the text and figures are given as mean ± s.e.m. Statistical comparisons were made using the Student's unpaired t test unless otherwise noted. P < 0.05 was considered as significant. Curve fitting was performed using least-square procedures implemented in Ky Plot 4.0 (KyensLab Inc., Tokyo, Japan).

Results

Presynaptic action potential and K⁺ currents at different postnatal stages

Figure 1A illustrates action potentials evoked in calyceal presynaptic terminals by afferent fibre stimulation at different postnatal periods. During postnatal development from P7 to P14 the action potential duration became markedly shorter as previously reported (Taschenberger & von Gersdorff, 2000; Fedchyshyn & Wang, 2005). However, no further change was observed from P14 to P21. The mean action potential half-width was 487 ± 29 μs at P7–8 (n = 9), 244 ± 6 μs at P13–15 (n = 9) and 279 ± 9 μs at P19–21 (n = 5). During postnatal development from P7 to P21, neither the action potential amplitude nor resting membrane potential significantly changed.

Figure 1.

Developmental changes in the calyceal action potential and I_pK A, developmental shortening in the presynaptic action potential waveform. Left, action potentials were evoked by afferent fibre stimulation. Sample traces at P7, P14 and P20 are normalized with peak amplitude and superimposed. Right, the action potential half-width (duration at 50% of amplitude) at P7–8, P13–15 and P19–21. As the half-width of action potentials elicited by fibre stimulation was not significantly different from those evoked by a brief (0.5–1 ms) depolarizing current injection, both data were pooled in this plot. *Significant difference (one-way ANOVA post hoc Duncan's multiple range test (MRT), P < 0.01). B, developmental increase in I_pK. I_pK was evoked by a 50 ms depolarizing pulse (command voltage protocol shown in the bottom). Left, sample traces at various command voltages (superimposed). Right, the current–voltage (I–V) relationships of I_pK at P7–8 (•), P13–15 (○) and P19–21 (∇). C, developmental acceleration in the rise time of I_pK. Left, sample traces at 0 mV showing the rising phase of I_pK at different ages (superimposed after peak amplitude normalization). Right, voltage dependence of the 10–90% rise time at P7–8, P13–15 and P19–21. *Significant difference among data from three age groups (two-way ANOVA, P < 0.01).

The action potential duration is primarily determined by the properties of K⁺ and Na⁺ channels (Hodgkin & Huxley, 1952). To clarify developmental changes in presynaptic K⁺ channels, we evoked presynaptic K⁺ current (I_pK) in calyces under whole-cell voltage-clamp (Forsythe, 1994) at different postnatal periods. As shown in Fig. 1B, the I_pK amplitude became markedly larger during development from P7–8 to P13–15. Like the action potential duration, the I_pK amplitude did not further increase from P13–15 to P19–21. Another prominent change was observed for the activation time course of I_pK, which became significantly faster from P7–8 to P13–15, and remained stable thereafter (Fig. 1C). The rise time (10–90%) of I_pK (at 0 mV) was 2.1 ± 0.3 ms at P7–8 (n = 15), 1.2 ± 0.1 ms at P13–15 (n = 6) and 1.2 ± 0.1 ms at P19–21 (n = 5). The I_pK rise time showed a voltage dependence and was faster at more positive potentials. As both developmental changes in I_pK and presynaptic action potential duration occur during the second postnatal week, in the following studies, we focused on the developmental changes from P7–8 to P13–15.

Because of large current amplitudes, the I_pK amplitude often exceeded the amplifier range (±20 nA) when a command potential above +20 mV was given, particularly in more mature calyces. To improve voltage-clamp performance, we reduced I_pK amplitudes by reducing K⁺ concentration in recording pipette solutions by 75% (see Methods). In this low [K⁺]_i (32.5 mm) solution, I_pK could be recorded, for depolarizing command pulses of up to +40 mV. At +40 mV the I_pK amplitude was 3.7 ± 0.5 nA at P7–8 (n = 8) and 14.1 ± 1.0 nA at P13–15 (n = 9) (Fig. 2A), which were corrected for series resistance (see Methods) to be 4.2 ± 0.7 nA at P7–8 and 24.7 ± 3.9 nA at P13–15. During development the membrane capacitance of calyceal terminal increased from 9.9 ± 1.4 pF (n = 8) to 16.7 ± 1.5 pF (n = 9) (Fig. 2B) because of an increase in the surface membrane area (Wimmer et al. 2006). The K⁺ current densities, estimated from the corrected I_pK amplitude at +40 mV divided by nerve terminal capacitance, were 0.46 ± 0.08 nA pF⁻¹ at P7–8 and 1.51 ± 0.02 nA pF⁻¹ at P13–15, indicating a 3.3-fold increase during the second postnatal week.

Developmental changes in individual K⁺ current components

We next asked which K⁺ current component is responsible for this developmental increase in I_pK, as K⁺ currents in the calyx nerve terminal are composed of multiple components (Ishikawa et al. 2003). The main component is sensitive to 4-aminopyridine (4-AP), and comprises the high-voltage-activated (HVA) and tetraethylammonium (TEA)-sensitive Kv3 currents, and the low-voltage-activated (LVA) and margatoxin (MgTX)-sensitive Kv1 currents. The 4-AP-insensitive minor component comprises Ca²⁺-induced large-conductance K⁺ (BK) currents, and as yet unidentified slowly activating K⁺ currents.

4-AP (5 mm) blocked a large proportion of I_pK at P7–8 as at P13–15 (Fig. 3A). After 4-AP application, the high-voltage-activated Ca²⁺ currents appeared as inward currents, which had been masked by larger outward K⁺ currents (Ishikawa et al. 2003). The amplitude of Ca²⁺ currents remained similar during the second postnatal week (data not shown; see also Fedchyshyn & Wang, 2005). Thus the developmental increase in I_pK was caused predominantly by an increase in the 4-AP-sensitive component. The BK channel-specific blocker iberiotoxin (IbTX; 100 nm) attenuated outward currents remaining in the presence of 4-AP (Fig. 3B). The density of BK currents at +20 mV, estimated from the IbTX-sensitive difference currents, was 48 ± 12 pA pF⁻¹ at P7–8 (n = 5) and 40 ± 12 pA pF⁻¹ at P13–15 (n = 5) (Fig. 3C). Thus BK current density remained similar during the second postnatal week. The rise time (10–90%) of BK currents at P13–15 (1.5 ± 0.2 ms, n = 5) was not significantly different from that at P7–8 (1.8 ± 0.1 ms, n = 5, P = 0.20).

Parallel developmental changes in Kv3 and Kv1 currents in the nerve terminal

After blocking Ca²⁺ currents and BK currents with Cd²⁺ (100 μm), K⁺ currents were pharmacologically dissected. Bath-application of TEA (1 mm) attenuated K⁺ currents (Fig. 4A), and additional application of the scorpion peptide MgTX (10 nm) further attenuated the K⁺ currents. Currents remaining after applications of Cd²⁺ (100 μm), TEA (1 mm) and MgTX (10 nm) showed delayed rectifying properties (Fig. 4Ac) and were blocked by 10 mm TEA (Ishikawa et al. 2003), suggesting that they were K⁺ currents. Because of a lack of type-specific K⁺ channel blocker, we could not identify these currents. From these experiments the 1 mm TEA-sensitive Kv3 and MgTX-sensitive Kv1 components were isolated as difference currents. In this protocol, changing the order of blocker applications has no effect on the amplitude of difference currents, excluding an overlap in their blocking effects (Ishikawa et al. 2003). It also argues against possible errors arising from current subtraction, such as time-dependent change in series resistance or current amplitude-dependent change in the voltage-clamp condition. The TEA-sensitive Kv3 component underwent a 3-fold increase in density from P7–8 to P13–15 (Table 1). Similarly the MgTX-sensitive Kv1 component underwent a 3.1-fold increase, and the remaining currents underwent a 2.2-fold increase in densities from P7–8 to P13–15. Consequently, the proportion of Kv3 and Kv1 conductance relative to total K⁺ conductance remained essentially the same during the second postnatal week (Fig. 4B). When compared at the maximal conductance given by the sigmoid fits, the Kv3 proportion was 58 ± 2% of total K⁺ conductance (G_total) at P7–8 (n = 8) and 55 ± 5% at P14–15 (n = 6), whereas the Kv1 proportion was 24 ± 2% at P7–8 and 28 ± 4% at P14–15. The activation curves of Kv3 also remained essentially the same during development. Although the half-activation voltage of Kv1 currents slightly shifted toward positive potential, this difference was statistically insignificant (Table 1).

Table 1.

Voltage-gated K⁺ current components at the calyx of Held presynaptic terminal

	Age of animals	Amplitude (nA)	Current density (pA pF−1)	G/Gtotal (%)	V1/2	k (mV)
Kv3 currents	P7–8	2.7 ± 0.5	286 ± 59	58.2 ± 2.4	1.2 ± 1.9	9.8 ± 1.6
	P13–15	13.6 ± 3.5	847 ± 192	55.4 ± 4.7	−2.9 ± 4.1	7.8 ± 2.8
Kv3 currents	P7–8	1.1 ± 0.2	118 ± 31	23.9 ± 1.6	−34.5 ± 3.9	7.9 ± 4.0
	P13–15	6.2 ± 2.0	361 ± 111	27.7 ± 4.1	−20.1 ± 9.1	12.8 ± 7.6
Remaining currents	P7–8	0.7 ± 0.1	76 ± 15	16.5 ± 1.3	−2.3 ± 3.9	12.3 ± 3.1
	P13–15	2.7 ± 0.3	164 ± 17	13.1 ± 1.7	−16.1 ± 8.8	19.5 ± 6.6

The current amplitude was measured at +47 mV in the presence of Cd²⁺ (100 μm). The number of data is 8 at P7–8 and 6 at P13–15.

We next asked whether activation kinetics of Kv3 and Kv1 currents change during development. The rise times of Kv3 and Kv1 currents were both faster at more positive potentials, with stronger voltage dependence being observed for Kv3 currents compared with Kv1 currents. The rise time (10–90%) of Kv3 currents became three times faster from P7–8 (1.7 ± 0.2 ms, n = 8, at +40 mV) to P13–15 (0.5 ± 0.1 ms, n = 6) (Fig. 5A). The rise time of Kv1 currents also became faster from P7–8 (4.3 ± 0.7 ms, n = 8, at 0 mV) to P13–15 (2.0 ± 0.2 ms, n = 6) (Fig. 5B).

K⁺ currents in outside-out patch membrane excised from calyces

Further to attain optimal voltage-clamp performance we recorded K⁺ currents in outside-out patches excised from calyx nerve terminals at P7–8 and P13–15 (Fig. 6A). K⁺ currents were observed in all patches examined. The K⁺ current amplitude in excised patches was 248 ± 59 pA at P7–8 (n = 13) and 610 ± 166 pA at P13–15 (n = 10), indicating a 2.5-fold increase during development, which is comparable to the developmental increase in the K⁺ current density in whole-cell recordings (Fig. 2B). Consistent with whole-cell recordings, the activation kinetics of K⁺ currents in excised patches became faster during the second postnatal week, with the rise time (10–90%) at 0 mV being 4.6 ± 0.4 ms at P7–8 (n = 9) and 2.8 ± 0.4 ms at P13–15 (n = 8) (P < 0.01) (Fig. 6B). In excised patches, despite better voltage-clamp conditions, the rise time of K⁺ currents was slower than that of whole-terminal I_pK in both age groups (paired Student's t test, P < 0.03). Similar phenomena have been reported for K⁺ currents (Scannevin & Trimmer, 1997; Maguire et al. 1998) and Na⁺ currents (Shcherbatko et al. 1999; Leão et al. 2005), and are thought to arise from a cytoskeletal deformation by patch excision (Scannevin & Trimmer, 1997; Maguire et al. 1998; Shcherbatko et al. 1999).

Figure 6.

K⁺ currents recorded from outside-out patches excised from calyceal terminals Pipette solution contained 143.5 mm K⁺ for both whole-terminal and excised patch recordings in this figure. A, developmental increase in K⁺ currents in excised patches. K⁺ currents were evoked by 50 ms depolarizing pulses (command voltage protocol shown in the bottom). Sample traces at various command voltages (superimposed) and I–V relationships of K⁺ currents at P7–8 (•) and P13–15 (○). B, developmental acceleration in the rise time of K⁺ currents in excised patches. Left, sample traces at a command potential of +20 mV (superimposed after peak amplitude normalization). Right, the 10–90% rise time of K⁺ current in excised patches (filled bars) and whole-terminal I_pK( open bars) at P7–8 and P13–15. Command potential was 0 mV. *Significant difference in patch K⁺ currents between P7–8 and P13–15 (P < 0.05). C, Kv3 and Kv1 compositions in excised patches at P7–8 and P13–15. Kv3 currents (black) could be recorded in all patches examined. Kv1 currents (grey) were observed in none of five patches at P7–8, but recorded in 8 out of 13 patches at P13–15 (a). D, developmental acceleration in the activation kinetics of Kv3 currents in excised patches. Left, sample traces at P7 and P14 (at +20 mV, normalized with peak amplitude and superimposed). Right, the 10–90% rise time of Kv3 currents (at +20 mV) at P7–8 (filled bar) and P13–15 (open bar). *Significant difference (P < 0.05).

Immunocytochemical studies at P9 rat calyces suggest that Kv3 channels are expressed in the terminal, whereas Kv1 channels are expressed in the transition zone between the axon and terminal (Dodson et al. 2003). Consistently, in all excised patches examined at P7–8 calyces (n = 5), MgTX (10 nm) had no effect, whereas TEA (1 mm) attenuated K⁺ currents (by 59 ± 14%) (Fig. 6C). However, at P13–15, MgTX (10 nm) attenuated K⁺ currents (by 29 ± 7%) in 5 out of 13 patches, suggesting that Kv1 channels are expressed, albeit at a low density, within calyceal terminals. The TEA-sensitive Kv3 currents were recorded from all patches at both P7–8 and P13–15. As in whole-cell recordings (Fig. 5A), the rise time (10–90%) of Kv3 currents in excised patches became 2.2-fold faster from P7–8 (2.4 ± 0.7 ms, n = 5, at +20 mV) to P13–15 (1.1 ± 0.2 ms, n = 12) (Fig. 6D).

Functional outcome of the developmental increase in Kv1 and Kv3 currents

We next investigated the functional outcome of the developmental changes in presynaptic Kv3 and Kv1 currents. Action potentials were evoked by a train of fibre stimulations at room temperature. At P7–8, presynaptic action potentials followed inputs up to 200 Hz (30 stimuli), but started to show failures at 300 Hz (Fig. 7). At P13–15, however, presynaptic spikes followed afferent inputs up to 400 Hz with no failure. Although TEA (1 mm) blocks both Kv3 and BK channels (Coetzee et al. 1999), the BK channel-specific blocker IbTX (100 nm) had no effect on the spiking fidelity either at P7–8 or P13–15. Furthermore, raising EGTA concentration in recording pipettes to 5 mm had no effect on the spiking fidelity. Irrespective of whether IbTX was present, TEA (1 mm) markedly increased failures at 400 Hz at P13–15 (n = 5, P < 0.01). However, this effect of TEA was less significant at P7–8 (Fig. 7B and C). In the presence of TEA spiking fidelity at P7–8 calyces was similar to that at P13–15 calyces (Fig. 7C), suggesting that developmental changes in Kv3 currents contribute to the establishment of high-fidelity presynaptic spiking in response to high-frequency inputs. We next investigated whether the developmental changes in Kv1 currents contribute to action potential generation in calyceal nerve terminals. A strong sustained presynaptic depolarization (100 ms) by a current injection (up to 240 pA) evoked only several action potentials at the beginning of the depolarization at P13–15 calyces (3.7 ± 1.1, n = 5) (Fig. 8A), in all calyces examined as previously reported (Ishikawa et al. 2003). In contrast, at P7–8 calyces, action potentials occurred in a burst (21.8 ± 2.5, n = 6) throughout the depolarization (Fig. 8A). The threshold current for action potential generation increased from 73 ± 10 pA at P7–8 (n = 6) to 123 ± 12 pA at P13–15 (n = 6, P < 0.01) (Fig. 8B). After blocking Kv1 channels with MgTX, these age differences disappeared. In the presence of MgTX (10 nm), a sustained depolarization gave rise to a burst of action potentials in P13–15 calyces (Fig. 8A, see also Ishikawa et al. 2003). The number of spikes generated by a given current was similar to that in P7–8 calyces (Fig. 8B). Furthermore, the firing threshold for calyces was similar between P7–8 and P13–15. These results suggest that a developmental increase in Kv1 currents raises firing threshold and converts the spiking behaviour of the nerve terminal from tonic to phasic.

Figure 8.

Developmental increase in the stability of nerve terminals by Kv1 upregulation A, left panel, presynaptic action potentials elicited by a sustained (100 ms) depolarizing current in the absence (left column) or presence (right column) of MgTX (10 nm) at P7 and P14. MgTX had no effect on the resting membrane potential at both ages. Right panel, number of action potentials (ordinates) evoked by the depolarizing current pulse of various intensities (up to 240 pA, abscissa) in the absence (○) or presence (•) of MgTX. B, left: the number of action potentials (ordinate) evoked by the maximal depolarizing current injection (240 pA) in the presence or absence of MgTX at P7–8 and P13–15. Right, the minimal magnitude of injection currents (100 ms) required for generating action potentials, in the presence or absence of MgTX at P7–8 (filled bars) and P13–15 (open bars). *Significant difference (P < 0.01, one-way ANOVA post hoc Duncan's MRT).

Discussion

Our results indicate that the density and kinetics of K⁺ currents in calyceal terminals become higher and faster during the second postnatal week, thereby contributing to shortening of the action potential duration. Both Kv3 and Kv1 channels are involved in these changes and contribute to the establishment of reliable and stable presynaptic firings in response to high-frequency inputs.

Developmental changes in presynaptic K⁺ currents

During embryonic and postnatal development, Kv3 channels transcripts are upregulated in various CNS regions in rats (Perney et al. 1992; Gurantz et al. 2000), as well as in Xenopus (Gurantz et al. 2000). Expression of Kv1 channels also increases during development in culture (Gurantz et al. 1996; Grosse et al. 2000). In line with these reports, at the calyx of Held nerve terminal, both Kv3 and Kv1 whole-terminal currents underwent a 3-fold increase in density during the second postnatal week. Consistently, K⁺ currents recorded from patches excised from calyceal terminals underwent a 2.5-fold increase in density during the second postnatal week. In contrast, BK currents, which play little role in synaptic transmission at the calyx of Held (Ishikawa et al. 2003), did not undergo developmental change at this nerve terminal, unlike in cerebral (Kang et al. 1996) and cerebellar (Kang et al. 1996; Muller et al. 1998) neuronal somata. In agreement with our observations, the number of Kv3.1 immunogold particles, which are predominantly located on the non-synaptic side of calyceal terminals, undergoes 4-fold increase from P9 to P16 (Elezgarai et al. 2003).

The rise time of K⁺ currents became markedly faster during development at the calyx of Held, as in cultured Xenopus spinal neurons (Harris et al. 1988; O'Dowd et al. 1988). Although both Kv1 and Kv3 currents underwent developmental acceleration in activation kinetics, Kv3 currents have much faster rise time and higher density than Kv1 currents at +40 mV, to which action potentials normally reach. Therefore Kv3 currents are likely to determine the rise time of total I_pK. The developmental acceleration in Kv3 currents was not secondary to changes in their voltage dependence (Fig. 5), unlike the modulatory effect of casein kinase on the voltage dependence of Kv3 channel kinetics (Macica & Kaczmarek, 2001). Kv3 channels are classified into four subtypes having different activation kinetics (Rudy et al. 1999). Developmental reorganization of Kv3 channel subtypes toward faster activation kinetics might underlie acceleration in Kv3 current rise time. Recombinant channels of Kv3.1a and Kv3.1b mRNA splicing variants have similar kinetics (Yokoyama et al. 1989; Kanemasa et al. 1995), whereas Kv3.3 and Kv3.4 channels have relatively fast activation kinetics among Kv3 families (Rudy et al. 1999). It remains to be seen whether Kv3.3 or Kv3.4 channels increase their expression during development at the calyceal terminals.

Contribution of presynaptic K⁺ currents to action potential duration

Presynaptic action potentials at the calyx of Held become shorter in duration during the second postnatal week (Taschenberger & von Gersdorff, 2000; Fedchyshyn & Wang, 2005). Our results indicate that this developmental change reaches a steady level at around P14 (Fig. 1A). During the second postnatal week, I_pK increased in density and became faster in activation kinetics. Concomitantly, presynaptic Na⁺ currents became faster in inactivation kinetics (Supplemental Fig. 1; Leão et al. 2005). All these changes are likely to contribute to the developmental shortening of action potential duration. Blocking Kv3 channels by TEA (1 mm) prolongs action potential duration (Wang & Kaczmarek, 1998), whereas blocking Kv1 channels by MgTX (Ishikawa et al. 2003) or dendrotoxin (Dodson et al. 2003) has no such effect. Although both Kv3 and Kv1 currents undergo developmental changes, Kv3 currents have higher density (Fig. 4 and Table 1) and faster rise time kinetics (Fig. 5) compared with Kv1 currents. Taken together, these results suggest that developmental changes of Kv3 channels, rather than Kv1 channels, contribute to developmental shortening of action potential duration.

It has been reported that Kv1 immunoreactivity is predominantly located in the transition zone between the axon and calyceal terminal in P9 rats (Dodson et al. 2003). Consistently excised patches from P7–8 calyces showed no MgTX-sensitive Kv1 current. However, at P13–15 calyces, 5 out of 13 patches showed Kv1 current components (Fig. 6C). These results suggest that Kv1 channels undergo redistribution from the axon to the nerve terminal during postnatal development. This developmental redistribution of Kv1 channels, together with its increase in current density (Fig. 4), may contribute to stabilizing mature nerve terminals. Regarding the stabilizing role of Kv1 channels, it has been postulated that the firing behaviour during sustained depolarization depends upon axonal length (Dodson et al. 2003). However, at P13–15 calyces, in the absence of MgTX, we did not see tonic firing during sustained depolarization, as previously reported (Ishikawa et al. 2003), whereas tonic firing was consistently observed at P7–8 calyces.

Besides developmental changes in K⁺ currents and Na⁺ currents, morphological reformation of the nerve terminal (Kandler & Friauf, 1993) and redistribution of ion channels might affect presynaptic action potential duration. Morphological reformation of calyces from a spoon-shaped to a finger-like structure can accelerate extracellular K⁺ clearance, thereby potentially shortening action potential waveform (Clay, 2005), particularly during high-frequency spiking.

Physiological implications for the developmental changes in presynaptic K⁺ currents

The duration of the presynaptic action potential is directly correlated with the magnitude of Ca²⁺ influx, which triggers transmitter release (Katz & Miledi, 1969; Augustine, 1990; Borst & Sakmann, 1999). Therefore, the developmental shortening of the action potential can contribute to the developmental decrease in transmitter release probability observed at the rat calyx of Held during the second postnatal week (Iwasaki & Takahashi, 2001; Taschenberger et al. 2002). Developmental shortening of presynaptic action potential duration may also synchronize quantal release, shorten synaptic latency, and reduce its jitter (Borst & Sakmann, 1999; Fedchyshyn & Wang, 2005), thereby contributing to establishment of precisely timed synaptic transmission.

Reliable high-frequency synaptic transmission at the calyx of Held is essential for sound localization (Oertel, 1997). Mature calyceal synapses follow inputs of several hundred Hertz at room temperature (Fig. 7B, see also Wu & Kelly, 1993; Taschenberger & von Gersdorff, 2000), whereas transmission in response to high-frequency inputs is unreliable in immature calyces (Futai et al. 2001). During postnatal development, high-fidelity transmission at high frequency is acquired through developmental changes in both presynaptic and postsynaptic elements. Postsynaptically, developmental downregulation of NMDA receptors, which depends upon auditory activity, contributes to acquisition of high-fidelity transmission up to 100 Hz (Futai et al. 2001). Developmental shortening in the decay time of EPSCs (Taschenberger & von Gersdorff, 2000; Futai et al. 2001; Joshi & Wang, 2002; Yamashita et al. 2003; Koike-Tani et al. 2005), caused by speeding of deactivation and desensitization kinetics of AMPA receptor channels (Koike-Tani et al. 2005), may further contribute to high-fidelity transmission above 100 Hz (Joshi et al. 2004).

Presynaptically, developmental changes in Kv3 channels contribute to shortening of action potential duration. Furthermore, developmental stabilization of nerve terminals by the Kv1 channel upregulation and redistribution, suppresses aberrant firings. These changes, together with a developmental decrease in the recovery time of Na⁺ channels from inactivation (Leão et al. 2005), would enable calyceal terminals to fire reliably in response to high-frequency inputs, thereby contributing to the establishment of high-fidelity synaptic transmission.

Supplementary material

Online supplemental material for this paper can be accessed at:

http://jp.physoc.org/cgi/content/full/jphysiol.2007.128702/DC1 and http://www.blackwell-synergy.com/doi/suppl/10.1113/jphysiol.2007.128702