Abstract
The clearance of Ca2+ from nerve terminals is critical for determining the build-up of residual Ca2+ after repetitive presynaptic activity. We found previously that K+-dependent Na+/Ca2+ exchangers (NCKXs) show polarized distributions in axon terminals of supraoptic magnocellular neurons and play a major role in Ca2+ clearance. The role of NCKXs in presynaptic terminals, however, has not been studied. We investigated the contribution of NCKX in conjunction with other Ca2+ clearance mechanisms at the calyx of Held by analyzing the decay of Ca2+ transients evoked by depolarizing pulses. Inhibition of Na+/Ca2+ exchange by replacing external Na+ with Li+ decreased the Ca2+ decay rate by 68%. Selective inhibition of NCKX by replacing internal K+ with TEA+ (tetraethylammonium) or Li+ decreased the Ca2+ decay rate by 42%, and the additional inhibition of the K+-independent form of Na+/Ca2+ exchanger (NCX) by reducing external [Na+] caused an additional decrease by 26%. Inhibition of plasma membrane Ca2+-ATPase (PMCA) decreased the Ca2+ decay rate by 23%, whereas inhibition of SERCA (smooth endoplasmic reticulum Ca2+-ATPase) had no effect. The contribution of mitochondria was negligible for small Ca2+ transients but became apparent at [Ca2+]i > 2.5 μm, when Na+/Ca2+ exchange became saturated. Mitochondrial contribution was also observed when the duration of Ca2+ transients was prolonged by inhibiting Na+/Ca2+ exchangers or by increasing Ca2+ buffers. These results suggest that, in response to small Ca2+ transients (<2 μm), Ca2+ loads are cleared from the calyx of Held by NCKX (42%), NCX (26%), and PMCA (23%), and that mitochondria participate when the Ca2+ load is larger or prolonged.
Introduction
In presynaptic axon terminals, calcium plays a key role in activity-dependent changes in synaptic strength as well as for the release of neurotransmitter. Incoming Ca2+ during action potentials (APs) rapidly equilibrates with Ca2+ buffers and then decays by the concerted action of different Ca2+ clearance mechanisms (CCMs): plasma membrane Ca2+-ATPase (PMCA), Na+/Ca2+ exchanger (Na/CaX), smooth endoplasmic reticulum Ca2+-ATPase (SERCA), and mitochondria. Recent studies have shown that presynaptic CCMs have a strong influence on presynaptic residual calcium, which has been proposed as an underlying mechanism for short-term synaptic plasticity. It has been reported that presynaptic residual calcium after a period of high-frequency activity is caused by Ca2+ influx through a reverse-mode Na/Ca exchange or by a delayed Ca2+ release from mitochondria subsequent to the Ca2+ uptake (Tang and Zucker, 1997; Zhong et al., 2001). Investigating the quantitative role of different CCMs in nerve terminals will thus provide a basic understanding of the mechanisms underlying short-term plasticity.
There is evidence for presynaptic roles of PMCA in retinal bipolar cells (Zenisek and Matthews, 2000), for Na/CaX in cultured hippocampal neurons (Reuter and Porzig, 1995) and in cerebellar granule cells (Regehr, 1997), and for mitochondria in sympathetic ganglion (Peng, 1998). Previous studies on presynaptic CCMs aimed to determine the contribution of specific mechanisms but have not provided a comprehensive picture of how different CCMs interact with each other as a function of the cytosolic [Ca2+] attained after a stimulus. Considering that these mechanisms have different properties in terms of the affinity and the clearance capacity for Ca2+, intracellular calcium load might be one of the important factors that affect the relative contribution of CCMs (Fierro et al., 1998; Suzuki et al., 2002; Kim et al., 2003). In this respect, quantitative studies are required to elucidate the extent to which each CCM contributes depending on calcium load.
At the calyx of Held, a mammalian giant presynaptic terminal, [Ca2+]i can be measured quantitatively using Ca2+ indicator dye introduced via a patch pipette. Recently, it has been reported that mitochondria contribute to the Ca2+ clearance at the calyx of Held and that mitochondrial Ca2+ uptake affects recovery from synaptic depression (Billups and Forsythe, 2002). However, Chuhma and Ohmori (2002) showed that the Ca2+ extrusion rate was reduced by inhibition of Na/CaX at the same preparation. Functional and immunocytochemical evidence indicates the existence of Na/CaX in the presynaptic terminals of other mammalian CNS synapses (Reuter et al., 1995; Regehr, 1997). Moreover, we recently described the role of a K+-dependent form of Na+/Ca2+ exchanger (NCKX) in peptidergic axon terminals and found that its distribution is polarized to the axon terminals in hypothalamic magnocellular neurons (Lee et al., 2002; Kim et al., 2003). The role of NCKX, however, has not yet been studied in CNS presynaptic terminals, probably because of the inaccessibility of most nerve terminals for intracellular ion exchange experiments. Here, we tested whether NCKX contributes to Ca2+ clearance at the calyx of Held and elucidated the dependence of the relative contributions of Na/CaX and other CCMs, especially mitochondria, on the intracellular calcium load.
Materials and Methods
Preparation of brainstem slices. Transverse brainstem slices containing the medial nucleus of the trapezoid body (MNTB) were prepared from 8- to 10-d-old Sprague Dawley rats (17 ± 4 g). Rats were decapitated, and brainstems were chilled in ice-cold low-calcium artificial CSF (aCSF) containing the following (in mm): 125 NaCl, 25 NaHCO3, 2.5 KCl, 1.25 NaH2PO4, 2.5 MgCl2, 0.5 CaCl2, 25 glucose, 0.4 Na ascorbate, 3 myo-inositol, 2 Na pyruvate, at pH 7.4, when saturated with carbogen (95% O2, 5% CO2), and with an osmolarity of ∼320 mOsm. Isolated brainstems were glued onto the stage of a vibratome (VT1000S; Leica, Wetzlar, Germany), and 150- to 200-μm-thick transverse brainstem slices were cut from caudal to rostral in the same solution. Slices containing the MNTB (approximately four to five slices) were incubated at 37°C for 30 min in normal aCSF (as low-calcium aCSF above with 1 mm MgCl2 and 2 mm CaCl2) and thereafter maintained at room temperature (23-25°C) until required.
Electrophysiological recordings. Whole-cell patch-clamp recordings of calyces of Held were made under visual control using differential interference illumination in an upright microscope (BX50WI; Olympus, To-Tokyo, Japan). Calcium influx was evoked by applying depolarizing pulses in voltage-clamp mode. Patch pipettes with a resistance of 4.5-5.5 MΩ were used for recordings. The standard K+ pipette solution contained the following (in mm): 120 K gluconate, 30 KCl, 20 HEPES, 4 MgATP, 4 Na ascorbate, and 0.3 NaGTP at pH 7.3 (adjusted with KOH). For a K+-free pipette solution, K gluconate and KCl in the K+ pipette solution were replaced with equimolar tetraethylammonium (TEA)-Cl. Alternatively, to rule out the possibility of organic cations inhibiting the K+-independent form of Na/CaX (NCX), K+ ions in the K+ pipette solution were replaced with Li+ ions (Blaustein, 1977). Recordings were made in calyces of Held using an EPC-9 amplifier (HEKA Elektronik, Lambrecht, Germany). During recordings, series resistances were compensated up to 85% with a 10 μs lag time. Recordings were terminated when the series resistance exceeded 30 MΩ. Experiments were performed at 35 ± 1°C. All chemicals were obtained from Sigma (St. Louis, MO), except 5(6)-carboxyeosin diacetate (CE) and fura-2FF from Fluka (Buchs, Switzerland) and fura-4F from Molecular Probes (Eugene, OR).
Cytosolic Ca2+ measurements. The procedures for cytosolic Ca2+ measurement in slices have been described previously in detail (Kim et al., 2003). Ca2+ concentrations were measured by fluorescence imaging. Cells were loaded with fura-4F or fura-2FF (pentapotassium salts; 50 μm each) via patch electrodes. For fluorescence excitation, we used a polychromatic light source (xenon lamp based; Polychrome-II; TILL Photonics, Gräfelfing, Germany), which was coupled to the epiillumination port of an upright microscope (BX50; Olympus) via a quartz light guide and a UV condenser. Imaging was performed using a 60× water immersion objective (numerical aperture, 0.9; LUMPlanFl; Olympus) and an air-cooled slow-scan CCD camera (SensiCam; PCO, Kelheim, Germany). The monochromator and the CCD camera were controlled by a personal computer, running a custom-made software programmed with MicroSoft Visual C++ (version 6.0).
The ratio (R = Fiso/F380) of fluorescence at the isosbestic wavelength (360 nm; Fiso) to that at 380 nm (F380) was converted to [Ca2+]i according to the following equation: (1)
Calibration parameters were determined by “in-cell” calibration as described by Lee et al. (2000). Rmin values were measured using a Ca2+-free pipette solution containing 10 mm BAPTA. Rmax values were obtained from in vitro measurements, because calyces of Held did not endure internal dialysis with high CaCl2 (10 mm). The values for the calibration ratio at intermediate [Ca2+]i were measured using a pipette solution containing 5 mm BAPTA and 3.5 mm CaCl2 ([Ca2+]i ≈ 540 nm) for fura-4F and a pipette solution containing 80 mm DPTA and 10 mm CaCl2 ([Ca2+]i ≈ 10 μm) for fura-2FF. The effective dissociation constant of fura-2 (Keff) was calculated by measuring the fluorescence ratio at these intermediate [Ca2+]i and by rearranging Equation 1 for Keff. The Kd values of fura-4F and fura-2FF were calculated as 0.79 and 3.1 μm, respectively, from Kd = Keff · (α + Rmin)/(α + Rmax), where α is the iso-coefficient (Zhou and Neher, 1993). To increase time resolution and minimize the photobleaching effect, we adopted a single-wavelength protocol (Helmchen et al., 1996; Lee et al., 2000) and pixels were binned by 8 × 8, which allowed exposure times of 5 ms. Images taken at 40 Hz with a single-wavelength excitation at 380 nm (F380) were preceded and followed by images taken with excitation at isosbestic wavelengths (360 nm). Fiso (isosbestic fluorescence) values were linearly interpolated between points just before and after the period of excitation at 380 nm.
Calculation of total Ca2+ clearance rate. We measured the decay rate of free Ca2+ from the time derivative of the decay phase of a Ca2+ transient (d[Ca2+]i/dt). Because genuine Ca2+ clearance rate of a cell is represented by the clearance rate of total calcium (d[Ca2+]T/dt) rather than by decay rate of free calcium, d[Ca2+]i/dt was converted to d[Ca2+]T/dt using the following relationship: (2)
where κS and κB represent the calcium binding ratios of the endogenous and exogenous buffers, respectively (Neher and Augustine, 1992; Kim et al., 2003).
When the decay phase of a Ca2+ transient is fitted with the following biexponential function: (3)
the Ca2+ decay rate at the peak of the Ca2+ transient can be represented by a time derivative of the biexponential function estimated at t = 0 as follows: (4)
where A and λ represent amplitude and rate constant (inverse of time constant) of each component, respectively. In addition, according to Equation 2, the total Ca2+ clearance rate at the peak of a Ca2+ transient can be calculated by the following: (5)
Calculation of Ca2+ decay rate constant at the peak (λt = 0). Although the decay phase of Ca2+ transients in this study were best fitted with the biexponential function, each (fast or slow) component of the biexponential fit does not necessarily reflect the activity of a single entity of Ca2+ clearance mechanism, because the Ca2+ clearance caused by each CCM was not constant but is nonlinearly dependent on the [Ca2+]i excursion from resting level (Δ[Ca2+]i) (see Figs. 1 D,2 Bb, dotted lines). Thus, for statistical comparison of Ca2+ clearance measured from different cells and/or different experimental conditions, the decay rate at the peak was normalized by the peak Δ[Ca2+]i level (Δ[Ca2+]peak), resulting in a Ca2+ decay rate constant at the peak (λt = 0), which is defined as follows: (-d[Ca2+]i/dt)t = 0/Δ[Ca2+]peak (Kim et al., 2003). In the particular case in which the Ca2+ decay is biexponential, λt = 0 can be calculated by the following equation: (6)
Although the relationship between d[Ca2+]i/dt and Δ[Ca2+]i is not linear over the entire range of Δ[Ca2+]i, the relationship is almost linear when Δ[Ca2+]i is between 0.6 and 2.5 μm (see Figs. 1, 2, 3, 4, plots of d[Ca2+]T/dt vs Δ[Ca2+]i). In this range of Δ[Ca2+]i, λt = 0 is constant, independent of Δ[Ca2+]i, especially when peak Δ[Ca2+]i levels vary within a relatively narrow range. Thus, we adjusted the depolarizing pulse such that the peak Δ[Ca2+]i level of a Ca2+ transient fell between 2 and 2.5 μm, and we used λt = 0 as a parameter to statistically compare the Ca2+ clearance rates.
Calculation of relative contribution of a clearance mechanism. The contribution of a clearance mechanism, Φ, to the entire Ca2+ clearance at a given Δ[Ca2+]i level has been described previously (Kim et al., 2003). From the time derivatives (d[Ca2+]i/dt) of the decay phases of two Ca2+ transients before and after treatment with an inhibitor of Φ, total Ca2+ clearance rate (d[Ca2+]T/dt) was calculated and plotted as a function of Δ[Ca2+]i (Eq. 2). The difference between the polynomial fit (fcontrol) to the d[Ca2+]T/dt curve under control condition and that (finhibitor) in the presence of the inhibitor was regarded as the contribution made by Φ, and the relative contribution made by Φ(RΦ) to the entire Ca2+ clearance mechanism can be expressed as follows: (7)
Data analysis. Data were analyzed using IgorPro (version 4.1; WaveMetrics, Lake Oswego, OR). Statistical data are expressed as mean ± SEM, and n indicates the number of cells studied. The significance of differences between Ca2+ decay rate constants was evaluated using paired or nonpaired Student's t test using a significance level of 0.05.
Results
We investigated Ca2+ clearance mechanisms at the calyx of Held (Fig. 1A) by analyzing the decay of Ca2+ transients obtained from fluorescent Ca2+ imaging. Calyces of Held were loaded via a patch pipette with an internal solution containing 50 μm fura-4F or fura-2FF, and Ca2+ transients were evoked by applying a short depolarizing pulse (from a holding potential of -80 or -70 mV to 0 mV; 50 ms in duration). The decay phases of the Ca2+ transients were well fitted by biexponential functions (χ2 < 0.05), and the fitting parameters in various experimental conditions are summarized in Table 1. The clearance is defined as a decay rate constant (τ-1) in the case of a monoexponential decay. To obtain the parameter representing the Ca2+ clearance from biexponential Ca2+ decay rates, we adopted a weighted average of two decay rate constants (λt = 0) (see Materials and Methods), which is mathematically the same as the decay rate at the peak divided by the amplitude of a biexponential Ca2+ transient [(-d[Ca2+]i/dt)t = 0/Δ[Ca2+]peak]. Thus, we refer to λt = 0 as a Ca2+ decay rate constant at the peak (see Materials and Methods). Because the relative amplitude of the fast component was >80% in this study, the value for λt = 0 is quite close to τfast-1 (Fig. 1C).
Na+/Ca2+ exchange is a major CCM at the calyx of Held To investigate the contribution of Na/CaXs to Ca2+ clearance, we examined the effect of [Na+]o reduction (replacement of the 125 mm NaCl in the bathing solution with equimolar LiCl) on the Ca2+ decay. The calyces were stimulated by a depolarization pulse (50 ms) to 0 mV in voltage-clamp mode. With the standard K+ pipette solution (see Materials and Methods), [Na+]o reduction significantly reduced the Ca2+ decay rate (Fig. 1Ba). Mean values for fast (τfast) and slow (τslow) time constants in control condition (Fig. 1Ba, black line) were 74.2 ± 4.1 and 975.4 ± 108.7 ms (n = 7), respectively. After [Na+]o reduction (Fig. 1Ba, gray line), the increase in the fast time constant (τfast) was more significant (218.7 ± 13.5 ms; n = 7; paired t test, p < 0.01) than the increase in the slow time constant (τslow; 1428.2 ± 236.0 ms; n = 7; paired t test, p > 0.05) (Table 1). The preferential deceleration (approximately threefold) of the fast decay phase under the low [Na+]o condition suggests that Na/CaX is a Ca2+ clearance mechanism of high-capacity and low-affinity at the calyx of Held.
We next tested whether the Na/CaX activity depends on intracellular K+. For this purpose, we recorded the Ca2+ transients using K+-free pipette solutions. Figure 1, Bb and Bc, shows the representative Ca2+ transients recorded with Li+ and TEA+ pipette solutions, respectively. The Ca2+ transients in normal [Na+]o (black line) and after the Na+o replacement (gray line) were superimposed for each recording condition. In normal [Na+]o condition, the decay rate constants (λt = 0) of the Ca2+ transients recorded with Li+ and TEA+ pipettes were 7.18 ± 0.27 s-1 (n = 6) and 6.69 ± 0.22 s-1 (n = 3), respectively (Fig. 1C). These values are significantly lower than that with the K+ pipette (12.01 ± 0.72 s-1; p < 0.01) (Fig. 1C). The slower decay rate constant in the K+-free internal condition suggests that NCKXs contribute to Ca2+ clearance at the calyx of Held. Under the K+-free internal condition, [Na+]o reduction further decreased the Ca2+ decay rate (Fig. 1Bb,Bc, gray traces). When the Na+ gradient was reduced, the decay rate constants were reduced to 3.12 ± 0.17 s-1 in the Li+ pipette condition and to 3.09 ± 0.25 s-1 in the TEA+ pipette condition, which are similar to that in the K+ pipette condition (3.80 ± 0.37 s-1; p > 0.05) (Fig. 1C). These results show that Na/CaX activity still remained even in the absence of internal K+, suggesting that both NCX and NCKX contribute to the calcium clearance at the calyx of Held.
To quantify the relative contributions of NCX and NCKX, the decrease in the Ca2+ decay rate constant (λt = 0) caused by low [Na+]o in the K+ pipette condition was compared with those in the K+-free pipette condition. The relative contributions of NCKX according to the equations (λt = 0,K_Na - λt = 0,Li_Na)/λt = 0,K_Na and (λt = 0,K_Na -λt = 0,TEA_Na)/λt = 0,K_Na were estimated to be ∼41 and 45%, respectively (the first subscripts, K, Li, and TEA, represent major internal cations, and the second subscripts, Na and Li, represent major external cations). Similarly, the contributions of NCX according to (λt = 0,Li_Na -λt = 0,K_Li)/λt = 0,K_Na and (λt = 0,TEA_Na - λt = 0,K_Li)/λt = 0,K_Na were 28.2 and 24.0%, respectively. NCKX and NCX were responsible for ∼42 and 26% of total Ca2+ clearance, respectively. The overall contribution of Na/CaXs, including both NCX and NCKX, was calculated using (λt = 0,K_Na - λt = 0,K_Li)/λt = 0,K_Na to be 68.6 ± 1.7%.
To estimate the relative contribution of Na/CaX to the entire Ca2+ clearance as a function of the Δ[Ca2+]i level, the total Ca2+ clearance rate (-d[Ca2+]T/dt) was calculated from the time derivative of the decay phase of a Ca2+ transient according to Equation 2, using a κS value of 40 (Helmchen et al., 1997) (Fig. 1D). From the polynomial fits to -d[Ca2+]T/dt measured before (fK_Na) (Fig. 1D, black solid line) and after (fK_Li) (Fig. 1D, gray solid line) [Na+]o reduction in the K+ pipette condition, we calculated the relative contribution of Na/CaX according to the equation (fK_Na - fK_Li)/fK_Na (Fig. 1D, open circles, right axis). The relative contribution of Na/CaX increased as Δ[Ca2+]i increased to reach a maximum value (69%) at Δ[Ca2+]i > 400 nm. We conclude that Na/CaX is the primary CCM in the calyx of Held in response to brief Δ[Ca2+]i elevations in the range of 1-2 μm. Our experiments under different internal ionic conditions suggest that Na/CaX is a major Ca2+ clearance mechanism, and that both NCX and NCKX contribute to Na/CaX in the calyx of Held.
Calyces of Held in young rats [postnatal day 8 (P8) to P10] can follow discharge rates up to 200 Hz (Borst et al., 1995), and the calyx of Held in adult mice receives and follows higher input frequencies of up to 600 Hz (Wu and Kelly, 1993). We examined the role of Na/CaX in Ca2+ clearance when Ca2+ transients were evoked by repetitive depolarization pulses (2 ms in duration) at frequencies of 10, 20, or 50 Hz for 1 s. The Δ[Ca2+]i during the 10 or 20 Hz stimulation underwent a fast rising phase and reached a plateau within 300 ms under control condition (Fig. 1E, black lines), but at 50 Hz stimulation, the fast rising phase was followed by a slower rising phase. Linear summation of Ca2+ transients predicts that the time required to reach the steady state is determined by the decay time constant of a single Ca2+ transient (Regehr et al., 1994; Helmchen et al., 1996). Consistently, after [Na+]o reduction, the Δ[Ca2+]i level continued to build up throughout the train of pulses (1 s), and thus Δ[Ca2+]i at the end of the stimulation was greatly increased (Fig. 1E, gray lines). At a stimulation frequency of 50 Hz, the amplitudes of Δ[Ca2+]i before and after [Na+]o reduction were 1.88 ± 0.37 and 4.95 ± 0.19 μm, respectively (n = 3) (Fig. 1E). These results indicate that Na/CaXs dampen presynaptic Ca2+ load very efficiently. Such efficient Ca2+ clearance seems to be essential for presynaptic axon terminals to transduce high-frequency spike input to synaptic release of neurotransmitter and to maintain presynaptic Ca2+ homeostasis.
Contribution of SERCA and PMCA pumps to presynaptic Ca2+ clearance
To investigate the contributions of SERCA and PMCA to Ca2+ clearance, we examined the effects of specific inhibitors of SERCA and PMCA on the Ca2+ decay rate measured with the K+ pipette solution containing 50 μm fura-4F.
Inhibition of SERCA by the application of 2 μm thapsigargin (TG) to the bath had no effect on Ca2+ transients (Fig. 2Aa). The Ca2+ transients obtained before and after SERCA inhibition were superimposable. Another SERCA inhibitor, 5 μm cyclopiazonic acid (CPA), also had no effect (Fig. 2Aa, inset). The total Ca2+ clearance rate (-d[Ca2+]T/dt) as a function of Δ[Ca2+]i showed no difference before and after SERCA inhibition (Fig. 2Ab). These results suggest that SERCA contributes little to Ca2+ clearance, in agreement with previous work at the calyx of Held (Billups and Forsythe, 2002; Chuhma and Ohmori, 2002).
The contribution of the PMCA to Ca2+ clearance was assayed using 40 μm CE, a blocker of PMCA (Gatto and Milanick, 1993; Bassani et al., 1995). The Ca2+ decay was slightly slowed by CE (Fig. 2Ba). After PMCA inhibition, both the fast and slow time constants of Ca2+ decay were significantly increased (n = 6; paired t test; p < 0.01) (Table 1). Because CE can inhibit SERCA as well as PMCA (Fierro et al., 1998), 2 μm TG was preapplied to inhibit SERCA in some experiments, but results obtained with and without pretreatment of TG were similar. In contrast to Na/CaX, the relative contribution of PMCA was higher at lower Δ[Ca2+]i levels and decreased as Δ[Ca2+]i increased (Fig. 2Bb, gray open circles, right axis), which indicates that PCMA is a CCM of low capacity and high affinity. Figure 2Bc summarizes the effect of carboxyeosin on Ca2+ decay rate constants. The relative contribution of PMCA calculated according to the equation (λt = 0,control - λt = 0,CE)/λt = 0,control was 23.8 ± 1.7% at Δ[Ca2+]i of 1 μm.
Role of mitochondria in presynaptic Ca2+ clearance
The contribution of mitochondrial Ca2+ uptake to presynaptic CCM was investigated using the protonophore carbonyl cyanide m-chlorophenylhydrazone (CCCP), which dissipates mitochondrial membrane potential and inhibits mitochondrial Ca2+ uptake. The application of CCCP had no effect on the resting Ca2+ concentration (100 ∼ 150 nm). In the K+ pipette condition, the bath application of 2 μm CCCP had no effect on Ca2+ transients (Fig. 3Aa). However, when NCKX was inhibited by using a TEA+ pipette solution, CCCP significantly slowed the Ca2+ decay rate (Fig. 3Ab). Plots of -d[Ca2+]T/dt as a function of Δ[Ca2+]i before and after CCCP application are superimposed in Figure 3B, and C summarizes the effect of CCCP on the Ca2+ decay rate constants with K+ and TEA+ pipette conditions.
We further tested whether the mitochondrial contribution to the Ca2+ clearance occurred when Na/CaX was inhibited. Using the K+ pipette, Ca2+ decay was slowed by [Na+]o reduction (Fig. 3D; black line, normal aCSF; gray line, low Na+ aCSF). The addition of 2 μm CCCP to the low [Na+]o bath solution led to an additional slowing of Ca2+ decay (Fig. 3D, gray dotted line). This result contrasted with the finding that CCCP had no effect on the Ca2+ decay rate measured with a K+ pipette in normal aCSF (Fig. 3Aa). When we reintroduced external Na+ in the presence of CCCP, the Ca2+ transient was completely restored to the control condition level (Fig. 3D; black line, normal aCSF; black dotted line, normal aCSF plus 2 μm CCCP), indicating that mitochondria do not contribute to Ca2+ clearance when Na/CaX is fully functional. Consistently, -d[Ca2+]T/dt plots calculated from control (normal aCSF) and recovery (normal aCSF plus 2 μm CCCP) data overlapped almost completely (Fig. 3E). The mean values for λt = 0 in each condition are compared in Figure 2F. Whereas λt = 0 was significantly reduced by CCCP when Na/CaX was inhibited (p < 0.01) (Fig. 3F), λt = 0 measured when [Na+]o was reintroduced did not differ from that determined under the control condition despite the continued presence of CCCP (p = 0.85). These results imply that Na/CaX clears Ca2+ loads more readily than mitochondria, and that mitochondria, which compete for incoming Ca2+ with Na/CaX, take part in Ca2+ clearance only when Na/CaX is inhibited. In addition, the contribution of Na/CaX to Ca2+ clearance may be higher than our estimate (68%), because the inhibition of Na/CaX could be partially compensated by mitochondria under low [Na+]o condition.
Dependence of mitochondrial Ca2+ clearance on Δ[Ca2+]i
Studies of mitochondrial Ca2+ uptake suggest that mitochondrial Ca2+ uptake requires a high local Ca2+ concentration in the micromolar range because of the low affinity of mitochondria for Ca2+ (Brini, 2003). Thus, we examined the mitochondrial contribution to Ca2+ clearance at higher Δ[Ca2+]i levels. To induce various peak Δ[Ca2+]i of up to 6 μm, we increased [Ca2+]o to 5 mm, and applied a single-step pulse (50 ms in duration) of various depolarization levels. To avoid saturation of Ca2+ indicator dye, 50 μm fura-2FF (Kd ≈ 3.1 μm) was used instead of fura-4F. Ca2+ decay rates measured using fura-2FF were more rapid than those measured with the same concentration of fura-4F, reflecting its lower affinity for Ca2+ (Table 1). When peak Δ[Ca2+]i was <2.5 μm, CCCP had no noticeable effect on Ca2+ transients (Fig. 4A, left traces), which is consistent with the results in Figure 3A. However, when peak Δ[Ca2+]i was higher than 3 μm, CCCP slowed Ca2+ decay, and this effect of CCCP became more obvious as peak Δ[Ca2+]i was further increased (Fig. 4A, middle and right).
We analyzed 11 pairs of Ca2+ transients obtained from five different calyces. The relative contribution of mitochondria (Rmito) in each pair of Ca2+ transients (before and after the CCCP treatment) was calculated according to the equation (λt= 0,control - λt = 0,CCCP)/λt = 0,control, and plotted as a function of Δ[Ca2+]i (Fig. 4B, triangles). The mean value for Rmito in the range of Δ[Ca2+]i > 2.5 μm (filled triangles) was significantly higher than that in the lower range (Fig. 4B, open triangles) (p < 0.01). These results indicate that mitochondria begin to take part in Ca2+ clearance when Δ[Ca2+]i is >2.5 μm.
For the same set of Ca2+ transients, we estimated total Ca2+ clearance rate at the peak [(-d[Ca2+]T/dt)t = 0] according to Equation 5. Values for (-d[Ca2+]T/dt)t = 0 before (black circles) and after (gray circles) CCCP treatment are plotted as a function of peak Δ[Ca2+]i in Figure 4C. In addition, the data of -d[Ca2+]T/dt obtained with fura-4F in lower Δ[Ca2+]i range (<1 μm) were superimposed (open circles). The composite graph of -d[Ca2+]T/dt provides the dependence of -d[Ca2+]T/dt on Δ[Ca2+]i over a wider range than the similar graph in Figure 3B. Although no effect of CCCP was observed at Δ[Ca2+]i <2 μm, CCCP downward-shifted the -d[Ca2+]T/dt curve in the higher range of Δ[Ca2+]i (Fig. 4C, gray circles). In contrast, the -d[Ca2+]T/dt values in the presence of thapsigargin (2 μm, crosses) were not different from the control condition, indicating that the SERCA did not contribute to Ca2+ clearance over the entire Δ[Ca2+]i range. It was also noted that the -d[Ca2+]T/dt curve in the presence of CCCP showed saturation in the high Δ[Ca2+]i range. Considering that Na/CaX is the major CCM in such high Δ[Ca2+]i range, the -d[Ca2+]T/dt curve might represent the saturation of Na/CaX activity (Fig. 4C, gray curve). In the absence of CCCP, the -d[Ca2+]T/dt curve became more linear, indicating that the saturation of Na/CaX activity is partially compensated by the activation of mitochondrial Ca2+ uptake at Δ[Ca2+]i levels >2.5 μm.
Mitochondrial Ca2+ clearance and intracellular buffer concentration
We investigated the mechanisms whereby the inhibition or saturation of Na/CaX can render mitochondrial Ca2+ uptake active. We hypothesized that the inhibition or the saturation of Na/CaX allows mitochondria to be exposed to high cytosolic [Ca2+] for a longer duration. To test whether a longer exposure to high [Ca2+]i is necessary for mitochondrial Ca2+ uptake in the calyx of Held, we slowed the Ca2+ decay rate by increasing the [fura-2FF] in the pipette solution instead of inhibiting Na/CaX, and examined the effect of CCCP on Ca2+ transients. Moreover, to avoid the saturation of Na/CaX, the peak Δ[Ca2+]i was adjusted to 2∼2.5 μm by varying the membrane potential during the 50 ms depolarizations. At 50 μm fura-2FF (Fig. 5Aa), CCCP had little effect on Ca2+ transients. However, at higher [fura-2FF]i levels (Fig. 5Ab, 200 μm; Ac, 400 μm), the effects of CCCP on Ca2+ transients became evident. We obtained similar results when mitochondrial Ca2+ uptake was inhibited by 10 μm tetraphenylphosphonium (TPP+), which has no effect on ATP production. Because the inhibition of the Ca2+ decay rate caused by CCCP was not statistically different from that caused by TPP+, two data sets were pooled for statistical analysis. The mitochondrial contribution to the Ca2+ clearance at various concentrations of fura-2FF is summarized as a bar graph in Figure 5B, which shows that the effects of CCCP or TPP+ increased as the fura-2FF concentration increased. The relative contribution of mitochondria to total Ca2+ clearance measured with 50, 200, and 400 μm fura-2FF were 8.4 ± 1.9, 24.1 ± 4.9, and 31.1 ± 5.3%, respectively. These results suggest that the slowed Ca2+ decay is a direct cause of mitochondrial Ca2+ uptake when Na/CaX is inhibited.
Discussion
Ca2+ signaling in axon terminals plays a crucial role in synaptic function. The Ca2+ clearance rate in axon terminals not only determines the duration and amplitude of a Ca2+ transient but also affects short-term plasticity by a mechanism involving residual calcium. In the present study, we quantified the contribution made by each putative CCM at the calyx of Held and demonstrate the role of NCKX in presynaptic Ca2+ clearance. In addition, we found that (1) Na/CaX, comprised of NCKX and NCX, is a major CCM in the calyx of Held; (2) PMCA is responsible for 23% of the Ca2+ clearance when Δ[Ca2+]i is <1 μm; and (3) mitochondria contribute to Ca2+ clearance only when Na/CaX is inhibited or when the Ca2+ load is high enough to saturate Na/CaX capacity.
Role of Na/CaX in the calyx of Held
Although calyces of Held are extremely large presynaptic terminals, their thin cuplike structure constitutes a high surface-to-volume ratio (Sätzler et al., 2002). In neuronal compartments with a high surface-to-volume ratio, CCMs in the plasma membrane may clear cytosolic Ca2+ more efficiently than other CCMs that sequestrate Ca2+ into intracellular organelles. Various axon terminals effectively clear Ca2+ using plasma membrane CCMs such as Na/CaX and PMCA (Gill et al., 1981; Sanchez-Armass and Blaustein, 1987; Reuter et al., 1995; Morgans et al., 1998; Zenisek et al., 2000; Chuhma and Ohmori, 2002; Lee et al., 2002). Our results indicate that Na/CaX is a primary CCM at the calyx of Held. Moreover, mitochondria play a supplementary role when Na/CaX is inhibited or its activity is saturated (Fig. 4). This interaction between Na/CaX and mitochondria is similar to that between PMCA and mitochondria in the presynaptic terminals of retinal bipolar neurons, in which PMCAs are the primary CCM and mitochondrial Ca2+ uptake is observed only when PMCAs are inhibited (Zenisek et al., 2000).
The calyx of Held synapse plays a pivotal role in sound localization, especially for high-frequency sounds (Oertel, 1999; von Gersdorff and Borst, 2002). Adult calyces of Held have been known to receive and follow APs of high frequency up to 600 Hz (Taschenberger and von Gersdorff, 2000). It is thought that, for accurate sound localization, calyces of Held relay high-frequency inputs to postsynaptic MNTB neurons with high fidelity. In order for a calyx of Held to synchronize repetitive electrical events with transmitter release, presynaptic Ca2+ transients triggered by APs should be cleared immediately during an interspike interval so that the subsequent AP triggers a discrete Ca2+ transient. Na/CaX, which has a higher capacity for Ca2+ extrusion than other CCMs, seems to meet these requirements. Our results in Figure 1E show what happens to presynaptic Δ[Ca2+]i in the absence of Na/CaX when high-frequency APs invade the calyx of Held. When we stimulated the calyx of Held with trains of 2 ms depolarizing pulses at 10, 20, or 50 Hz, Δ[Ca2+]i reached a steady state in the range of 0.5∼2 μm within 300 ms in the control condition, but the inhibition of Na/CaX caused a continued buildup of [Ca2+]i, which lasted for the stimulation duration (1 s).
The range of spatially averaged Ca2+ signals observed in this study (0.5-6 μm) contrasts with the local [Ca2+]i reached transiently at the sites of vesicle fusion during a presynaptic AP, which has been estimated at 10-25 μm for the calyx of Held (Bollmann et al., 2000; Schneggenburger and Neher, 2000). Thus, Na/CaX dampens the buildup of residual Ca2+, but the termination of the local Ca2+ signal for transmitter release is probably achieved by cytosolic Ca2+ diffusion away from presynaptic active zones, after presynaptic Ca2+ channels close.
One would expect that presynaptic Na+ influx during prolonged high-frequency trains of APs might dissipate transmembrane Na+ gradient, which in turn would compromise the Na/CaX activity. Recently, it was reported that voltage-gated Na+ channels are almost absent from the calyceal terminal and highly concentrated in the unmyelinated axonal heminode (Leao et al., 2005). The exclusion of Na+ channels from the calyx might help to avoid a local build-up of [Na+]i in the presynaptic terminal (but see Engel and Jonas, 2005).
Role of mitochondria in the calyx of Held
The present study shows that, at the calyx of Held, Na/CaX is the main CCM after brief bursts of presynaptic APs, which result in Δ[Ca2+]i ≤ 2.5 μm. However, mitochondria are expected to contribute to Ca2+ clearance when the function of Na/CaX is compromised or saturated.
The role of mitochondria in Ca2+ clearance observed here is somewhat different from that presented in a previous study, which used the same preparation (Billups and Forsythe, 2002). These authors reported that the inhibition of mitochondrial Ca2+ sequestration significantly slowed Ca2+ decay when [Na+]o was not lowered. However, the experimental conditions used in the present study differ from those of Billups and Forsythe (2002) in several respects. First, these authors used a Cs+ pipette solution containing a relatively high [Na+]i (34 mm), whereas a K+ pipette solution with a low [Na+]i (4 mm) was used in the present study. Assuming a single Eyring barrier and a 3 Na+:1 Ca2+ stoichiometry for NCX, Ca2+ efflux via NCX (JNCX) is heavily dependent on the transmembrane Na+ gradient as the following thermodynamic relationship suggests:
where Em, F, R, and T are membrane potential, the Faraday constant, the gas constant, and temperature, respectively. When [Ca2+]i = 2 μm and Em = -70 mV, JNCX is lower at [Na+]i = 34 mm than when [Na+]i = 4 mm by a factor of 6.65. In addition, we measured the outward current (reverse mode) of NCKX2 (IN-CKX) heterologously expressed in HEK293 cells and found that Cs+ only partially substitutes for K+ (M.-H. Kim, W.-K. Ho, and S.-H. Lee, unpublished observation), and thus we used K+ pipettes instead of Cs+ to observe Ca2+ transients under more physiological conditions. As our data suggest, the mitochondrial contribution could have been exaggerated under the high [Na+]i and K+-free condition, which is unfavorable for the function of Na/CaX. Second, in the study by Billups and Forsythe (2002), Δ[Ca2+]i induced by repetitive depolarization pulses at frequencies of 100 Hz for 40 ms was ∼8 μm. Under this condition, mitochondria could contribute to Ca2+ clearance, because Na/CaX activity might be saturated. Third, Billups and Forsythe (2002) used a higher concentration of intracellular Ca2+ buffer (i.e., 200 μm fura-2FF plus 200 μm EGTA in the pipette solution). Because the Ca2+ decay rate is inversely proportional to the calcium binding ratio, a higher concentration of exogenous Ca2+ buffer would lead to an increase in the mitochondrial contribution to Ca2+ clearance, as shown in Figure 5. Borst et al. (1995) reported that the EPSC amplitudes evoked by the stimulation of axon fibers were unchanged when presynaptic terminals were perfused with 50 μm BAPTA (or 200 μm EGTA). However, Helmchen et al. (1997) suggested that most endogenous buffer in the calyx of Held might be immobile. In the present study, when the Δ[Ca2+]i of a Ca2+ transient triggered by a single AP was 500 nm, our exogenous Ca2+ buffer condition (50 μm fura-4F) is in between those of these two studies in terms of the calcium binding ratio, κB (41.3 for 50 μm BAPTA; 32.8 for 50 μm fura-4F). In contrast, 200 μm fura-2FF plus 200 μm EGTA, the buffer condition used by Billups and Forsythe (2002), is higher (κB ≈ 102) than known physiological buffer conditions (κS = 40) (Helmchen et al., 1997). Nevertheless, a recent immunohistochemical study revealed that the mobile Ca2+ buffers calretinin and parvalbumin are present in the presynaptic terminals of the MNTB. In addition, the expression of calretinin is not homogenous, and the proportion of calretinin-positive calyces increases during postnatal development (Felmy and Schneggenburger, 2004). In view of the fact that higher Ca2+ buffers favor mitochondrial Ca2+ clearance, it is possible that mitochondria play a more important role for Ca2+ clearance with ongoing developmental maturation of calyces of Held.
Physiological implications
Recent studies suggest that CCMs might be involved in posttetanic potentiation (PTP) by a mechanism of residual calcium. At the crayfish neuromuscular junction, posttetanic slow Ca2+ release from mitochondria subsequent to Ca2+ uptake during tetanic stimulation was proposed as an underlying mechanism for residual calcium (Tang and Zucker, 1997). Because posttetanic Ca2+ release is preceded by mitochondrial Ca2+ uptake during tetanic stimulation, the conditions required for mitochondrial Ca2+ uptake are of particular interest in terms of understanding presynaptic residual calcium and the associated short-term synaptic plasticity induced by high-frequency activity. In addition, an involvement of the Na/CaX in the residual calcium has also been suggested, because PTP and presynaptic Ca2+ accumulation at the crayfish neuromuscular junction are promoted by the reverse-mode Na/CaX (Zhong et al., 2001). Functional and immunocytochemical evidence support the existence of Na/CaX in presynaptic terminals in mammalian central synapses (Reuter et al., 1995; Regehr, 1997). The role of Na/CaX in the mammalian synapses is further supported by a study that showed that paired-pulse facilitation and PTP are enhanced in the mice lacking NCX2 (Jeon et al., 2003). We showed that the relative contributions of Na/CaX and of mitochondria to Ca2+ clearance are not static, but interact dynamically depending on Δ[Ca2+]i level. These results suggest that high-frequency activity, which allows mitochondria to take up cytosolic Ca2+, could cause short-term plastic changes at the calyx of Held synapse. Indeed, PTP of transmitter release has been observed recently at the calyx of Held (Habets and Borst, 2005; Korogod et al., 2005).
Footnotes
This work was supported by Grant M103KV010012-03K2201-01220 from the Brain Research Center of the 21st Century Frontier Research Program and Grant for National Research Laboratory 2004-02433 from the Ministry of Science and Technology, Republic of Korea. M.-H.K. was a postgraduate student supported by Program BK21 from the Ministry of Education.
Correspondence should be addressed to Dr. Suk-Ho Lee, Department of Physiology, Seoul National University College of Medicine, Chongno-Ku, Yongon-Dong 28, Seoul 110-799, Korea. E-mail: leesukho{at}snu.ac.kr.
Copyright © 2005 Society for Neuroscience 0270-6474/05/256057-09$15.00/0