- Email: [email protected]
The Role of Mitochondria in Presynaptic Calcium Handling at a Ribbon Synapse
Neuron, Vol. 25, 229–237, January, 2000, Copyright 2000 by Cell Press
The Role of Mitochondria in Presynaptic Calcium Handling at a Ribbon Synapse David Zenisek† and Gary Matthews* Department of Neurobiology and Behavior State University of New York Stony Brook, New York 11794
Summary Mitochondria are thought to be important in clearing calcium from synaptic terminals. It is unclear, however, whether the principal role of mitochondria in presynaptic calcium handling is to take up Ca2ⴙ directly or to fuel Ca2ⴙ removal by other mechanisms. We used patch clamp techniques and fluorescence imaging to examine calcium clearance mechanisms, including mitochondrial uptake, in single synaptic terminals of retinal bipolar neurons. We found that extrusion through the ATP-dependent Ca2ⴙ pump of the plasma membrane is the dominant form of Ca2ⴙ removal in the synaptic terminal. Calcium uptake into mitochondria was sometimes evident with large Ca2ⴙ loads but was consistently observed only when plasma membrane extrusion was inhibited. We conclude that mitochondria act primarily as an energy source in clearance of Ca2ⴙ from bipolar cell synaptic terminals.
Introduction Internal calcium plays a number of roles in synaptic terminals in addition to neurotransmitter release, such as recruitment of vesicle pools and activation of ionic conductances. Information about the mechanisms regulating Ca2⫹ concentration is therefore important for understanding synaptic function from a number of perspectives. Several mechanisms have been proposed to be important in presynaptic Ca2⫹ regulation, including Na⫹–Ca2⫹ exchange, uptake into mitochondria, and plasma membrane Ca2⫹-ATPase (PMCA). However, opportunities for direct study of calcium handling in single terminals are rare, and most of what is known about Ca2⫹ regulation in neurons stems from studies of neuronal somata (e.g., Thayer and Miller, 1990; Friel and Tsien, 1994; Werth and Thayer, 1994) and from studies of calcium transport in squid giant axon (e.g., Baker, 1972; Blaustein and Russell, 1975; DiPolo, 1978; DiPolo and Beauge´, 1983). We have exploited the large size of synaptic terminals of bipolar neurons from goldfish retina to study directly the regulation of [Ca2⫹]i in single terminals. These giant synaptic terminals (diameter ≈ 10 m) are amenable to physiological manipulation and allow direct measurement of Ca2⫹ influx and [Ca2⫹]i, using combined patch
* To whom correspondence should be addressed (e-mail: gary.g. [email protected]). † Present address: Vollum Institute, Oregon Health Sciences University, Portland, Oregon 97201.
clamp and Ca2⫹ imaging techniques. The results demonstrate that PMCA is of primary importance in regulating basal Ca2⫹ and in clearing Ca2⫹ after depolarizationevoked influx. Mitochondrial Ca2⫹ uptake plays some role, especially when [Ca2⫹]i exceeds 800 nM, but the principal role of mitochondria in Ca2⫹ extrusion in bipolar cell synaptic terminals seems to be production of ATP to fuel the Ca2⫹ pump. Results Calcium Transients Evoked by Depolarization in Retinal Bipolar Neurons A typical large-terminal bipolar neuron from goldfish retina is shown in Figure 1A. Synaptic terminals of these neurons have a high density of slowly inactivating, dihydropyridine-sensitive Ca2⫹ channels (Heidelberger and Matthews, 1992; Tachibana et al., 1993) that drive synaptic release of glutamate (Tachibana and Okada, 1991; Tachibana et al., 1993; von Gersdorff et al., 1998). To examine calcium clearance in the synaptic terminal, single cells were patch clamped to control Ca2⫹ current, and Ca2⫹ concentration was monitored in the terminal and soma using fluorescent Ca2⫹ indicator (usually bisfura-2) introduced through the patch pipette. In keeping with the predominant localization of the Ca2⫹ channels in the synaptic terminal (Heidelberger and Matthews, 1992), depolarization evoked a large, rapid increase in [Ca2⫹]i in the terminal but a much smaller [Ca2⫹]i increase in the soma (Figure 1B). The small Ca2⫹ increase in the soma was also due to Ca2⫹ influx through Ca2⫹ channels (data not shown), which are present in somata, albeit at a much lower density than in terminals (Heidelberger and Matthews, 1992; Mennerick et al., 1997). After depolarization, [Ca2⫹]i returned rapidly to baseline in the synaptic terminal, and recovery was well described by a single exponential time constant of ⵑ1 s. Figure 1C shows calcium responses in the synaptic terminal for depolarizations of increasing duration, and Figure 1D summarizes the speed of calcium decay for different amounts of Ca2⫹ influx. The waveform of calcium decay in the synaptic terminal was similar for both large and small Ca2⫹ loads, but the rate of clearance slowed with increasing Ca2⫹ load. The half-time of Ca2⫹ decay increased from 0.6 ⫾ 0.04 s for calcium influx of ⬍20 pC (mean ⫾ SEM, n ⫽ 48) to 1.3 ⫾ 0.06 s (n ⫽ 9) for calcium influx of ⬎150 pC (Figure 1D). Thus, Ca2⫹ removal in the synaptic terminal has a high capacity and can sustain rapid clearance after large Ca2⫹ influx. Although less Ca2⫹ entered the soma than the terminal during depolarization, recovery was consistently slower in the soma (Figure 1D). In experiments described in the next section, calcium clearance was studied in intact bipolar neurons loaded with fura-2AM and depolarized by high external potassium. In whole-cell recordings, the rate of calcium clearance was faster than in intact bipolar neurons, where the average half-time of Ca2⫹ decay for responses under control conditions was 3.8 ⫾ 0.4 s (mean ⫾ SEM, n ⫽
Neuron 230
Figure 1. Calcium Transients Elicited by Activation of Calcium Current in the Synaptic Terminal and Soma of a Retinal Bipolar Neuron (A) Typical appearance of an isolated goldfish bipolar neuron. Cells were filled with bisfura-2 (60–100 M) through a whole-cell patch pipette placed on either the soma or the terminal. (B) Calcium transients recorded in the synaptic terminal (solid line) and soma (dashed line) of a bipolar neuron filled with bis-fura-2. The timing of a 900 ms depolarization from ⫺60 mV to 0 mV is given by the arrow. The small Ca2⫹ transient in the soma was due to activation of somatic Ca2⫹ channels, which are present at much lower density than in the synaptic terminal (Heidelberger and Matthews, 1992). (C) Calcium transients elicited in a synaptic terminal by depolarizations from ⫺60 mV to 0 mV for 5 s (solid line), 270 ms (dashed line), and 70 ms (gray line). Traces were aligned so that pulse offset coincides. (D) Time course of Ca2⫹ recovery as a function of total Ca2⫹ influx in bipolar cell terminals (closed squares) and somata (open circles). Half decay time (t1/2) of Ca2⫹ transients provided an index of recovery time course. Calcium charge entering during depolarization (abscissa) was calculated by integrating leak-subtracted Ca2⫹ current; 100 pC of charge corresponds to ⵑ0.5 fmol of Ca2⫹. The amount of influx was varied by changing the duration of depolarization.
26). Because high-K⫹ depolarization typically elevated [Ca2⫹]i to high levels for 10 s or more, the slower Ca2⫹ decay in intact cells may reflect in part the slowing of clearance after large calcium loads (see Figure 1D). Also, differences in experimental conditions associated with whole-cell voltage clamp, such as internal dialysis, may contribute to differences in the rate of calcium clearance between intact cells and patch-clamped cells. To estimate the endogenous rate of calcium clearance in the synaptic terminal in the absence of the buffering action of added Ca2⫹ indicator, time constants of recovery were also measured with varying amounts of exogenous Ca2⫹ buffer capacity. By extrapolating the relation between recovery time constant and buffer capacity to the y intercept, we estimated a time constant of 0.7 s in the absence of added buffer. PMCA Plays a Central Role in Calcium Clearance Recent results point to an important role for PMCA in Ca2⫹ extrusion from synaptic terminals of retinal photoreceptors (Krizaj and Copenhagen, 1998; Morgans et al., 1998), and immunostaining reveals PMCA labeling of bipolar cell terminals in goldfish retina (Morgans et al., 1998). Therefore, we examined the possible role of PMCA in calcium clearance in bipolar cell synaptic terminals. To ensure that intracellular conditions were disturbed minimally, we tested the effect of PMCA inhibitors applied extracellularly to bipolar cells loaded with membrane-permeant fura-2AM. Calcium was elevated by external superfusion with high-K⫹ solution. Figure 2A shows that recovery of [Ca2⫹]i to baseline was eliminated by external application of 1 mM La3⫹, an inhibitor of PMCA (Carafoli, 1991; Herrington et al., 1996). When La3⫹ was applied during the falling phase of the calcium response, the decay of [Ca2⫹]i abruptly ceased (Figure
2B), and [Ca2⫹]i remained at a sustained plateau during application of the PMCA inhibitor. The inset in Figure 2B summarizes the action of La3⫹ on the rate of calcium clearance, which slowed more than 100-fold in the presence of La3⫹ (1.2 ⫾ 0.8 nM/s versus 133 ⫾ 30 nM/s without La3⫹). In paired comparisons, the decay rate in the presence of La3⫹ was 0.8% ⫾ 0.4% of the control rate of clearance in the same synaptic terminal (mean ⫾ SEM, n ⫽ 15). After La3⫹ was removed, [Ca2⫹]i recovered fully to baseline (Figures 2A and 2B), and La3⫹ did not affect basal [Ca2⫹]i in the absence of depolarization-evoked Ca2⫹ influx (data not shown). Together with the fact that La3⫹ is a potent Ca2⫹ channel blocker, these observations suggest that La3⫹ influx was minimal. Therefore, external La3⫹ likely acts by inhibiting mechanisms of calcium clearance, rather than by binding to fura-2 within the terminal and producing a false Ca2⫹ signal. Although La3⫹ affects both PMCA and Na⫹–Ca2⫹ exchange (Kimura et al., 1986), evidence presented later will demonstrate that Na⫹–Ca2⫹ exchange plays little role in calcium clearance in bipolar cell synaptic terminals (see Figure 6). Thus, we attribute the potent slowing of Ca2⫹ recovery by La3⫹ to inhibition of PMCA. Calcium extrusion via PMCA is also blocked by elevated external pH (DiPolo and Beauge´, 1982; Kobayashi and Tachibana, 1995). In bipolar cell terminals, calcium clearance slowed dramatically when extracellular pH was elevated from 7.4 to 8.4, as illustrated in Figures 2C and 2D. As with La3⫹, elevation of external pH during the falling phase of the Ca2⫹ response rapidly inhibited calcium clearance after high-K⫹ depolarization, producing a sustained plateau of elevated Ca2⫹ (Figure 2D). In paired comparisons, the rate of decay of [Ca2⫹]i in pH 8.4 external solution slowed to 3.4% ⫾ 1.7% of the control rate (mean ⫾ SEM, n ⫽ 11). The average decay rate was 9.7 ⫾ 5 nM/s at pH 8.4 versus 293 ⫾ 80 nM/s at pH 7.4 (see inset of Figure 2D).
Mitochondria in Presynaptic Calcium Handling 231
Figure 2E shows the residual rate of decay of [Ca2⫹]i plotted against the plateau Ca2⫹ concentration during application of PMCA inhibitor (La3⫹ or elevated external pH). The rate of decay did not depend on the level of plateau Ca2⫹ up to ⵑ800 nM. Above 800 nM, the residual rate of decay increased somewhat, which might indicate operation of an alternative Ca2⫹ removal mechanism at higher Ca2⫹ levels. Results presented later suggest that this residual mechanism might be uptake into mitochondria. However, even the most rapid rate of recovery observed during PMCA inhibition was ⵑ10-fold slower than the control rate of clearance over the same range of Ca2⫹ concentration.
Figure 2. Calcium Clearance in the Synaptic Terminal Is Slowed by Inhibition of PMCA Bipolar neurons were loaded with fura-2AM and depolarized by local superfusion with 80 mM external K⫹. All signals were recorded from the synaptic terminal. (A) Two consecutive calcium responses were evoked by high-K⫹ depolarization at the times indicated by bars below the trace. After the second application of high K⫹, an external solution containing 1 mM La3⫹ was applied to the terminal at the indicated time. (B) Superimposed falling phases of calcium responses from the same terminal with (thick line) and without (thin line) application of 1 mM La3⫹ at the indicated time. A different cell from that in (A) is shown. The inset summarizes the average rate of Ca2⫹ decay for control responses and in the presence of 1 mM La3⫹ from 15 synaptic terminals. Note the logarithmic scale. (C) Elevation of extracellular pH from 7.4 to 8.4 slowed calcium recovery in the synaptic terminal after high-K⫹ depolarization. Timing high-K⫹ external solution is indicated by the bars below the trace. (D) Superimposed falling phases of a control calcium response (thin line) and during application of elevated pH (thick line) in the same synaptic terminal. A different cell from that in (C) is shown. The
Manipulations of Internal ATP Also Affect Calcium Clearance We also examined the effect on calcium clearance of intracellular manipulations that affect Ca2⫹ extrusion via PMCA, such as dialysis via whole-cell patch pipettes with internal solution lacking ATP. Without exogenous ATP, Ca2⫹ recovery following activation of Ca2⫹ current slowed progressively during the experiment. Figure 3A summarizes results from four bipolar neurons dialyzed with ATP-free solution (circles), and the inset in Figure 3A shows sample Ca2⫹ transients recorded at early and late times during dialysis. Time-dependent slowing of Ca2⫹ transients was not observed in control experiments in which the pipette solution contained ATP (Figure 3A, squares). These results demonstrate that reduced availability of ATP retards recovery of [Ca2⫹]i to baseline following Ca2⫹ influx, which is consistent with a contribution of PMCA to calcium clearance. As a further test of the role of PMCA, bipolar neurons were dialyzed in whole-cell recordings with internal solution containing ATP (2 mM) ⫹ 1 mM sodium orthovanadate, an inhibitor of PMCA (Barrabin et al., 1980; Bond and Hudgins, 1980; Rossi et al., 1981). As it diffused into the terminal, orthovanadate markedly slowed Ca2⫹ transients (Figure 3A, triangles; n ⫽ 3). This action is consistent with the effects of external PMCA inhibitors shown in Figure 2 and provides further confirmation that PMCA plays an important role in Ca2⫹ removal from the synaptic terminal. Replacing exogenous ATP in the whole-cell pipette solution with the nonhydrolyzable ATP analog ATP-␥S offers an alternative method of inhibiting PMCA. Figure 3B shows superimposed Ca2⫹ transients recorded at different times after onset of dialysis with internal solution containing 10 mM ATP-␥S, which progressively interfered with recovery of [Ca2⫹]i to the usual resting level after activation of Ca2⫹ current. In ten cells dialyzed with ATP-␥S, baseline [Ca2⫹]i increased to an average of 850 nM from the normal baseline of ⵑ150 nM. This action of ATP-␥S is another indication that ATP-dependent extrusion is important in returning [Ca2⫹]i in the synaptic terminal to low levels after influx.
inset summarizes the results from 11 similar experiments (note the logarithmic scale). (E) The observed Ca2⫹ decay rate during application of La3⫹ (closed squares) or pH 8.4 (open squares) is plotted as a function of the average plateau [Ca2⫹]i during application.
Neuron 232
Inhibition of endogenous ATP production by mitochondria also retarded calcium clearance in the synaptic terminal. Oligomycin, an inhibitor of mitochondrial ATP synthase, had effects similar to ATP-␥S, abolishing recovery of Ca2⫹ to normal resting level after activation of Ca2⫹ current and elevating basal [Ca2⫹]i (data not shown).
Figure 3. Calcium Regulation in the Synaptic Terminal Is Affected by Manipulations of Intracellular ATP (A) Slowing of calcium clearance in the synaptic terminal during whole-cell dialysis with a pipette solution containing no ATP (circles) or 1 mM orthovanadate ⫹ 2 mM ATP (triangles). Orthovanadate is an inhibitor of PMCA. The closed squares show results from control cells in which the pipette solution included 2 mM ATP. The patch pipette was placed on the soma to slow dialysis of the terminal. The time course of dialysis was monitored by the rise of fluorescence as bis-fura-2 entered the terminal, and the abscissa (dialysis index) is the fluorescence as a percent of its asymptotic level. The ordinate shows the half-time (t1/2) of decay of Ca2⫹ following depolarizing voltage clamp stimuli. The inset shows two superimposed Ca2⫹ responses obtained at the times indicated in the legend during dialysis with ATP-free solution. (B) Superimposed responses to depolarization during dialysis with an internal solution containing 10 mM ATP-␥S. The arrow indicates timing of 270 ms depolarizations from ⫺60 mV to 0 mV. The legend gives the time after break-in for each of the responses. (C) Effect of the protonophore CCCP (10 M) on Ca2⫹ removal in a cell dialyzed with 10 mM ATP-␥S. The trace shows responses observed after [Ca2⫹]i reached a stable elevated plateau (note elevated baseline), and the calcium recovery observed after depolarization represents the residual component of Ca2⫹ removal remaining after blockade of ATP-dependent removal. Thin and thick arrows show timing of 70 ms and 270 ms depolarizations from ⫺60 mV to 0 mV, respectively. The gray bar indicates timing of CCCP application. The inset shows superimposed responses to 70 ms depolarizations before (thin trace) and during (thick trace) CCCP application.
The Role of Mitochondrial Uptake in Calcium Clearance In Figure 3B, some degree of residual Ca2⫹ removal remained after ATP-␥S had taken full effect, and the terminal was still able to regulate [Ca2⫹]i, albeit at an elevated level. Giant synaptic terminals of goldfish bipolar neurons contain large numbers of mitochondria (von Gersdorff et al., 1996), and mitochondrial Ca2⫹ uptake is thought to be particularly significant at high [Ca2⫹]i, such as the elevated baseline seen with ATP-␥S (Nicholls and Scott, 1980; Nicholls and A˚kerman, 1982). Therefore, we examined whether mitochondria are responsible for the residual Ca2⫹ regulation observed when PMCA was inhibited. Figure 3C shows that residual Ca2⫹ recovery in the presence of ATP-␥S was strongly inhibited by CCCP, which blocks mitochondrial Ca2⫹ uptake by disrupting the mitochondrial proton gradient. Similar results were observed in seven experiments. Thus, uptake into mitochondria apparently accounts for the residual calcium clearance observed at elevated [Ca2⫹]i in the presence of ATP-␥S. Although mitochondrial uptake contributes to Ca2⫹ regulation when plasma membrane extrusion is inhibited, it remains to be established whether mitochondrial Ca2⫹ uptake also plays a role when PMCA is active. In neuronal somata and adrenal chromaffin cells, CCCP releases Ca2⫹ accumulated by mitochondria during a prior Ca2⫹ load, producing a large rebound increase in cytoplasmic [Ca2⫹]i (Thayer and Miller, 1990; Budd and Nicholls, 1996a; Herrington et al., 1996). This action of CCCP has been taken as evidence favoring mitochondrial contribution to calcium clearance in these cells. To determine whether mitochondria also take up Ca2⫹ during calcium clearance in bipolar cell synaptic terminals, we applied CCCP to voltage-clamped terminals after Ca2⫹ influx triggered by activation of Ca2⫹ current. Figure 4A shows an example of an experiment in which CCCP successfully released Ca2⫹ from an internal source—presumably mitochondria—after activation of Ca2⫹ current. In this instance, CCCP produced a moderate elevation of [Ca2⫹]i after brief depolarizations (first two arrows in Figure 4A). After a 5 s depolarization that drove [Ca2⫹]i to ⬎2 M, CCCP elicited a large rebound increase in [Ca2⫹]i. Similar release of Ca2⫹ by CCCP after prolonged depolarization was observed in only 6 out of 19 cells, however. Thus, CCCP revealed Ca2⫹ accumulation by mitochondria in the synaptic terminal inconsistently, even after large and prolonged Ca2⫹ influx. In whole-cell recordings, internal dialysis may remove cellular factors needed for mitochondrial metabolism, which may explain the inconsistent release of Ca2⫹ by CCCP in whole-cell recordings. When a cocktail designed to promote mitochondrial metabolism was added to the pipette solution (see Experimental Procedures), CCCP still failed to release Ca2⫹ after large Ca2⫹ loads
Mitochondria in Presynaptic Calcium Handling 233
Figure 4. Mitochondrial Ca2⫹ Uptake after Depolarization-Evoked Ca2⫹ Influx (A) Calcium responses in the synaptic terminal of a bipolar neuron under whole-cell voltage clamp. The arrows of increasing size show timing of depolarizations from ⫺60 mV to 0 mV for 70, 270, or 470 ms, and the large arrowhead indicates a 5 s depolarization. CCCP (10 M) was applied at the gray bars in a Ca2⫹-free external solution, to ensure that any changes in [Ca2⫹]i represent release of internal Ca2⫹. (B) Calcium responses in the synaptic terminal of four different bipolar neurons loaded with fura-2AM. The black bars below the traces show timing of high-K⫹ depolarization, and the gray bars indicate application of CCCP. CCCP released Ca2⫹ from an internal source in the top two cells, but not in the lower two. (C) Summary of Ca2⫹ release induced by CCCP in bipolar neurons loaded with fura-2AM. The rise in [Ca2⫹]i induced by CCCP is plotted as a function of the preceding Ca2⫹ load activated by high-K⫹ depolarization. (D) Superimposed falling phases of Ca2⫹ responses in a synaptic terminal under control conditions (thin trace) and during application of CCCP (thick trace). CCCP was applied just before the high-K⫹ external solution was removed. The inset summarizes Ca2⫹ decay rates from 14 synaptic terminals with and without CCCP.
in four cells. However, to ensure that mitochondria were perturbed as little as possible, we also tested CCCP’s action on intact bipolar neurons loaded with fura-2AM. Figure 4B shows sample Ca2⫹ responses, and Figure 4C summarizes the results from synaptic terminals in which CCCP was applied after Ca2⫹ influx stimulated by high-K⫹ depolarization. As in dialyzed cells, CCCP had a variable effect on [Ca2⫹]i, although a larger fraction of intact cells responded to CCCP (63% of intact cells versus 32% of dialyzed cells). The amount of Ca2⫹ released by CCCP did not correlate well with the size of the preceding Ca2⫹ load (Figure 4C), and so the variability in response to CCCP cannot be accounted for by variation in Ca2⫹ load. We conclude that when PMCA is functional, bipolar cell synaptic terminals are frequently capable of clearing large Ca2⫹ loads without detectable accumulation of Ca2⫹ in mitochondria, at least as indicated by CCCP-induced Ca2⫹ release. In other instances, a variable fraction of the Ca2⫹ load was subsequently releasable by CCCP. The reason for this variability in apparent mitochondrial uptake is unclear. In bipolar neurons loaded with fura-2AM, we also examined whether CCCP affects the rate of Ca2⫹ decay after high-K⫹ depolarization in the range of [Ca2⫹]i ⬎500 nM, where previous results suggest mitochondrial uptake to be most prevalent. CCCP was applied though a second puffer pipette just before the high-K⫹ superfusion pipette was removed, so that CCCP was present throughout the falling phase of the Ca2⫹ response. The rate of Ca2⫹ decay was then measured from the peak
of the Ca2⫹ response (1000–1600 nM) to 500 nM with and without CCCP. In contrast with the potent effects of PMCA inhibitors on calcium clearance (see Figure 2), CCCP did not consistently affect Ca2⫹ decay rate in the range of ⬎500 nM, as shown in Figure 4D. In 14 experiments, the rate of Ca2⫹ removal in the presence of CCCP was 113% ⫾ 13% of the control rate.
Mitochondrial Depolarization Monitored with Rhodamine-123 Because CCCP commonly had little or no effect on [Ca2⫹]i in synaptic terminals, it was necessary to confirm that CCCP did indeed affect the mitochondria. For this purpose, bipolar neurons were loaded with the fluorescent dye rhodamine-123, which accumulates in normally polarized mitochondria and is released when the mitochondria depolarize. Quenching reduces the fluorescence of the accumulated dye in polarized mitochondria, and overall fluorescence increases as the dye is released and diffuses throughout the cell upon mitochondrial depolarization (Johnson et al., 1981; Duchen et al., 1990). By dissipating the mitochondrial proton gradient, CCCP should depolarize mitochondria and increase rhodamine-123 fluorescence. Figure 5B shows that CCCP evoked a large, rapid increase in rhodamine123 fluorescence in both the synaptic terminal and soma. Similar results were observed in 16 of 17 cells. The average increase in fluorescence was 62% ⫾ 19% in terminals and 37% ⫾ 6% in somata. Thus, CCCP
Neuron 234
Figure 5. Rhodamine-123 Fluorescence Reveals Changes in Mitochondrial Membrane Potential Produced by CCCP and Ca2⫹ Influx (A) Images illustrating redistribution of rhodamine-123 fluorescence in a bipolar cell synaptic terminal (top row) and a cone photoreceptor (bottom row) in response to mitochondrial inhibition with 10 M CCCP. On the left are bright-field images of the cells viewed through the intensifier of the CCD camera used to acquire the fluorescence images. Scale bars, 10 m. (B) The protonophore CCCP rapidly depolarizes mitochondria in the soma (thin line) and synaptic terminal (thick line) of a bipolar neuron. At the arrow, 10 M CCCP was applied, producing a large increase in rhodamine-123 fluorescence. Fluorescence traces are normalized relative to prestimulus baseline. (C) Effect of depolarizing voltage clamp stimuli on rhodamine-123 fluorescence. Strong Ca2⫹ influx, elicited by a 5 s depolarizing pulse from ⫺60 mV to 0 mV (first arrow), sometimes increased rhodamine123 fluorescence in the synaptic terminal (thick line) but not in the soma (thin line). Briefer depolarization of the bipolar neuron (870 ms; second arrow) produced a smaller increase in fluorescence.
reliably and rapidly depolarizes mitochondria in bipolar neurons. Mitochondria in bipolar cell terminals are not evenly distributed and instead form a tightly packed cluster (von Gersdorff et al., 1996). In the resting synaptic terminal, rhodamine-123 fluorescence was also nonuniformly distributed (Figure 5A), as expected for a cluster of mitochondria. After application of CCCP, fluorescence spread evenly throughout the terminal, which is consistent with release of the dye upon mitochondrial depolarization induced by CCCP. Figure 5A also shows that rhodamine-123 fluorescence was localized to the ellipsoid region of cone photoreceptors. The ellipsoid is a
prominent cluster of mitochondria in the photoreceptor inner segment, and application of CCCP caused fluorescence to spread evenly throughout the cone (Figure 5A). Thus, the rhodamine-123 signal behaves as expected for a mitochondrial marker. Because mitochondria depolarize when Ca2⫹ is taken up (e.g., Duchen, 1992; Loew et al., 1994; Duchen et al., 1998), rhodamine-123 fluorescence was also used to monitor changes in mitochondrial membrane potential associated with Ca2⫹ influx in bipolar cell synaptic terminals. Activation of Ca2⫹ current in bipolar neurons loaded with rhodamine-123 produced an appreciable change in fluorescence only with prolonged Ca2⫹ current, and then only in a minority of the cells examined. Figure 5C shows an example of a cell that did show a detectable increase in rhodamine-123 fluorescence in the synaptic terminal in response to 5 s and 870 ms depolarizing voltage clamp pulses. In 18 bipolar neurons in which similar responses were observed, 5 s depolarizing pulses increased fluorescence in the terminal by 23% ⫾ 3%. However, a detectable response to 5 s stimuli was observed in only 18 of 38 cells tested. Depolarizing voltage clamp pulses briefer than ⵑ500 ms were without effect on rhodamine-123 fluorescence, even when fluorescence was averaged over 15 stimuli to decrease noise. No consistent responses to depolarization were observed in the soma. The small and inconsistent effect of large amounts of Ca2⫹ influx on the membrane potential of mitochondria in the synaptic terminal suggests either that mitochondria are able to take up a large calcium load without appreciable loss of membrane potential or that calcium uptake by mitochondria is not appreciable. Given the previous results showing little effect of CCCP on calcium clearance and little release of Ca2⫹ by CCCP after calcium loads, the latter explanation seems more plausible. Naⴙ–Ca2ⴙ Exchange Plays Little Role in Presynaptic Calcium Clearance To test whether Na⫹–Ca2⫹ exchange contributes to calcium clearance in bipolar cell terminals, we examined the effect of removing external Na⫹ on Ca2⫹ removal. Figure 6A shows that replacing external sodium with choline had no effect on the rate of decay of [Ca2⫹]i after depolarization. Similar results were obtained in six cells. As shown in Figure 6B, sodium removal also did not affect recovery of [Ca2⫹]i after high-K⫹ depolarization in intact bipolar cells loaded with fura-2AM. After sodium removal, the Ca2⫹ decay rate averaged 176 ⫾ 14 nM/s (mean ⫾ SEM, n ⫽ 10), compared with 213 ⫾ 24 nM/s in the same set of intact cells with normal external sodium (see inset of Figure 6B). This difference is not statistically significant (p ⬍ 0.22; t test for paired samples). Thus, we conclude that Na⫹–Ca2⫹ exchange contributes little to calcium clearance in bipolar cell synaptic terminals. Discussion Taken as a whole, our results suggest that the dominant role of mitochondria in Ca2⫹ handling in the bipolar cell synaptic terminal is to provide ATP to drive Ca2⫹ efflux via PMCA. Direct Ca2⫹ uptake into mitochondria was
Mitochondria in Presynaptic Calcium Handling 235
Figure 6. Removing External Sodium Has Little Effect on Calcium Clearance in the Synaptic Terminal (A) Calcium transients evoked in the synaptic terminal by 5 s depolarizations from ⫺60 mV to 0 mV in normal (thin trace) and sodiumfree external solution (thick trace). (B) Superimposed falling phases of Ca2⫹ responses evoked by highK⫹ depolarization in a bipolar neuron loaded with fura-2AM. The thin trace shows a control response, and the thick trace shows a response in sodium-free external solution, applied immediately after removal of the high-K⫹ solution. The inset shows average Ca2⫹ decay rates under control conditions and in sodium-free solution for ten synaptic terminals.
inconsistently observed, even after large and prolonged elevation of [Ca2⫹]i. Several lines of evidence support these conclusions. Attempts to release Ca2⫹ from presynaptic mitochondria after influx of Ca2⫹ through voltage-gated channels met with limited success, even after prolonged elevation of Ca2⫹ to high levels. Measurement of mitochondrial membrane potential, using rhodamine123, also failed to provide evidence for Ca2⫹ uptake into mitochondria during prolonged depolarization in over half of the terminals examined. Inhibition of mitochondrial Ca2⫹ uptake had little effect on the rate of calcium clearance in the synaptic terminal, except when PMCA was inactivated. By contrast, inhibition of PMCA has a large effect on calcium clearance. Two different extracellular inhibitors of PMCA—La3⫹ and elevated external pH—dramatically slowed calcium clearance to 1%–3% of normal, suggesting that PMCA accounts for ⬎95% of calcium clearance under normal conditions. Intracellular inhibitors of ATP-dependent extrusion (vanadate, ATP␥S) slowed Ca2⫹ removal and increased basal Ca2⫹. Thus, the role of mitochondria in Ca2⫹ handling in the synaptic terminal differs from chromaffin cells, where mitochondrial uptake has been shown to be important (Herrington et al., 1996). Because large uptake of Ca2⫹ into mitochondria may be toxic, reliance on ATP-dependent extrusion to the exclusion of mitochondrial Ca2⫹ uptake may be necessary at synaptic terminals, which experience larger Ca2⫹ loads than neuronal somata. Prevention of mitochondrial Ca2⫹ uptake is known to protect against excitotoxic effects of glutamate in somata (Budd and Nicholls, 1996b; Castilho et al., 1998; Stout et al., 1998), and
the amount of Ca2⫹ influx that triggers excitotoxicity is similar to that occurring in tonic synaptic terminals during depolarization. Therefore, avoiding Ca2⫹ uptake into mitochondria may be required to prevent an excitotoxic response in bipolar cell synaptic terminals, which experience large and prolonged elevation of internal Ca2⫹ during tonic signaling. Mitochondria in bipolar cell synaptic terminals form a discrete cluster near the junction with the axon (von Gersdorff et al., 1996; see also Figure 5A). This segregation of mitochondria may help to isolate them from Ca2⫹ influx and thus minimize Ca2⫹ uptake. Recent work using x-ray microanalysis showed that the amount of Ca2⫹ taken up by a mitochondrion declined with distance from the plasma membrane (Pivovarova et al., 1999). Thus, clustering of mitochondria at a distance from sites of influx may explain in part the reduced role of mitochondrial uptake in calcium clearance in bipolar cell synaptic terminals. In our experiments, Ca2⫹ uptake into mitochondria was detected in a subset of terminals, evidenced either by CCCP-induced Ca2⫹ release (see Figure 4) or by mitochondrial depolarization during Ca2⫹ influx (see Figure 5). A possible explanation is that a partial breakdown of mitochondrial segregation occurred in those instances. Available evidence suggests that at least some Ca2⫹ uptake by mitochondria occurs under physiological conditions in conventional synaptic terminals, particularly during prolonged stimuli. For example, Tang and Zucker (1997) suggest that efflux of Ca2⫹ accumulated by mitochondria during tetanic stimulation is responsible for posttetanic potentiation, which implies that mitochondrial Ca2⫹ uptake plays an important role in synaptic plasticity. Also, David et al. (1998) detected increased mitochondrial Ca2⫹ during trains of action potentials at physiological rates in motor nerve terminals. Therefore, the relative dominance of ATP-dependent Ca2⫹ extrusion over mitochondrial uptake observed in bipolar cell synaptic terminals may not hold for smaller, conventional synaptic terminals. Removal of external sodium did not affect the rate of Ca2⫹ removal in bipolar cell synaptic terminals, which suggests that Na⫹–Ca2⫹ exchange plays little role in presynaptic Ca2⫹ extrusion. In this regard, bipolar cell terminals are similar to hair cells (Tucker and Fettiplace, 1995) and photoreceptor terminals (Krizaj and Copenhagen, 1998; Morgans et al., 1998), in which ATP-dependent pumps are also more important than Na⫹–Ca2⫹ exchange in Ca2⫹ extrusion. Hair cells, photoreceptors, and bipolar cells are all tonic neurons that release neurotransmitter for sustained periods under physiological conditions. Therefore, a preference for ATP-dependent Ca2⫹ removal over Na⫹–Ca2⫹ exchange may be a general feature of tonic synaptic terminals. Internal sodium accumulation may slow Ca2⫹ extrusion by the Na⫹–Ca2⫹ exchanger during repetitive or prolonged stimulation (Mulkey and Zucker, 1992; Regehr, 1997). In tonic neurons, then, reliance on mechanisms other than Na⫹–Ca2⫹ exchange may be required to maintain Ca2⫹ extrusion in the face of prolonged Ca2⫹ influx. We estimated that in whole-cell recordings, Ca2⫹ decays with a time constant of about 0.7 s in the absence of exogenous Ca2⫹ buffer, compared with a time constant of ⵑ10 s under comparable conditions in adrenal
Neuron 236
chromaffin cells (Neher and Augustine, 1992). Although bipolar cell synaptic terminals clear calcium relatively rapidly, calcium recovery at other synaptic terminals is even faster than in bipolar cells. The time constant of calcium decay was 45 ms following single action potentials in the calyx of Held in the medial nucleus of the trapezoid body (Helmchen et al., 1997), 150 ms in cerebellar granule cell synapses (Regehr and Atluri, 1995), and 30 ms in hippocampal synapses (Sinha et al., 1997). This enhanced speed of recovery in other terminals may reflect the more favorable surface-to-volume ratio of the smaller synaptic terminals compared with the large bulbous terminals of bipolar neurons. Calcium clearance mechanisms in bipolar cell terminals also have high capacity for removing ongoing Ca2⫹ loads. In our experiments under whole-cell voltage clamp, the half-decay time of Ca2⫹ transients approximately doubled (from 0.6 to 1.3 s) as the Ca2⫹ load increased from 5 pC to over 200 pC. By contrast, in crayfish motor nerve terminals, the time constant for Ca2⫹ recovery following high-frequency stimulus trains was 20- to 40-fold slower than after low-frequency stimuli (Delaney et al., 1989). A similar large slowing of Ca2⫹ removal after strong stimuli has also been observed in other neuronal preparations (Nohmi et al., 1992; Tatsumi and Katayama, 1993; Steunkel, 1994; Regehr, 1997). ATP-dependent Ca2⫹ extrusion underlies the ability of bipolar cell terminals to rapidly extrude large Ca2⫹ loads. The efficient extrusion of Ca2⫹ by plasma membrane pumps may also help explain why mitochondria do not take up substantial amounts of Ca2⫹ in bipolar cell synaptic terminals, even during prolonged influx.
filter wheel. Spatially averaged fluorescence measurements were obtained from rectangular areas of interest encompassing the synaptic terminal and soma, using BL-10 acquisition and analysis software (BioLase). Alternatively, in some experiments fluorescence was measured using a photomultiplier tube, which viewed the preparation through a pinhole centered on the synaptic terminal. In some experiments in which patch clamp recording was not used, bipolar cells were loaded with calcium indicator by incubation for 8–15 min in normal external solution containing 5 M fura-2AM, followed by thorough washing and a 20 min rest period. For rhodamine-123 experiments, cells were incubated for 20 min in Ringer solution containing 10 g/ml rhodamine-123, followed by thorough washing. Calcium concentration was calculated from fura-2 or bis-fura-2 fluorescence at excitation wavelengths of 337 and 384 nm, using calibration constants obtained by filling cells with highly buffered concentrations of Ca2⫹ (Heidelberger and Matthews, 1992). To estimate calcium clearance rate, a straight line was fitted to [Ca2⫹]i in the range from 400 nM to 1200 nM during the falling phase of calcium responses. Chemicals were obtained from Sigma (CCCP, oligomycin, ATP␥S, Na2ATP, GTP), Research Biochemicals (sodium orthovanadate), or Molecular Probes (fura-2, bis-fura-2, rhodamine-123). CCCP was dissolved as a concentrated stock in DMSO and diluted to yield working solutions containing no more than 0.1% DMSO, which did not by itself affect bipolar neurons. CCCP was applied in Ca2⫹-free external solution containing no added calcium and 0.15 mM EGTA. Impermeant drugs were applied by internal dialysis through the whole-cell patch pipette. Externally acting drugs were applied by local superfusion through continuously flowing puffer pipettes placed near the recorded cell. Acknowledgments This work was supported by NIH Grant EY03821. Received June 21, 1999; revised November 18, 1999. References
Experimental Procedures Dissociated bipolar neurons from goldfish retina were prepared as described previously (Heidelberger and Matthews, 1992). Bipolar neurons were recognized based on their distinctive morphology (Figure 1A). Experiments were carried out at 21⬚C–24⬚C within 2–4 hr of dissociation. The external Ringer solution for recordings contained (in mM): 115 NaCl, 2.5 KCl, 1 MgCl2, 2.5 CaCl2, 10 glucose, and 10 HEPES (pH 7.4 with NaOH). Choline chloride replaced NaCl in sodium-free external solution. In experiments using high external K⫹ to depolarize cells, KCl was increased to 80 mM, with a corresponding reduction in NaCl. For whole-cell recordings, the pipette solution typically contained (in mM): 100 cesium gluconate, 50 HEPES, 2 Na2ATP, 3 MgCl2, and 0.5 GTP (pH 7.2 with CsOH). In some experiments, the pipette solution was supplemented with 2 mM sodium pyruvate, 2 mM malate, 1 mM NaH2PO4, 4 mM ATP, and 0.5 mM cAMP, a mixture intended to aid in maintaining mitochondrial metabolism in wholecell recordings. Pipette solutions were treated with Calcium Sponge (Molecular Probes) to remove contaminating Ca2⫹. ATP content was varied and fluorescent calcium indicator dyes (typically bis-fura-2 at 70–100 M) were added to the pipette solution as indicated in descriptions of individual experiments. Voltage clamp and current recordings were carried out using an EPC-9 patch clamp amplifier (HEKA). Patch pipettes were placed on either the soma or the terminal of isolated bipolar neurons (Figure 1A). When recordings of terminal calcium current are made with the pipette on the soma, experimental results and electrical modeling of the bipolar neuron (Mennerick et al., 1997) show that the resistance of the connecting axon introduces a small voltage error, which is of no consequence for the experiments reported here. For fluorescence measurements, images of cells were acquired at up to 30 frames/s with an intensified CCD camera (Xybion). Excitation light was provided by two tunable nitrogen/dye lasers (Laser Science) or in some cases by a xenon arc lamp coupled to a rotating
Baker, P.F. (1972). Transport and metabolism of calcium ions in nerve. Prog. Biophys. Mol. Biol. 24, 177–223. Barrabin, H., Garrahan, P.J., and Rega, A.F. (1980). Vanadate inhibition of the Ca2⫹-ATPase from human red cell membranes. Biochim. Biophys. Acta 600, 796–804. Blaustein, M.P., and Russell, J.M. (1975). Sodium-calcium exchange and calcium-calcium exchange in internally dialyzed squid giant axons. J. Membr. Biol. 24, 285–312. Bond, G.H., and Hudgins, P. (1980). Inhibition of the red cell Ca2⫹ATPase by vanadate. Biochim. Biophys. Acta 600, 781–790. Budd, S.L., and Nicholls, D.G. (1996a). A reevaluation of the role of mitochondria in neuronal Ca2⫹ homeostasis. J. Neurochem. 66, 403–411. Budd, S.L., and Nicholls, D.G. (1996b). Mitochondrial calcium regulation and acute glutamate excitotoxicity in cultured cerebellar granule cells. J. Neurochem. 67, 2282–2291. Carafoli, E. (1991). Calcium pump of the plasma membrane. Physiol. Rev. 71, 129–153. Castilho, R.F., Hansson, O., Ward, M.W., Budd, S.L., and Nicholls, D.G. (1998). Mitochondrial control of acute glutamate excitotoxicity in cultured cerebellar granule cells. J. Neurosci. 18, 10277–10286. David, G., Barrett, J.N., and Barrett, E.F. (1998). Evidence that mitochondria buffer physiological Ca2⫹ loads in lizard motor nerve terminals. J. Physiol. 509, 59–65. Delaney, K.R., Zucker, R.S., and Tank, D.W. (1989). Calcium in motor nerve terminals associated with posttetanic potentiation. J. Neurosci. 9, 3558–3567. DiPolo, R. (1978). Ca pump driven by ATP in squid axons. Nature 274, 390–392. DiPolo, R., and Beauge´, L. (1982). The effect of pH on Ca2⫹ extrusion mechanisms in dialysed squid axons. Biochim. Biophys. Acta 688, 237–245.
Mitochondria in Presynaptic Calcium Handling 237
DiPolo, R., and Beauge´, L. (1983). The calcium pump and sodiumcalcium exchange in squid axons. Annu. Rev. Physiol. 45, 313–324. Duchen, M.R. (1992). Ca(2⫹)-dependent changes in the mitochondrial energetics in single dissociated mouse sensory neurons. Biochem. J. 283, 41–50. Duchen, M.R., Valdeolmillos, M., O’Neill, S.C., and Eisner, D.A. (1990). Effects of metabolic blockade on the regulation of intracellular calcium in dissociated mouse sensory neurons. J. Physiol. 424, 411–426. Duchen, M.R., Leyssens, A., and Crompton, M. (1998). Transient mitochondrial depolarizations reflect focal sarcoplasmic reticular calcium release in single rat cardiomyocytes. J. Cell Biol. 142, 975–988. Friel, D.D., and Tsien, R.W. (1994). An FCCP-sensitive Ca2⫹ store in bullfrog sympathetic neurons and its participation in stimulusevoked changes in [Ca2⫹]i. J. Neurosci. 14, 4007–4024. Heidelberger, R., and Matthews, G. (1992). Calcium influx and calcium current in single synaptic terminals of goldfish retinal bipolar neurons. J. Physiol. 207, 623–633. Helmchen, F., Borst, J.G.G., and Sakmann, B. (1997). Calcium dynamics associated with a single action potential in a presynaptic terminal. Biophys. J. 72, 1458–1471. Herrington, J., Park, Y.B., Babcock, D.F., and Hille, B. (1996). Dominant role of mitochondria in clearance of large Ca2⫹ loads from rat adrenal chromaffin cells. Neuron 16, 219–228. Johnson, L.V., Walsh, M.L., Bockus, B.J., and Chen, L.B. (1981). Monitoring of relative mitochondrial membrane potential in living cells by fluorescence microscopy. J. Cell Biol. 88, 526–535. Kimura, J., Noma, A., and Irisawa, H. (1986). Na-Ca exchange current in mammalian heart cells. Nature 319, 596–597. Kobayashi, K., and Tachibana, M. (1995). Ca2⫹ regulation in the presynaptic terminals of goldfish retinal bipolar cells. J. Physiol. 483, 79–94. Krizaj, D., and Copenhagen, D.R. (1998). Compartmentalization of calcium extrusion mechanisms in the outer and inner segments of photoreceptors. Neuron 21, 249–256. Loew, L.M., Carrington, W., Tuft, R.A., and Fay, F.S. (1994). Physiological cytosolic Ca2⫹ transients evoke concurrent mitochondrial depolarizations. Proc. Natl. Acad. Sci. USA 91, 12579–12583. Mennerick, S., Zenisek, D., and Matthews, G. (1997). Static and dynamic membrane properties of large-terminal bipolar cells from goldfish retina: experimental test of a compartment model. J. Neurophysiol. 78, 51–62. Morgans, C.W., El Far, O., Berntson, A., Wassle, H., and Taylor, W.R. (1998). Calcium extrusion from mammalian photoreceptor terminals. J. Neurosci. 18, 2467–2474. Mulkey, R.M., and Zucker, R.S. (1992). Posttetanic potentiation at the crayfish neuromuscular junction is dependent on both intracellular calcium and sodium ion concentration. J. Neurosci. 12, 4327– 4336. Neher, E., and Augustine, G.J. (1992). Calcium gradients and buffers in bovine chromaffin cells. J. Physiol. 450, 273–301. Nicholls, D.G., and A˚kerman, K.E.O. (1982). Mitochondrial calcium transport. Biochim. Biophys. Acta 683, 57–88. Nicholls, D.G., and Scott, I.D. (1980). The role of mitochondria in the regulation of calcium ion transport in synaptosomes. Biochem. Soc. Trans. 8, 264–266. Nohmi, M., Hua, S.Y., and Kuba, K. (1992). Intracellular calcium dynamics in response to action potentials in bullfrog sympathetic ganglion cells. J. Physiol. 458, 171–190. Pivovarova, N.B., Hongpaisan, J., Andrews, S.B., and Friel, D.D. (1999). Depolarization-induced mitochondrial Ca accumulation in sympathetic neurons: spatial and temporal characteristics. J. Neurosci. 19, 6372–6384. Regehr, W.G. (1997). Interplay between sodium and calcium dynamics in granule cell presynaptic terminals. Biophys. J. 73, 2476–2488. Regehr, W.G., and Atluri, P. (1995). Calcium transients in cerebellar granule cell presynaptic terminals. Biophys. J. 68, 2156–2170.
Rossi, J.P.F.C., Garrahan, P.J, and Rega, A.F. (1981). Vanadate inhibition of active Ca2⫹ transport across human red cell membranes. Biochim. Biophys. Acta 648, 145–150. Sinha, S.R., Wu, W.G., and Saggau, P. (1997). Presynaptic calcium dynamics and transmitter release evoked by single action potentials at mammalian central synapses. Biophys. J. 73, 637–651. Steunkel, E.L. (1994). Regulation of intracellular calcium and calcium buffering properties of rat isolated neurohypophysial nerve endings. J. Physiol. 481, 251–271. Stout, A.K., Raphael, H.M., Kanterewicz, B.I., Klann, E., and Reynolds, I.J. (1998). Glutamate-induced neuron death requires mitochondrial calcium uptake. Nat. Neurosci. 1, 366–373. Tachibana, M., and Okada, T. (1991). Release of endogenous excitatory amino acids from ON-type bipolar cells isolated from the goldfish retina. J. Neurosci. 11, 2199–2208. Tachibana, M., Okada, T., Arimura, T., Kobayashi, K., and Piccolino, M. (1993). Dihydropyridine-sensitive calcium current mediates neurotransmitter release from bipolar cells of the goldfish retina. J. Neurosci. 13, 2898–2909. Tang, Y., and Zucker, R.S. (1997). Mitochondrial involvement in posttetanic potentiation of synaptic transmission. Neuron 18, 483–491. Tatsumi, H., and Katayama, Y. (1993). Regulation of the intracellular free calcium concentration in acutely dissociated neurones from rat nucleus basalis. J. Physiol. 464, 165–181. Thayer, S.A., and Miller, R.J. (1990). Regulation of the intracellular free calcium concentration in single rat dorsal root ganglion neurons in vitro. J. Physiol. 425, 85–115. Tucker, T., and Fettiplace, R. (1995). Confocal imaging and calcium extrusion in turtle hair cells. Neuron 15, 1323–1335. von Gersdorff, H., Vardi, E., Matthews, G., and Sterling, P. (1996). Evidence that vesicles on the synaptic ribbon of retinal bipolar neurons can be rapidly released. Neuron 16, 1221–1227. von Gersdorff, H., Sakaba, T., Berglund, K., and Tachibana, M. (1998). Submillisecond kinetics of glutamate release from a sensory synapse. Neuron 21, 1177–1188. Werth, J.L., and Thayer, S.A. (1994). Mitochondria buffer physiological calcium loads in cultured rat dorsal root ganglion cells. J. Neurosci. 14, 348–356.
The Role of Mitochondria in Presynaptic Calcium Handling at a Ribbon Synapse David Zenisek† and Gary Matthews* Department of Neurobiology and Behavior State University of New York Stony Brook, New York 11794
Summary Mitochondria are thought to be important in clearing calcium from synaptic terminals. It is unclear, however, whether the principal role of mitochondria in presynaptic calcium handling is to take up Ca2ⴙ directly or to fuel Ca2ⴙ removal by other mechanisms. We used patch clamp techniques and fluorescence imaging to examine calcium clearance mechanisms, including mitochondrial uptake, in single synaptic terminals of retinal bipolar neurons. We found that extrusion through the ATP-dependent Ca2ⴙ pump of the plasma membrane is the dominant form of Ca2ⴙ removal in the synaptic terminal. Calcium uptake into mitochondria was sometimes evident with large Ca2ⴙ loads but was consistently observed only when plasma membrane extrusion was inhibited. We conclude that mitochondria act primarily as an energy source in clearance of Ca2ⴙ from bipolar cell synaptic terminals.
Introduction Internal calcium plays a number of roles in synaptic terminals in addition to neurotransmitter release, such as recruitment of vesicle pools and activation of ionic conductances. Information about the mechanisms regulating Ca2⫹ concentration is therefore important for understanding synaptic function from a number of perspectives. Several mechanisms have been proposed to be important in presynaptic Ca2⫹ regulation, including Na⫹–Ca2⫹ exchange, uptake into mitochondria, and plasma membrane Ca2⫹-ATPase (PMCA). However, opportunities for direct study of calcium handling in single terminals are rare, and most of what is known about Ca2⫹ regulation in neurons stems from studies of neuronal somata (e.g., Thayer and Miller, 1990; Friel and Tsien, 1994; Werth and Thayer, 1994) and from studies of calcium transport in squid giant axon (e.g., Baker, 1972; Blaustein and Russell, 1975; DiPolo, 1978; DiPolo and Beauge´, 1983). We have exploited the large size of synaptic terminals of bipolar neurons from goldfish retina to study directly the regulation of [Ca2⫹]i in single terminals. These giant synaptic terminals (diameter ≈ 10 m) are amenable to physiological manipulation and allow direct measurement of Ca2⫹ influx and [Ca2⫹]i, using combined patch
* To whom correspondence should be addressed (e-mail: gary.g. [email protected]). † Present address: Vollum Institute, Oregon Health Sciences University, Portland, Oregon 97201.
clamp and Ca2⫹ imaging techniques. The results demonstrate that PMCA is of primary importance in regulating basal Ca2⫹ and in clearing Ca2⫹ after depolarizationevoked influx. Mitochondrial Ca2⫹ uptake plays some role, especially when [Ca2⫹]i exceeds 800 nM, but the principal role of mitochondria in Ca2⫹ extrusion in bipolar cell synaptic terminals seems to be production of ATP to fuel the Ca2⫹ pump. Results Calcium Transients Evoked by Depolarization in Retinal Bipolar Neurons A typical large-terminal bipolar neuron from goldfish retina is shown in Figure 1A. Synaptic terminals of these neurons have a high density of slowly inactivating, dihydropyridine-sensitive Ca2⫹ channels (Heidelberger and Matthews, 1992; Tachibana et al., 1993) that drive synaptic release of glutamate (Tachibana and Okada, 1991; Tachibana et al., 1993; von Gersdorff et al., 1998). To examine calcium clearance in the synaptic terminal, single cells were patch clamped to control Ca2⫹ current, and Ca2⫹ concentration was monitored in the terminal and soma using fluorescent Ca2⫹ indicator (usually bisfura-2) introduced through the patch pipette. In keeping with the predominant localization of the Ca2⫹ channels in the synaptic terminal (Heidelberger and Matthews, 1992), depolarization evoked a large, rapid increase in [Ca2⫹]i in the terminal but a much smaller [Ca2⫹]i increase in the soma (Figure 1B). The small Ca2⫹ increase in the soma was also due to Ca2⫹ influx through Ca2⫹ channels (data not shown), which are present in somata, albeit at a much lower density than in terminals (Heidelberger and Matthews, 1992; Mennerick et al., 1997). After depolarization, [Ca2⫹]i returned rapidly to baseline in the synaptic terminal, and recovery was well described by a single exponential time constant of ⵑ1 s. Figure 1C shows calcium responses in the synaptic terminal for depolarizations of increasing duration, and Figure 1D summarizes the speed of calcium decay for different amounts of Ca2⫹ influx. The waveform of calcium decay in the synaptic terminal was similar for both large and small Ca2⫹ loads, but the rate of clearance slowed with increasing Ca2⫹ load. The half-time of Ca2⫹ decay increased from 0.6 ⫾ 0.04 s for calcium influx of ⬍20 pC (mean ⫾ SEM, n ⫽ 48) to 1.3 ⫾ 0.06 s (n ⫽ 9) for calcium influx of ⬎150 pC (Figure 1D). Thus, Ca2⫹ removal in the synaptic terminal has a high capacity and can sustain rapid clearance after large Ca2⫹ influx. Although less Ca2⫹ entered the soma than the terminal during depolarization, recovery was consistently slower in the soma (Figure 1D). In experiments described in the next section, calcium clearance was studied in intact bipolar neurons loaded with fura-2AM and depolarized by high external potassium. In whole-cell recordings, the rate of calcium clearance was faster than in intact bipolar neurons, where the average half-time of Ca2⫹ decay for responses under control conditions was 3.8 ⫾ 0.4 s (mean ⫾ SEM, n ⫽
Neuron 230
Figure 1. Calcium Transients Elicited by Activation of Calcium Current in the Synaptic Terminal and Soma of a Retinal Bipolar Neuron (A) Typical appearance of an isolated goldfish bipolar neuron. Cells were filled with bisfura-2 (60–100 M) through a whole-cell patch pipette placed on either the soma or the terminal. (B) Calcium transients recorded in the synaptic terminal (solid line) and soma (dashed line) of a bipolar neuron filled with bis-fura-2. The timing of a 900 ms depolarization from ⫺60 mV to 0 mV is given by the arrow. The small Ca2⫹ transient in the soma was due to activation of somatic Ca2⫹ channels, which are present at much lower density than in the synaptic terminal (Heidelberger and Matthews, 1992). (C) Calcium transients elicited in a synaptic terminal by depolarizations from ⫺60 mV to 0 mV for 5 s (solid line), 270 ms (dashed line), and 70 ms (gray line). Traces were aligned so that pulse offset coincides. (D) Time course of Ca2⫹ recovery as a function of total Ca2⫹ influx in bipolar cell terminals (closed squares) and somata (open circles). Half decay time (t1/2) of Ca2⫹ transients provided an index of recovery time course. Calcium charge entering during depolarization (abscissa) was calculated by integrating leak-subtracted Ca2⫹ current; 100 pC of charge corresponds to ⵑ0.5 fmol of Ca2⫹. The amount of influx was varied by changing the duration of depolarization.
26). Because high-K⫹ depolarization typically elevated [Ca2⫹]i to high levels for 10 s or more, the slower Ca2⫹ decay in intact cells may reflect in part the slowing of clearance after large calcium loads (see Figure 1D). Also, differences in experimental conditions associated with whole-cell voltage clamp, such as internal dialysis, may contribute to differences in the rate of calcium clearance between intact cells and patch-clamped cells. To estimate the endogenous rate of calcium clearance in the synaptic terminal in the absence of the buffering action of added Ca2⫹ indicator, time constants of recovery were also measured with varying amounts of exogenous Ca2⫹ buffer capacity. By extrapolating the relation between recovery time constant and buffer capacity to the y intercept, we estimated a time constant of 0.7 s in the absence of added buffer. PMCA Plays a Central Role in Calcium Clearance Recent results point to an important role for PMCA in Ca2⫹ extrusion from synaptic terminals of retinal photoreceptors (Krizaj and Copenhagen, 1998; Morgans et al., 1998), and immunostaining reveals PMCA labeling of bipolar cell terminals in goldfish retina (Morgans et al., 1998). Therefore, we examined the possible role of PMCA in calcium clearance in bipolar cell synaptic terminals. To ensure that intracellular conditions were disturbed minimally, we tested the effect of PMCA inhibitors applied extracellularly to bipolar cells loaded with membrane-permeant fura-2AM. Calcium was elevated by external superfusion with high-K⫹ solution. Figure 2A shows that recovery of [Ca2⫹]i to baseline was eliminated by external application of 1 mM La3⫹, an inhibitor of PMCA (Carafoli, 1991; Herrington et al., 1996). When La3⫹ was applied during the falling phase of the calcium response, the decay of [Ca2⫹]i abruptly ceased (Figure
2B), and [Ca2⫹]i remained at a sustained plateau during application of the PMCA inhibitor. The inset in Figure 2B summarizes the action of La3⫹ on the rate of calcium clearance, which slowed more than 100-fold in the presence of La3⫹ (1.2 ⫾ 0.8 nM/s versus 133 ⫾ 30 nM/s without La3⫹). In paired comparisons, the decay rate in the presence of La3⫹ was 0.8% ⫾ 0.4% of the control rate of clearance in the same synaptic terminal (mean ⫾ SEM, n ⫽ 15). After La3⫹ was removed, [Ca2⫹]i recovered fully to baseline (Figures 2A and 2B), and La3⫹ did not affect basal [Ca2⫹]i in the absence of depolarization-evoked Ca2⫹ influx (data not shown). Together with the fact that La3⫹ is a potent Ca2⫹ channel blocker, these observations suggest that La3⫹ influx was minimal. Therefore, external La3⫹ likely acts by inhibiting mechanisms of calcium clearance, rather than by binding to fura-2 within the terminal and producing a false Ca2⫹ signal. Although La3⫹ affects both PMCA and Na⫹–Ca2⫹ exchange (Kimura et al., 1986), evidence presented later will demonstrate that Na⫹–Ca2⫹ exchange plays little role in calcium clearance in bipolar cell synaptic terminals (see Figure 6). Thus, we attribute the potent slowing of Ca2⫹ recovery by La3⫹ to inhibition of PMCA. Calcium extrusion via PMCA is also blocked by elevated external pH (DiPolo and Beauge´, 1982; Kobayashi and Tachibana, 1995). In bipolar cell terminals, calcium clearance slowed dramatically when extracellular pH was elevated from 7.4 to 8.4, as illustrated in Figures 2C and 2D. As with La3⫹, elevation of external pH during the falling phase of the Ca2⫹ response rapidly inhibited calcium clearance after high-K⫹ depolarization, producing a sustained plateau of elevated Ca2⫹ (Figure 2D). In paired comparisons, the rate of decay of [Ca2⫹]i in pH 8.4 external solution slowed to 3.4% ⫾ 1.7% of the control rate (mean ⫾ SEM, n ⫽ 11). The average decay rate was 9.7 ⫾ 5 nM/s at pH 8.4 versus 293 ⫾ 80 nM/s at pH 7.4 (see inset of Figure 2D).
Mitochondria in Presynaptic Calcium Handling 231
Figure 2E shows the residual rate of decay of [Ca2⫹]i plotted against the plateau Ca2⫹ concentration during application of PMCA inhibitor (La3⫹ or elevated external pH). The rate of decay did not depend on the level of plateau Ca2⫹ up to ⵑ800 nM. Above 800 nM, the residual rate of decay increased somewhat, which might indicate operation of an alternative Ca2⫹ removal mechanism at higher Ca2⫹ levels. Results presented later suggest that this residual mechanism might be uptake into mitochondria. However, even the most rapid rate of recovery observed during PMCA inhibition was ⵑ10-fold slower than the control rate of clearance over the same range of Ca2⫹ concentration.
Figure 2. Calcium Clearance in the Synaptic Terminal Is Slowed by Inhibition of PMCA Bipolar neurons were loaded with fura-2AM and depolarized by local superfusion with 80 mM external K⫹. All signals were recorded from the synaptic terminal. (A) Two consecutive calcium responses were evoked by high-K⫹ depolarization at the times indicated by bars below the trace. After the second application of high K⫹, an external solution containing 1 mM La3⫹ was applied to the terminal at the indicated time. (B) Superimposed falling phases of calcium responses from the same terminal with (thick line) and without (thin line) application of 1 mM La3⫹ at the indicated time. A different cell from that in (A) is shown. The inset summarizes the average rate of Ca2⫹ decay for control responses and in the presence of 1 mM La3⫹ from 15 synaptic terminals. Note the logarithmic scale. (C) Elevation of extracellular pH from 7.4 to 8.4 slowed calcium recovery in the synaptic terminal after high-K⫹ depolarization. Timing high-K⫹ external solution is indicated by the bars below the trace. (D) Superimposed falling phases of a control calcium response (thin line) and during application of elevated pH (thick line) in the same synaptic terminal. A different cell from that in (C) is shown. The
Manipulations of Internal ATP Also Affect Calcium Clearance We also examined the effect on calcium clearance of intracellular manipulations that affect Ca2⫹ extrusion via PMCA, such as dialysis via whole-cell patch pipettes with internal solution lacking ATP. Without exogenous ATP, Ca2⫹ recovery following activation of Ca2⫹ current slowed progressively during the experiment. Figure 3A summarizes results from four bipolar neurons dialyzed with ATP-free solution (circles), and the inset in Figure 3A shows sample Ca2⫹ transients recorded at early and late times during dialysis. Time-dependent slowing of Ca2⫹ transients was not observed in control experiments in which the pipette solution contained ATP (Figure 3A, squares). These results demonstrate that reduced availability of ATP retards recovery of [Ca2⫹]i to baseline following Ca2⫹ influx, which is consistent with a contribution of PMCA to calcium clearance. As a further test of the role of PMCA, bipolar neurons were dialyzed in whole-cell recordings with internal solution containing ATP (2 mM) ⫹ 1 mM sodium orthovanadate, an inhibitor of PMCA (Barrabin et al., 1980; Bond and Hudgins, 1980; Rossi et al., 1981). As it diffused into the terminal, orthovanadate markedly slowed Ca2⫹ transients (Figure 3A, triangles; n ⫽ 3). This action is consistent with the effects of external PMCA inhibitors shown in Figure 2 and provides further confirmation that PMCA plays an important role in Ca2⫹ removal from the synaptic terminal. Replacing exogenous ATP in the whole-cell pipette solution with the nonhydrolyzable ATP analog ATP-␥S offers an alternative method of inhibiting PMCA. Figure 3B shows superimposed Ca2⫹ transients recorded at different times after onset of dialysis with internal solution containing 10 mM ATP-␥S, which progressively interfered with recovery of [Ca2⫹]i to the usual resting level after activation of Ca2⫹ current. In ten cells dialyzed with ATP-␥S, baseline [Ca2⫹]i increased to an average of 850 nM from the normal baseline of ⵑ150 nM. This action of ATP-␥S is another indication that ATP-dependent extrusion is important in returning [Ca2⫹]i in the synaptic terminal to low levels after influx.
inset summarizes the results from 11 similar experiments (note the logarithmic scale). (E) The observed Ca2⫹ decay rate during application of La3⫹ (closed squares) or pH 8.4 (open squares) is plotted as a function of the average plateau [Ca2⫹]i during application.
Neuron 232
Inhibition of endogenous ATP production by mitochondria also retarded calcium clearance in the synaptic terminal. Oligomycin, an inhibitor of mitochondrial ATP synthase, had effects similar to ATP-␥S, abolishing recovery of Ca2⫹ to normal resting level after activation of Ca2⫹ current and elevating basal [Ca2⫹]i (data not shown).
Figure 3. Calcium Regulation in the Synaptic Terminal Is Affected by Manipulations of Intracellular ATP (A) Slowing of calcium clearance in the synaptic terminal during whole-cell dialysis with a pipette solution containing no ATP (circles) or 1 mM orthovanadate ⫹ 2 mM ATP (triangles). Orthovanadate is an inhibitor of PMCA. The closed squares show results from control cells in which the pipette solution included 2 mM ATP. The patch pipette was placed on the soma to slow dialysis of the terminal. The time course of dialysis was monitored by the rise of fluorescence as bis-fura-2 entered the terminal, and the abscissa (dialysis index) is the fluorescence as a percent of its asymptotic level. The ordinate shows the half-time (t1/2) of decay of Ca2⫹ following depolarizing voltage clamp stimuli. The inset shows two superimposed Ca2⫹ responses obtained at the times indicated in the legend during dialysis with ATP-free solution. (B) Superimposed responses to depolarization during dialysis with an internal solution containing 10 mM ATP-␥S. The arrow indicates timing of 270 ms depolarizations from ⫺60 mV to 0 mV. The legend gives the time after break-in for each of the responses. (C) Effect of the protonophore CCCP (10 M) on Ca2⫹ removal in a cell dialyzed with 10 mM ATP-␥S. The trace shows responses observed after [Ca2⫹]i reached a stable elevated plateau (note elevated baseline), and the calcium recovery observed after depolarization represents the residual component of Ca2⫹ removal remaining after blockade of ATP-dependent removal. Thin and thick arrows show timing of 70 ms and 270 ms depolarizations from ⫺60 mV to 0 mV, respectively. The gray bar indicates timing of CCCP application. The inset shows superimposed responses to 70 ms depolarizations before (thin trace) and during (thick trace) CCCP application.
The Role of Mitochondrial Uptake in Calcium Clearance In Figure 3B, some degree of residual Ca2⫹ removal remained after ATP-␥S had taken full effect, and the terminal was still able to regulate [Ca2⫹]i, albeit at an elevated level. Giant synaptic terminals of goldfish bipolar neurons contain large numbers of mitochondria (von Gersdorff et al., 1996), and mitochondrial Ca2⫹ uptake is thought to be particularly significant at high [Ca2⫹]i, such as the elevated baseline seen with ATP-␥S (Nicholls and Scott, 1980; Nicholls and A˚kerman, 1982). Therefore, we examined whether mitochondria are responsible for the residual Ca2⫹ regulation observed when PMCA was inhibited. Figure 3C shows that residual Ca2⫹ recovery in the presence of ATP-␥S was strongly inhibited by CCCP, which blocks mitochondrial Ca2⫹ uptake by disrupting the mitochondrial proton gradient. Similar results were observed in seven experiments. Thus, uptake into mitochondria apparently accounts for the residual calcium clearance observed at elevated [Ca2⫹]i in the presence of ATP-␥S. Although mitochondrial uptake contributes to Ca2⫹ regulation when plasma membrane extrusion is inhibited, it remains to be established whether mitochondrial Ca2⫹ uptake also plays a role when PMCA is active. In neuronal somata and adrenal chromaffin cells, CCCP releases Ca2⫹ accumulated by mitochondria during a prior Ca2⫹ load, producing a large rebound increase in cytoplasmic [Ca2⫹]i (Thayer and Miller, 1990; Budd and Nicholls, 1996a; Herrington et al., 1996). This action of CCCP has been taken as evidence favoring mitochondrial contribution to calcium clearance in these cells. To determine whether mitochondria also take up Ca2⫹ during calcium clearance in bipolar cell synaptic terminals, we applied CCCP to voltage-clamped terminals after Ca2⫹ influx triggered by activation of Ca2⫹ current. Figure 4A shows an example of an experiment in which CCCP successfully released Ca2⫹ from an internal source—presumably mitochondria—after activation of Ca2⫹ current. In this instance, CCCP produced a moderate elevation of [Ca2⫹]i after brief depolarizations (first two arrows in Figure 4A). After a 5 s depolarization that drove [Ca2⫹]i to ⬎2 M, CCCP elicited a large rebound increase in [Ca2⫹]i. Similar release of Ca2⫹ by CCCP after prolonged depolarization was observed in only 6 out of 19 cells, however. Thus, CCCP revealed Ca2⫹ accumulation by mitochondria in the synaptic terminal inconsistently, even after large and prolonged Ca2⫹ influx. In whole-cell recordings, internal dialysis may remove cellular factors needed for mitochondrial metabolism, which may explain the inconsistent release of Ca2⫹ by CCCP in whole-cell recordings. When a cocktail designed to promote mitochondrial metabolism was added to the pipette solution (see Experimental Procedures), CCCP still failed to release Ca2⫹ after large Ca2⫹ loads
Mitochondria in Presynaptic Calcium Handling 233
Figure 4. Mitochondrial Ca2⫹ Uptake after Depolarization-Evoked Ca2⫹ Influx (A) Calcium responses in the synaptic terminal of a bipolar neuron under whole-cell voltage clamp. The arrows of increasing size show timing of depolarizations from ⫺60 mV to 0 mV for 70, 270, or 470 ms, and the large arrowhead indicates a 5 s depolarization. CCCP (10 M) was applied at the gray bars in a Ca2⫹-free external solution, to ensure that any changes in [Ca2⫹]i represent release of internal Ca2⫹. (B) Calcium responses in the synaptic terminal of four different bipolar neurons loaded with fura-2AM. The black bars below the traces show timing of high-K⫹ depolarization, and the gray bars indicate application of CCCP. CCCP released Ca2⫹ from an internal source in the top two cells, but not in the lower two. (C) Summary of Ca2⫹ release induced by CCCP in bipolar neurons loaded with fura-2AM. The rise in [Ca2⫹]i induced by CCCP is plotted as a function of the preceding Ca2⫹ load activated by high-K⫹ depolarization. (D) Superimposed falling phases of Ca2⫹ responses in a synaptic terminal under control conditions (thin trace) and during application of CCCP (thick trace). CCCP was applied just before the high-K⫹ external solution was removed. The inset summarizes Ca2⫹ decay rates from 14 synaptic terminals with and without CCCP.
in four cells. However, to ensure that mitochondria were perturbed as little as possible, we also tested CCCP’s action on intact bipolar neurons loaded with fura-2AM. Figure 4B shows sample Ca2⫹ responses, and Figure 4C summarizes the results from synaptic terminals in which CCCP was applied after Ca2⫹ influx stimulated by high-K⫹ depolarization. As in dialyzed cells, CCCP had a variable effect on [Ca2⫹]i, although a larger fraction of intact cells responded to CCCP (63% of intact cells versus 32% of dialyzed cells). The amount of Ca2⫹ released by CCCP did not correlate well with the size of the preceding Ca2⫹ load (Figure 4C), and so the variability in response to CCCP cannot be accounted for by variation in Ca2⫹ load. We conclude that when PMCA is functional, bipolar cell synaptic terminals are frequently capable of clearing large Ca2⫹ loads without detectable accumulation of Ca2⫹ in mitochondria, at least as indicated by CCCP-induced Ca2⫹ release. In other instances, a variable fraction of the Ca2⫹ load was subsequently releasable by CCCP. The reason for this variability in apparent mitochondrial uptake is unclear. In bipolar neurons loaded with fura-2AM, we also examined whether CCCP affects the rate of Ca2⫹ decay after high-K⫹ depolarization in the range of [Ca2⫹]i ⬎500 nM, where previous results suggest mitochondrial uptake to be most prevalent. CCCP was applied though a second puffer pipette just before the high-K⫹ superfusion pipette was removed, so that CCCP was present throughout the falling phase of the Ca2⫹ response. The rate of Ca2⫹ decay was then measured from the peak
of the Ca2⫹ response (1000–1600 nM) to 500 nM with and without CCCP. In contrast with the potent effects of PMCA inhibitors on calcium clearance (see Figure 2), CCCP did not consistently affect Ca2⫹ decay rate in the range of ⬎500 nM, as shown in Figure 4D. In 14 experiments, the rate of Ca2⫹ removal in the presence of CCCP was 113% ⫾ 13% of the control rate.
Mitochondrial Depolarization Monitored with Rhodamine-123 Because CCCP commonly had little or no effect on [Ca2⫹]i in synaptic terminals, it was necessary to confirm that CCCP did indeed affect the mitochondria. For this purpose, bipolar neurons were loaded with the fluorescent dye rhodamine-123, which accumulates in normally polarized mitochondria and is released when the mitochondria depolarize. Quenching reduces the fluorescence of the accumulated dye in polarized mitochondria, and overall fluorescence increases as the dye is released and diffuses throughout the cell upon mitochondrial depolarization (Johnson et al., 1981; Duchen et al., 1990). By dissipating the mitochondrial proton gradient, CCCP should depolarize mitochondria and increase rhodamine-123 fluorescence. Figure 5B shows that CCCP evoked a large, rapid increase in rhodamine123 fluorescence in both the synaptic terminal and soma. Similar results were observed in 16 of 17 cells. The average increase in fluorescence was 62% ⫾ 19% in terminals and 37% ⫾ 6% in somata. Thus, CCCP
Neuron 234
Figure 5. Rhodamine-123 Fluorescence Reveals Changes in Mitochondrial Membrane Potential Produced by CCCP and Ca2⫹ Influx (A) Images illustrating redistribution of rhodamine-123 fluorescence in a bipolar cell synaptic terminal (top row) and a cone photoreceptor (bottom row) in response to mitochondrial inhibition with 10 M CCCP. On the left are bright-field images of the cells viewed through the intensifier of the CCD camera used to acquire the fluorescence images. Scale bars, 10 m. (B) The protonophore CCCP rapidly depolarizes mitochondria in the soma (thin line) and synaptic terminal (thick line) of a bipolar neuron. At the arrow, 10 M CCCP was applied, producing a large increase in rhodamine-123 fluorescence. Fluorescence traces are normalized relative to prestimulus baseline. (C) Effect of depolarizing voltage clamp stimuli on rhodamine-123 fluorescence. Strong Ca2⫹ influx, elicited by a 5 s depolarizing pulse from ⫺60 mV to 0 mV (first arrow), sometimes increased rhodamine123 fluorescence in the synaptic terminal (thick line) but not in the soma (thin line). Briefer depolarization of the bipolar neuron (870 ms; second arrow) produced a smaller increase in fluorescence.
reliably and rapidly depolarizes mitochondria in bipolar neurons. Mitochondria in bipolar cell terminals are not evenly distributed and instead form a tightly packed cluster (von Gersdorff et al., 1996). In the resting synaptic terminal, rhodamine-123 fluorescence was also nonuniformly distributed (Figure 5A), as expected for a cluster of mitochondria. After application of CCCP, fluorescence spread evenly throughout the terminal, which is consistent with release of the dye upon mitochondrial depolarization induced by CCCP. Figure 5A also shows that rhodamine-123 fluorescence was localized to the ellipsoid region of cone photoreceptors. The ellipsoid is a
prominent cluster of mitochondria in the photoreceptor inner segment, and application of CCCP caused fluorescence to spread evenly throughout the cone (Figure 5A). Thus, the rhodamine-123 signal behaves as expected for a mitochondrial marker. Because mitochondria depolarize when Ca2⫹ is taken up (e.g., Duchen, 1992; Loew et al., 1994; Duchen et al., 1998), rhodamine-123 fluorescence was also used to monitor changes in mitochondrial membrane potential associated with Ca2⫹ influx in bipolar cell synaptic terminals. Activation of Ca2⫹ current in bipolar neurons loaded with rhodamine-123 produced an appreciable change in fluorescence only with prolonged Ca2⫹ current, and then only in a minority of the cells examined. Figure 5C shows an example of a cell that did show a detectable increase in rhodamine-123 fluorescence in the synaptic terminal in response to 5 s and 870 ms depolarizing voltage clamp pulses. In 18 bipolar neurons in which similar responses were observed, 5 s depolarizing pulses increased fluorescence in the terminal by 23% ⫾ 3%. However, a detectable response to 5 s stimuli was observed in only 18 of 38 cells tested. Depolarizing voltage clamp pulses briefer than ⵑ500 ms were without effect on rhodamine-123 fluorescence, even when fluorescence was averaged over 15 stimuli to decrease noise. No consistent responses to depolarization were observed in the soma. The small and inconsistent effect of large amounts of Ca2⫹ influx on the membrane potential of mitochondria in the synaptic terminal suggests either that mitochondria are able to take up a large calcium load without appreciable loss of membrane potential or that calcium uptake by mitochondria is not appreciable. Given the previous results showing little effect of CCCP on calcium clearance and little release of Ca2⫹ by CCCP after calcium loads, the latter explanation seems more plausible. Naⴙ–Ca2ⴙ Exchange Plays Little Role in Presynaptic Calcium Clearance To test whether Na⫹–Ca2⫹ exchange contributes to calcium clearance in bipolar cell terminals, we examined the effect of removing external Na⫹ on Ca2⫹ removal. Figure 6A shows that replacing external sodium with choline had no effect on the rate of decay of [Ca2⫹]i after depolarization. Similar results were obtained in six cells. As shown in Figure 6B, sodium removal also did not affect recovery of [Ca2⫹]i after high-K⫹ depolarization in intact bipolar cells loaded with fura-2AM. After sodium removal, the Ca2⫹ decay rate averaged 176 ⫾ 14 nM/s (mean ⫾ SEM, n ⫽ 10), compared with 213 ⫾ 24 nM/s in the same set of intact cells with normal external sodium (see inset of Figure 6B). This difference is not statistically significant (p ⬍ 0.22; t test for paired samples). Thus, we conclude that Na⫹–Ca2⫹ exchange contributes little to calcium clearance in bipolar cell synaptic terminals. Discussion Taken as a whole, our results suggest that the dominant role of mitochondria in Ca2⫹ handling in the bipolar cell synaptic terminal is to provide ATP to drive Ca2⫹ efflux via PMCA. Direct Ca2⫹ uptake into mitochondria was
Mitochondria in Presynaptic Calcium Handling 235
Figure 6. Removing External Sodium Has Little Effect on Calcium Clearance in the Synaptic Terminal (A) Calcium transients evoked in the synaptic terminal by 5 s depolarizations from ⫺60 mV to 0 mV in normal (thin trace) and sodiumfree external solution (thick trace). (B) Superimposed falling phases of Ca2⫹ responses evoked by highK⫹ depolarization in a bipolar neuron loaded with fura-2AM. The thin trace shows a control response, and the thick trace shows a response in sodium-free external solution, applied immediately after removal of the high-K⫹ solution. The inset shows average Ca2⫹ decay rates under control conditions and in sodium-free solution for ten synaptic terminals.
inconsistently observed, even after large and prolonged elevation of [Ca2⫹]i. Several lines of evidence support these conclusions. Attempts to release Ca2⫹ from presynaptic mitochondria after influx of Ca2⫹ through voltage-gated channels met with limited success, even after prolonged elevation of Ca2⫹ to high levels. Measurement of mitochondrial membrane potential, using rhodamine123, also failed to provide evidence for Ca2⫹ uptake into mitochondria during prolonged depolarization in over half of the terminals examined. Inhibition of mitochondrial Ca2⫹ uptake had little effect on the rate of calcium clearance in the synaptic terminal, except when PMCA was inactivated. By contrast, inhibition of PMCA has a large effect on calcium clearance. Two different extracellular inhibitors of PMCA—La3⫹ and elevated external pH—dramatically slowed calcium clearance to 1%–3% of normal, suggesting that PMCA accounts for ⬎95% of calcium clearance under normal conditions. Intracellular inhibitors of ATP-dependent extrusion (vanadate, ATP␥S) slowed Ca2⫹ removal and increased basal Ca2⫹. Thus, the role of mitochondria in Ca2⫹ handling in the synaptic terminal differs from chromaffin cells, where mitochondrial uptake has been shown to be important (Herrington et al., 1996). Because large uptake of Ca2⫹ into mitochondria may be toxic, reliance on ATP-dependent extrusion to the exclusion of mitochondrial Ca2⫹ uptake may be necessary at synaptic terminals, which experience larger Ca2⫹ loads than neuronal somata. Prevention of mitochondrial Ca2⫹ uptake is known to protect against excitotoxic effects of glutamate in somata (Budd and Nicholls, 1996b; Castilho et al., 1998; Stout et al., 1998), and
the amount of Ca2⫹ influx that triggers excitotoxicity is similar to that occurring in tonic synaptic terminals during depolarization. Therefore, avoiding Ca2⫹ uptake into mitochondria may be required to prevent an excitotoxic response in bipolar cell synaptic terminals, which experience large and prolonged elevation of internal Ca2⫹ during tonic signaling. Mitochondria in bipolar cell synaptic terminals form a discrete cluster near the junction with the axon (von Gersdorff et al., 1996; see also Figure 5A). This segregation of mitochondria may help to isolate them from Ca2⫹ influx and thus minimize Ca2⫹ uptake. Recent work using x-ray microanalysis showed that the amount of Ca2⫹ taken up by a mitochondrion declined with distance from the plasma membrane (Pivovarova et al., 1999). Thus, clustering of mitochondria at a distance from sites of influx may explain in part the reduced role of mitochondrial uptake in calcium clearance in bipolar cell synaptic terminals. In our experiments, Ca2⫹ uptake into mitochondria was detected in a subset of terminals, evidenced either by CCCP-induced Ca2⫹ release (see Figure 4) or by mitochondrial depolarization during Ca2⫹ influx (see Figure 5). A possible explanation is that a partial breakdown of mitochondrial segregation occurred in those instances. Available evidence suggests that at least some Ca2⫹ uptake by mitochondria occurs under physiological conditions in conventional synaptic terminals, particularly during prolonged stimuli. For example, Tang and Zucker (1997) suggest that efflux of Ca2⫹ accumulated by mitochondria during tetanic stimulation is responsible for posttetanic potentiation, which implies that mitochondrial Ca2⫹ uptake plays an important role in synaptic plasticity. Also, David et al. (1998) detected increased mitochondrial Ca2⫹ during trains of action potentials at physiological rates in motor nerve terminals. Therefore, the relative dominance of ATP-dependent Ca2⫹ extrusion over mitochondrial uptake observed in bipolar cell synaptic terminals may not hold for smaller, conventional synaptic terminals. Removal of external sodium did not affect the rate of Ca2⫹ removal in bipolar cell synaptic terminals, which suggests that Na⫹–Ca2⫹ exchange plays little role in presynaptic Ca2⫹ extrusion. In this regard, bipolar cell terminals are similar to hair cells (Tucker and Fettiplace, 1995) and photoreceptor terminals (Krizaj and Copenhagen, 1998; Morgans et al., 1998), in which ATP-dependent pumps are also more important than Na⫹–Ca2⫹ exchange in Ca2⫹ extrusion. Hair cells, photoreceptors, and bipolar cells are all tonic neurons that release neurotransmitter for sustained periods under physiological conditions. Therefore, a preference for ATP-dependent Ca2⫹ removal over Na⫹–Ca2⫹ exchange may be a general feature of tonic synaptic terminals. Internal sodium accumulation may slow Ca2⫹ extrusion by the Na⫹–Ca2⫹ exchanger during repetitive or prolonged stimulation (Mulkey and Zucker, 1992; Regehr, 1997). In tonic neurons, then, reliance on mechanisms other than Na⫹–Ca2⫹ exchange may be required to maintain Ca2⫹ extrusion in the face of prolonged Ca2⫹ influx. We estimated that in whole-cell recordings, Ca2⫹ decays with a time constant of about 0.7 s in the absence of exogenous Ca2⫹ buffer, compared with a time constant of ⵑ10 s under comparable conditions in adrenal
Neuron 236
chromaffin cells (Neher and Augustine, 1992). Although bipolar cell synaptic terminals clear calcium relatively rapidly, calcium recovery at other synaptic terminals is even faster than in bipolar cells. The time constant of calcium decay was 45 ms following single action potentials in the calyx of Held in the medial nucleus of the trapezoid body (Helmchen et al., 1997), 150 ms in cerebellar granule cell synapses (Regehr and Atluri, 1995), and 30 ms in hippocampal synapses (Sinha et al., 1997). This enhanced speed of recovery in other terminals may reflect the more favorable surface-to-volume ratio of the smaller synaptic terminals compared with the large bulbous terminals of bipolar neurons. Calcium clearance mechanisms in bipolar cell terminals also have high capacity for removing ongoing Ca2⫹ loads. In our experiments under whole-cell voltage clamp, the half-decay time of Ca2⫹ transients approximately doubled (from 0.6 to 1.3 s) as the Ca2⫹ load increased from 5 pC to over 200 pC. By contrast, in crayfish motor nerve terminals, the time constant for Ca2⫹ recovery following high-frequency stimulus trains was 20- to 40-fold slower than after low-frequency stimuli (Delaney et al., 1989). A similar large slowing of Ca2⫹ removal after strong stimuli has also been observed in other neuronal preparations (Nohmi et al., 1992; Tatsumi and Katayama, 1993; Steunkel, 1994; Regehr, 1997). ATP-dependent Ca2⫹ extrusion underlies the ability of bipolar cell terminals to rapidly extrude large Ca2⫹ loads. The efficient extrusion of Ca2⫹ by plasma membrane pumps may also help explain why mitochondria do not take up substantial amounts of Ca2⫹ in bipolar cell synaptic terminals, even during prolonged influx.
filter wheel. Spatially averaged fluorescence measurements were obtained from rectangular areas of interest encompassing the synaptic terminal and soma, using BL-10 acquisition and analysis software (BioLase). Alternatively, in some experiments fluorescence was measured using a photomultiplier tube, which viewed the preparation through a pinhole centered on the synaptic terminal. In some experiments in which patch clamp recording was not used, bipolar cells were loaded with calcium indicator by incubation for 8–15 min in normal external solution containing 5 M fura-2AM, followed by thorough washing and a 20 min rest period. For rhodamine-123 experiments, cells were incubated for 20 min in Ringer solution containing 10 g/ml rhodamine-123, followed by thorough washing. Calcium concentration was calculated from fura-2 or bis-fura-2 fluorescence at excitation wavelengths of 337 and 384 nm, using calibration constants obtained by filling cells with highly buffered concentrations of Ca2⫹ (Heidelberger and Matthews, 1992). To estimate calcium clearance rate, a straight line was fitted to [Ca2⫹]i in the range from 400 nM to 1200 nM during the falling phase of calcium responses. Chemicals were obtained from Sigma (CCCP, oligomycin, ATP␥S, Na2ATP, GTP), Research Biochemicals (sodium orthovanadate), or Molecular Probes (fura-2, bis-fura-2, rhodamine-123). CCCP was dissolved as a concentrated stock in DMSO and diluted to yield working solutions containing no more than 0.1% DMSO, which did not by itself affect bipolar neurons. CCCP was applied in Ca2⫹-free external solution containing no added calcium and 0.15 mM EGTA. Impermeant drugs were applied by internal dialysis through the whole-cell patch pipette. Externally acting drugs were applied by local superfusion through continuously flowing puffer pipettes placed near the recorded cell. Acknowledgments This work was supported by NIH Grant EY03821. Received June 21, 1999; revised November 18, 1999. References
Experimental Procedures Dissociated bipolar neurons from goldfish retina were prepared as described previously (Heidelberger and Matthews, 1992). Bipolar neurons were recognized based on their distinctive morphology (Figure 1A). Experiments were carried out at 21⬚C–24⬚C within 2–4 hr of dissociation. The external Ringer solution for recordings contained (in mM): 115 NaCl, 2.5 KCl, 1 MgCl2, 2.5 CaCl2, 10 glucose, and 10 HEPES (pH 7.4 with NaOH). Choline chloride replaced NaCl in sodium-free external solution. In experiments using high external K⫹ to depolarize cells, KCl was increased to 80 mM, with a corresponding reduction in NaCl. For whole-cell recordings, the pipette solution typically contained (in mM): 100 cesium gluconate, 50 HEPES, 2 Na2ATP, 3 MgCl2, and 0.5 GTP (pH 7.2 with CsOH). In some experiments, the pipette solution was supplemented with 2 mM sodium pyruvate, 2 mM malate, 1 mM NaH2PO4, 4 mM ATP, and 0.5 mM cAMP, a mixture intended to aid in maintaining mitochondrial metabolism in wholecell recordings. Pipette solutions were treated with Calcium Sponge (Molecular Probes) to remove contaminating Ca2⫹. ATP content was varied and fluorescent calcium indicator dyes (typically bis-fura-2 at 70–100 M) were added to the pipette solution as indicated in descriptions of individual experiments. Voltage clamp and current recordings were carried out using an EPC-9 patch clamp amplifier (HEKA). Patch pipettes were placed on either the soma or the terminal of isolated bipolar neurons (Figure 1A). When recordings of terminal calcium current are made with the pipette on the soma, experimental results and electrical modeling of the bipolar neuron (Mennerick et al., 1997) show that the resistance of the connecting axon introduces a small voltage error, which is of no consequence for the experiments reported here. For fluorescence measurements, images of cells were acquired at up to 30 frames/s with an intensified CCD camera (Xybion). Excitation light was provided by two tunable nitrogen/dye lasers (Laser Science) or in some cases by a xenon arc lamp coupled to a rotating
Baker, P.F. (1972). Transport and metabolism of calcium ions in nerve. Prog. Biophys. Mol. Biol. 24, 177–223. Barrabin, H., Garrahan, P.J., and Rega, A.F. (1980). Vanadate inhibition of the Ca2⫹-ATPase from human red cell membranes. Biochim. Biophys. Acta 600, 796–804. Blaustein, M.P., and Russell, J.M. (1975). Sodium-calcium exchange and calcium-calcium exchange in internally dialyzed squid giant axons. J. Membr. Biol. 24, 285–312. Bond, G.H., and Hudgins, P. (1980). Inhibition of the red cell Ca2⫹ATPase by vanadate. Biochim. Biophys. Acta 600, 781–790. Budd, S.L., and Nicholls, D.G. (1996a). A reevaluation of the role of mitochondria in neuronal Ca2⫹ homeostasis. J. Neurochem. 66, 403–411. Budd, S.L., and Nicholls, D.G. (1996b). Mitochondrial calcium regulation and acute glutamate excitotoxicity in cultured cerebellar granule cells. J. Neurochem. 67, 2282–2291. Carafoli, E. (1991). Calcium pump of the plasma membrane. Physiol. Rev. 71, 129–153. Castilho, R.F., Hansson, O., Ward, M.W., Budd, S.L., and Nicholls, D.G. (1998). Mitochondrial control of acute glutamate excitotoxicity in cultured cerebellar granule cells. J. Neurosci. 18, 10277–10286. David, G., Barrett, J.N., and Barrett, E.F. (1998). Evidence that mitochondria buffer physiological Ca2⫹ loads in lizard motor nerve terminals. J. Physiol. 509, 59–65. Delaney, K.R., Zucker, R.S., and Tank, D.W. (1989). Calcium in motor nerve terminals associated with posttetanic potentiation. J. Neurosci. 9, 3558–3567. DiPolo, R. (1978). Ca pump driven by ATP in squid axons. Nature 274, 390–392. DiPolo, R., and Beauge´, L. (1982). The effect of pH on Ca2⫹ extrusion mechanisms in dialysed squid axons. Biochim. Biophys. Acta 688, 237–245.
Mitochondria in Presynaptic Calcium Handling 237
DiPolo, R., and Beauge´, L. (1983). The calcium pump and sodiumcalcium exchange in squid axons. Annu. Rev. Physiol. 45, 313–324. Duchen, M.R. (1992). Ca(2⫹)-dependent changes in the mitochondrial energetics in single dissociated mouse sensory neurons. Biochem. J. 283, 41–50. Duchen, M.R., Valdeolmillos, M., O’Neill, S.C., and Eisner, D.A. (1990). Effects of metabolic blockade on the regulation of intracellular calcium in dissociated mouse sensory neurons. J. Physiol. 424, 411–426. Duchen, M.R., Leyssens, A., and Crompton, M. (1998). Transient mitochondrial depolarizations reflect focal sarcoplasmic reticular calcium release in single rat cardiomyocytes. J. Cell Biol. 142, 975–988. Friel, D.D., and Tsien, R.W. (1994). An FCCP-sensitive Ca2⫹ store in bullfrog sympathetic neurons and its participation in stimulusevoked changes in [Ca2⫹]i. J. Neurosci. 14, 4007–4024. Heidelberger, R., and Matthews, G. (1992). Calcium influx and calcium current in single synaptic terminals of goldfish retinal bipolar neurons. J. Physiol. 207, 623–633. Helmchen, F., Borst, J.G.G., and Sakmann, B. (1997). Calcium dynamics associated with a single action potential in a presynaptic terminal. Biophys. J. 72, 1458–1471. Herrington, J., Park, Y.B., Babcock, D.F., and Hille, B. (1996). Dominant role of mitochondria in clearance of large Ca2⫹ loads from rat adrenal chromaffin cells. Neuron 16, 219–228. Johnson, L.V., Walsh, M.L., Bockus, B.J., and Chen, L.B. (1981). Monitoring of relative mitochondrial membrane potential in living cells by fluorescence microscopy. J. Cell Biol. 88, 526–535. Kimura, J., Noma, A., and Irisawa, H. (1986). Na-Ca exchange current in mammalian heart cells. Nature 319, 596–597. Kobayashi, K., and Tachibana, M. (1995). Ca2⫹ regulation in the presynaptic terminals of goldfish retinal bipolar cells. J. Physiol. 483, 79–94. Krizaj, D., and Copenhagen, D.R. (1998). Compartmentalization of calcium extrusion mechanisms in the outer and inner segments of photoreceptors. Neuron 21, 249–256. Loew, L.M., Carrington, W., Tuft, R.A., and Fay, F.S. (1994). Physiological cytosolic Ca2⫹ transients evoke concurrent mitochondrial depolarizations. Proc. Natl. Acad. Sci. USA 91, 12579–12583. Mennerick, S., Zenisek, D., and Matthews, G. (1997). Static and dynamic membrane properties of large-terminal bipolar cells from goldfish retina: experimental test of a compartment model. J. Neurophysiol. 78, 51–62. Morgans, C.W., El Far, O., Berntson, A., Wassle, H., and Taylor, W.R. (1998). Calcium extrusion from mammalian photoreceptor terminals. J. Neurosci. 18, 2467–2474. Mulkey, R.M., and Zucker, R.S. (1992). Posttetanic potentiation at the crayfish neuromuscular junction is dependent on both intracellular calcium and sodium ion concentration. J. Neurosci. 12, 4327– 4336. Neher, E., and Augustine, G.J. (1992). Calcium gradients and buffers in bovine chromaffin cells. J. Physiol. 450, 273–301. Nicholls, D.G., and A˚kerman, K.E.O. (1982). Mitochondrial calcium transport. Biochim. Biophys. Acta 683, 57–88. Nicholls, D.G., and Scott, I.D. (1980). The role of mitochondria in the regulation of calcium ion transport in synaptosomes. Biochem. Soc. Trans. 8, 264–266. Nohmi, M., Hua, S.Y., and Kuba, K. (1992). Intracellular calcium dynamics in response to action potentials in bullfrog sympathetic ganglion cells. J. Physiol. 458, 171–190. Pivovarova, N.B., Hongpaisan, J., Andrews, S.B., and Friel, D.D. (1999). Depolarization-induced mitochondrial Ca accumulation in sympathetic neurons: spatial and temporal characteristics. J. Neurosci. 19, 6372–6384. Regehr, W.G. (1997). Interplay between sodium and calcium dynamics in granule cell presynaptic terminals. Biophys. J. 73, 2476–2488. Regehr, W.G., and Atluri, P. (1995). Calcium transients in cerebellar granule cell presynaptic terminals. Biophys. J. 68, 2156–2170.
Rossi, J.P.F.C., Garrahan, P.J, and Rega, A.F. (1981). Vanadate inhibition of active Ca2⫹ transport across human red cell membranes. Biochim. Biophys. Acta 648, 145–150. Sinha, S.R., Wu, W.G., and Saggau, P. (1997). Presynaptic calcium dynamics and transmitter release evoked by single action potentials at mammalian central synapses. Biophys. J. 73, 637–651. Steunkel, E.L. (1994). Regulation of intracellular calcium and calcium buffering properties of rat isolated neurohypophysial nerve endings. J. Physiol. 481, 251–271. Stout, A.K., Raphael, H.M., Kanterewicz, B.I., Klann, E., and Reynolds, I.J. (1998). Glutamate-induced neuron death requires mitochondrial calcium uptake. Nat. Neurosci. 1, 366–373. Tachibana, M., and Okada, T. (1991). Release of endogenous excitatory amino acids from ON-type bipolar cells isolated from the goldfish retina. J. Neurosci. 11, 2199–2208. Tachibana, M., Okada, T., Arimura, T., Kobayashi, K., and Piccolino, M. (1993). Dihydropyridine-sensitive calcium current mediates neurotransmitter release from bipolar cells of the goldfish retina. J. Neurosci. 13, 2898–2909. Tang, Y., and Zucker, R.S. (1997). Mitochondrial involvement in posttetanic potentiation of synaptic transmission. Neuron 18, 483–491. Tatsumi, H., and Katayama, Y. (1993). Regulation of the intracellular free calcium concentration in acutely dissociated neurones from rat nucleus basalis. J. Physiol. 464, 165–181. Thayer, S.A., and Miller, R.J. (1990). Regulation of the intracellular free calcium concentration in single rat dorsal root ganglion neurons in vitro. J. Physiol. 425, 85–115. Tucker, T., and Fettiplace, R. (1995). Confocal imaging and calcium extrusion in turtle hair cells. Neuron 15, 1323–1335. von Gersdorff, H., Vardi, E., Matthews, G., and Sterling, P. (1996). Evidence that vesicles on the synaptic ribbon of retinal bipolar neurons can be rapidly released. Neuron 16, 1221–1227. von Gersdorff, H., Sakaba, T., Berglund, K., and Tachibana, M. (1998). Submillisecond kinetics of glutamate release from a sensory synapse. Neuron 21, 1177–1188. Werth, J.L., and Thayer, S.A. (1994). Mitochondria buffer physiological calcium loads in cultured rat dorsal root ganglion cells. J. Neurosci. 14, 348–356.