INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In the vertebrate retina, amacrine cells make synapses onto bipolar cells, ganglion cells, and other amacrine cells, and the output of these synapses can be critical for shaping the visual signal. With a culture system containing identified GABAergic amacrine cells, we have begun to elucidate the physiological properties of the synapses formed between these cells. Unlike many other CNS neurons, but like retinal bipolar cells and photoreceptors, amacrine cells use L-type Ca²⁺ channels at their synapses to control neurotransmitter release (Gleason et al. 1994). One of the consequences of using L-type Ca²⁺ channels at the synapse is that both the Ca²⁺ channels and the exocytotic machinery are sensitive to cytosolic Ca²⁺. It has been established that L-type Ca²⁺ channels are inactivated by the Ca²⁺-calmodulin complex (Peterson et al. 1999). Thus we would predict that the effects of Ca²⁺ buffering in the presynaptic terminal represents a balance between limiting exocytosis and promoting the activity of L-type Ca²⁺ channels.

At least five Ca²⁺-buffering mechanisms are thought to coexist in most cells: the plasma membrane Ca²⁺ ATPase and Na-Ca exchanger, endoplasmic reticulum Ca²⁺ATPases, cytosolic Ca²⁺-buffering proteins, and mitochondria. In amacrine cells, it has been estimated that ~60% of a Ca²⁺ load admitted through voltage-gated Ca²⁺ channels is removed by the plasma membrane Na-Ca exchanger (Gleason et al. 1995). This fraction was unaffected by inhibition of Ca²⁺ATPases, indicating that the balance of the Ca²⁺ must be taken up by temporary Ca²⁺ removal systems within the cell, possibly mitochondria, prior to export. We have also demonstrated that in the absence of Na-Ca exchanger activity, synaptic transmission between amacrine cells is dramatically prolonged (Gleason et al. 1994). However, the role of the other Ca²⁺-buffering mechanisms in synaptic transmission has not previously been explored.

Although it has long been understood that significant Ca²⁺ flux occurs across the mitochondrial membranes, more recent studies have demonstrated that this flux can have a significant impact on the amplitude and time course of Ca²⁺ transients in neurons (Friel and Tsien 1994; Thayer and Miller 1990; Werth and Thayer 1994). Although the outer mitochondrial membrane is freely permeable to Ca²⁺ ions, the inner mitochondrial membrane contains machinery that controls the flux of Ca²⁺ into and out of the matrix (for review, see Babcock and Hille 1998; Duchen 1999; Simpson and Russell 1998). The Ca²⁺ uniporter transports Ca²⁺down its electrochemical gradient into the mitochondrial matrix and is dependent on the mitochondrial membrane potential generated by the proton gradient across the inner mitochondrial membrane. The Na-Ca exchanger on the inner mitochondrial membrane usually moves Ca²⁺ back into the cytoplasm and is the primary efflux mechanism thought to operate during normal cellular function. Another pathway for Ca²⁺ efflux is the permeability transition pore. This conductance is activated at relatively high mitochondrial matrix Ca²⁺ concentration, and its opening may be key in Ca²⁺-mediated cell death.

Measurements of relative Ca²⁺ concentration and electrophysiological techniques were employed to reveal the role of mitochondrial Ca²⁺ buffering in retinal amacrine cells and their synapses. The protonophore p-trifluoromethoxy-phenylhydrazone (FCCP) was used to dissipate the proton gradient across the inner mitochondrial membrane, disabling uniporter activity and preventing Ca²⁺ uptake by the mitochondria. Because the role of mitochondria in buffering cytosolic Ca²⁺ transients has not previously been examined in amacrine cells, we first use cytosolic Ca²⁺ measurements to determine how Ca²⁺ transients in amacrine cells are shaped by mitochondrial Ca²⁺ buffering. We then examine the impact of mitochondrial Ca²⁺ buffering on synaptic transmission using perforated-patch voltage-clamp recordings.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Cell culture

Chick retinal cells were dissociated and cultured as previously described (Gleason et al. 1993). Cells were maintained in culture at 37°C under 5% CO₂ atmosphere until they were used for experiments, 7-14 days after plating. For both Ca²⁺ measurement and electrophysiology experiments, amacrine cells were identified based on their morphology. Cells with relatively large cell bodies (10-15 µM) bearing two to five primary processes have been previously identified as amacrine cells using both immunocytochemical and physiological criteria (Gleason et al. 1993; Huba and Hofmann 1990, 1991; Huba et al. 1992).

Ca²+ measurements

Cells were loaded with 2 µM fluo-3 acetoxymethylester (AM, Molecular Probes, Eugene, OR) for 1 h at room temperature and in the dark. At the end of the loading period, cells were washed with normal external solution (see following text) and maintained in the dark until they were used, routinely within 20 min. Culture dishes were mounted on the stage of an upright Nikon Optiphot II microscope, and the cells were visualized using a Noran laser confocal imaging system with a 40× water-immersion objective. Fluo-3 fluorescence was visualized using the 488-nm laser line with a 515-nm long-pass filter. Image and data acquisition were controlled by the InterVision (Noran) software and hardware. Images consisting of an average of eight frames were collected every 5 s. Shorter shuttering intervals often resulted in damage to the cells as determined by inspection with transmitted illumination at the end of an experiment. Because our shuttering rate was relatively slow (0.2 Hz), faster sampling (0.33 Hz) was done on a subset of cells to test whether we were accurately measuring the peak fluorescence increases. When the shutter was opened and fluorescence was measured every 3 s, 10-s applications of 100 mM K⁺ generated responses with 5-10 data points at the peak of the response, indicating that peak fluorescence amplitudes could be accurately measured with a 0.2-Hz sampling rate. Mean pixel intensity was sampled over a rectangular region (~25 µM²) of the cell body at 28 Hz. Peak pixel intensity values collected during shutter openings were subsequently extracted from the raw data using the peak detection analysis portion of the Origin software package (Microcal). Background fluorescence was sampled from a cell-free region of the field and remained constant over the duration of the experiments (10-20 min). For analysis of the spatial and temporal aspects of FCCP-dependent Ca²⁺ elevations (Fig. 7), fluorescence intensity values were obtained with a sampling rate of 28 Hz and without shuttering. For display purposes, most of the fluorescence intensity values were normalized to baseline values and reported as F/F_o.

All solutions were delivered through gravity-driven bath perfusion (3-4 ml/min). Ten seconds of the delay between agonist application and the cellular response was estimated to be due to the dead space in the perfusion line. Composition of the normal external solution was (in mM) 5.3 KCl, 135.0 NaCl, 3.0 CaCl₂, 0.41 MgCl₂, 5.6 glucose, and 3.0 HEPES, pH 7.4. Fifty and 100 mM K external solutions were made by replacement of NaCl. External solutions were adjusted to a final pH of 7.4 with NaOH. FCCP was dissolved in DMSO to form a 10 mM stock and then added to external solution for a final concentration of 1 µM.

Electrophysiological recording

Tissue culture dishes were mounted on the stage of an Olympus IX70 inverted microscope. A reference Ag/AgCl pellet served to ground the dish. Patch electrodes were pulled from thick-walled borosilicate glass (1.5 mm OD, 0.86 mm ID; Sutter Instruments, Novato, CA) on a Flaming-Brown micropipette puller (Sutter Instruments). Electrode resistance values ranged from 3 to 5 M. Most voltage-clamp recordings were made using the perforated-patch technique (Horn and Marty 1988). The pipette solution contained 200 µg/ml amphotericin B. Internal solutions consisted of the following (in mM): solution A (high Cl): 12.0 Cs acetate, 133.0 CsCl, 1.0 NaCl, 2.0 MgCl₂, 0.1 CaCl₂, 1.1 EGTA, and 10.0 HEPES; solution B (low Cl): 135.0 Cs acetate, 10.0 CsCl, 1.0 NaCl, 2.0 MgCl₂, 0.1 CaCl₂, 1.1 EGTA, and 10.0 HEPES. Voltage-clamp experiments performed in the ruptured-patch configuration used the following internal solution: 10.0 Cs acetate, 100.0 CsCl, 1.0 NaCl, 2.0 MgCl₂, 0.1 CaCl₂, 1.1 EGTA, 10.0 HEPES, 3.0 K ATP, 1.0 Na ATP, and 20.0 phosphocreatine and 50.0 IU/ml creatine phosphokinase. Current-clamp experiments were done using K⁺-containing internal solution: 150.0 K acetate, 10.0 KCl, 2.0 MgCl₂, 0.1 CaCl₂, 10.0 HEPES, 1.1 EGTA, and 1.0 Na ATP. All internal solutions were adjusted to pH 7.4 with CsOH.

For perforated-patch experiments, gigaohm seals were obtained by briefly hyperpolarizing the patch (140 mV) prior to gaining electrical access to the cell. Within 5 min of obtaining a seal, the resistance had usually fallen below 50 M and stabilized. Series resistance was monitored throughout all experiments. and if the series resistance was unstable during the recording, data from that cell were eliminated from the data set. The measured liquid junction potentials between the normal external and internals A and B were 2 and 7 mV, respectively. No corrections for errors due to series resistance or junction potentials were applied to these data.

External solutions were delivered by gravity-flow perfusion (2-3 ml/min) from one of two types of inlets. For some electrophysiology experiments (Figs. 3 and 5), a 1-mm diameter inlet tube was positioned ~5 mm from the cell. Solutions were manually switched upstream from the inlet. We estimated from dye experiments that the dead space in the perfusion accounted for ~7-9 s of the delay between test solution application and the cellular response. For all other experiments (Figs. 2 and 7-9), a double-barreled inlet was positioned ~3 mm from the cell. One of the barrels contained normal external and the second was manually switchable between test solutions. The inlet was positioned such that turning off the normal solution would allow the plume of test solution to quickly (<1 s) cover the cell. Based on control experiments with dye-containing FCCP solution, we determined that the delay between FCCP arrival and elevation of cytosolic Ca²⁺ (estimated by increased exchanger activity or stimulation of exocytosis) was <5 s. External solutions consisted of the following (in mM): normal external: 116.7 NaCl, 5.3 KCl, 20.0 TEA Cl, 3.0 CaCl₂, 0.41 MgCl₂, 5.6 glucose, and 3.0 HEPES. Ba external: normal external with 3 mM BaCl₂ replacing 3 mM CaCl₂. Li external: normal external with LiCl₂ replacing NaCl₂. Ca²⁺-free external: normal external with no added Ca²⁺. Bis-(o-aminophenoxy)-N,N,N',N'-tetraacetic acid (BAPTA) loading of cells was achieved by incubating cultures in 10 µM BAPTA AM (Biomol, Plymouth Meeting, PA) for 30 min at 37°C.

Unless synaptic currents were being recorded, 10 µM bicuculline methiodide (Research Biochemicals, Natick, MA) was added to the external solutions to block receptor-mediated GABA_A currents arising from autapses (Frerking et al. 1995; Gleason et al. 1993). In all voltage-clamp experiments, tetrodotoxin (150 nM, Alomone Labs, Jerusalem, Israel) was included in external solutions to block currents through voltage-gated Na⁺ channels and TEA was included to block voltage-gated K⁺ channels. Statistical analyses were done using the paired t-test and are reported as means ± SE. Unless otherwise specified, all reagents were obtained from Sigma (St. Louis, MO). All experiments were carried out at room temperature (24-27°C).

Specificity of FCCP

We chose to use FCCP for these experiments because it is fast acting and rapidly reversible and its effects are well characterized. Importantly, both carbonyl cyanide m-chlorophenylhydrazone (CCCP) and FCCP have been demonstrated to inhibit mitochondrial Ca²⁺ elevations in intact cells (Boitier et al. 1999; David et al. 1998; Herrington et al. 1996; Ricken et al. 1998). Although there is evidence that prolonged exposure (>10 min) of these compounds can have secondary effects in invertebrate neurons (Jensen and Rehder 1991), in numerous other preparations, brief exposures to low concentrations (1-5 µM) have effects on cytosolic Ca²⁺ consistent with perturbation of mitochondrial function exclusively (Babcock et al. 1996; David and Barrett 2000; David et al. 1998; Friel and Tsien 1994; Herrington et al. 1996; Scotti et al. 1999; Werth and Thayer 1994; White and Reynolds 1995, 1997). Furthermore, we have tested the effect of pretreatment of cells with antimycin A1(1 µM), an electron transport chain inhibitor, and oligomycin (1 µM), an ATPsynthase inhibitor. We found that in all cells tested (n = 9), this pretreatment occluded the FCCP-dependent Ca²⁺ elevations that normally occur in 78% of these cells (see RESULTS) indicating that the effects of FCCP on cytosolic Ca²⁺ are due to its action on mitochondria exclusively (Colegrove et al. 2000; David and Barrett 2000).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Mitochondria buffer Ca²+ loads in amacrine cells

To determine whether mitochondrial Ca²⁺ uptake normally plays a role in buffering Ca²⁺ loads in amacrine cells, cultured amacrine cells were loaded with fluo-3, and fluorescence was monitored at the cell body. Cytosolic Ca²⁺ elevations were elicited by depolarizing the cells with high K⁺ external solution, and mean fluorescence intensity was sampled from cell bodies. Most experiments were performed using 50 mM K⁺ external but the effects of 100 mM K⁺ external were examined on a subset of amacrine cells. Under normal conditions, depolarization produced two kinetic classes of Ca²⁺ responses in amacrine cells (Fig. 1, A-C). In response to 50 mM K⁺, all cells exhibited an initial steep rise in cytosolic Ca²⁺ (n = 43). After reaching a peak, calcium concentration began to decline quickly; however, some cells (7%) had a second, slower plateau phase of recovery (Fig. 1, A and C). The plateau response was more prevalent in cells depolarized by 100 mM K⁺ external (39%; n = 23).

View larger version (21K): [in this window] [in a new window]   Fig. 1. Mitochondria buffer Ca2+ elevations induced by depolarization. Cells were loaded with fluo-3 Ca2+ indicator and images collected every 5 s. A-D: vertical arrows denote initiation of high K+ application and gray boxes denote duration of p-trifluoromethoxy-phenylhydrazone (FCCP) application. A: a 10-s application of 100 mM K+ generates a cytosolic Ca2+ elevation. After the initial rapid clearance of calcium, a plateau phase of Ca2+ concentration is observed. In the presence of 1 µM FCCP, the initial clearance of calcium is slowed and the plateau phase is lost. B: in a different cell from A, application of 50 mM K+ for 10 s produces a Ca2+ transient with no plateau phase. In the presence of FCCP, recovery to baseline Ca2+ levels is slowed. Note that in this cell, an FCCP-dependent Ca2+ increase is observed (arrow) prior to depolarization. C: addition of FCCP immediately after a 30-s application of 100 mM K+ produces a plateau phase in a cell that did not exhibit a pronounced plateau phase in response to depolarization alone. D: a 35-s application of FCCP produces a transient increase in cytosolic Ca2+ in the absence of any previous depolarization. A subsequent 10 s application of 50 mM K+ also produces a Ca2+ elevation.

To quantify the role of mitochondria in buffering amacrine cell calcium loads, the peaks and durations of the Ca²⁺ elevations in response to depolarization were measured and compared in cells before and after the Ca²⁺ uniporter was blocked by the addition of 1 µM FCCP. Only data from cells stimulated with 50 mM K⁺ for 10 s were included in these analyses. The duration of the response was measured at half height to exclude the plateau phase data in the measurements. When mitochondrial Ca²⁺ uptake was impaired, we observed a significantly longer duration of Ca²⁺ elevations in response to FCCP (Fig. 1, A and B; normal, 47.1 ± 5.0 s; FCCP, 72.3 ± 6.6 s; n = 42; P < 0.0006). In amacrine cells that normally produced a plateau phase, inhibition of mitochondrial Ca²⁺ uptake always removed that component of the response (Fig. 1A). This is consistent with the plateau phase being generated by release of Ca²⁺ from mitochondria subsequent to uptake via the uniporter. In cells without a plateau phase, the response duration increased, and no other consistent effect on the waveform of the response was observed (Fig. 1B). In contrast to observations in other neurons (Friel and Tsien 1994; Thayer and Miller 1990; Werth and Thayer 1994), peak response amplitudes were not significantly different in the presence of FCCP [control, 183.6 ± 7.8% (of baseline values); FCCP, 189.6 ± 8.0%, P = 0.14; n = 41]. The underlying cause for this difference will be explored in a later section.

If mitochondria normally take up Ca²⁺ during depolarization, then application of FCCP after the depolarization should impair uniporter activity and shift the balance between uptake and release, resulting in a cytosolic calcium elevation. When FCCP was added after the depolarization, during the recovery phase, either generation or enhancement of a plateau phase was produced in all cells examined (Fig. 1C; n = 57). These results indicate that in response to depolarization, mitochondria in amacrine cells can sequester a significant amount of Ca²⁺ and that with some Ca²⁺ loads, subsequent release of mitochondrial calcium can have a substantial and long-lasting effect on cytosolic Ca²⁺ levels in these cells.

In some cells, an increase in Ca²⁺ levels was detectable in the presence of FCCP before addition of the K⁺ stimulus (Fig. 1B; arrow), indicating that inhibition of the mitochondrial uniporter alone results in cytosolic Ca²⁺ elevations. To test this, previously unstimulated cells were exposed to FCCP. By itself, FCCP generated cytosolic Ca²⁺ increases in 78% of cells examined (Fig. 1D; n = 51). On average, the peak amplitudes of FCCP-dependent Ca²⁺ elevations were 43.0% (±9.2; n = 23) of the peak amplitudes of Ca²⁺ elevations produced in the same cell by depolarization with 50 mM K⁺. These results imply that under resting conditions, mitochondria store a measurable amount of Ca²⁺ and/or that a constant leak of Ca²⁺ into the cytosol is normally offset by mitochondrial uniporter activity.

Relatively small Ca²+ loads can be buffered by mitochondria

With these measurements of cytosolic Ca²⁺, we have established that mitochondrial Ca²⁺ buffering plays a role in clearing cytosolic Ca²⁺ elevations in response to prolonged (10 s) depolarizations. In amacrine cells, we have the opportunity to measure Ca²⁺ influx through voltage-gated Ca²⁺ channels and to compare that to subsequent efflux on the plasma membrane Na-Ca exchanger (Gleason et al. 1995). Thus it is possible to determine whether mitochondria normally remove a fraction of a Ca²⁺ load imposed by relatively brief activation of voltage-gated Ca²⁺ channels by asking whether the fraction of the load removed by the exchanger is larger when mitochondrial Ca²⁺ uptake is inhibited. For these experiments, cells were voltage clamped in the perforated-patch configuration and depolarized from 70 to 0 mV for 700 ms. To estimate the fraction of the load removed by the exchanger, the inward current was integrated during the voltage step and during the subsequent exchange current (for an example, see Fig. 8B). The ratio of the charge moved on the exchanger to the charge entering during the voltage step was calculated (n = 7). As previously reported (Gleason et al. 1995), under control conditions, the fraction of a load removed by the exchanger varied considerably from cell to cell, ranging in value from 25 to almost 100%. For three of seven cells analyzed, blocking mitochondrial Ca²⁺ uptake shifted the normal balance and increased the fraction of the load removed by the Na-Ca exchanger (by 15, 42, and 70%, respectively). It is probably significant that the three cells whose ratios were shifted were also the three cells with the lowest control ratios, indicating that the plasma membrane Na-Ca exchanger normally played less of a role in these cells. These data suggest that in a subset of amacrine cells, mitochondrial Ca²⁺ buffering normally contributes to Ca²⁺ handling on this time frame.

Depolarization-induced Ca²+ current inactivation limits the amplitude of high-K⁺-dependent Ca²⁺ elevations

In our initial examination of the effects of inhibiting mitochondrial Ca²⁺ uptake, we found that depolarization in the presence of FCCP did not significantly increase the amplitude of Ca²⁺ elevations. This was unexpected because we would predict that normally, mitochondrial Ca²⁺ uptake would be removing some fraction of the Ca²⁺ entering while Ca²⁺ concentration is reaching its peak amplitude and that blocking this uptake would increase the amplitude of the Ca²⁺ elevation. However, a confounding factor for these cells is that the first high-K⁺ depolarization almost always produces a larger response than a second high-K⁺ depolarization in these cells, even in the absence of FCCP application (50 mM K⁺, 24 ± 0.76% reduction, n = 10; 100 mM K⁺, 50 ± 0.05% reduction, n = 10). This is may be due to the expression of primarily L-type voltage-gated Ca²⁺ channels in these cells that are susceptible to Ca²⁺-dependent inactivation. To examine this possibility, we attempted to mimic ten second applications of 50 and 100 mM K⁺ with a "test depolarization" in an electrophysiology experiment so that we could observe the effects of this sort of depolarization on the Ca²⁺ current amplitude.

Using ruptured patch whole cell recordings in the current-clamp configuration, we recorded the voltage changes engendered by either 50 or 100 mM K⁺ external solution. The averaged voltage responses to high K⁺ were used to establish the time course of the voltage excursions. Amplitude information from these recordings was disregarded because it is unlikely that we could mimic the conditions of an intact cell in this recording configuration. Instead we used peak amplitudes of 30 and 10 mV to approximate the predicted values derived from calculations of E_K in 50 and 100 mM K⁺ externals. The time course and predictions of voltage amplitude were used to construct the waveform of the test depolarization shown in Fig. 2C. To determine whether amacrine cell Ca²⁺ currents were being inactivated by pulses of high external K⁺, Ca²⁺ currents were recorded in the perforated-patch configuration before and after the test depolarization. Ca²⁺ currents were elicited by a voltage step from 70 to 10 mV for 250 ms (Fig. 2A). These voltage steps were applied once every 60 s. Thirty seconds after the second voltage step, the test depolarization (mimicking high K⁺ application) depicted in Fig. 2C was applied. Following this, the Ca²⁺ current was retested (once every 60 s) for 7 min to encompass the time span normally separating high K⁺ pulses in the fluo-3 experiments. As shown in Fig. 2B, the mean peak Ca²⁺ current amplitudes were initially reduced by ~50 and 70% by depolarizations to 30 and 10 mV, respectively. The degree of inhibition declined with time, but even 7 min after the test depolarization, the effects on the Ca²⁺ current amplitudes were substantial (30 mV protocol, ~14% inhibited; 10 mV protocol, ~25% inhibited). These results indicate that Ca²⁺ channel inactivation has the potential to mask an increase in amplitude of depolarization-dependent Ca²⁺ elevations in the presence of FCCP. Although the magnitude of Ca²⁺ influx is not the only factor determining free cytosolic Ca²⁺ concentration, we used our data on the time course of Ca²⁺ current inactivation to generate a correction factor to offset the effect of inactivation on the amplitude of the Ca²⁺ elevation. With this manipulation, we estimated what probably represents the upper limit of the role of mitochondrial Ca²⁺ uptake in determining the peak Ca²⁺ elevation engendered by depolarization. Using the value for sustained inhibition (average of last 3 data points for Ca²⁺ amplitude after 50 mM K⁺), we scaled the data collected with 50 mM K⁺ stimulation in the presence of FCCP. Re-analysis with scaled amplitude values gave FCCP values significantly larger than control (normal, 183.6 ± 7.8%; FCCP, 216.16 ± 9.2%, P < 0.0001, n = 41) suggesting that mitochondrial Ca²⁺ buffering normally does limit the amplitude of depolarization-induced Ca²⁺ elevations.

View larger version (18K): [in this window] [in a new window]   Fig. 2. Ten-second depolarizations have long-term inhibitory effects on calcium current amplitudes. A: data from a single amacrine cell recorded in the perforated-patch voltage-clamp configuration. The cell was held at 70 mV and stepped to 10 mV for 250 ms once every minute. Between a and b, the test depolarization depicted in C was applied to the cell to mimic application of high K+. a-c: obtained at the time points indicated in B. B: the time course for the changes in peak current amplitude is shown. , averaged data from 5 cells that were depolarized to 30 mV with the protocol in C. , averaged data from 4 cells that were depolarized to 10 mV with a protocol similar to the one shown in C. C: the voltage protocol applied and the inward current elicited in the same cell as shown in A.

Inhibition of the mitochondrial uniporter stimulates Na-Ca exchange in amacrine cells

Having established that mitochondrial Ca²⁺ buffering can shape the time course of Ca²⁺ elevations in amacrine cells, electrophysiological methods were used to examine whether Ca²⁺-dependent processes were also under the influence of mitochondrial function. Initial recordings from single, isolated amacrine cells revealed that application of 1 µM FCCP usually caused a reversible change in the holding current when cells were held at 70 mV (Fig. 3A). This current was not elicited by DMSO alone (n = 4). Two possible sources of the inward current were considered. One possibility was that FCCP either directly or indirectly activated a plasma membrane conductance. The other possibility was that the electrogenic plasma membrane Na-Ca exchanger (Gleason et al. 1995) was being activated by the FCCP-dependent increases in cytosolic Ca²⁺ that were seen for most cells in the Ca²⁺ measurement experiments.

View larger version (22K): [in this window] [in a new window]   Fig. 3. Inhibition of mitochondrial Ca2+ uptake activates the plasma membrane Na-Ca exchanger and reveals Ca2+ efflux from internal sources. A: a whole cell, perforated-patch recording from a representative isolated amacrine cell. Application of 1 µM FCCP induces an inward current that is sustained over the duration of FCCP application. B: in a different cell from A, the FCCP-dependent inward current is elicited by 1 µM FCCP. This current is abolished by subsequent replacement of external Na+ with Li+. The dependence of the inward current on external Na+ is consistent with the electrogenic activity of the plasma membrane Na-Ca exchanger. On removal of FCCP and replacement of Li+ with Na+, the inward current rebounds before returning to baseline. C: data from another isolated amacrine cell for which external Ca2+ was removed 30 s prior to application of FCCP. The presence of the FCCP-dependent Na-Ca exchange current is not dependent on the influx of Ca2+ across the plasma membrane. D: in the absence of any previous stimulus, application of 1 µM FCCP stimulates an elevation in cytosolic Ca2+. Prolonged application of FCCP (2.5 min) generates a Ca2+ elevation that is sustained well above baseline over the duration of exposure to FCCP. E: data are shown for a different amacrine cell that is first stimulated with FCCP (1.5 min) in normal external solution. After the initial Ca2+ transient, the external solution was changed to Ca2+-free external. Subsequent application of FCCP (1.5 min) also produced a Ca2+ elevation, indicating an intracellular Ca2+ source. The 2nd response was generally smaller that the first regardless of whether it was elicited in normal external or 0 mM Ca2+ external. F: data from a different cell in an experiment similar to that shown in B but where the first response was elicited in 0 Ca2+.

To determine whether the plasma membrane Na-Ca exchanger generated this FCCP-dependent current, the effect of replacing external Na⁺ with Li⁺, a manipulation well established to inhibit exchanger function, was examined. Replacement of external Na⁺ with Li⁺ reversibly abolished this inward current (Fig. 3B) in all cells examined (n = 15). On return to normal (Na⁺-containing, FCCP-free) external solution, we observed a transient increase in inward current before it returned to baseline values. The appearance of a transient current is consistent with the recovery of normal cytosolic Ca²⁺ levels once the plasma membrane Na-Ca exchanger has been re-enabled. These results strongly suggest that the FCCP-dependent inward current was due to the electrogenic activity of the plasma membrane Na-Ca exchanger.

If the FCCP-dependent current is due to the activity of the Na-Ca exchanger working to transport Ca²⁺ originating from the inside of the cell, then this current should persist in the absence of external Ca²⁺. In 0 mM Ca²⁺ external solution and with no previous Ca²⁺ load imposed, the FCCP-dependent inward current was still observed (Fig. 3C). When the FCCP-dependent current amplitude was measured in both normal and in Ca²⁺-free external in the same cell, there was only a 24 ± 10.5% (n = 5) reduction in current amplitude in the absence of external Ca²⁺. The persistence of this current in the absence of extracellular Ca²⁺ indicates that external Ca²⁺ is not the primary source of Ca²⁺contributing to the production of the current. Additionally, blockade of voltage-gated Ca²⁺ channels specifically with 200 µM Cd²⁺ was ineffective in reducing the amplitude of the FCCP-dependent current (not shown; FCCP, 28.3 ± 5.0 pA; FCCP/Cd²⁺, 27.8 ± 5.6 pA; n = 4, P = 0.7).

These data lead to two predictions that could be tested with Ca²⁺ measurements. First, because the inward current can remain above baseline for minutes in the continued presence of FCCP, we would expect that prolonged (>2 min) applications of FCCP would produce sustained rather than transient Ca²⁺ elevations in amacrine cells. This was tested by applying FCCP to cells for 2.5 min and measuring cytosolic Ca²⁺. Although the amplitude of the Ca²⁺ elevation declined in many cells over the 2.5-min period, Ca²⁺ did remain well above basal levels for the duration of the application in all cells examined (n = 19; Fig. 3D). The persistence of the FCCP-induced Ca²⁺ elevation indicates that the mitochondria themselves are probably not the sole source of the Ca²⁺.

A second prediction from our electrophysiology experiments would be that FCCP-dependent Ca²⁺ elevations could be generated in zero external Ca²⁺. For these experiments, FCCP was applied in the presence and absence of external Ca²⁺ (Fig. 3, E and F). Seven of nine cells examined produced Ca²⁺ elevations in the absence of external Ca²⁺. The lack of response in two of the cells is probably due to the loss of Ca²⁺ from internal stores that can occur fairly quickly when these cells are bathed in Ca²⁺-free external (not shown). The smaller amplitude of the FCCP response in zero Ca²⁺ shown in Fig. 3E does not necessarily reflect a large contribution from extracellular Ca²⁺ because when the zero Ca²⁺ responses were measured first, they were often (5 of 8 cells, Fig. 3F) larger than the responses elicited in normal external Ca²⁺. These results support a scenario where in the absence of uniporter activity, cytosolic Ca²⁺ concentration rises and stimulates an efflux of Ca²⁺ via the plasma membrane Na-Ca exchanger. Furthermore, the persistence of the cytosolic Ca²⁺ elevation in the absence of external Ca²⁺ indicates that the source of Ca²⁺ is largely internal.

Ca²+ elevations are not due to ATP depletion

Because FCCP dissipates the proton gradient across the inner mitochondrial membrane, ATP production via oxidative phosphorylation is also inhibited and, under these conditions, the ATPsynthase will consume ATP. It has been demonstrated in several neuronal cell types that the glycolytic pathway can maintain the ATP/ADP ratio for tens of minutes (Kauppinen and Nicholls 1986; Peng 1998; Werth and Thayer 1994; White and Reynolds 1995). Nonetheless, it was necessary to establish that the FCCP-dependent Ca²⁺ elevations we were detecting were not due to loss of ATP and inhibition of the Ca²⁺ ATPases. Oligomycin was used to directly inhibit the activity of the mitochondrial ATPsynthase. FCCP and oligomycin were co-applied to see whether, in the absence of ATP consumption, an FCCP-dependent rise in cytosolic Ca²⁺ would be observed (Colegrove et al. 2000). All cells producing Ca²⁺ elevations in response to FCCP produced similar Ca²⁺ elevations in response to FCCP in the presence of oligomycin (n = 21; Fig. 4A). To address whether inhibition of the ATPsynthase alone would generate Ca²⁺ elevations, oligomycin was applied to cells for 3 min (Fig. 4B). In these experiments, oligomycin-dependent Ca²⁺ elevations were never observed (n = 21), indicating that on the time frame of our experiments, ATP depletion alone was not sufficient to generate an elevation in cytosolic Ca²⁺. As an additional control, we recorded from amacrine cells in the ruptured-patch configuration with an internal solution containing 4 mM ATP and an ATP regeneration system. Under these conditions, all cells examined still exhibited the FCCP-dependent Na-Ca exchange current (n = 6, not shown) indicating that ATP depletion is not the source of this current.

View larger version (12K): [in this window] [in a new window]   Fig. 4. FCCP-dependent Ca2+ elevations are not due to ATP depletion. A: a previously unstimulated amacrine cell was exposed to FCCP in the presence of 1 µM oligomycin to determine if the rise in Ca2+ was due to ATP consumption by the ATP synthase. Even when the synthase was inhibited by oligomycin, FCCP still evoked an elevation in cytosolic Ca2+ in all cells tested (n = 21). Further stimulation with FCCP alone indicates that cells can produce a similar response to FCCP in the presence or absence of oligomycin. B: to determine if inhibition of ATP production alone produces an elevation in cytosolic Ca2+, amacrine cells were exposed to oligomycin only for 3 min. Over that time period, no elevation of calcium was seen, although subsequent cytosolic Ca2+ elevations were elicited in response to FCCP in all cells tested (n = 21). These data indicate that the FCCP-dependent Ca2+ elevations are not due to inhibition of ATP-dependent processes.

Inhibition of mitochondrial Ca²+ buffering stimulates exocytosis

We have demonstrated that disruption of mitochondrial Ca²⁺ uptake produces Ca²⁺ elevations in amacrine cells; however, because the imaging data were collected from cell bodies only and because the FCCP-dependent Na-Ca exchange current was recorded from the whole cell, these experiments do not address any site-specific mitochondrial function in amacrine cells. Specifically, we wanted to know if mitochondria were important at synaptic release sites, so we asked whether FCCP-induced Ca²⁺ elevations could stimulate exocytosis. We recorded from amacrine cells in contact with three to five other unclamped amacrine cells (Fig. 5). Previous studies have demonstrated that after ~8 days in culture, amacrine cells can make functional Ca²⁺-dependent GABAergic synapses with themselves (autapses) and with other amacrine cells that employ postsynaptic GABA_A receptors (Frerking et al. 1995; Gleason et al. 1993). When cells were held at 0 mV, synaptic currents (presumably mostly autaptic) were usually observed (Fig. 5A). In these experiments, the Cl equilibrium potential is estimated to be about 57 mV (internal solution B), and accordingly, the synaptic currents were outward at 0 mV. Exposure to FCCP generated a dramatic increase in synaptic activity. At 0 mV, no FCCP-dependent inward currents were seen; consistent with the voltage dependence of the Na-Ca exchanger previously described for amacrine cells (Gleason et al. 1995). When cells were held at 70 mV, quantal events were rare until FCCP was applied (Fig. 5B). For these experiments, a high Cl internal solution (solution A) was used that gave a calculated Cl equilibrium potential of 2 mV and thus inward synaptic currents at 70 mV. As expected from previous experiments, application of FCCP at this holding potential was accompanied by the inward Na-Ca exchange current. The increase in synaptic activity followed the development of the Na-Ca exchange current as would be expected if this current is tracking cytosolic Ca²⁺ levels near release sites. Another consequence of disrupting mitochondrial function is that acidification of the cytosol can occur, which has been shown to stimulate Ca²⁺-independent exocytosis (Drapeau and Nachshen 1988). To address this possibility, experiments were also conducted on cells loaded with 10 µM BAPTA-AM to minimize Ca²⁺ elevations. With BAPTA-loaded cells, FCCP failed to elicit exocytosis, indicating that the release stimulated by FCCP was Ca²⁺ dependent (n = 4, not shown).

View larger version (35K): [in this window] [in a new window]   Fig. 5. Inhibition of mitochondrial Ca2+ buffering stimulates exocytosis. A: a recording from a single amacrine cell in contact with other, unclamped amacrine cells. The cell was recorded with internal solution B and held at 0 mV. Prior to application of FCCP, quantal events are detectable and most likely arise from autaptic input stimulated by influx through voltage-gated Ca2+ channels. In the presence of FCCP, no inward current is generated which is consistent with the voltage-dependence of the Na-Ca exchanger. After FCCP application, the frequency of quantal events increases dramatically. Inset: a segment of the current (denoted by dotted lines) displayed in A. B: recording from another amacrine cell contacting other unclamped amacrine cells. This cell was held at 70 mV with solution B in the pipette. At 70, quantal events were rare but in the presence of FCCP, the inward exchange current develops and quantal events appear. Inset: a segment of data from this recording displayed on a longer time scale. These results indicate that in the absence of mitochondrial Ca2+ buffering, cytosolic Ca2+ levels increase in the presynaptic cells and exocytosis is either enhanced (A) or initiated (B), depending on the holding potential of the cell. Note that most of the delay for the increase in exchanger current is due to the dead space in the perfusion system (see METHODS). C and D: for this analysis, a data set from an amacrine cell connected to other, unclamped amacrine cells was selected for analysis on the following criteria. First, quantal events in normal solution had to be detectable but not so frequent that they were indistinguishable from one another. Second, a substantial Na-Ca exchange current had to be present (10 pA) to avoid counting events during solution changes. These data were recorded at 40 mV with solution A internally. C: peaks of quantal events were manually identified and counted from 20 discontinuous traces (5 s in duration) in normal external and from 15 traces in FCCP. *, a statistical difference (P = 0.0007, unpaired t-test) between normal and FCCP quantal frequency. D: peak quantal current amplitudes were measured from these same recordings. Peaks were measured only if the onset of the quantal current began at baseline to avoid considering multiple, nearly coincident quantal events as one. These data indicate that although the frequency of quantal events is increased in FCCP, their amplitude distribution is unchanged.

To determine if the effects of FCCP were entirely presynaptic, we analyzed the effects of FCCP on GABA-gated current amplitude. GABA-gated (100 µM) whole cell currents were not significantly affected by FCCP (normal, 75.4 ± 6.7 pA; FCCP, 70.2 ± 5.8 pA, P = 0.18, n = 5). To quantify the effects of FCCP on postsynaptic GABA_A receptors specifically, we analyzed quantal events recorded from an amacrine cell connected to other unclamped amacrine cells. The cell that provided the data in Fig. 5 (C and D) was selected for analysis because quantal events were frequent enough in normal external to generate a reasonable sample size but not so frequent as to preclude accurate measurements of individual quanta. First we confirmed that the frequency of the quantal events increased significantly in the presence of FCCP (Fig. 5C; P = 0.0007). We then analyzed the amplitude distribution of the quantal events and found that the distribution was unaffected by the presence of FCCP (Fig. 5D). These results indicated that the effects of FCCP were presynaptic.

These results indicate either that mitochondria normally keep Ca²⁺ concentration low at release sites or that in the absence of mitochondrial Ca²⁺ buffering, Ca²⁺ diffuses from elsewhere, possibly the cell body, out to active sites on the processes. To discriminate between these two possibilities, we collected fluorescence intensity data from cell bodies and multiple sites in the processes of six amacrine cells. In all cases, perfusion of FCCP generated Ca²⁺ elevations in cell bodies. In some cells (3 of 6, Fig. 6A), Ca²⁺ elevations were also seen at all of the sites recorded in the processes. The other three cells had both responding and nonresponding sites (Fig. 6B). Comparisons of the time courses of Ca²⁺ elevations in cell bodies and processes indicate that Ca²⁺ elevations in processes do not follow those measured in at the cell body but instead are initiated coincidently (Fig. 6A, inset). This, as well as the absence of Ca²⁺ elevations in some sites in processes, provides evidence that FCCP-dependent Ca²⁺ elevations occur locally and do not propagate. Given these results, we favor the hypothesis that mitochondria normally maintain low Ca²⁺ at presynaptic terminals and that disruption of this mechanism leads to exocytosis.

View larger version (21K): [in this window] [in a new window]   Fig. 6. Cytolosic Ca2+ elevations elicited by FCCP are local. Measurements of fluorescence intensity over time are shown for 2 fluo-3-loaded amacrine cells. Data were collected with a sampling rate of 28 Hz. A: a representative example of recordings from an amacrine cell body and one of its processes (*). FCCP was applied just prior to the beginning of the record and remained on for 1 min. Inset: the fluorescence data from the process have been scaled to the average of the values during the first 4 s of the recording at the cell body. · · · , the baseline. The increases in fluorescence begin at the same time in both the cell body and the process. B: fluorescence measurements from a different amacrine cell body and process (*). FCCP application is the same as in A. In this cell, the FCCP produced an elevation in the cell body but not at the site recorded from in the process. The data in A and B have been smoothed with 10 point adjacent averaging for display purposes.

Inhibition of mitochondrial Ca²+ uptake alters evoked synaptic transmission

The role of mitochondrial Ca²⁺ buffering during evoked synaptic transmission was examined by recording from isolated pairs of amacrine cells. For these experiments, both cells of a pair were voltage clamped and held at 70 mV. The presynaptic cell was depolarized to 10 mV by a 100-ms voltage step. In the presence of FCCP, the postsynaptic currents were reversibly reduced in amplitude (Fig. 7A). Because the currents are so noisy, a useful measure of synaptic transmission is to integrate the postsynaptic current to measure the charge transferred at the synapse. The postsynaptic currents at these synapses normally exhibit considerable variability in peak amplitude and duration (Gleason et al. 1993). Because of this intrinsic variability, we attempted to apply FCCP multiple times to each cell pair to confirm our observations. In one pair of amacrine cells, we were able to apply FCCP four times and on each occasion, the charge transfer was reduced (Fig. 7B). The effect of FCCP was examined on six cell pairs and in each case, we observed a reduction in the postsynaptic currents. On average, inhibition of the mitochondrial uniporter significantly reduced the charge transfer at the synapse (Fig. 7C; P = 0.0001). These results indicate that although loss of mitochondrial Ca²⁺ buffering can increase cytosolic Ca²⁺ and provoke exocytosis in an unstimulated cell, inhibition of this same process has the opposite effect during evoked synaptic transmission.

View larger version (28K): [in this window] [in a new window]   Fig. 7. Inhibition of mitochondrial Ca2+ buffering affects the magnitude of evoked synaptic transmission. A: postsynaptic currents recorded from an isolated pair of voltage-clamped amacrine cells. The presynaptic cell is held at 70 mV and stepped to 10 mV for 100 ms. The postsynaptic cell is held at 70 mV with internal solution A in the pipette. The FCCP-sensitive exchange current has been subtracted so that the evoked currents can be compared. In the presence of FCCP, the postsynaptic currents are reduced. B and C: because the postsynaptic currents are noisy and can be quite variable, the effect of FCCP was quantified by measuring the total charge of the postsynaptic current measured from the beginning of the presynaptic voltage step to the end of the record (~1.2 s) and normalizing this to the 1st record in normal external. In B, the normalized charge transfer is plotted over time for a cell pair that was recorded for ~15 min and had 4 exposures to FCCP (). Although the amount of charge transferred at the synapse was quite variable, FCCP was consistently inhibitory. C: normalized data for 6 cell pairs where postsynaptic currents recorded just prior to FCCP application are compared with postsynaptic currents recorded during FCCP application and for 4 cell pairs after washout of the FCCP. The normalization precludes a statistical comparison between normal and FCCP data but a comparison between FCCP and wash data shows them to be significantly different (P = 0.0002, unpaired t-test).

If inhibition of Ca²⁺ uptake into mitochondria was the sole outcome of FCCP application, then we might expect that in the presence of FCCP, presynaptic Ca²⁺ levels would be enhanced and that the charge transfer during synaptic transmission would be augmented rather than diminished during synaptic transmission. One implication of this inhibitory effect of FCCP is that FCCP is either directly or indirectly inhibiting some other aspect of synaptic function.

Inhibition of mitochondrial Ca²+ uptake inhibits voltage-gated Ca²⁺ currents

These amacrine cells primarily express L-type voltage-gated Ca²⁺ channels that have been previously demonstrated to control evoked synaptic transmission between these cells (Gleason et al. 1993). Because it is well established that L-type Ca²⁺ channels undergo Ca²⁺-dependent inactivation (Imredy and Yue 1992, 1994; Peterson et al. 1999; Zong et al. 1994), we considered the possibility that the Ca²⁺ elevations provoked by inhibition of mitochondrial Ca²⁺ uptake might be sufficient in magnitude to contribute to Ca²⁺-dependent inactivation of the channels. Unfortunately, recordings of presynaptic Ca²⁺ currents from pairs of amacrine cells during synaptic transmission are often contaminated by autaptic currents, thus compromising accurate measurements of current amplitude. However, in those amacrine cell pairs that had low levels of autaptic currents, FCCP-dependent decreases in presynaptic Ca²⁺ current amplitude were discernable. To confirm this, Ca²⁺ currents were recorded from single, isolated amacrine cells in the presence of 10 µM bicuculline to block GABA_A receptor activation at autapses during depolarization. For initial experiments, cells were stepped from 70 to 0 mV for 500 ms to activate voltage-gated Ca²⁺ channels. In all cases, application of FCCP produced inhibition of the voltage-dependent Ca²⁺ current (38 ± 8%, n = 8).

Inhibition of mitochondrial Ca²+ uptake also inhibits Ba²⁺ currents

It is well known that Ba²⁺ carries a larger current through L-type Ca²⁺ channels than Ca²⁺ does but is a poor substitute for Ca²⁺ on the plasma membrane Na-Ca exchanger. As expected, when Ba²⁺ was present externally, the voltage-dependent inward current was larger than in external Ca²⁺. However, the FCCP-dependent plasma membrane Na-Ca exchange current was not significantly enhanced (Ca²⁺, 45.3 ± 8.9 pA; Ba²⁺, 41.0 ± 6.5 pA; n = 4; P = 0.36; Fig. 8). This observation agrees with our experiments in Cd²⁺ and provides additional evidence that the FCCP-dependent inward current is not due to the flux of Ca²⁺ through L-type Ca²⁺ channels. In the presence of FCCP, the voltage-dependent inward Ba²⁺ current was also inhibited (Fig. 8, A and C), consistent with the hypothesis that cytosolic Ca²⁺ elevations are largely derived from internal stores and can be sufficient to inhibit these channels.

View larger version (17K): [in this window] [in a new window]   Fig. 8. Inward currents through voltage-gated Ca2+ channels are inhibited in FCCP. A-D: gray lines denote the current recorded in the presence of FCCP. These recordings were made in the perforated-patch recording configuration. A: in normal external containing 3 mM Ca2+, a 500-ms depolarization from 70 to 0 mV produced an inward current followed by a prolonged tail current that was due to the activity of the plasma membrane Na-Ca exchanger. In the presence of FCCP, the inward exchange current is present and the Ca2+ current is reduced in amplitude. Offsets in baseline current here and in C are due to the FCCP-dependent exchange current. B: the Na-Ca exchange tail current from A is shown. Here, the FCCP-dependent exchange current has been subtracted from the tail recorded in FCCP so that the portion of the exchange current due to the preceding depolarization could be isolated. The dotted line represents the level of the leak current in normal external. C: currents from the same cell were recorded in external solution with 3 mM Ba2+ replacing Ca2+. As expected, the Ba2+ currents are larger than the Ca2+ currents but in the presence of FCCP, the inward Na-Ca exchange current appears and the voltage-dependent Ba2+ current is reduced in amplitude. D: the Na-Ca exchange tail currents are depicted as in B but were greatly reduced in external Ba2+. The small tail currents that persist are presumably due to the accumulation of Ca2+ that occurs in the cytosol during the voltage step when the exchanger is inactive. This effect is larger when mitochondrial Ca2+ uptake is inhibited with FCCP.

Under normal conditions, when Ca²⁺ enters during a voltage step, the charge transferred during the subsequent plasma membrane Na-Ca exchange tail current is directly related to the charge transferred during the preceding voltage-dependent Ca²⁺ current. Accordingly, the Na-Ca exchange tail current in FCCP is smaller (after subtracting the FCCP-dependent component) than the Na-Ca exchange tail current in normal external (Fig. 8B). In external Ba²⁺, the Na-Ca exchange tail current is nearly abolished because Ba²⁺ is not easily transported on the exchanger. However, in the presence of FCCP, although the inward Ba²⁺ current is reduced, the Na-Ca exchange tail current is actually enhanced (Fig. 8D). This observation is consistent with FCCP-dependent Ca²⁺ accumulation (probably leakage from stores) occurring during the 500-ms voltage step to 0 mV when the Na-Ca exchanger is not working to transport Ca²⁺ out of the cell.

Ca²+ influx can influence the magnitude and time course of the FCCP-dependent inhibition of Ca²⁺ currents

To test the hypothesis that inhibition of mitochondrial Ca²⁺ uptake enhances Ca²⁺-dependent inactivation of Ca²⁺ current, we looked at the effects of FCCP over time with two different Ca²⁺-loading conditions. Cells were depolarized to 0 mV for either 100 or 700 ms. The onsets of the voltage step protocols were separated by 1 min. Regardless of the duration of the voltage step, the Ca²⁺ current was always suppressed in the presence of FCCP (Fig. 9; n = 15), and this effect persisted for minutes after FCCP was washed out of the bath. The duration of the effect indicated that the inhibition of the current is not a direct effect of FCCP on the channels because we know from our other experiments that FCCP washes out of the cells within seconds. When cells were depolarized for 100 ms, the effect of FCCP took longer to develop, was less pronounced, and had a shorter duration than when cells were depolarized for 700 ms (Fig. 9C). Thus the recent history of Ca²⁺ influx (previous control voltage steps) determined the magnitude and the time course of the FCCP effect, providing further evidence that cytosolic Ca²⁺ was mediating the inhibition. Control experiments done on the same time frame, either without any treatment (n = 3) or with pulses of 0.01% DMSO alone (n = 4), did not produce decreases in Ca²⁺ current amplitude (not shown). These results do not rule out the possibility that mitochondria normally play a role in shaping the time course of the presynaptic Ca²⁺ elevation during synaptic transmission. Instead, they point to an additional role of mitochondria in helping to offset the effects of Ca²⁺-dependent inactivation on the Ca²⁺ channels.

View larger version (12K): [in this window] [in a new window]   Fig. 9. FCCP-dependent Ca2+ current inhibition is enhanced by increased Ca2+ influx. For these experiments Ca2+ currents were elicited at 1-min intervals by depolarization from 70 to 0 mV for either 100 (A) or 700 (B) ms. Five identical voltage steps were delivered, separated by 5 s. Only data from the 1st of these voltage steps are shown in this figure. A and B: Ca2+ currents recorded from 2 different isolated amacrine cells. Recordings were made with solution B internal. For both recordings, the FCCP-dependent exchange current has been subtracted from the currents recorded at 70 mV but not during the voltage steps to 0 mV when the exchanger is known not to function. C: the effects of FCCP on peak Ca2+ current amplitude are shown over the time course of a 9-min (100-ms steps, n = 6) or 13-min (700-ms steps, n = 9) experiment. Approximately 1.5 min after removal of FCCP, the current amplitude began to recover; however, with this duration of FCCP application, complete recovery was usually not achieved.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Mitochondria and the control of resting cytosolic Ca²+

Our results indicate that regulation of Ca²⁺-dependent processes in amacrine cells, including exocytosis and voltage-gated Ca²⁺ channel function, depends on the Ca²⁺ uptake capabilities of mitochondria. Measurements of cytosolic Ca²⁺ concentration demonstrated that in the absence of a previously imposed Ca²⁺ load, FCCP stimulated a Ca²⁺ elevation in 78% of amacrine cells examined. This observation suggests that mitochondria store some Ca²⁺ at rest and that uniporter function is required to maintain this store. Because FCCP-dependent Ca²⁺ elevations are not usually transient in the continued presence of FCCP, it is unlikely that the mitochondria themselves are the sole source of Ca²⁺. One possibility is that in amacrine cells, mitochondrial Ca²⁺ uptake normally functions to offset a leak of Ca²⁺ into the cytosol. If mitochondria do normally offset a Ca²⁺ leak, then the relative insensitivity of this leak to removal of extracellular Ca²⁺ implies that its source is intracellular, possibly the endoplasmic reticulum (ER). Previous studies on Ca²⁺ transport into the ER and sarcoplasmic reticulum (SR) have demonstrated that elimination of trans-ER or -SR proton gradients with protonophores do not inhibit the uptake of Ca²⁺ into these organelles (Bode et al. 1994; Levy et al. 1990). These observations argue against the possibility that there is a direct effect of FCCP on Ca²⁺ transport into the ER. Instead, we propose that FCCP causes a release of mitochondrial Ca²⁺, and then, in the absence of mitochondrial buffering, the leak of Ca²⁺ from the ER elevates resting cytosolic Ca²⁺. In support of this hypothesis, preliminary experiments indicate that pretreatment of the cells with thapsigargin reduces the duration of FCCP-dependent Ca²⁺ elevations (unpublished observations).

In other preparations, mitochondrial Ca²⁺ uptake does not usually become significant until cytosolic Ca²⁺ levels reach ~500 nM (Friel and Tsien 1994; Herrington et al. 1996; Werth and Thayer 1994). A recent study in sympathetic neurons, however, demonstrated mitochondrial Ca²⁺ uptake at significantly lower (200-300 nM) cytosolic Ca²⁺ concentrations (Colegrove et al. 2000). Previous quantitative Ca²⁺ imaging of these amacrine cells demonstrated resting cytosolic Ca²⁺ levels between 50 and 100 nM (Borges et al. 1996), but it may be that mitochondria are sensing higher local Ca²⁺ levels. Consistent with this interpretation, both anatomical (Rizzuto et al. 1998; Simpson et al. 1997) and physiological interactions between mitochondria and IP₃ receptor sites on the endoplasmic reticulum have been demonstrated, and the physiological studies have suggested that mitochondria can buffer the local Ca²⁺ efflux from activated IP₃ receptors in nonneuronal cells (Csordás et al. 1999; Hajnóczky et al. 1999). As suggested by Duchen (1999), it might be that this local buffering is required to maintain Ca²⁺ levels in the permissive range for IP₃ receptor activation. The organization of mitochondria and ER in amacrine cells is unknown, but imaging of the subcellular architecture should be possible with a combination of antibodies and organelle-specific dyes.

Mitochondria as activity sensors

Because of the glutamatergic input that amacrine cells receive from bipolar cells, amacrine cells in the intact retina are potentially subjected to substantial and fairly long-lasting Ca²⁺ elevations. Two mechanisms have now been identified that make a significant contribution to recovery from Ca²⁺ elevations in amacrine cells: the plasma membrane Na-Ca exchanger (Gleason et al. 1994) and mitochondria. One benefit of using mitochondria as a Ca²⁺-buffering organelle would be that a Ca²⁺ elevation in mitochondria increases the activity of several mitochondrial dehydrogenases leading to an increase in NADH levels (Rizzuto et al. 1994; Rutter et al. 1998). As originally proposed by Denton and McCormack (1980), transmission of a Ca²⁺ signal to the mitochondria could allow metabolic activity to be matched to the current energy demands of the cell. Activation of glutamate receptors and depolarization produce an influx of Na⁺ as well as Ca²⁺. Transmission of the Ca²⁺ signal to mitochondria could ensure that sufficient ATP is available to power the Na-K ATPase, thus keeping Na⁺ levels in the cell low enough to maintain the Ca²⁺ export activity of the Na-Ca exchanger.

Mitochondria at the amacrine cell synapse

Our results also demonstrate that mitochondria are important regulators of cytosolic Ca²⁺ at the amacrine cell synapse. The ability of FCCP to induce Ca²⁺-dependent exocytosis from amacrine cells implies that normal mitochondrial function serves to limit the rate of exocytosis in the absence of presynaptic depolarization. Furthermore, we demonstrate this function for the first time at an inhibitory central synapse. FCCP- and Ca²⁺-dependent exocytosis has also been previously demonstrated at excitatory synapses between hippocampal neurons (Scotti et al. 1999) and at the frog neuromuscular junction (Alnaes and Rahamimoff 1975; Molgo and Pecot-Dechavassine 1988). Evidence for the importance of mitochondrial Ca²⁺ buffering in regulation of evoked synaptic transmission has also been found at several peripheral synapses including the frog (Alnaes and Rahamimoff 1975), lizard (David et al. 1998) and crayfish (Tang and Zucker 1997) neuromuscular junctions as well as in bullfrog sympathetic ganglia (Peng 1998). Additionally, synaptic activity has been shown to stimulate the Ca²⁺-dependent activation of a mitochondrial conductance, possibly the Ca²⁺ uniporter, at the squid giant synapse (Jonas et al. 1999). The experiments reported here have all been conducted at room temperature (24-27°C). David and Barrett (2000) have demonstrated that in mouse motor nerve terminals the role of mitochondria in presynaptic Ca²⁺ buffering is enhanced at higher, more physiological temperatures. Thus it is important to consider that the effects of mitochondrial Ca²⁺ buffering on synaptic transmission between amacrine cells may actually be underestimated with our recording conditions.

The full role of mitochondrial calcium uptake during evoked synaptic transmission between amacrine cells remains unresolved. From our initial examination of the waveforms of depolarization-induced Ca²⁺ elevations, we saw that the effects of inhibiting mitochondrial Ca²⁺ uptake depended on the time course of the initial control response (compare Fig. 1, A to B). If the time course of a depolarization-induced Ca²⁺ elevation at the cell body is a reflection of the time course at the synapse, then for cells like those in Fig. 1A, mitochondrial Ca²⁺ uptake could amplify synaptic transmission by trapping Ca²⁺ at the synapse, then re-releasing it into the presynaptic cytoplasm via the mitochondrial Na-Ca exchanger. For cells like the one shown in Fig. 1B, mitochondrial Ca²⁺ uptake shortens the duration of the Ca²⁺ elevation and could limit the time course of synaptic transmission.

Although we were unable to dissect the role of mitochondrial Ca²⁺ uptake during evoked synaptic transmission, we did observe that in the absence of mitochondrial Ca²⁺ buffering, Ca²⁺ currents in amacrine cells are more susceptible to inactivation. A similar observation has been reported for Ca²⁺ currents in pituitary gonadotrophs treated with CCCP (Kaftan et al. 2000). In chromaffin cells, Ca²⁺-dependent inhibition of L-type calcium channels is also augmented in the absence of mitochondrial Ca²⁺uptake (Hernández-Guijo et al. 2001). Because these amacrine cells employ L-type Ca²⁺ channels at their synapses (Gleason et al. 1993), it would be attractive to speculate that synaptic mitochondria are generally important at synapses containing Ca²⁺-sensitive voltage-gated Ca²⁺ channels. However, in another neuron that employs L-type channels at its synapses, the retinal bipolar cell, it has been demonstrated that mitochondria play only a minor role in Ca²⁺ buffering. Instead, at these terminals, the primary Ca²⁺-buffering mechanism is the plasma membrane ATPase (Zenisek and Matthews 2000). Anatomy is probably a factor in determining the relative importance of mitochondrial Ca²⁺ buffering at a particular synapse. For example, in bipolar terminals, mitochondria occur in a cluster near the end of the axon fairly distant from the synaptic ribbon (Dowling and Boycott 1966; von Gersdorff et al. 1996), whereas in profiles of presynaptic amacrine cell processes, mitochondria are generally situated closer to the release sites and can occupy more than up to one-half of the cross sectional area of the process (Dowling and Boycott 1966). Thus the mechanisms that regulate the distribution of mitochondria may play an important role in determining the functional properties of synaptic transmission between amacrine cells.