Control of secretion by mitochondria depends on the size of the local [Ca²⁺] after chromaffin cell stimulation

Introduction

It has been known for many years that mitochondria play a central role in energy metabolism and the production of ATP (Duchen, 1999). During cell activation, some mitochondria take up Ca²⁺ from local cytosolic [Ca²⁺] ([Ca²⁺]_c) microdomains generated by the activation of nearby Ca²⁺ channels located either at endomembranes or at the plasma membrane (Rizzuto et al., 1993, 1994; Brini et al., 1997). Using mitochondrially targeted aequorins with different Ca²⁺ affinities to measure mitochondrial [Ca²⁺] ([Ca²⁺]_M), we have shown (Montero et al., 2000) that activation of Ca²⁺ entry by high K⁺ depolarization or acetylcholine produced a heterogeneous response of [Ca²⁺]_M in chromaffin cells. Whereas half of mitochondria responded with moderate increases in [Ca²⁺]_M up to 2–3 µm, a very fast Ca²⁺ uptake was observed in the remainder; these mitochondria, probably located closer to the Ca²⁺ entry sites, reached near millimolar [Ca²⁺] in a few seconds. In addition, caffeine-induced release of Ca²⁺ from the endoplasmic reticulum (ER) also induced a fast Ca²⁺ uptake into the same population of mitochondria, suggesting that voltage-dependent Ca²⁺ channels, ryanodine receptors and mitochondria are colocalized.

These functional units could be relevant to determine the rate and extent of Ca²⁺ delivery to the secretory machinery. Our observations (Montero et al., 2000) are compatible with the idea that the subpopulation of mitochondria located nearby the plasmalemma observe and buffer Ca²⁺ from high [Ca²⁺] microdomains occurring near the inner mouth of plasma membrane Ca²⁺ channels. From the rate of Ca²⁺ uptake into these mitochondria, the presence of subplasmalemmal [Ca²⁺]_c microdomains of 20–40 µm was inferred. These concentrations are close to those reached at subplasmalemmal exocytotic sites during physiological cell depolarization of bovine chromaffin cells (Augustine & Neher, 1992; Neher, 1998). Mitochondria are well suited to control [Ca²⁺]_c in this range because of the high K_m of the mitochondrial Ca²⁺ uniporter (Uceda et al., 1995; Csordás et al., 1999; Montero et al., 2000). This large capacity and low affinity Ca²⁺ uptake mechanism might therefore serve the function of controlling the amount of Ca²⁺ available to trigger secretion. In fact, blockade of mitochondrial Ca²⁺ uptake using protonophores or mitochondrial poisons potentiated catecholamine secretion strongly (Giovanucci et al., 1999; Montero et al., 2000).

In this report we extend our study by taking advantage of the presence of different types of voltage-dependent Ca²⁺ channels in bovine chromaffin cells. Detailed pharmacological studies have found that bovine chromaffin cells contain different proportions of L-type, N-type and P/Q-type Ca²⁺ channels (López et al., 1994; Albillos et al., 1996; Lomax et al., 1997). Selective inhibition of these channels enabled the study of the conditions required for modulation of catecholamine secretion by mitochondria. Mitochondrial Ca²⁺ uptake starts when [Ca²⁺]_c reaches > 0.5 µm (Herrington et al., 1996; Colegrove et al., 2000); however, we show here that the fast mitochondrial Ca²⁺ uptake required to modulate secretion starts only at [Ca²⁺]_c > 4–5 µm. This sharp threshold, due to the sigmoidal dependence of the rate of mitochondrial Ca²⁺ uptake on [Ca²⁺]_c (Montero et al., 2000), could transform a relatively small difference in Ca²⁺ entry through the plasma membrane into a large difference in terms of mitochondrial Ca²⁺ uptake.

Materials and methods

Results

Depolarization of chromaffin cells using high K⁺ induces activation of plasma membrane voltage-dependent Ca²⁺ channels and Ca²⁺ entry. The contribution of different types of Ca²⁺ channels to this entry of Ca²⁺ has been estimated previously by measuring the effect of selective Ca²⁺ channel inhibitors on whole-cell, voltage-dependent Ca²⁺ currents (López et al., 1994; Albillos et al., 1996). The conclusion of these studies was that the Ca²⁺ current is carried 20% by L-type Ca²⁺ channels, 30% by N-type Ca²⁺-channels and 50% by P/Q-type Ca²⁺channels. Opening of Ca²⁺ channels produces a transient [Ca²⁺]_c peak, and the contribution of different types of Ca²⁺ channels to these peaks has also been estimated by measuring the effect of selective Ca²⁺ channel inhibitors on the high K⁺-induced [Ca²⁺]_c peaks measured with fura-2 (Lomax et al., 1997). One problem of this approach lies in the relatively high affinity for Ca²⁺ of fura-2, which provides little sensitivity at [Ca²⁺]_c above 2 µm. For that reason, in the experiments in Fig. 1 we used the lower Ca²⁺ affinity dye fura-4F, which has a Ca²⁺ sensitivity of up to 7 µm. In Fig. 1a, cells were sequentially stimulated with high K⁺ three times to obtain reproducible [Ca²⁺]_c increases. The cells were then treated with 3 µm nisoldipine (L-channel inhibitor) and 1 µmω-conotoxin GVIA (N-channel inhibitor) for 20 min. Subsequent stimulation with high K⁺ induced [Ca²⁺]_c peaks that were decreased by about 50% with respect to those of the controls. In four similar experiments, inhibition was 36 ± 6% (mean ± SΕΜ). Following this, cells were treated with 3 µmω-conotoxin MVIIC (P/Q-/N-channel inhibitor) for an additional 20-min period. This treatment essentially abolished the [Ca²⁺]_c peaks induced by the previous high K⁺ stimulations. Control experiments showed that high K⁺-induced [Ca²⁺]_c peaks were stable over periods of > 30 min of periodic stimulation every 2–5 min (data not shown; see also Alonso et al., 1999). In Fig. 1b, after the initial three control high K⁺-induced [Ca²⁺]_c peaks, cells were incubated with 3 µmω-conotoxin MVIIC for 20 min. This treatment decreased subsequent high K⁺-induced [Ca²⁺]_c peaks by nearly 80%. In four similar experiments, inhibition was 73 ± 7% (mean ± SΕM). These results are consistent with those obtained by using whole-cell patch-clamp techniques, and suggest that the [Ca²⁺]_c peak induced by high K⁺ pulses is constituted by 50–60% Ca²⁺ entry through P/Q-type channels, and about 20% through each of L- or N-type channels. Finally, in Fig. 1c, we show that the inhibition by ω-conotoxin MVIIC of high K⁺-induced [Ca²⁺]_c peaks was partially reversed by the selective and potent nondihydropyridine activator of L-type Ca²⁺ channels FLP64176 (McKechnie et al., 1989; Villarroya et al., 1999). The average [Ca²⁺]_c peak obtained in the presence of both the toxin and FPL64176 was 2.0 ± 0.1 µm compared to 0.88 ± 0.04 µm in the presence of the toxin alone (mean ± SEM, 53 single cells). The L-channel activator, however, produced little effect on the high K⁺-induced [Ca²⁺]_c peaks in control cells (3.8 ± 0.1 µm with FPL64176 and 3.4 ± 0.1 µm in the absence of FPL64176; mean ± SΕΜ, 53 single cells). This could be attributed in part to saturation of fura-4F (81% at this [Ca²⁺]), although we have been able to detect previously potentiation of high-K⁺-induced [Ca²⁺]_c peaks in these cells with the protonophore CCCP (Montero et al., 2001). Therefore, activation of Ca²⁺ entry through l-type Ca²⁺ channels probably contributes little to the global high K⁺-induced [Ca²⁺]_c peak, except when non-L-channels are inhibited.

We then studied the effect of these selective Ca²⁺ channel inhibitors on high K⁺-induced [Ca²⁺]_M increase. We expected effects proportional to the inhibition of the [Ca²⁺]_c peaks; however, the results were very different. Figure 2a shows that high K⁺ stimulation induced a fast and reversible mitochondrial Ca²⁺ uptake in control cells, as reported previously (Montero et al., 2000). This large mitochondrial Ca²⁺ uptake can only be measured using mitochondrially targeted low Ca²⁺ affinity mutated aequorin, reconstituted with coelenterazine n (AEQ3; Montero et al., 2000). This type of aequorin, originally designed to measure [Ca²⁺] in the ER (Montero et al., 1995, 1997), is able to measure [Ca²⁺] as high as 1 mm. The figure shows the effect of three consecutive stimulations, two 10 s K⁺ pulses and a final 30 s K⁺ stimulation. When cells were treated with 3 µm nisoldipine and 1 µmω-conotoxin GVIA (inhibition of L- and N-type Ca²⁺ channels), the high K⁺-induced [Ca²⁺]_M increase was not modified significantly (Fig. 2b). However, when cells were treated with 3 µmω-conotoxin MVIIC (inhibition of P-, Q- and N-type Ca²⁺ channels; Fig. 2c), the increase in [Ca²⁺]_M induced by 10 s K⁺ pulses was completely abolished, whereas a 30 s high K⁺ pulse produced a small response. Figure 2d also shows that the activator of L-type Ca²⁺ channels, FLP64176 restored mitochondrial Ca²⁺ uptake in the presence of ω-conotoxin MVIIC. The results shown in 1, 2 are combined and summarized as a bar chart in Fig. 3. The figure shows mean peak [Ca²⁺]_c-values (white bars) and [Ca²⁺]_M-values (black bars) under the different conditions.

We have shown previously (Montero et al., 2000) that the amplitude of the [Ca²⁺]_c hot-spot generated around mitochondria after cell stimulation can be estimated from the rate of increase in [Ca²⁺]_M. The rate was very similar in control cells and in cells treated with L- and N-type Ca²⁺ channel inhibitors (67 ± 2 µm/s, n = 17 and 63 ± 6 µm/s, n = 10, respectively, after a 30 s high K⁺ stimulation). These rates correspond to a local [Ca²⁺]_c of nearly 20 µm (Montero et al., 2000); by contrast, the rate of [Ca²⁺]_M increase induced by the same stimulus in cells treated with ω-conotoxin MVIIC was only 10 ± 1 µm/s (n = 7). From the dependence of the rate of [Ca²⁺]_M increase on the extramitochondrial [Ca²⁺] (Montero et al., 2000), we can calculate that mitochondria in this case were surrounded by a [Ca²⁺]_c of about 6 µm. Perfusion of FPL64176 restored the high rate of mitochondrial Ca²⁺ uptake induced by high K⁺ in ω-conotoxin MVIIC-treated cells (45 ± 4 µm/s, n = 6).

Wild-type aequorin reconstituted with coelenterazine n (AEQ2; Montero et al., 2000) is almost completely consumed after the first high K⁺ stimulation in mitochondria close to plasma membrane Ca²⁺ channels. Subsequent stimulation, not only with high K⁺ but also with caffeine, produces little further aequorin consumption (Montero et al., 2000). This effect is not due to emptying of the Ca²⁺ store of the ER. Direct measurements of [Ca²⁺] inside the ER showed that the steady-state level of [Ca²⁺] in this organelle was restored rapidly after any of these stimuli (Alonso et al., 1999). In fact, caffeine evokes cytosolic [Ca²⁺]_c peaks of normal height after both high K⁺ or caffeine stimulation (Alonso et al., 1999). The lack of effect on [Ca²⁺]_M with caffeine added as the second stimulus can therefore be explained only on the basis of fast aequorin consumption in a specific population (about 50%) of mitochondria. The rest of the mitochondria respond to high K⁺ or caffeine with much lower [Ca²⁺]_M peaks that hardly consume any AEQ2 (Montero et al., 2000). This implies that Ca²⁺ entering through plasma membrane Ca²⁺ channels induces a large Ca²⁺ uptake in the same mitochondria that undergo a large Ca²⁺ uptake after activation by caffeine of ryanodine receptors. Once aequorin in these mitochondria is completely consumed during the first high K⁺ stimulation, further addition of caffeine will not produce any further signal. This indicates the colocalization of plasma membrane Ca²⁺ channels and ER ryanodine receptors. Figure 4a and b shows this effect. If caffeine is added as the first stimulus, it produces a large [Ca²⁺]_M peak; if it is added 2 min after a 10 s pulse with high K⁺, it produces a much smaller effect. Cell treatment with nisoldipine (Fig. 4c), ω-conotoxin GVIA (Fig. 4d), or both together (Fig. 4e) did not modify this pattern of response. However, when cells were treated with ω-conotoxin MVIIC (Fig. 4f), the [Ca²⁺]_M peak induced by high K⁺ was reduced considerably. Consistently, subsequent addition of caffeine produced a large [Ca²⁺]_M peak similar to the control (Fig. 4a). The small increase in [Ca²⁺]_M induced by the 10 s high K⁺ stimulation is compatible with the apparent lack of effect of this stimulus shown in Fig. 2c, because of the low Ca²⁺-affinity of AEQ3. The higher resolution at low [Ca²⁺] obtained here using AEQ2 enabled the precise measurement of the rate of increase in [Ca²⁺]_M after a 10 s stimulation with high K⁺ of ω-conotoxin MVIIC-treated cells. The rate obtained (0.7 ± 0.4 µm/s) corresponds to a [Ca²⁺]_c around mitochondria of slightly > 4 µm (Montero et al., 2000).

Figure 5 shows mean data obtained in experiments similar to those of Fig. 4, using both 10 s (black bars) or 30 s (white bars) high K⁺ stimuli. The height of the bars corresponds to the percentage of aequorin consumed by the first addition of high K⁺, with respect to the total aequorin consumed by the two consecutive stimulations with high K⁺ and caffeine. These values show how good the inhibitors are at blocking Ca²⁺ entry close to both mitochondria and ryanodine receptors from the ER. The data confirm that inhibition of non-L-type channels specifically reduces Ca²⁺ entry close to the ryanodine receptors. A 10 s high K⁺ pulse consumed nearly 90% of high [Ca²⁺]_M aequorin in control cells, so that the subsequent caffeine-induced [Ca²⁺]_M peak consumed only 10%. Treatment with nisoldipine, ω-conotoxin GVIA or both did not modify these values. Instead, treatment with ω-conotoxin MVIIC dramatically reduced aequorin consumption induced by the initial 10 s high K⁺ pulse. Closed bars show similar data, but calculated from experiments where the high K⁺ pulse was 30 s. Regarding control cells, there is a barely significant increase in the percentage of AEQ2 consumption induced by high K⁺. Again here, treatment with nisoldipine, ω-conotoxin GVIA, or both had no effect. However, a 30 s stimulation with high K⁺ consumed more than half of the AEQ2 in subplasmalemmal mitochondria even in the presence of ω-conotoxin MVIIC.

The data in 1-5 show that mitochondria take up Ca²⁺ only when it enters the cell through P/Q-type Ca²⁺ channels or when Ca²⁺ entry through L-type channels is enhanced. If our hypothesis is correct and these functional units control exocytosis, then we should find a correlation between mitochondrial Ca²⁺ uptake and potentiation of secretion by protonophores; this was the case. Figure 6 compares secretion evoked by pulses of high K⁺ in the absence and in the presence of the protonophore CCCP (2 µm). Representative traces are shown in panel (a) and averages of several experiments are shown in panels (b and c). In the controls, peaks of catecholamine secretion were enhanced approximately twofold after 90 s of prior superfusion with CCCP. The enhancement by CCCP persisted in cells treated with either nisoldipine (3 µm) or nisoldipine plus ω-conotoxin GVIA (1 µm), but was prevented by treatment with ω-conotoxin MVIIC (3 µm; Fig. 6a and b). Therefore, CCCP potentiates secretion when Ca²⁺ enters through P/Q-type channels, but it is unable to do so when Ca²⁺ enters only through L-type channels. The question was now whether potentiation by CCCP would be restored by enhancing Ca²⁺ entry through L-type channels. To stimulate Ca²⁺ influx through these channels we enhanced their opening by using FPL64176. In the continued presence of 0.3 µm FPL64176, the secretory response to high K⁺ pulses increased 2–3-times and CCCP further enhanced the secretory response approximately twofold (Fig. 6c). In this case, pretreatment with ω-conotoxin MVIIC (right bars) did not modify secretion, and the response was still potentiated by CCCP.

Discussion

We have reported previously (Montero et al., 2000) the presence in bovine chromaffin cells of functional units composed of plasma membrane voltage-dependent Ca²⁺ channels, ER ryanodine receptors and mitochondria, associated to exocytotic sites. In this paper we have extended our study to provide further evidence for the role of mitochondria in the control of secretion, and to investigate the conditions in which this modulation takes place. We found that high K⁺ stimulation induces a [Ca²⁺]_c peak of about 4 µm, whereas a population of mitochondria senses a microdomain of 20 µm[Ca²⁺]. When non-L Ca²⁺ channels were inhibited with ω-conotoxin MVIIC, high K⁺ stimulation induced a [Ca²⁺]_c peak of only 1 µm and mitochondria sensed a local [Ca²⁺]_c microdomain of about 4 µm. This local [Ca²⁺]_c is below the threshold for fast mitochondrial Ca²⁺ uptake, and so little Ca²⁺ was taken up by mitochondria under these conditions. Consistently, CCCP did not potentiate catecholamine secretion in cells treated with ω-conotoxin MVIIC. The L-channel activator FPL64176 partially recovered the high K⁺-induced [Ca²⁺]_c response and fully restored mitochondrial Ca²⁺ uptake in the presence of ω-conotoxin MVIIC. Consistently, catecholamine secretion was increased and it again became sensitive to stimulation with CCCP. All of these results show a close correlation between mitochondrial Ca²⁺ uptake and potentiation of secretion by CCCP. In addition, the lack of effect of CCCP in ω-conotoxin MVIIC-treated cells demonstrates that its potentiation of stimulated secretion in control cells is not due to a nonspecific effect of this drug in the secretory machinery. Rather, it suggests that CCCP potentiates secretion in control cells by abolishing mitochondrial Ca²⁺ uptake.

The entry of Ca²⁺ through only P/Q-type channels (with L- and N-type channels inhibited) was enough to trigger fast mitochondrial Ca²⁺ uptake, which was indistinguishable from that obtained with all the channels active. Therefore, suppression of nearly 40% of Ca²⁺ entry did not significantly modify mitochondrial Ca²⁺ uptake. Consistently, catecholamine secretion was potentiated in a similar manner by CCCP under those conditions. These results could be interpreted to conclude that P/Q-type channels are more specifically coupled to the functional units containing mitochondria and ER ryanodine receptors, whereas L- or N-type channels would be situated further away. In fact, we have recently reported functional evidence suggesting that Q-type Ca²⁺ channels may be located closer to the secretory sites than L-type Ca²⁺ channels (Lara et al., 1998). Alternatively, these results may also be interpreted on the basis of the presence of a sharp threshold for fast mitochondrial Ca²⁺ uptake. The Ca²⁺ uptake by mitochondria has a sigmoidal dependence on [Ca²⁺]_c, so that it remains very slow until [Ca²⁺]_c rises to > 4 µm, and then the rate increases almost linearly until saturation is reached at [Ca²⁺]_c > 40–50 µm (Montero et al., 2000). This threshold might condition a nearly all-or-nothing response of mitochondrial Ca²⁺ uptake, depending on the percentage of inhibition of Ca²⁺ entry. High K⁺ stimulation generates a local [Ca²⁺]_c microdomain of about 20 µm, estimated from the rate of mitochondrial Ca²⁺ uptake (Montero et al., 2000). Assuming that the magnitude of that microdomain is proportional to Ca²⁺ entry, 70% inhibition of Ca²⁺ entry would give microdomains of 6 µm, the local [Ca²⁺]_c we estimate after 30 s high K⁺-stimulation in the presence of ω-conotoxin MVIIC (Fig. 2). If the local [Ca²⁺]_c decreases any further, the rate of [Ca²⁺]_M increase drops dramatically. 2, 3 show that this rate decreases 20-fold when the local [Ca²⁺]_c is reduced from 6 to 4 µm. Therefore, 40% inhibition (that expected from L- and N-type channel inhibition) of Ca²⁺ entry might produce relatively small changes in mitochondrial Ca²⁺ uptake but 70–80% inhibition could nearly abolish it. This alternative view is also reinforced by several findings. First, potentiation of Ca²⁺ entry through L-type Ca²⁺ channels with FPL64176 restored both mitochondrial Ca²⁺ uptake and stimulation by CCCP of catecholamine release in ω-conotoxin MVIIC-treated cells. Second, FPL64176 strongly increased high K⁺-induced secretion in control cells, whereas the [Ca²⁺]_c peak was modified very little, and this stimulated secretion was hardly inhibited by ω-conotoxin MVIIC, although [Ca²⁺]_c peaks were reduced by 37% (from 3.8 µm to 2.4 µm). These puzzling data reflect a dissociation between the magnitude of global [Ca²⁺]_c peaks and the secretory rate, and could be interpreted as meaning that activated L-type channels are also closely coupled to secretion in these cells.

In conclusion, although we cannot exclude the possibility that a particular type of Ca²⁺ channel could have a preferential spatial relationship with the functional units, our data stress the point that mitochondrial fast Ca²⁺ uptake has a very sharp threshold, at around 5 µm[Ca²⁺]_c. Although mitochondrial Ca²⁺ uptake starts at much lower [Ca²⁺]_c values (about 0.5 µm; see Herrington et al., 1996; Colegrove et al., 2000), translation of [Ca²⁺]_c changes into [Ca²⁺]_M changes is not linear. Below 4 µm[Ca²⁺]_c, the rate of mitochondrial Ca²⁺ uptake is low (< 1 µm/s) but at 7 µm[Ca²⁺]_c, the rate increases to about 20 µm/s, and rates above 100 µm/s can be obtained when the local [Ca²⁺]_c reaches 40 µm (Montero et al., 2000). If [Ca²⁺]_c changes do not reach 4–5 µm, the accompanying changes in [Ca²⁺]_M are relatively small, although they may be enough to activate mitochondrial metabolism. Instead, if the local [Ca²⁺]_c reaches over these values, fast [Ca²⁺]_M uptake is triggered and mitochondria could act as a Ca²⁺ sink. As a consequence, mitochondria can modulate secretion (or, in general, subplasmalemmal [Ca²⁺]_c) only when the stimuli raise the local [Ca²⁺]_c above that threshold. The local [Ca²⁺]_c known to be required for catecholamine secretion in chromaffin cells or for neurotransmitter secretion in neurons are well above that threshold (Neher, 1998). Even though these large [Ca²⁺] are very short-lasting, [Ca²⁺]_M could increase in an accumulative way during a train of depolarizing pulses because the release of Ca²⁺ from mitochondria is relatively slow (Montero et al., 2000). Physiological stimulation of the adrenal gland by acetylcholine generates spikes of action potentials, and a burst of action potentials may be evoked in response to repetitive discharges of acetylcholine from the splanchnic nerve. This means that mitochondria are perfectly suited to take up Ca²⁺ and modulate secretion in response to physiological stimuli. Whether, and to what extent, this modulation actually operates under physiological conditions will require further study.

Acknowledgements

Financial support from Dirección General de Enseñanza Superior to J.A. (PM98/0142) and J.G.-S. (PB97/0474), from Dirección General de Investigación Científica y Técnica (PB94/0150) and Janssen-Cilag to A.G.G., and from Junta de Castilla y León to J.A. (VA19/99) and to M.T.A. (VA62/99) is gratefully acknowledged. I.C.I. and A.A. hold fellowships from Ministerio de Educación y Ciencia.

Abbreviations