Perinuclear, perigranular and sub-plasmalemmal mitochondria have distinct functions in the regulation of cellular calcium transport

Abstract

We have identified three distinct groups of mitochondria in normal living pancreatic acinar cells, located (i) in the peripheral basolateral region close to the plasma membrane, (ii) around the nucleus and (iii) in the periphery of the granular region separating the granules from the basolateral area. Three-dimensional reconstruction of confocal slices showed that the perigranular mitochondria form a barrier surrounding the whole of the granular region. Cytosolic Ca²⁺ oscillations initiated in the granular area triggered mitochondrial Ca²⁺ uptake mainly in the perigranular area. The most intensive uptake occurred in the mitochondria close to the apical plasma membrane. Store-operated Ca²⁺ influx through the basolateral membrane caused preferential Ca²⁺ uptake into sub-plasmalemmal mitochondria. The perinuclear mitochondria were activated specifically by local uncaging of Ca²⁺ in the nucleus. These mitochondria could isolate nuclear and cytosolic Ca²⁺ signalling. Photobleaching experiments indicated that different groups of mitochondria were not luminally connected. The three mitochondrial groups are activated independently by specific spatiotemporal patterns of cytosolic Ca²⁺ signals and can therefore participate in the local regulation of Ca²⁺ homeostasis and energy supply.

Introduction

Much recent Ca²⁺ signalling work has dealt with the importance of the mitochondria. Contrary to conclusions from earlier work, it is now clear that mitochondria can and do take up Ca²⁺ during normal Ca²⁺ signalling events and then release it much more slowly (Pozzan et al., 1994, 2000; Duchen, 2000; Rutter and Rizzuto, 2000). Ca²⁺ in the mitochondria is of importance for the regulation of metabolism, since three dehydrogenases of the Krebs cycle are modulated by the Ca²⁺ concentration in the micromolar range (Denton and McCormack, 1990; McCormack et al., 1990). Repetitive cytosolic Ca²⁺ spikes induce mitochondrial Ca²⁺ spikes. Each spike is sufficient to cause a maximal transient activation of the Ca²⁺-sensitive mitochondrial dehydrogenases and above a certain frequency there is sustained activation of mitochondrial metabolism (Hajnoczky et al., 1995). Agonist stimulation, evoking cytosolic and mitochondrial Ca²⁺ signals, elicits increases in both the mitochondrial and cytosolic ATP concentrations that depend on both the amplitude of the Ca²⁺ rise inside the mitochondria and the availability of mitochondrial substrates. Thus, mitochondrial Ca²⁺ plays a direct role in driving ATP production (Jouaville et al., 1999). Mitochondrial Ca²⁺ uptake is also important for the regulation of other processes. In chromaffin cells, stimulation triggers fast millimolar mitochondrial Ca²⁺ transients and these can modulate secretion (Montero et al., 2000). In pancreatic acinar cells, active mitochondria localized in a belt surrounding the granular region are important for confining cytosolic Ca²⁺ signals initiated in this region to this part of the cell, where secretion is regulated (Tinel et al., 1999; Straub et al., 2000). It has been shown recently that respiring mitochondria determine the pattern of activation and inactivation of the store-operated Ca²⁺ inflow that is of importance for refilling endoplasmic reticulum (ER) stores with Ca²⁺ lost during agonist stimulation (Gilabert and Parekh, 2000).

It has been proposed that the localization of mitochondria with respect to Ca²⁺ sources can be a determinant of their Ca²⁺ uptake kinetics (Lawrie et al., 1996). It would be desirable to test this critically in intact polarized cells, where different groups of mitochondria could be clearly identified in relation to other subcellular structures and in particular in relation to specific subcellular Ca²⁺ release sites. Here we have investigated, in normal polarized pancreatic acinar cells, how mitochondria positioned in specific subcellular locations respond to various patterns of cytosolic Ca²⁺ ([Ca²⁺]_i) signals. We identified three distinct groups of mitochondria located (i) in the peripheral basolateral region very close to the plasma membrane, (i) around the nucleus and (iii) in the periphery of the granular region separating the apical granular area from the basolateral part of the cell. When [Ca²⁺]_i was increased uniformly throughout the cell, by global uncaging of caged Ca²⁺, we found no regional differences in mitochondrial Ca²⁺ uptake. In experiments where we used UV laser-induced local bleaching of an intra-mitochondrial dye, we saw no evidence of luminal connectivity between mitochondria in different regions. [Ca²⁺]_i oscillations initiated in the granular area triggered mitochondrial Ca²⁺ uptake mainly in the perigranular area. Ca²⁺ influx through the basolateral membrane, generated by emptying intracellular Ca²⁺ stores, elicited Ca²⁺ uptake into nearby peripheral basolateral mitochondria. Perinuclear mitochondria were specifically activated by local Ca²⁺ release in the nucleus. Distinct groups of separate mitochondria, due to their strategic positioning within the cell, can therefore sense and shape [Ca²⁺]_i signals locally to help the cell regulate activities in specific subcellular compartments and ensure adequate local energy supply.

Results

The relative positions of granules, mitochondria, ER and nucleus in the living pancreatic acinar cells

The polarization of pancreatic acinar cells with the ER and the nucleus in the basolateral part and the granules in the apical pole was well maintained even after isolation of single cells, as shown in Figure 1. There was heavy staining with ERtracker in the basolateral areas surrounding the granular region, except in the nucleus (n = 6) (Figure 1A). The granules were localized in the apical pole close to the tiny apical (luminal) membrane. High resolution confocal images of NADH autofluorescence showed the locations of the mitochondria, which were predominantly grouped in a central perigranular area and in a peripheral area close to the plasma membrane (n = 26) (Figure 1B). This is similar to what we have shown previously with the help of MitoTracker Green (Tinel et al., 1999) and tetramethyl rhodamine ethyl ester (Raraty et al., 2000) fluorescence. In some experiments we used both MitoTracker Green FM, to check the mitochondrial distribution, and Hoechst 33342, to visualize the nuclei, simultaneously. In these cells we saw that there was also a group of mitochondria close to and sometimes surrounding the nucleus (Figure 1C) (n = 10).

Fig. 1. Localization of endoplasmic reticulum (ER), nuclei and mitochondria in living pancreatic acinar cells. (A) Endoplasmic reticulum. The transmitted light picture of a single isolated acinar cell is shown in (a). In (b), the fluorescence image obtained with ERtracker is shown. The ER is densely packed outside the granular and nuclear areas. (B) Mitochondria. Transmitted light picture in (a) and autofluorescence (NADH) image in (b). The mitochondria are localized principally as a belt surrounding the granular region and as a peripheral ring just under the plasma membrane. Lengths of red bars in (A, a) and (B, a) represent 10 µm. (C) Nuclei and mitochondria. (a) Transmitted light picture showing the typical structure of two connected acinar cells with the secretory (zymogen) granules in the central part surrounding what in the intact organ would be the lumen. The nuclei are stained red with Hoechst 33342 and the fluorescence image is superimposed on the transmitted light picture. In the cell on the left, the focal plane goes through one nucleus, but it is just possible to detect the presence of a second nucleus. (b) MitoTracker Green fluorescence image. The strongest staining is localized as a ring surrounding the granular area, with some staining at the periphery as well as around the three nuclei (two in the left and one in the right cell). The mitochondria surrounding the nuclei seem to be aligned along the surfaces of these organelles.

Mitochondrial Ca²⁺ uptake following acetylcholine (ACh)-elicited release of Ca²⁺ from the endoplasmic reticulum

To image mitochondria and measure the mitochondrial Ca²⁺ concentration ([Ca²⁺]_m), we used the Ca²⁺-sensitive dye Rhod-2, which is concentrated in compartments with a highly negative voltage (Pinton et al., 2001). The transmitted light picture of a cell under investigation is shown in Figure 2A. The mitochondrial distribution is disclosed in Figure 2B. After exposing the cell to Rhod-2-AM, the fluorescence intensity was generally weak, in the absence of stimulation, as shown in Figure 2C1. However, after ACh (10 µM) application the characteristic localization of mitochondria was revealed gradually and finally the Rhod-2 fluorescence image (Figure 2C6) was similar to the picture obtained imaging the NADH autofluorescence (Figure 2B). This suggests that the initial low Rhod-2 fluorescence intensity was due to the relatively low resting [Ca²⁺]_m and not caused by insufficient mitochondrial dye loading. Mitochondrial Ca²⁺ uptake occurred, at first, in the lateral area close to the apical membrane and thereafter spread to the basolateral part of the cell (Figure 2C and D). Mitochondrial Ca²⁺ uptake was rapid and the maximal [Ca²⁺]_m was reached within ∼5 s as seen in Figure 2D, which shows the changes of fluorescence intensity measured in three differently located mitochondria marked in Figure 2A (black, red and green circles). The average time lag between the mitochondrial Ca²⁺ rise in the lateral area close to the apical membrane (black trace) and at the base of the cell (green trace) was 2.4 ± 0.2 s (n = 7).

Fig. 2. Mitochondrial Ca²⁺ uptake, measured with Rhod-2, following maximal ACh (10 µM)-elicited Ca²⁺ release from the endoplasmic reticulum. (A) Transmitted light picture showing the cell under investigation, with the three areas of interest identified by the three coloured circles. The length of the red bar represents 10 µm. (B) Autofluorescence image showing mitochondrial localization. (C) Six images showing mitochondrial Ca²⁺ concentration. Image 1 shows the situation just before the start of stimulation. Image 2 shows that immediately after ACh application, there was Ca²⁺ uptake into mitochondria very close to the apical membrane. A little later, the whole perigranular mitochondrial belt was revealed (images 3 and 4) and finally (images 5 and 6) all mitochondria in the cell had taken up Ca²⁺. (D) The time course of the mitochondrial Ca²⁺ uptake in the three regions identified in (A) using the appropriate colour coding. The main part of the figure represents a short time segment of the more complete traces shown in the inset. The time scale relates to the expanded traces labeled with the numbers 1–6, corresponding to the images in (C).

In general, the mitochondrial Ca²⁺ uptake in response to an ER-derived rise in the cytosolic Ca²⁺ concentration has been shown to occur immediately (Rizzuto et al., 2000). However, in a recent study on pancreatic acinar cells, it has been reported that there is a substantial delay of ∼10 s, after the initiation of the cytosolic Ca²⁺ signal, before a measurable rise in the mitochondrial Ca²⁺ concentration is seen (Gonzalez et al., 2000). In view of these surprising results, we carried out experiments in which changes in the Ca²⁺ concentration in the cytosol and the mitochondria were monitored simultaneously, by measurement of fluo-4 and Rhod-2 fluorescence, in order to evaluate how quickly the mitochondrial Ca²⁺ uptake mechanism responded to the cytosolic Ca²⁺ signal. We saw no evidence of a substantial delay between the initiation of the cytosolic and mitochondrial Ca²⁺ signals. Figure 3 shows the result from one of these experiments. ACh elicited a cytosolic Ca²⁺ wave, which was initiated in the granular pole and then moved towards the base and reached the basal region within ∼2 s. As seen in Figure 3, the initial rise in [Ca²⁺]_m coincided with the initial rise in [Ca²⁺]_i, although the peak [Ca²⁺]_m was attained later than the peak [Ca²⁺]_i. Six experiments of this type gave similar results.

Fig. 3. Simultaneous measurements of cytosolic (fluo-4) and mitochondrial (Rhod-2) Ca²⁺ concentration changes after ACh stimulation. (A) The transmitted light picture shows the structure of the acinar doublet with the three colour-coded circles indicating the regions of interest. (B) The time courses of the ACh-elicited changes in the cytosolic Ca²⁺ concentrations in the granular (blue) and basal (red) regions, as well as the mitochondrial Ca²⁺ concentration in the perigranular region (green). In the inset below, showing the initial part of the three curves on an expanded time scale, it is seen that the mitochondrial Ca²⁺ uptake started immediately after the initial rise in the cytosolic Ca²⁺ concentration in the granular area.

The ER is the main Ca²⁺ storage organelle as well as a main supplier of Ca²⁺ to the mitochondria. Therefore, we measured [Ca²⁺]_m, the Ca²⁺ concentration in the ER ([Ca²⁺]_ER) and [Ca²⁺]_i simultaneously in the same cell. To do this, we exposed cells to Rhod-2-AM and the low affinity Ca²⁺-sensitive dye Mag-fluo-4-AM (K_d = 22 µM). Thereafter, we introduced a patch pipette (whole-cell recording configuration) to wash out the cytosolic dye components. The time course of changes in [Ca²⁺]_i was assessed by measurement of the Ca²⁺-dependent current (Petersen, 1992). Since the ER is mainly located in the basolateral part of the cell (Figure 1A), the decrease in [Ca²⁺]_ER was observed mainly in the basolateral region (Figure 4), as demonstrated previously (Park et al., 2000). As discussed extensively (Park et al., 2000; Petersen et al., 2001), this is entirely consistent with the well established finding that the [Ca²⁺]_i rise is initiated in, and often confined to, the apical granular pole. The Ca²⁺ release occurs from very tiny extensions of the basolateral ER into the apical pole, which is dominated by granules. The [Ca²⁺]_m change was mainly in the perigranular (particularly marked in the lateral area close to the apical membrane) and peripheral areas (Figure 4A, a). The Mag-fluo-4 fluorescence intensity in the perigranular (mitochondrial) area was initially slightly increased after application of ACh, which could be due to some of the dye being trapped in the mitochondria. This phenomenon was abolished by application of mitochondrial metabolic blockers (1 µM oligomycin and 1 µM rotenone; n = 6) or dialysis of a 10 mM BAPTA/2 mM Ca²⁺ mixture through the patch pipette (n = 17). We plotted the time course of the ACh-elicited changes in [Ca²⁺]_i (Ca²⁺-dependent Cl^– current), [Ca²⁺]_m (Rhod-2 fluorescence) and [Ca²⁺]_ER (Mag-fluo-4 fluorescence) together in Figure 4B. After ACh stimulation, [Ca²⁺]_i and [Ca²⁺]_m increased immediately, even before there was a measurable decrease in [Ca²⁺]_ER. After washout of ACh, [Ca²⁺]_i returned much more quickly to the resting level than [Ca²⁺]_ER and [Ca²⁺]_m. The mitochondria released Ca²⁺ and the ER refilled with Ca²⁺ without provoking any Ca²⁺-dependent current, indicating the possibility of Ca²⁺ recycling between mitochondria and nearby ER elements. The mitochondrial Ca²⁺ release following the Ca²⁺ uptake was very slow (see Figure 4). This phenomenon was similar in intact Rhod-2-loaded cells (n = 12) and cells dialysed via a patch pipette (n = 6). The time course was somewhat variable from cell to cell, but usually it took >10 min to restore [Ca²⁺]_m to the pre-stimulation level after a period of supramaximal ACh (10 µM) stimulation. The mitochondrial Ca²⁺ uptake was particularly marked in the lateral area close to the apical membrane (Figure 4A, a), which is very close to the primary ER Ca²⁺ release site (trigger zone) (Kasai et al., 1993; Mogami et al., 1997; Cancela et al., 2000). Also, after washout of ACh, the Ca²⁺ extrusion was particularly slow in these mitochondria.

Fig. 4. Simultaneous measurements of the ACh-elicted Ca²⁺ concentration changes in the ER and the mitochondria. (A) Images from a single cell. (a) Rhod-2 fluorescence images taken at 132.6 s intervals showing the accumulation and subsequent loss of Ca²⁺ from the mitochondria following a short period of ACh stimulation. The first image was obtained just before the start of stimulation with ACh. (b) Mag-fluo-4 fluorescence images showing the gradual loss of Ca²⁺ from the ER in response to ACh stimulation and the subsequent re-accumulation. The first image was taken just before the start of stimulation. The interval between images was 132.6 s. (B) Time course of ACh-elicited changes in the Ca²⁺ concentrations in the mitochondria (Rhod-2, red), the ER (Mag-fluo-4, blue) and the cytosol (whole-cell Ca²⁺-dependent current, black).

In the pancreatic acinar cell, it is well known that low doses of ACh or cholecystokinin (CCK) generate [Ca²⁺]_i oscillations, which always start close to the apical membrane and are mostly confined to the granular area (Thorn et al., 1993). It has been suggested that the perigranular mitochondria provide a Ca²⁺ buffer barrier, which prevents or reduces the degree of Ca²⁺ signal spreading to the basolateral region (Tinel et al., 1999). However, so far there has been no direct demonstration of Ca²⁺ uptake into the perigranular mitochondria during Ca²⁺ spiking. We therefore evoked [Ca²⁺]_i oscillations with a relatively low dose of ACh and observed [Ca²⁺]_m changes in different parts of the cell. To monitor the [Ca²⁺]_i oscillations we measured the Ca²⁺-dependent current. In the experiment illustrated in Figure 5, ACh (50 nM) elicited repetitive [Ca²⁺]_i spikes. The individual spikes were initially short-lasting and largely confined to the apical granular area (i.e. large cytosolic Ca²⁺ concentration gradient between apical and basolateral regions during the spike), but gradually became broader and global. The initial cytosolic Ca²⁺ spike was associated with a marked Ca²⁺ uptake into the perigranular mitochondria. In the Rhod-2 fluorescence images shown in Figure 5, the mitochondrial perigranular belt was invisible before the first ACh-elicited Ca²⁺ spike (image 1), but at the height of the spike the perigranular mitochondrial belt was clearly delineated (image 2). After the spike, the image of the perigranular belt disappeared (image 3). At the height of the first spike there was little sign of Ca²⁺ uptake into the mitochondria in other parts of the cell. Later, when the cytosolic Ca²⁺ spikes became broader and global, Ca²⁺ accumulated both in the perigranular belt and in the peripheral sub-plasmalemmal region (image 4). This was due to the relatively slow Ca²⁺ liberation from the mitochondria. During each spike, Ca²⁺ was taken up and then subsequently released, but the release was so slow that when the next spike occurred, all Ca²⁺ taken up during the previous spike had not yet been liberated. In this phase, the cytosolic Ca²⁺ spike was associated with mitochondrial Ca²⁺ uptake both in the perigranular belt and in the peripheral region (image 5). There was a visible, but minor, reduction in the mitochondrial Ca²⁺ concentration following the spike (image 6). Finally, a high concentration of ACh elicited a marked increase in the fluorescence intensity in the perigranular belt, the peripheral region and near the nucleus (image 7). Five experiments of this type all gave similar results.

Fig. 5. Region-specific mitochondrial Ca²⁺ uptake following cytosolic Ca²⁺ oscillations. The coloured traces in the main panel show the time courses of the Ca²⁺ concentration changes in mitochondria localized as shown in the transmitted light image. The colour coding of the traces corresponds to the coloured circles in the transmitted light picture. The length of the red bar in the transmitted light picture corresponds to 10 µm. The black trace in the main panel represents the Ca²⁺-dependent whole-cell current, which shows the time course of the changes in the cytosolic Ca²⁺ concentration. Stimulation with a low ACh concentration elicited repetitive cytosolic Ca²⁺ spikes. The Rhod-2 fluorescence images taken before, during and after the first ACh-elicited cytosolic Ca²⁺ spike (images 1–3) show that the main mitochondrial Ca²⁺ uptake during this spike occurred in the perigranular region. Since the release of Ca²⁺ taken up into the mitochondria is relatively slow, the mitochondria did not liberate all the Ca²⁺ accumulated during a single spike before the next spike occurred. Therefore, during this experiment, Ca²⁺ accumulated gradually in the mitochondria. In relation to the last spike in the series, the corresponding Rhod-2 fluorescence images are shown (images 4–6). It can be seen that even before the spike started there was considerable fluorescence both from the most apically located mitochondria in the perigranular ring and also from the peripheral ring close to the plasma membrane. During the spike there was increased fluorescence in all mitochondrial regions, which then declined after the spike. Finally, the cell was stimulated maximally by a high ACh concentration, causing a marked increase in the cytosolic Ca²⁺ concentration, as seen in the electrophysiological trace and futher mitochondrial Ca²⁺ accumulation (image 7).

In view of the functional importance of the perigranular mitochondrial belt, we investigated to what extent these mitochondria form a complete barrier around the granules. Using serial confocal slicing through Rhod-2-stained cells, after a maximal ACh-elicited Ca²⁺ signal had caused substantial Ca²⁺ uptake into the mitochondria, it was possible to see that the mitochondria essentially cover the whole surface of the granular area, like the skin on a football (Figure 6).

Fig. 6. Serial confocal sectioning reveals essentially complete perigranular mitochondrial belt. An acinar cell triplet (inset transmitted light image) was loaded with Rhod-2 and stimulated maximally with ACh. After the peak response, several confocal sections (see schematic illustration) revealed the almost complete coverage of the granular region by mitochondria. The thickness of the confocal sections was 3.5 µm.

Mitochondrial Ca²⁺ uptake in response to uncaging of caged Ca²⁺ in the cytosol and the nucleus

In order to investigate the time course of mitochondrial Ca²⁺ uptake in different regions, we tried to abruptly increase [Ca²⁺]_i globally or locally by uncaging of Ca²⁺ from the caged Ca²⁺ compound o-nitrophenyl (NP)-EGTA. To prevent Ca²⁺ induction of Ca²⁺ release from the ER, we incubated the cells in Ca²⁺-free solution containing 1 µM thapsigargin for 10 min. Thereafter, we introduced a patch pipette (whole-cell configuration) containing 1.2–1.5 mM NP-EGTA. Figure 7 shows an example of such an experiment (n = 5). The transmitted light picture of the cell with the three regions of interest identified by circles in different colours and the outline of the experimental arrangement is shown in Figure 7A, a and b. When Ca²⁺ was uncaged throughout all regions of the cell, to produce a uniform global rise in [Ca²⁺]_i, we observed that mitochondrial Ca²⁺ uptake occurred simultaneously everywhere and reached a peak after ∼4 s (Figure 7A, c). [Ca²⁺]_i was monitored by measurement of the Ca²⁺-sensitive whole-cell current. The initial part of Figure 7A, c is expanded in Figure 7A, d, where it can be seen more clearly that there was no detectable difference in the kinetics of the mitochondrial Ca²⁺ uptake in the different parts of the cell. The regions of interest are marked in Figure 7A, a with colours corresponding to the traces in Figure 7A, c and d. These experiments demonstrate the high rate of mitochondrial Ca²⁺ uptake and the absence of a significant delay to the start of Ca²⁺ accumulation in the mitochondria. The peak of the Ca²⁺-dependent current did, however, occur earlier than the peak of the mitochondrial Ca²⁺ concentration. Although it has been proposed that mitochondria possess a Ca²⁺-induced Ca²⁺ release (CICR) mechanism in vitro (Ichas et al., 1997), the second uncaging of NP-EGTA consistently failed to evoke signs of CICR from the mitochondria in our intact cells. In all the 11 cells tested, the second uncaging of caged Ca²⁺ resulted in further mitochondrial Ca²⁺ uptake, rather than release (Figure 7A, c).

Fig. 7. Local and global uncaging of caged Ca²⁺ causes local and global mitochondrial Ca²⁺ uptake. (A) Global uncaging. (a) Transmitted light picture showing the three colour-coded regions of interest. Length of red bar corresponds to 10 µm. (b) Schematic diagram indicating recording configuration and experimental arrangement. (c) Time course of mitochondrial Ca²⁺ uptake, in the three colour-coded regions identified in (A, a), following two separate global cytosolic Ca²⁺ uncagings (arrowheads). The uncagings cause large increases in the cytosolic Ca²⁺ concentration (monitored as Ca²⁺-dependent current). It is seen that this is associated with a uniform increase in the Rhod-2 fluorescence in all regions. (d) An expanded time scale showing the first part of the response to the first uncaging of Ca²⁺. (B) The effect of local Ca²⁺ uncaging in the basal region. (a) Transmitted light picture with the three colour-coded regions of interest identified. (b) The mitochondrial Ca²⁺ concentration changes, in response to the local basal (green area) Ca²⁺ uncaging, in the three regions (colour coded as in the transmitted light picture). It can be seen that only the mitochondria near the basal plasma membrane react initially, but later there are rather small increases also in the perigranular region.

We also uncaged Ca²⁺ locally in the basal region, outlined in green in Figure 7B, a, and measured the mitochondrial Ca²⁺ uptake in three different areas (green, red and black). In this case the mitochondrial Ca²⁺ uptake was localized to the basal area (green) in which the uncaging had occurred. In the two other areas there were initially no signs of mitochondrial Ca²⁺ uptake, although subsequent Ca²⁺ uncagings induced small rises in [Ca²⁺]_m in the perigranular region (red) (Figure 7B, b) (n = 6).

Next, we loaded Rhod-2-AM and NP-EGTA-AM into the same cell and uncaged Ca²⁺ in a very small area within the nucleus. As shown in Figure 8A, a and b, after repetitive Ca²⁺ uncaging in a small area within the nucleus (red dotted circle), the perinuclear mitochondria (blue area) were the first to take up Ca²⁺. Later, mitochondria in the lateral area of the cell (green circle) took up Ca²⁺. Figure 8A, c shows simultaneous measurements of the Ca²⁺ concentration changes in the nucleoplasm and the cytosol in the granular area as well as in the mitochondria in the perinuclear and lateral regions. The [Ca²⁺]_m rise in the lateral region was considerably delayed as compared with the [Ca²⁺]_m rise in the perinuclear area, demonstrating the buffering power of the perinuclear mitochondria. Similar results were obtained from six cells. These data explain why the nucleus is not easily affected by short-lasting cytosolic Ca²⁺ changes (Gerasimenko et al, 1996; Cancela et al., 2000), whereas the steady-state Ca²⁺ concentration in the nucleoplasm reflects the cytosolic Ca²⁺ concentration reasonably well (Brini et al., 1993; Mogami et al., 1998).

Fig. 8. Local Ca²⁺ uncaging in the nucleus results in specific Ca²⁺ uptake into the mitochondria located just around the nucleus and local bleaching experiment indicates that different mitochondrial groups may not communicate directly. (A) Repetitive nuclear Ca²⁺ uncagings result in increasing Ca²⁺ uptake exclusively into the perinuclear mitochondrial ring, until suddenly towards the end of the experiment, all mitochondria take up Ca²⁺. (a) The transmitted light picture shows the site of Ca²⁺ uncaging (red broken circle) and the colour-coded regions of interest. Length of red bar corresponds to 10 µm. (b) Series of Rhod-2 fluorescence images showing increasing Ca²⁺ concentration in the perinuclear mitochondrial ring (images 1–5). The last couple of uncagings result in a generalized increase in the cytosolic Ca²⁺ concentration, which causes major Ca²⁺ uptake into the mitochondria, also in other regions (images 6 and 7). (c) Graphs representing the time courses of the Ca²⁺ concentration changes in the cytosol and the mitochondria in different regions following the repetitive Ca²⁺ uncagings in the nucleus (arrowheads). (B) An area of a cell represented by the broken red rectangle in (a) was bleached and the changes in the Rhod-2 fluorescence intensity, which had just been increased by uncaging of caged Ca²⁺, were monitored. Length of red bar represents 10 µm. (b) Graph showing that there is a dramatic decrease in the fluorescence intensity in the bleached area (red), which is not transmitted to the neighbouring region (black). (c) Fluorescence images obtained at the times indicated in (b).

Dye bleaching experiments

The results presented so far show that mitochondria in distinct regions can react to local changes in the cytosolic Ca²⁺ concentration without necessarily influencing [Ca²⁺]_m in other regions. We therefore tested the possible luminal connection between mitochondria in different areas. Figure 8B, a shows the transmitted light picture of a cell under investigation. We observed a marked rise in [Ca²⁺]_m in all the regions after global cytosolic Ca²⁺ uncaging. Thereafter, we bleached rapidly one area of the cell (red dotted rectangle in Figure 8B, a and c) using the maximum power of our UV laser (351, 364 nm line). This resulted in complete bleaching of mitochondrial fluorescence in this part of the cell. There was no restoration of fluorescence intensity and the fluorescence intensities in the neighbouring peripheral basal as well as perigranular mitochondria were not changed at all (Figure 8B, b and c, similar data obtained in seven cells). Although the possibility cannot be excluded that UV laser-induced damage could in some manner isolate otherwise communicating organelles, the simplest interpretation of our data indicates that in normal pancreatic acinar cells the mitochondria are not luminally connected.

Mitochondrial Ca²⁺ uptake specifically due to Ca²⁺ entry from external solution

The result shown in Figure 7B, demonstrating that local Ca²⁺ uncaging in the basal area causes selective Ca²⁺ uptake in mitochondria in that region, might indicate that Ca²⁺ entry across the basal membrane could selectively affect the peripheral basolaterally located mitochondria. Ca²⁺ entry through the basal membrane can refill the intracellular ER Ca²⁺ store after agonist-induced depletion (Mogami et al., 1997) and it has been shown recently that respiring mitochondria determine the pattern of activation and inactivation of the store-operated Ca²⁺ current (Gilabert and Parekh, 2000). We therefore investigated directly whether mitochondria in the peripheral basolateral area, which is very close to the plasma membrane, could preferentially respond to store-operated Ca²⁺ influx. Ca²⁺ was liberated from the ER store by exposing the cells to a Ca²⁺-free external solution containing a high concentration (0.5 µM) of the ER Ca²⁺ pump inhibitor thapsigargin for a relatively short period (3 min). This results in a substantial reduction in the ER Ca²⁺ concentration (Mogami et al., 1998). When the external Ca²⁺-free solution is then replaced by a solution containing a high Ca²⁺ concentration (10 mM), Ca²⁺ entry should occur through store-operated channels (Parekh and Penner, 1997). The result of such an experiment is illustrated in Figure 9. Admission of the external solution with the high Ca²⁺ concentration resulted initially in a selective rise in [Ca²⁺]_m in the basolateral region of the cell very close to the plasma membrane (Figure 9). Later, there was also a more modest rise in [Ca²⁺]_m in the perigranular region. At the end of the experiment, a supramaximal concentration of ACh was applied. This resulted in a rise in [Ca²⁺]_m that was most marked near the apical (luminal) membrane. The relatively minor effect of ACh stimulation indicates that refilling of the ER store had not occurred during the period of Ca²⁺ entry, since the effect of thapsigargin is irreversible. On the other hand, the fact that ACh did induce a rise in [Ca²⁺]_m near the apical membrane (green circle, green curve, image 5) indicates that the relatively short period of thapsigargin exposure had not completely depleted the ER store of Ca²⁺. Complete depletion of the ER store takes 10–15 min. We performed five separate experiments of the type shown in Figure 9, all with similar results.

Fig. 9. Basolateral mitochondrial Ca²⁺ uptake due to store-operated Ca²⁺ entry. In this series of experiments, cells were exposed to a Ca²⁺-free solution containing 0.5 µM thapsigargin for 3 min to partially deplete internal Ca²⁺ stores. A high external Ca²⁺ concentration (10 mM) was then introduced, allowing Ca²⁺ entry through store-operated Ca²⁺ channels. (A) Series of Rhod-2 fluorescence images showing increasing Ca²⁺ accumulation in the peripheral mitochondria during the period of Ca²⁺ entry (images 1–4) and finally the increase in the Ca²⁺ concentration in the apically located mitochondria after ACh stimulation (image 5). The numbers correspond in time to those shown in (C). (B) Transmitted light image of the cell under investigation with the three colour-coded regions identified. Length of red bar corresponds to 10 µm. (C) The three coloured traces represent mitochondrial Ca²⁺ measurements (Rhod-2) in the three correspondingly colour-coded regions identified in (B). It is seen that by far the most marked rise in the mitochondrial Ca²⁺ concentration occurred in the region very close to the basal membrane. At the end of the experiment, a high dose of ACh is applied, causing further increase in the mitochondrial Ca²⁺ concentration, but this time most marked in the area very close to the apical membrane [see also image 5 in (A)].

Discussion

Figure 10 summarizes our findings. In addition to the perigranular mitochondria described previously (Tinel et al., 1999; Straub et al., 2000), we have identified two separate groups of mitochondria located in the peripheral basolateral area close to the plasma membrane and in the perinuclar area. We have now shown that each of these three groups of mitochondria responds specifically to cytosolic Ca²⁺ signals in their immediate environment. It is not entirely clear to what extent electrical and chemical coupling between mitochondria occurs in different cell types. A recent study of COS-7 cells indicates that as far as electrical coupling is concerned, the situation may be dynamic, with mitochondrial units spontaneously connecting and disconnecting (De Giorgi et al., 2000). Our local bleaching experiments (Figure 8) indicate that the Ca²⁺-sensitive dye Rhod-2 may not move between mitochondrial areas. At least in normal pancreatic acinar cells, the mitochondria in different regions are apparently not sufficiently luminally connected to allow passage of small molecules along chemical gradients. It is therefore convenient to discuss the function of each of the three mitochondrial groups separately.

Fig. 10. Schematic drawing illustrating the locations of the different mitochondrial groups together with directions of Ca²⁺ transport and indications of ATP supply. For further details see text.

The perigranular mitochondria

The perigranular mitochondrial belt was described by Tinel et al. (1999), but here we have for the first time shown directly that cytosolic Ca²⁺ spiking leads to Ca²⁺ uptake into these mitochondria (Figure 5), thereby directly demonstrating their ability to perform a barrier function. We have now also identified a specific subgroup of perigranular mitochondria situated very close to the apical plasma membrane. These mitochondria are close to the primary Ca²⁺ release sites in the granular area (trigger zone) (Kasai et al., 1993; Thorn et al., 1993; Mogami et al., 1997; Cancela et al., 2000) and are particularly sensitive to the early stages of agonist-induced Ca²⁺ release (Figure 2). This is consistent with data showing that even in the case of supramaximal stimulation, when the [Ca²⁺]_i signal spreads rapidly all over the cell, the [Ca²⁺]_i rise is largest in the apical region (Ito et al., 1997).

Under physiological stimulation, the amounts of Ca²⁺ released from the ER during each spike are so small that the fall in [Ca²⁺]_ER is normally undetectable (Park et al., 2000). Nevertheless, our new results show that these short-lasting cytosolic Ca²⁺ spikes, which are largely confined to the granular region, cause clearly measurable Ca²⁺ uptake in the mitochondria located most apically in the perigranular belt (Figure 5). This indicates close contacts between the Ca²⁺ release sites in the ER and the apical mitochondria, as already described in other systems (Rizzuto et al., 1998). During each local [Ca²⁺]_i spike in the granular region, there is an exocytotic response (capacitance increase) (Maruyama and Petersen, 1994). The apical mitochondria are therefore likely to be most important for stimulus–metabolism coupling under physiological conditions. This makes sense, since exocytotic secretion, which requires a high ATP concentration (Baker and Knight, 1978), is exclusively taking place across the apical membrane (Palade, 1975).

The peripheral mitochondria near the basal plasma membrane

We have demonstrated that store-operated Ca²⁺ entry, which is important for refilling empty ER stores with Ca²⁺ (Parekh and Penner, 1997), is sensed by the peripheral basolateral mitochondria. This is in agreement with the recent demonstration (Gilabert and Parekh, 2000) that respiring mitochondria play a role both in the activation and inactivation phases of the store-operated Ca²⁺ entry. This requires Ca²⁺ uptake into mitochondria placed very close to the Ca²⁺ entry channels (Figure 9). This process would increase ATP production, which would be helpful locally, as Ca²⁺ entering the cell is taken up into the ER by powerful Ca²⁺-activated ATPases (Mogami et al., 1997). Ca²⁺ uptake into both the ER and the mitochondria would tend to reduce [Ca²⁺]_i near the Ca²⁺ entry sites and therefore help diminish the strong negative feedback effect of Ca²⁺ on the open state probability of the store-operated Ca²⁺ channels (Parekh and Penner, 1997), which would otherwise effectively stop entry.

The perinuclear mitochondria

To what extent the Ca²⁺ concentrations in the nucleoplasm and the cytosol can be regulated independently is still controversial (Rogue and Malviya, 1999). The nuclear pore complexes are normally permeable to Ca²⁺ (Brini et al., 1993; Gerasimenko et al., 1995; Lipp et al., 1997) but, at least temporarily, discrepancies between the Ca²⁺ levels inside and outside the nuclear envelope can occur (Rogue and Malviya, 1999). Intermediate doses of ACh can initiate cytosolic Ca²⁺ signals in the granular area that spread to the basal part of the cell, except the nucleus (Gerasimenko et al., 1996). The perinuclear ring of mitochondria (Figure 1C), which takes up Ca²⁺ specifically when Ca²⁺ is uncaged in the nucleus (Figure 8A), could play a role in protecting the nucleus against invasion of [Ca²⁺]_i signals. The nucleus has potential for generating its own Ca²⁺ signals (Gerasimenko et al., 1995; Malviya and Rogue, 1998). The perinuclear mitochondrial barrier could play a role by helping to confine such signals, initiated by release through the inner nuclear membrane, to the nucleoplasm and at the same time supply ATP locally. Ca²⁺ signals in the perinuclear mitochondria could also be important for the regulation of the energetics of nuclear transport and possibly play a role in triggering release of activators of apoptosis from perinuclear mitochondria.

Conclusion

The spatio-temporal pattern of [Ca²⁺]_i signals depends on agonist type and concentration (Petersen et al., 1991) and this is important for understanding differential physiological and pathophysiological regulation (Petersen et al., 1994; Parekh, 2000). Our new results show that different groups of functionally unconnected regional mitochondria can sense [Ca²⁺]_i signals in their immediate environment (Figure 10). The most restricted [Ca²⁺]_i signals in space and time are the repetitive short-lasting Ca²⁺ spikes that are confined to the granular region (Thorn et al., 1993), because of the perigranular mitochondrial Ca²⁺ buffer barrier (Figures 5 and 10). These spikes control exocytotic secretion (Maruyama and Petersen, 1994) through the apical membrane (Palade, 1975), and the local ATP supply from the apical part of the perigranular ring is therefore most likely to be of major physiological significance. Ca²⁺ for stimulus–secretion coupling is ultimately derived from the external solution and enters the acinar cell across the basal membrane. We have shown previously (Mogami et al., 1997; Park et al., 2000) that Ca²⁺ is transported across the cell, shielded from the cytosol, by an operational ER tunnel. Ca²⁺ is therefore always available for release from the ER extensions in the granular area. Ca²⁺ entry into the basal part of the ER tunnel depends on thapsigargin-sensitive Ca²⁺ pumps in the ER (Mogami et al., 1997). The peripheral mitochondria, which take up Ca²⁺ specifically during store-operated Ca²⁺ entry (Figures 9 and 10), must therefore play an important role in supplying the ER Ca²⁺ uptake machinery with the necessary ATP. Finally, the nucleus may have special problems in relation to Ca²⁺ signalling. During normal secretory events, it is not desirable to involve the nucleus. Indeed, we have shown previously that when the Ca²⁺ signal escapes from the granular region, the nuclear area is still somewhat protected against invasion by Ca²⁺ waves (Gerasimenko et al., 1996). The perinuclear mitochondrial ring (Figures 8 and 10) is likely to play an important role in this protection. The perinuclear mitochondria could also play an important role in confining signals generated inside the nucleus to this organelle.

Materials and methods

Cell preparation

Fresh mouse pancreatic acinar cells were isolated after collagenase treatment, as previously described (Osipchuk et al., 1990) and used within 2 h. All experiments were performed at room temperature (22–24°C). Mag-fluo-4-AM, fluo-4-AM, MitoTracker Green FM, Hoechst 33342, Rhod-2-AM and NP-EGTA-AM were purchased from Molecular Probes, and the other chemicals were from Sigma Co.

Solutions

The extracellular bathing solution contained (in mM): NaCl 140, KCl 4.7, MgCl₂ 1.13, CaCl₂ 1, glucose 10 and HEPES 10. pH was adjusted to 7.3 by NaOH. We used two different intracellular pipette solutions. One contained (in mM): KCl 135, MgCl₂ 1.13, NaCl 20, HEPES 10, Na₂ATP 2, EGTA 0.1; pH was 7.2, adjusted by KOH. In the other solution we replaced 0.1 mM EGTA with 10 mM BAPTA + 2 mM Ca²⁺.

Dye-loading procedures

For visualization of mitochondria, isolated cells were incubated with 10 µM MitoTracker Green FM for 30 min at 37°C. For visualization of nuclei, Hoechst 33342 was added to the bathing solution on the microscope stage to a concentration of ∼100 µg/ml. To stain the ER, we incubated cells in a solution containing 0.3–1 µM ERtracker (Molecular Probes) for 15–30 min at 37°C. For imaging of Ca²⁺ in intracellular stores, cells were incubated with 4–6 µM Mag-fluo-4-AM and 0.01% pluronic acid for 20–30 min. For mitochondrial Ca²⁺ measurements, cells were incubated with Rhod-2-AM and 0.01% pluronic acid for 20–30 min at 37°C. For combined mitochondrial and cytosolic Ca²⁺ measurements, the cells were washed after loading of Rhod-2 and then incubated with 2.5 µM fluo-4-AM for 20 min at 22°C, For experiments with uncaging of caged Ca²⁺, NP-EGTA was loaded into the cells by incubation in a solution containing 10 µM NP-EGTA-AM for 30 min. To remove cytosolic indicators, we used the patch–clamp whole-cell configuration (Mogami et al., 1998). The same pipette was used to introduce calcium buffer into the cell. The patch pipette contained 0.1 mM EGTA or a 10 mM BAPTA/2 mM Ca²⁺ mixture. Some experiments with Mag-fluo-4 and/or Rhod-2 were performed on intact cells, without unloading the cytosolic calcium indicator.

Confocal imaging, photobleaching and uncaging

Confocal imaging was performed using a Zeiss LSM510 confocal system. This system allows rapidly alternating excitation of multiple fluorophores, thus reducing cross-talk. Also, accurate spatially restricted exposure of laser light can be achieved and this was used for the uncaging and bleaching experiments. Rhod-2 was excited with a 543 nm laser line and emission collected through a BP560-615 or LP560 filter. Fluo-4, Mag-fluo-4 and MitoTracker Green-FM were excited using 488 nm laser light. Emitted light was collected using a BP505-550 or LP505 filter. Hoechst 33342 was excited with UV laser light (351 nm/364 nm) and emission collected through a BP385-470 filter. We used near-maximal power of UV laser lines (351 nm/364 nm) to induce fast bleaching of Rhod-2 and adequate lower power for uncaging of NP-EGTA. Bleaching/uncaging were combined with simultaneous confocal imaging. For the image analysis, we used Zeiss confocal 510 image software as well as software developed by ourselves.

Electrophysiology

Standard patch–clamp whole-cell current recording (Hamill et al., 1981) was used. The electrophysiological recording of Ca²⁺-dependent current (Petersen, 1992) was made using the EPC-8 amplifier and Pulse software (HEKA). The pipette resistance was 2–3 MΩ. The detailed procedure has been described previously (Thorn and Petersen, 1992).