Stable Golgi-Mitochondria Complexes and Formation of Golgi Ca2+ Gradients in Pancreatic Acinar Cells*

We have determined the localization of the Golgi with respect to other organelles in living pancreatic acinar cells and the importance of this localization to the establishment of Ca2+ gradients over the Golgi. Using confocal microscopy and the Golgi-specific fluorescent probe 6-((N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino)hexanoyl)sphingosine, we found Golgi structures localizing to the outer edge of the secretory granular region of individual acinar cells. We also assessed Golgi positioning in acinar cells located within intact pancreatic tissue using two-photon microscopy and found a similar localization. The mitochondria segregate the Golgi from lateral regions of the plasma membrane, the nucleus, and the basal part of the cytoplasm. The Golgi is therefore placed between the principal Ca2+ release sites in the apical region of the cell and the important Ca2+ sink formed by the peri-granular mitochondria. During acetylcholine-induced cytosolic Ca2+ signals in the apical region, large Ca2+ gradients form over the Golgi (decreasing from trans- to cis-Golgi). We further describe a novel, close interaction of the peri-granular mitochondria and the Golgi apparatus. The mitochondria and the Golgi structures form very close contacts, and these contacts remain stable over time. When the cell is forced to swell, the Golgi and mitochondria remain juxtaposed up to the point of cell lysis. The strategic position of the Golgi (closer to release sites than the bulk of the mitochondrial belt) makes this organelle receptive to local apical Ca2+ transients. In addition the Golgi is ideally placed to be preferentially supplied by ATP from adjacent mitochondria.

The pancreatic acinar cell is the classical model for studies of the secretory pathway. In the pancreatic acinar cell, Palade (1) first described the segregation, transport, and discharge of secretory proteins. Ca 2ϩ regulates processing of secretory proteins within (2) and transport along the secretory pathway (3)(4)(5). In turn many of the membrane-bound compartments of the secretory pathway serve as a source and a sink for Ca 2ϩ (6 -9). Ca 2ϩ signaling and the secretory pathway are inextricably tied in pancreatic acinar cells (10). Structural and functional polarity is an important aspect of this reciprocal relationship (11)(12)(13). Initiation of the secretory pathway occurs at the basolateral part of the cell with protein synthesis and terminates with Ca 2ϩ -dependent secretion at the apical membrane (14 -16).
The endoplasmic reticulum (ER) 1 and nucleus are located in the basolateral part of the cell, whereas the secretory granules (SGs) are located in the apical pole (1,11,17). The most striking polarization is that of the mitochondria with three distinct groupings: peri-granular, sub-plasmalemmal, and peri-nuclear (18 -22).
In this study we examined the localization of the Golgi with respect to other cellular organelles and the functional consequences of this localization. We found close Golgi-mitochondria contacts that remain remarkably stable over time and during different cellular perturbations. Finally, we visualized Ca 2ϩ gradients that are formed over the Golgi as a consequence of relative positioning of the Golgi and mitochondria.

EXPERIMENTAL PROCEDURES
Pancreatic Acinar Cell Preparation-Pancreata were obtained from male CD 1 mice (21-30 days old) as previously described (23) in accordance with the Animals (Scientific Procedures) Act, 1986. Undissociated tissue was immobilized and imaged as previously described (24). Single cells and clusters of acinar cells were obtained by brief (10-min) collagenase digestion of the pancreas, followed by gentle agitation with a pipette. For all experiments, isolated cells, pancreatic clusters, or pancreata were suspended in a standard HEPES-buffered physiological salt solution containing 140 mM NaCl, 4.7 mM KCL, 1.13 mM MgCl 2 , 10 mM HEPES, 10 mM glucose, 1 mM CaCl 2 , pH adjusted to 7.2 (with NaOH). In studies of Ca 2ϩ signaling, cells were stimulated by pressure application of acetylcholine (ACh) (see below).
Confocal Microscopy-Live cells were examined on a Leica TCS SP2 or Leica SP2-AOBS confocal microscope with a 63ϫ water immersion objective lens and a 1.2 NA. SGs were visualized using confocal reflectance (excitation, 543 nm; emission, 538 -548 nm). To visualize the Golgi, cells were loaded with 2.5 M NBD C 6 -ceramide for 15 min at 4°C. NBD C 6 -ceramide was excited by a 476 nm laser line, and emission was collected at 500 -550 nm. Cells were loaded with 50 nM Mito-Tracker Deep Red for 15 min at 37°C to visualize mitochondria. Mito-Tracker Deep Red was excited with a 633 nm laser line, and emission was collected above 650 nm. For ER labeling, cells were incubated with 500 nM BODIPY Texas Red thapsigargin for 15 min at room temperature and excited at 594 nm, and the emission was collected between 600 and 650 nm. Images of intracellular organelles were obtained with a confocal pinhole corresponding to 1 Airy unit. The Ca 2ϩ -sensitive indicator Fura Red was loaded into the cells in the membrane-permeant AM form (5 M for 30 min at room temperature) and imaged using a 488 nm laser line for excitation and 570 -650 nm for emission. Applications of agonists and analysis to calculate cytosolic Ca 2ϩ gradients were performed as described by Gerasimenko et al. (25). Briefly, ACh stimulation of pancreatic acinar cells was via a pipette filled with a 100 M ACh solution. The pipette was attached to a pressure injection system (Eppendorf, Hamburg, Germany); this system allowed short (0.1-1-s) applications of ACh. In some experiments local Ca 2ϩ signals were produced by stimulation with a low (10 nM) concentration of ACh. The formation of Ca 2ϩ gradients across the cells axis with respect to the Golgi was analyzed in the following way: cells in which ACh elicited apically localized Ca 2ϩ signal were selected for analysis. Line profiles of Fura Red fluorescence were taken across the cell from the apical to basolateral pole; the signal was then averaged for six such selected lines. The profile at rest was then subtracted from the stimulated profile (during peak of Ca 2ϩ transient) giving rise to the ⌬F (change in fluorescence) profile reflecting Ca 2ϩ changes along the apical to basolateral axis. This trace was then compared with the distribution of NBD C 6 -ceramide fluorescence recorded along the same line.
Chemicals-All fluorescent dyes were obtained from Molecular Probes (Eugene, OR). All other chemicals were from Sigma.

Localization of the Golgi and Its
Interaction with the Perigranular Mitochondrial Belt-Labeling of the Golgi using the Golgi-specific probe NBD C 6 -ceramide (26) in live cells reveals a large crescent-shaped organelle also located in the peri-granular part of the cell (Fig. 1A, I and II; supplemental Fig. S1; supplemental Video 1). A similar localization of the Golgi has been described previously (using immunofluorescence labeling) in fixed isolated pancreatic acinar cells (27,28). Using twophoton microscopy, we also investigated the positioning of the Golgi in cells located within intact (undissociated) pancreatic tissue or large pancreatic clusters and found a clearly resolved crescent-shaped organelle located in the peri-granular region (see supplemental Fig. S1). We were able to disassemble this structure by incubating the cells for 30 min with brefeldin A (5 g/ml; n ϭ 42; see supplemental Fig. S2), a compound known to disassemble the Golgi (29).
Previous studies have shown a dense belt of mitochondria around the SGs in pancreatic acinar cells (18 -22). Positioning of the Golgi in this region suggested a possible overlap of the two organelles and prompted us to investigate their relative positioning. Co-labeling of acinar cells with NBD C 6 -ceramide ( Fig. 1A, II) and MitoTracker Deep Red (Fig. 1A, III) shows that these two organelles are juxtaposed and sometimes interwoven ( Fig. 1A, IV). The peri-granular mitochondrial belt is positioned predominantly on the basolateral side of the Golgi (Fig. 1A, IV, representative of 132 cells). Magnification of this part reveals that the Golgi (Fig. 1B, I) has close contacts with the mitochondria from the peri-granular belt (Fig. 1B, II and III). In the regions of contacts, the distance between these organelles is so small that it is beyond the resolution of confocal microscopy ( Fig. 1B, III). Thus, the energy-dependent membrane trafficking steps from the ER to the Golgi and on through the Golgi stacks (1) should be well served by the observed close proximity of the Golgi to the ATP-producing mitochondria. The membrane trafficking steps also depend on Ca 2ϩ signals (4) and Ca 2ϩ gradients (5). Localization of the Golgi on the apical side of the mitochondrial belt could potentially expose the Golgi to frequent Ca 2ϩ elevations induced in this cell type by physiological doses of secretagogues (Refs. 8, 23, and 30; see also "The Golgi Localization and Cytosolic Ca 2ϩ Gradients" below).
Another interesting feature of Golgi and mitochondria positioning can be seen in a region adjacent to the apical part of the lateral plasma membrane. The white dashed line in Fig. 1C, IϪIV, depicts the lateral membrane. It is clear that the mitochondria (Fig. 1C, III), but not the Golgi (Fig. 1C, II), extend up to the surface of the lateral membrane (Fig. 1C, IV and supplemental Video 2). These peri-granular mitochondria (Fig. 1C, III) extend past the edge of the Golgi (Fig. 1C, IIϪIV). Comparison of the distribution of the two organelles ( Fig. 1) with the recently reported patterns of mitochondrial Ca 2ϩ accumulation suggests that the lateral mitochondria not covered by the Golgi are the mitochondria that participate in the initial Ca 2ϩ uptake (20). The region adjacent to the apical part of the lateral membrane was reported to contain Ca 2ϩ -releasing channels (reviewed in Ref. 31); therefore, the lateral "Golgi-free" mitochondria could potentially be important participants in the signaling events generated in this part the cell.
The Golgi-Mitochondria Complexes Are Stable Over Time and Resist Cell Perturbation-In many cell types mitochondria form a dynamic interconnected network (32,33). We used three-dimensional confocal reconstructions recorded over time to examine the stability of these close contacts formed between Golgi and mitochondria. Three-dimensional images of mitochondria and Golgi were recorded over time and analyzed. Despite small changes in cell shape and some dynamic movement of the two organelles, the regions in which they come into close contact remained stable over time (269 s) (Fig. 2, A and B To further test the stability of Golgi-mitochondria contacts we induced artificial swelling of the cells by replacing the normal extracellular solution with distilled water (Fig. 2, C and  D). The Golgi structures continued to be in close contact with mitochondria up to the point of cell lysis (Fig. 2D, III). At the time point immediately prior to lysis, the cell had swollen by 45 Ϯ 7% (increase in the area of the central optical section; n ϭ 7). During the cell expansion, up to the time of cell lysis, the Golgi and mitochondria moved simultaneously away from their initial location with regions of close contacts still present (n ϭ 7). After cell lysis in distilled water, the mitochondria lost most of their staining, rounded up, and disconnected from the Golgi and other organelles (data not shown).
Relative Positioning of Golgi and Other Organelles in Live Pancreatic Acinar Cells- Fig. 3 shows the relative positioning of intracellular organelles. Interestingly, in this cell type, the Golgi apparatus is not immediately adjacent to the nucleus (Fig. 3A). The mitochondria, which occupy a more basal location than the Golgi structures, create a narrow barrier between the Golgi and the nucleus (Fig. 3, A and C).
The bulk of the ER is found basally with respect to the Golgi and the mitochondria (Fig. 3B); however, the thin projections of the ER described previously (17,34) can be seen in the apical (with respect to Golgi) part of the cells (Fig. 3B). These ER projections account for the apical Ca 2ϩ release from the lumenally connected ER (34,35).
The Golgi surrounds the SGs, whereas the peri-granular mitochondrial belt engulfs the Golgi (Fig. 3C) from both the basal and lateral sides. This dense layered packing results in the Golgi occupying a position in the vicinity of the SGs on the apical (with respect to the majority of the mitochondria) part of the cell (Fig. 3C). The Ca 2ϩ released from the ER projections in the secretory granule area (Fig. 3B) should therefore first reach the Golgi before the bulk of the mitochondria can come into play and absorb this Ca 2ϩ via high capacity mitochondrial uniporters (36,37). We decided to verify this hypothesis by directly mapping Ca 2ϩ gradients developed during localized apical Ca 2ϩ signals with respect to the position of the Golgi.
The Golgi Localization and Cytosolic Ca 2ϩ Gradients-By co-loading NBD C 6 -ceramide and the Ca 2ϩ -sensitive dye Fura Red, we imaged apically restricted Ca 2ϩ signals with respect to the Golgi localization (Fig. 4). The transmitted image of the cell cluster is shown in Fig. 4A. The apical (red circle) and basal  IV) shows that the Golgi is located to the apical side of the peri-granular mitochondrial belt. The region in the blue box was enlarged for a more detailed examination of the relationship between the Golgi and mitochondria (B, I-III). Overlay of the Golgi and mitochondrial staining shows that these two organelles are juxtaposed around the apical-basal interface. C, I shows an enlarged part of the transmitted image. The white dashed line indicates the lateral membranes close to the apical region. Labeling of the Golgi (C, II) reveals that the proximal end of the Golgi is some way from the lateral membrane, whereas the peri-granular mitochondrial belt (C, III) encapsulates the Golgi (C, IV) extending right up to the lateral membrane (C, III and IV) and continuing for some distance along the lateral membrane.
(blue circle) regions of interest for Ca 2ϩ measurements are shown in Fig. 4A, I. The line along which the Ca 2ϩ gradient was plotted is also shown on the transmitted image (Fig. 4A, I). The Golgi staining can be seen on Fig. 4A, II. Fig. 4B shows an apically restricted Ca 2ϩ signal generated by pressure application of ACh (orange arrow). The dotted lines represent the time periods used for the calculation of the Ca 2ϩ gradient. The differences between resting and stimulated levels were calculated and are displayed in Fig. 4C, I (along the line crossing the Golgi), whereas the Golgi fluorescence profile measured along the same line is shown on Fig. 4C, II. By measuring the Ca 2ϩ level along a line across the cell (Fig. 4A, II) before the stimulation and at the time points of maximal Ca 2ϩ response, we could ascertain the Ca 2ϩ gradients across the Golgi (n ϭ 5). We found that during apical, physiologically relevant Ca 2ϩ signaling (23,30), the Golgi is exposed to Ca 2ϩ gradients that dissipate 1-2 m past the Golgi (Fig. 4C, II). For comparison, in separate experiments, we stained the Golgi and mitochondria (Fig. 4D) and drew similar line profiles as used in Fig. 4, AϪC, to compare the localization of the Golgi and mitochondria (n ϭ 10). The mitochondrial staining profile partially overlaps with the Golgi, but most of the mitochondrial staining was found within a 2-m region past the Golgi. Uptake of Ca 2ϩ by these mitochondria (20) probably accounts for the dissipation of the Ca 2ϩ gradient immediately past the Golgi.
The notion that mitochondrial Ca 2ϩ uptake plays a critical role in terminating local Ca 2ϩ signals and formation of perigranular Ca 2ϩ gradients is supported by our previous studies, in which inhibition of mitochondria with the protonophore carbonyl cyanide m-chlorophenylhydrazone or an inhibitor of the electron transport chain antimycin (or a mixture of antimycin and oligomycin) resulted in conversion of local Ca 2ϩ signaling to global Ca 2ϩ responses (18). Straub et al. (19) reported that inhibition of mitochondria with carbonyl cyanide 4 trifluoromethoxyphenylhydrazone resulted in globalization of Ca 2ϩ spikes induced by inositol 1,4,5-trisphosphate uncaging. Similar results were obtained using Ru360, an inhibitor of the mitochondrial uniporter (38). Direct measurements of mitochondrial Ca 2ϩ (20) and indirect evidence (an increase in NADH during brief Ca 2ϩ signals) (39) also support the barrier role of mitochondria.
In our current study we repeated experiments with inhibition of mitochondria by a mixture of rotenone and oligomycin (with additional staining for Golgi and visualization of mitochondria). We confirmed the previous results and were able to record the formation of global Ca 2ϩ signals appearing in the apical, peri-granular, and basal regions almost simultaneously (supplemental Fig. S4, n ϭ 4; an additional benefit of these experiments is that the position of the mitochondria is revealed by NADH autofluorescence, which is clearly seen to surround the Golgi). Because the Golgi is sandwiched between the Ca 2ϩ source (release sites in the apical region) and the Ca 2ϩ sink (mitochondrial uniporters), Ca 2ϩ gradients are formed over this organelle with the Ca 2ϩ concentration over the trans-Golgi higher than that over the cis-Golgi. Thus, in pancreatic acinar cells, the Golgi is regularly exposed to elevations in cytosolic Ca 2ϩ and to steep Ca 2ϩ gradients. DISCUSSION If the Golgi were on the basolateral side of the peri-granular mitochondrial belt, it would not be exposed to Ca 2ϩ gradients during local Ca 2ϩ transients because the peri-granular mitochondrial belt would shield it from the Ca 2ϩ elevations. By placing the high density of mitochondria behind (with respect to the site of Ca 2ϩ signal generation) and not in front of the Golgi, this cell type has evolved to support Ca 2ϩ signaling in the Golgi region. Mitochondria have also been shown to display barrier functions in other types of epithelial cells: in airway epithelia mitochondria create restricted Ca 2ϩ signaling domains (40), whereas in parotid acinar cells mitochondria congregate around the nucleus, delaying nuclear Ca 2ϩ entry and probably playing an important role in generation of ATP for nuclear processes (41). Relative positioning of the Golgi with respect to such polarized mitochondrial groups could be a subject of an interesting comparative investigation.
The relative positioning of the Golgi and mitochondria in pancreatic acinar cells creates additional functional polarity within the peri-granular mitochondrial belt. The trafficking steps from the transitional ER to the Golgi and through the Golgi compartments (1) should be well served by ATP because of the close proximity of mitochondria to these traffic routes (Fig. 1). Analogous to the interactions seen in other cells between the ER and mitochondria (32,42), close contacts between the Golgi and mitochondria could facilitate communication between these two organelles. It is important to note that mitochondria on the basal side of the mitochondrial belt in pancreatic acinar cells are surrounded by and in close proximity to ER strands (22). One could suggest a concept of organelle-specific positioning of mitochondria: close contacts could be formed by mitochondria and other (different) cellular organelles requiring high local ATP supply; the physical nature of the contacts is probably specific for each target organelle. Lateral mitochondria are not covered by the Golgi (Fig. 1C) and are therefore exposed to Ca 2ϩ signals generated locally in the apical part of the cell (20).
The Golgi can also serve as a functional Ca 2ϩ -releasing store, containing inositol 1,4,5-trisphosphate receptors and two types of Ca 2ϩ pumps (43,44). The Golgi therefore has the potential not only to benefit from Ca 2ϩ signals but also to shape these signals serving either as a Ca 2ϩ sink or as the amplifier of the Ca 2ϩ responses. The positioning of the Golgi in front of the mitochondrial barrier provides it with useful signaling commodities, Ca 2ϩ elevations and gradients, which are required for secretory cargo processing (2) and vesicular trafficking (3)(4)(5). In our previous study we found that the cytosolic Ca 2ϩ gradients could reach hundreds of nanomoles per micrometer when measured along the line drawn from the apical to the basal part of the acinar cell (25); in the current work we found that these cytosolic Ca 2ϩ signals cross the Golgi and dissipate within the peri-granular mitochondrial belt without spreading into the nucleus and the rest of the basal cytosol.

FIG. 4. The Golgi localization and cytosolic Ca 2؉ gradients.
Cells shown on the transmitted image (A, I) were loaded with Fura Red to measure cytosolic Ca 2ϩ changes and NBD C 6 -ceramide (A, II) to visualize the Golgi. B, the time of the pressure application of ACh (1 mM in the pipette solution) from the pipette located close to the cell surface is denoted by the arrow. The ACh-induced Ca 2ϩ increase (shown as a normalized reverse change of fluorescence of Fura Red) was restricted to the apical region (red circle in A, I and red trace in B), and no change was seen in the basal region (blue circle in A, I and blue trace in B). C, I, the fluorescence gradient along a line (pictured in A, I; line averaging and polygonal approximation were used to decrease the noise) at the peak of the response was compared with that at rest (time intervals are indicated by dashed lines on B). The gradient is shown in C, I. The Golgi fluorescence along the same line is presented in C, II. The relative distribution of Golgi (green) and mitochondria (black) staining along a line connecting the apical and basal parts of the cell (a different cell from that shown in AϪC) is shown in D.