The Endoplasmic Reticulum Can Act as a Functional Ca2+ Store in All Subcellular Regions of the Pancreatic Acinar Cell*

Stimulation of pancreatic acinar cells raises [Ca2+] i via Ca2+ release from inositol-1,4,5-trisphosphate (InsP3)-sensitive intracellular Ca2+ stores, generally considered to reside within the endoplasmic reticulum (ER). However, with physiological doses of cholinergic agonists, the [Ca2+] i increase is localized to the apical (secretory) pole of the cell, leading to suggestions that zymogen (secretory) granules themselves may constitute an InsP3-sensitive Ca2+ store responsible for localized Ca2+ release. We have therefore re-investigated whether the ER in pancreatic acinar cells is capable of acting as a functional Ca2+ store in all, or only some, cellular regions. In streptolysin O-permeabilized cells, the ER accumulated up to 25 mmol of 45Ca2+ per liter ER volume by an ATP-dependent, thapsigargin-sensitive, process. This tracer Ca2+ uptake was dependent on ambient (loading) [Ca2+], as was the intra-ER free [Ca2+], assessed by imaging the fluorescence of Magfura-2 within the Ca2+ stores. Comparison of free and total intra-ER [Ca2+] indicated that 200–300 Ca2+ions are bound within the ER lumen for every Ca2+ ion remaining free. Subcellular analysis showed that ER stores in all regions of the permeabilized cell took up Ca2+ at loading [Ca2+] between 60 nm and 1 μm. Thapsigargin released Ca2+ from stores in all cellular regions, as did InsP3. Immunofluorescence with antibodies against sarco(endo)plasmic reticulum-2b type Ca2+,Mg2+-ATPase or calreticulin confirmed that ER Ca2+ stores were present throughout the cytoplasm. In summary, these results clearly show that the endoplasmic reticulum can act as a functional Ca2+ store in all regions of the acinar cell, including the apical pole.

Intracellular Ca 2ϩ stores play a dominant role in Ca 2ϩ signaling in pancreatic acinar cells. Indeed, the Ca 2ϩ -mobilizing action of the intracellular messenger inositol-1,4,5-trisphosphate (InsP 3 ) 1 was first demonstrated using a permeabilized pancreatic acinar cell preparation (1). This initial work also identified the endoplasmic reticulum (ER) as the intracellular Ca 2ϩ store responsible for agonist-induced increases in [Ca 2ϩ ] i (1). Recently, however, zymogen granules have been proposed to act as a Ca 2ϩ store in pancreatic acinar cells. (2). This hypothesis stemmed initially from the observation that stimulation of pancreatic (and other) acinar cells with acetylcholine results in a polarized rise in cytosolic free [Ca 2ϩ ], with the [Ca 2ϩ ] i increase being initiated at the apical pole of the cell where zymogen granules are clustered (3)(4)(5)(6). Subsequently the rise in [Ca 2ϩ ] i spreads to the basal pole of the acinar cell (3)(4)(5)(6). The role of the proposed zymogen granule Ca 2ϩ store would be to act as the releasable Ca 2ϩ store responsible for the initiation of the intracellular Ca 2ϩ signal at the apical pole (2,7). Propagation of the increase in [Ca 2ϩ ] i toward other regions of the cell could then be mediated by the ER Ca 2ϩ stores, since the rough endoplasmic reticulum is found throughout the acinar cell (see e.g. Ref. 8). This zymogen granule Ca 2ϩ store model is attractive since it can explain the restriction of the [Ca 2ϩ ] i signal to only the luminal (apical) region when low physiological doses of cholinergic agonists are applied (7,9). However, there is as yet no conclusive evidence on whether zymogen granules are equipped with intracellular messenger-triggered Ca 2ϩ release. Indeed, recent evidence suggests that the report of InsP 3 -sensitive Ca 2ϩ release from granules (2) may be an artifact produced by the impurity of the zymogen granule preparation employed (10).
A further problem with the report of zymogen granule Ca 2ϩ stores (2) is that no evidence was found for a Ca 2ϩ uptake mechanism in the proposed zymogen granule Ca 2ϩ store. However, an active Ca 2ϩ -sequestering mechanism is essential to explain the refilling of Ca 2ϩ stores and hence the repetitive nature of the agonist-induced [Ca 2ϩ ] i transients. These conflicting results have prompted us to re-evaluate the suggestion that the endoplasmic reticulum can act as a functional Ca 2ϩ store in all subcellular regions of the pancreatic acinar cell. Our results demonstrate that the endoplasmic reticulum can indeed act as a Ca 2ϩ store in all subcellular regions, including the apical pole.

EXPERIMENTAL PROCEDURES
Cell Preparation-Small clusters of acinar cells were prepared from the pancreas of one 200-g male Sprague-Dawley rat by the same enzymatic digestion procedure described in Ref. 11. After isolation, cells were resuspended in a HEPES/Tris-buffered physiological saline containing 133 mM NaCl, 4.2 mM KCl, 1.0 mM CaCl 2 , 1.0 MgCl 2 , 5.8 mM glucose, 0.2 mg/ml soybean trypsin inhibitor, amino acids as in Eagle's minimal essential medium, 1% bovine serum albumin, and 10 mM HEPES. The pH of this medium was set at 7.4 with Tris. Cells were either used immediately or stored in 1-ml portions on ice until use.
Loading with Magfura-2 and Fluorescence Imaging-Cells were loaded with 5 M Magfura-2-AM for 30 min at 37°C, as described previously (11), and allowed to settle on a poly-L-lysine-coated glass coverslip which formed the bottom of a perfusion chamber. Fluorescence was imaged using a system based on an inverted epifluorescence Nikon Diaphot microscope and a slow scan CCD camera (Digital Pixel Ltd, Brighton, UK). Details of the imaging system were described * This work was supported by Grant 039225/Z/93 from the Wellcome Trust (to A. C. E.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  previously (11). The microscope objective was a Nikon 40 ϫ oil immersion lens (Numerical Aperture 1.3), which allowed images of a field 90 ϫ 135 m to be captured. For imaging Magfura-2 fluorescence in permeabilized cells, a 3 ϫ 3 binning was applied to the individual pixels on the image sensor to give a theoretical spatial resolution of 0.67 ϫ 0.67 m/pixel. Background-subtracted 340:380 images were calculated off-line.
Permeabilization and Perifusion of Pancreatic Acinar Cells-Before initiating permeabilization acinar cells were perifused with Ca 2ϩ uptake medium containing 135 mM KCl, 1.2 mM KH 2 PO 4 , 0.5 mM EGTA, 0.5 mM HEDTA, 0.5 mM nitrilotriacetic acid, and 20 mM HEPES/KOH, pH 7.1. The free Mg 2ϩ concentration was 0.9 mM and was adjusted as described by Schoenmakers et al. (12). SLO (0.4 IU/ml final concentration) was added directly to the perifusion chamber to permeabilize acinar cells as described previously (11). When permeabilization was achieved (as judged by a drop in fluorescence at the dye isosbestic wavelength) cells were re-perifused with the Ca 2ϩ uptake medium as described above but devoid of SLO. Perifusion continued for the duration of the experiment. All experiments were performed at room temperature.
Calcium uptake by intracellular Ca 2ϩ stores was initiated by superfusing cells with Ca 2ϩ uptake medium containing 1 mM ATP, free Ca 2ϩ concentration as indicated in the text and figures, and free Mg 2ϩ concentration of 0.9 mM (free divalent cation concentrations calculated according to Ref. 12). When free Ca 2ϩ concentration was set to 60, 100, or 200 nM, no mitochondrial inhibitors were included in the medium since mitochondrial Ca 2ϩ uptake has previously been shown not to occur at this ambient free Ca 2ϩ concentration (13). The mitochondrial Ca 2ϩ uptake inhibitors oligomycin (5 M) and antimycin (5 M) were included when a free Ca 2ϩ concentration of 1.0 M was applied. Calcium uptake was monitored for a period of 20 min with ratio images acquired at 15-s intervals.
Radioactive 45 Ca 2ϩ Uptake Experiments in Permeabilized Cells-Isolated pancreatic acinar cells in suspension were permeabilized by treatment with SLO (0.4 IU/ml) for 10 min as described previously (14), washed twice, resuspended in (Ca 2ϩ -free) Ca 2ϩ uptake medium (see above), and kept at 0°C until use. Cell density was adjusted to give a protein concentration of around 4 mg of protein/ml. At the beginning of the radiotracer uptake experiment 10 l of permeabilized cell suspension was added to 87 l of Ca 2ϩ uptake medium which contained 10 mM phosphocreatine, 10 units of creatine kinase/ml, 1 M thapsigargin or 1% (w/v) Me 2 SO, and 5 Ci of 45 Ca 2ϩ /ml. The free Mg 2ϩ (0.9 mM) and Ca 2ϩ (as indicated) concentrations were adjusted as described above. After 3 min, the incubation was started by adding 3 l of MgATP stock solution to give a final MgATP concentration of 1 mM. The incubation was terminated after 15 min by adding 1.0 ml of ice-cold stop solution containing 150 mM KCl, 5.0 mM MgCl 2 , 1.0 mM EGTA, and 20 mM Hepes/KOH, pH 7.1, and the suspension was rapidly filtered (GF/C glass microfiber filters, Whatman, Kent, UK). The filters were washed with 2 ϫ 1.0 ml of ice-cold stop solution, dissolved in scintillation fluid, and counted for radioactivity. Total Ca 2ϩ was calculated and expressed as either nmol/mg protein or as mol per liter of endoplasmic reticulum. For the former, cellular protein was determined (after treatment of the cells with 0.1% Triton X-100) with a commercial Coomassie Blue kit (Bio-Rad), using gamma globulin (Bio-Rad) as a standard. For the latter, cell density was determined using a hemocytometer (Weber Scientific International Ltd., Teddington, UK), and the volume of endoplasmic reticulum per incubation was calculated using the number of cells combined with morphological data on pancreatic acinar cells given by Bolender (8). Calcium actively stored by the endoplasmic reticulum was defined as the difference in total Ca 2ϩ retained on the filter after incubation in the absence and presence of the endoplasmic reticulum Ca 2ϩ uptake inhibitor thapsigargin.
Immunohistochemistry-A small drop of cell suspension in physiological saline (see above) was placed on a silane-coated slide, and the cells were allowed to adhere for 10 min in a moist environment. The cells were then fixed for 15 min in a freshly prepared paraformaldehyde solution (2% w/v in phosphate-buffered saline (PBS)). Cells were washed twice in PBS and were permeabilized using 1% (v/v) Triton dissolved in PBS. The slides were subsequently incubated with primary antibody or preimmune serum in the presence of 0.1% Triton and 1% normal goat serum for 1 h at room temperature. Rabbit polyclonal antiserum against calreticulin (15) was diluted 1:100 and rabbit polyclonal antisera to SERCA-2b Ca 2ϩ -ATPase (16 -18) were used diluted 1:1000. Slides were washed three times with PBS and were then incubated for 1 h at room temperature with FITC-conjugated swine antirabbit polyclonal antiserum diluted 1:200 in PBS containing 0.1% (v/v) Triton and 1% (v/v) normal goat serum. Cells were washed three times and mounted in glycerol, which contained SlowFade-Light to prevent photobleaching. Slides were analyzed using the same imaging system employed for Magfura-2 imaging (see above and Ref. 11), except that a FITC dichroic filter set (Chroma Technologies) was used to observe FITC fluorescence. In addition, the resolution of the cooled CCD camera was increased to its maximum by applying a binning of 1 ϫ 1 (i.e. no summation of pixels) to give a theoretical spatial resolution of 0. Normal goat serum and FITC-conjugated swine anti-rabbit immunoglobulins were from DAKO, Glostrup, Denmark. Other chemicals were obtained from the following suppliers: SLO from Difco; thapsigargin from Calbiochem; Ins(1, 4, 5)P 3 from Sigma; Magfura-2-AM and Slow-Fade Light from Molecular Probes; 45 Ca 2ϩ (20 mCi/ml) from NEN Life Science Products. All other chemicals were of analytical grade.

Total Ca 2ϩ Uptake by Endoplasmic Reticulum in Permeabi-
lized Pancreatic Acinar Cells-The total exchangeable Ca 2ϩ uptake capacity of the thapsigargin-sensitive intracellular Ca 2ϩ stores in permeabilized pancreatic acinar cells was determined using the radioactive 45 Ca 2ϩ technique. The effect of the ambient (loading) [Ca 2ϩ ] i on the total exchangeable uptake capacity of intracellular Ca 2ϩ stores was examined over the range 60 nM to 1.0 M. Total thapsigargin-sensitive Ca 2ϩ uptake rose from 1.71 nmol per mg of protein to 6.11 nmol per mg of protein. Fig. 1 shows the uptake data expressed as total thapsigargin-sensitive Ca 2ϩ uptake per volume of endoplasmic reticulum (see "Experimental Procedures"). This gives a total exchangeable Ca 2ϩ accumulation of 7-25 mmol per liter of endoplasmic reticulum.
Subcellular Imaging of Ca 2ϩ Uptake Mechanisms-As in our previous paper (11), we used the low affinity Ca 2ϩ indicator Magfura-2 to monitor the free Ca 2ϩ concentration inside Ca 2ϩ storage compartments in real time in individual cells. Calcium uptake was ATP-dependent and reached steady-state levels within 15 min. The Magfura-2 ratio at steady state was dependent on the ambient (loading) [Ca 2ϩ ] and rose from 0.79 to 2.08 as [Ca 2ϩ ] i was increased from 60 nM to 1 M (Fig. 2A). The initial rate of Ca 2ϩ uptake into the stores was very steeply dependent on ambient [Ca 2ϩ ], increasing more than 6-fold with the change in [Ca 2ϩ ] i ( Fig. 2A). Because of the greatly enhanced initial Ca 2ϩ uptake rate, the time taken to reach steady state [Ca 2ϩ ] lumen was reduced at higher [Ca 2ϩ ] i . Fig. 2B shows the relationship between steady-state free [Ca 2ϩ ] lumen (as indicated by the steady-state Magfura-2 ratio from Fig. 2A) and total thapsigargin-sensitive exchangeable stored Ca 2ϩ (taken from Fig. 1) in pancreatic acinar cells. The two parameters change in parallel over the entire range of loading [Ca 2ϩ ] studied, showing that increased Ca 2ϩ uptake at the higher loading [Ca 2ϩ ] i values does not saturate the intrastore Ca 2ϩ buffering system. In addition, the parallelism of total and free [Ca 2ϩ ] within the stores implies that the free [Ca 2ϩ ] lumen in any given part of the cell can be used to infer the total exchangeable Ca 2ϩ in the same compartment.
We proceeded to study Ca 2ϩ uptake in more detail by comparing uptake in apical and basal areas of acinar cells. This was achieved by applying the same method of subcellular regional analysis used in our previous study (11), i.e. selecting regions of interest in the apical area and in the basal area of the same cell(s). At 60 and 100 nM [Ca 2ϩ ] i , both initial uptake rates and steady-state ratio values were almost identical between apical and basal regions (Fig. 3, A and B). At higher values of [Ca 2ϩ ] i , namely 200 nM and 1 M, initial uptake rates were also similar for the two regions. Interestingly, however, the [Ca 2ϩ ] lumen reached at steady state at these higher values of [Ca 2ϩ ] i was noticeably greater in the basal area of the cell (Fig.  3, C and D).
Thapsigargin has been widely used in intact acinar cells to induce "global" [Ca 2ϩ ] i release (see e.g. Ref. 4). In contrast to the apical pole [Ca 2ϩ ] i signals observed with acetylcholine and thapsigargin (and other organellar Ca 2ϩ -ATPase inhibitors) evoke a homogeneous elevation in cytosolic [Ca 2ϩ ] i (4) or, in some cases, a [Ca 2ϩ ] i rise which is largest in the basolateral pole (6). In permeabilized cells, thapsigargin completely prevents ATP-driven 45 Ca 2ϩ accumulation (13). We tested whether thapsigargin could deplete previously loaded Ca 2ϩ stores in permeabilized pancreatic acinar cells. Fig. 4A shows that, as expected, thapsigargin caused a slow depletion of Ca 2ϩ stores which was similar in both the apical and basolateral poles. Prevention of Ca 2ϩ -ATPase activity by removal of ATP had essentially similar results (data not shown). The kinetics of thapsigargin-induced depletion were examined by normalizing the data shown in Fig. 4A to the total size of the ATP-sensitive Ca 2ϩ pool (i.e. taking the ratio before loading the stores in each experiment as 0% and the steady-state ratio following loading as 100%). This gave depletion curves for apical and basolateral stores which were not significantly different (data not shown), again indicating that thapsigargin has equal actions on Ca 2ϩ stores in the two cellular regions.
We have previously shown that thapsigargin enhances the rate of InsP 3 -evoked Ca 2ϩ release from loaded intracellular Ca 2ϩ stores and also converts apparent "quantal" release of Ca 2ϩ from stores into an essentially monophasic release process (11). Fig. 4B presents subcellular regional analysis of the effects of sequential addition of thapsigargin and InsP 3 . To facilitate comparison between the apical and basolateral poles, data were normalized as described above. It is again clear that the kinetics of Ca 2ϩ store depletion by thapsigargin and InsP 3 are identical for Ca 2ϩ stores in the apical and basolateral poles.
SERCA-2b Ca 2ϩ -ATPase and Calreticulin Distribution in Pancreatic Acinar Cells-We used immunohistochemistry to study the distribution of SERCA-2b Ca 2ϩ -ATPase and of calreticulin in pancreatic acinar cells. Both SERCA-2b Ca 2ϩ -ATPase (Fig. 5A) and calreticulin (Fig. 5B) appeared to be present in all regions of the cell. Immunostaining for the SERCA-2b Ca 2ϩ -ATPase was slightly weaker in the apical area than in other regions of the cell (Fig. 5A). Weak decoration with the Ca 2ϩ -ATPase antibody in the central portion of the basal areas of the cells indicated the presence of the nucleus, which was verified by staining DNA with 4,6-diamidino-2-phenylindole (results not shown). The polyclonal antibody against calreticulin decorated acinar cells in essentially the same pattern as observed with Ca 2ϩ -ATPase antiserum. The only major difference was that the anti-calreticulin staining was slightly more punctate. As with the SERCA-2b antibody, both the apical region and the nucleus appeared less decorated than the basolateral cytoplasm. Since SERCA-2B Ca 2ϩ -ATPase and calreticulin are both markers of the endoplasmic reticulum in pancreatic acinar cells, the less intense decoration of the apical cytoplasm probably reflects the fact that the apical region contains relatively less endoplasmic reticulum compared with other areas of the cell. This is well known from numerous morphological studies (see e.g. Ref. 8) and arises from the fact that a large part of the apical cytoplasm is occupied by zymogen granules.

FIG. 2. Effect of ambient free Ca 2؉ concentration on the Ca 2؉ content of intracellular stores in permeabilized pancreatic acinar cells.
Magfura-2-loaded pancreatic acinar cells were permeabilized by SLO treatment, and the Magfura-2 remaining inside the intracellular compartments was used to monitor store lumen Ca 2ϩ changes. A, for each individual experiment the averaged 340/380 nm ratio was computed for a field of permeabilized cells (between 10 and 16 cells depending on the experiment). Analysis of individual cells indicated that all cells showed ATP-dependent Ca 2ϩ uptake (data not shown, but see Ref. 11). After assessment of permeabilization cells were initially perifused with Ca 2ϩ uptake medium in the absence of Ca 2ϩ but with a free Mg 2ϩ concentration of 0.9 mM. Calcium uptake was initiated by perifusing the cells with a medium containing ATP and with free Ca 2ϩ buffered to 0.06, 0.1, 0.2, or 1 M. Calcium uptake was followed for 20 min in each experiment, with images acquired every 15 s. The data points show the mean Ϯ S.E. and are averages derived from between three and five individual experiments (for clarity standard errors are only shown every 150 s). B, shows the steady-state ratio value 20 min after initiation of Ca 2ϩ uptake (representing the steady-state free [Ca 2ϩ ] within the Ca 2ϩ stores) plotted as a function of the total thapsigargin-sensitive Ca 2ϩ uptake from Fig. 1. Points show the mean Ϯ S.E.

DISCUSSION
This study describes the Ca 2ϩ sequestering properties of intracellular Ca 2ϩ stores in pancreatic acinar cells at the subcellular level. In situ imaging of intracellular Ca 2ϩ stores revealed that Ca 2ϩ can be accumulated in all regions of this polarized cell type. The [Ca 2ϩ ] lumen levels reached at steady state were dependent on the ambient [Ca 2ϩ ] i used to load the stores. The total Ca 2ϩ taken up by the Ca 2ϩ stores also rose with increasing [Ca 2ϩ ] i . These observations show that increased Ca 2ϩ uptake does not result in saturation of the intravesicular Ca 2ϩ -buffering system, since both total Ca 2ϩ and [Ca 2ϩ ] lumen increased upon elevation of [Ca 2ϩ ] i . In addition, these data tend to suggest that a single type of intra-store Ca 2ϩ buffer with a single class of binding site is highly unlikely to account for all intra-luminal Ca 2ϩ buffering.
Detailed subcellular analysis of the Ca 2ϩ uptake process showed that ATP-driven Ca 2ϩ sequestration occurred in both apical and basal regions of the cell. The two subcellular regions did not differ in their capacity to accumulate Ca 2ϩ at lower values of [Ca 2ϩ ] i . At 0.2 and 1.0 M [Ca 2ϩ ] i , however, stores in the basal area were able to accumulate significantly more Ca 2ϩ than stores in the apical region, as judged by a higher steadystate Magfura ratio. Immunohistochemistry revealed a higher density of SERCA-2b Ca 2ϩ -ATPases in the basal area. However, this most probably reflects the simple fact that the basal area contains relatively more endoplasmic reticulum (8). We conclude that intracellular endoplasmic reticulum stores in the apical and basal area are able to accumulate Ca 2ϩ but that stores in the basal area have an increased capacity for storing Ca 2ϩ .
In our previous study we showed that, at an ambient [Ca 2ϩ ] i of 0.2 M, the steady-state intravesicular [Ca 2ϩ ] was 70 M (11). This figure can be compared with the measured total thapsigargin-sensitive Ca 2ϩ uptake under identical conditions, which was 19 mmol of Ca 2ϩ per liter of endoplasmic reticulum (Fig. 1B). From this comparison we conclude that around 270 Ca 2ϩ ions are bound for every free Ca 2ϩ ion. This calculation confirms that most stored Ca 2ϩ is buffered within intracellular Ca 2ϩ stores. Despite this heavy Ca 2ϩ buffering, large amounts of Ca 2ϩ are immediately accessible for rapid mobilization from these stores in both intact and permeabilized cell systems. This indicates that stored Ca 2ϩ is not tightly bound or trapped once accumulated.
Calreticulin is proposed to act as a major Ca 2ϩ buffer within Ca 2ϩ stores, as well as having an important role as an intra-ER molecular chaperone (19). The calreticulin molecule has the ability to bind up to 25 mol of Ca 2ϩ per mol with a low millimolar affinity. Pancreatic tissue is the richest known source of calreticulin (calreticulin content of the pancreas is 540 g/g of tissue (20)), presumably because the high protein synthesis rate of acinar cells imposes a requirements for molecular chaperones. Simple calculation reveals that the calreticulin concentration within the pancreatic endoplasmic reticulum is about 52 mol/liter ER. 2 Our estimate of [Ca 2ϩ ] lumen indicates that calreticulin will not be saturated under these conditions. Even if one assumes a [Ca 2ϩ ] lumen of 1 mM (instead of our 70 M 2 Eighty-two percent of pancreatic volume consists of acinar cells (8).
To simplify the calculation we have assumed that all calreticulin in the pancreas is contained in acinar cells.

FIG. 3. Comparison of ATP-driven Ca 2؉ uptake into Ca 2؉ stores in apical and basolateral regions of permeabilized pancreatic acinar cells.
For every field of cells analyzed to obtain the Ca 2ϩ uptake data in Fig. 2A, we selected those cells where apical and basal regions could be unambiguously distinguished on the basis of light microscopic imaging (between three and five cells depending on the experiment). A region of interest was then defined in the apical and basolateral region of each selected cell, and a time course of the averaged 340/380 ratio was then computed for apical regions and basolateral regions in each experiment. Finally, time courses from individual experiments were averaged to give the data shown. The data follow the same format as Fig. 2A  estimate), as has been suggested for other cell types on the basis of work with ER-targeted aequorin (21), calreticulin can only account for the binding of 3-4% of the total amount of stored Ca 2ϩ . This calculation shows that calreticulin may not be as important in organellar Ca 2ϩ buffering as originally proposed. This is in agreement with recent work on fibroblasts from calreticulin null mice, which showed no differences in ER Ca 2ϩ storage from wild-type cells (22). In fact, recent studies have provided evidence that calreticulin may have other func- Both the SERCA-2B Ca 2ϩ -ATPase and calreticulin appeared to be present in all regions of the acinar cell, with the strongest staining in the basolateral cytoplasm. The less intensely stained regions in the basolateral areas of the cells indicate the presence of the nucleus. Calreticulin distribution was noticeably more punctate than was SERCA-2B distribution. Both panels are representative of at least three experiments on different cell preparations. Results shown in A were obtained using SERCA-2b Ca 2ϩ -ATPase antiserum provided by Dr. F. Wuytack (Leuven) (16,17). Essentially identical results (not shown) were obtained with a different polyclonal antiserum provided by Dr. R. L. Dormer (Cardiff) (18). Scale bar ϭ 10 m.
FIG. 4. Thapsigargin, or thapsigargin followed by InsP 3 , releases Ca 2؉ from loaded intracellular Ca 2؉ stores in both apical and basolateral regions of permeabilized pancreatic acinar cells. Permeabilized pancreatic acinar cells were superfused with a solution containing ATP and 200 nM free [Ca 2ϩ ] for 15 min prior to the start of the record to enable ATP-driven Ca 2ϩ uptake to reach a steady state. Thapsigargin (1 M) was then added (in the continued presence of ATP and Ca 2ϩ ) to block SERCA Ca 2ϩ pumps and induce Ca 2ϩ leak from the stores. Ratio images were acquired every minute in A or every 15 s in B, and data were obtained as described in the legend to Fig. 3. The format follows that of Fig. 3, with the averaged Magfura-2 ratio ([Ca 2ϩ ] lumen ) shown for apical (broken lines) or basolateral (solid lines) regions. Bars show Ϯ S.E. A shows Ca 2ϩ release evoked by prolonged exposure to thapsigargin alone and represents analysis of 14 cells derived from three experiments on different cell preparations. B shows the effect of addition of InsP 3 after thapsigargin and represents analysis of 21 cells derived from three experiments on different cell preparations. In B the data were normalized to the maximal store size, defined as the difference between the ratio before and after Ca 2ϩ loading (see text for details). tions in Ca 2ϩ signaling than organellar Ca 2ϩ buffering. For instance, although calreticulin overexpression inhibited Ca 2ϩ waves in Xenopus oocytes, this action was found to be associated with the high affinity, low capacity, Ca 2ϩ -binding site, rather than with the low affinity, high capacity binding site which confers Ca 2ϩ buffering properties (23). In addition, recent overexpression studies in epithelial and fibroblast cell lines revealed that calreticulin may play an important role in the regulation of Ca 2ϩ influx (24,25). Finally, the knockout studies indicate that calreticulin plays a critical role in transmitting signals from integrin extracellular matrix receptors to the cell interior (22).
As already discussed above, we can use morphological data to express our permeabilized cell Ca 2ϩ uptake data in terms of Ca 2ϩ uptake per unit volume of rough endoplasmic reticulum. It is interesting to compare these data with Ca 2ϩ fluxes observed in intact acinar cell systems. In isolated intact pancreas, the neurotransmitter acetylcholine can mobilize 0.5 mol of Ca 2ϩ per g of tissue. 3 (26). Assuming the endoplasmic reticulum is the agonist-sensitive intracellular Ca 2ϩ storage site, this value suggests the InsP 3 -sensitive store holds approximately 2.6 mmol of Ca 2ϩ per liter of endoplasmic reticulum. Dormer et al. (27) reported that agonists mobilized around 3 nmol of Ca 2ϩ per mg of protein from intact acini, equal to approximately 1.8 mmol of Ca 2ϩ per liter rough endoplasmic reticulum. 4 In permeabilized pancreatic acinar cells, 60% of the thapsigarginsensitive Ca 2ϩ store is InsP 3 -sensitive (13,14). Although the corresponding value for the thapsigargin-sensitive store in intact pancreas is not known, the thapsigargin-sensitive Ca 2ϩ store is known to be bigger than the agonist-sensitive intracellular store (29). If we assume that the InsP 3 -sensitive store in intact tissues accounts for 60% of the thapsigargin-sensitive store, then the thapsigargin-sensitive intracellular Ca 2ϩ stores of intact cells would contain about 4 mmol of Ca 2ϩ per liter endoplasmic reticulum. These calculations show that intracellular Ca 2ϩ stores in permeabilized cell systems are not by any means reduced compared with those in intact cells. In fact, under the standard conditions used in our study (0.2 M free Ca 2ϩ ) intracellular Ca 2ϩ stores in permeabilized cells accumulate considerably more Ca 2ϩ than in intact cells. The most likely explanation for this difference is that the ambient free [Ca 2ϩ ] which we have used to load the stores is higher than the free [Ca 2ϩ ] i in unstimulated intact cells. At the lowest ambient (loading) [Ca 2ϩ ] used in this study, 60 nM, the stores accumulated around 6 mmol of Ca 2ϩ per liter endoplasmic reticulum. This suggests, albeit indirectly, that the cytosolic [Ca 2ϩ ] i under resting conditions in intact cells is below 60 nM.
It well established that the endoplasmic reticulum can be found in all regions of the pancreatic acinar cell, including the apical pole (e.g. Refs. 8 and 11). The ER is, therefore, a plausible candidate for an apical pole Ca 2ϩ store. The same is true for other acinar cell types, for instance lacrimal cells (see Ref. 6 for references). In our previous studies we have shown that all ATP-dependent Ca 2ϩ uptake in permeabilized acinar cells can be inhibited by the SERCA Ca 2ϩ -ATPase inhibitor thapsigargin, indicating that the endoplasmic reticulum is wholly responsible for ATP-driven Ca 2ϩ sequestration (13). The in situ imaging experiments in the present study confirm that ATPdriven Ca 2ϩ uptake can indeed be observed in all regions of the acinar cell. To seek further evidence that the endoplasmic reticulum can act as Ca 2ϩ stores in all cellular regions, we mapped the distribution of the SERCA-2B isoform of the microsomal Ca 2ϩ pump, known to be present in acinar cells (18), at low (non-confocal) resolution. Immunofluorescence demonstrated that SERCA-2B Ca 2ϩ -ATPases are present in all regions of the pancreatic acinar cell. In keeping with this result, addition of thapsigargin caused depletion of Ca 2ϩ stores in all cellular regions. We can therefore conclude that Ca 2ϩ can be actively reaccumulated by the endoplasmic reticulum in any area of the acinar cell. Calcium accumulation will occur during rises in [Ca 2ϩ ] i , regardless of whether the increase in [Ca 2ϩ ] i derives from intracellular stores or from influx from the extracellular environment.
A functional agonist-sensitive Ca 2ϩ store minimally requires an active Ca 2ϩ uptake mechanism, an intra-store Ca 2ϩ buffering system, and a Ca 2ϩ release mechanism. The present work clearly shows that the first two are ubiquitous in the acinar cell cytoplasm. In our previous work we demonstrated that the InsP 3 -sensitive release mechanism is also present in all subcellular regions of the acinar cell (11). This latter result is re-emphasized in Fig. 4B, which shows that thapsigargin alters the kinetics of InsP 3 -evoked Ca 2ϩ release in both apical and basolateral poles (contrast Fig. 4B with Fig. 6B of Ref. 11). This shows that the Ca 2ϩ store compartment(s) where the InsP 3evoked Ca 2ϩ release mechanism is located are loaded by a thapsigargin-sensitive uptake mechanism. It is thus clear that endoplasmic reticulum stores able to take up, store, and release Ca 2ϩ are present in the apical pole of pancreatic acinar cells.
The present study and the recent work of Mogami et al. (30) show unequivocally that the endoplasmic reticulum in the apical pole can act as a functional Ca 2ϩ store. Zymogen granules have also been suggested to be an InsP 3 -sensitive Ca 2ϩ storage compartment in this cellular region, a view based largely on work on an isolated pancreatic zymogen granule preparation (2). However, a very recent cell fractionation study concluded that zymogen granules did not show InsP 3 -sensitive Ca 2ϩ release (10). In fact, Yule et al. (10) showed that the zymogen granule preparation used by Gerasimenko et al. (2) was contaminated with mitochondria, nuclei, and endoplasmic reticulum. This "mixed" preparation showed Ca 2ϩ release on treatment with InsP 3 , although Ca 2ϩ release was not evoked by thapsigargin (2,10). 5 Further purification resulted in a homogenous zymogen granule preparation that had lost its InsP 3 sensitivity, although the granules did contain Ca 2ϩ , which could be released using Ca 2ϩ ionophores (10). Immunolocalization studies in intact cells also failed to detect any InsP 3 receptors in the granule region (32), and it has been suggested very recently that some zymogen granules may "acquire" InsP 3 sensitivity during isolation by fusion with nearby membrane domains containing high concentrations of InsP 3 receptors (33).
In addition to the recent work of Yule et al. (10) discounting the role of zymogen granules in [Ca 2ϩ ] i signaling, several older studies employing radioactive 45 Ca 2ϩ in isolated intact pancreas also specifically ruled out zymogen granules as an agonist-mobilizable Ca 2ϩ store (34 -36). In particular, zymogen granules were unable to exchange Ca 2ϩ for radioactive 45 Ca 2ϩ despite prolonged labeling protocols (34 -36). The authors of 3 Data from Fig. 2 of Ref. 26. The pancreas primarily consists of acinar cells (82% of its volume (8)), and we assume for the sake of simplicity that all agonist-induced Ca 2ϩ fluxes originate from acinar cells. 4 Calculated by assuming 1 g of pancreas contains 144 mg of protein (28). 5 From the lack of effect of thapsigargin, it might be argued that InsP 3 sensitivity in the mixed granule preparation resides in an organelle other than the endoplasmic reticulum. However, this argument must be regarded with caution, since it is well known that cell homogenization and purification of the microsomal fraction of many different tissues results in a separation of Ca 2ϩ release and accumulation sites. For instance, only 10% of the Ca 2ϩ accumulated by ATP-dependent uptake in pancreatic microsomes is released by InsP 3 (31), compared with a corresponding figure of 60% in permeabilized acinar cell preparations (13,14). these studies concluded that during protein processing the divalent cations Ca 2ϩ and Mg 2ϩ become trapped inside zymogen granules (34 -36). This is consistent with the classical observation that, during agonist stimulation, pancreatic enzyme secretion is tightly correlated with the secretion of Ca 2ϩ and Mg 2ϩ into the luminal space (where divalent cations are believed to be necessary for the activation of digestive enzymes) (37,38).
In conclusion, InsP 3 -sensitive Ca 2ϩ stores can only act as fully functional Ca 2ϩ stores when they are equipped with a Ca 2ϩ -pumping mechanism. This is mandatory in order for the stores to refill following depletion. The combination of techniques applied in this study to investigate organellar Ca 2ϩ uptake provides clear evidence that the endoplasmic reticulum can act as a functional Ca 2ϩ store in all subcellular regions of the pancreatic acinar cell.