Polarized Expression of Ca2+ Channels in Pancreatic and Salivary Gland Cells

In polarized epithelial cells [Ca2+] i waves are initiated in discrete regions and propagate through the cytosol. The structural basis for these compartmentalized and coordinated events are not well understood. In the present study we used a combination of [Ca2+] i imaging at high temporal resolution, recording of Ca2+-activated Cl− current, and immunolocalization by confocal microscopy to study the correlation between initiation and propagation of [Ca2+] i waves and localization of Ca2+ release channels in pancreatic acini and submandibular acinar and duct cells. In all cells Ca2+ waves are initiated in the luminal pole and propagate through the cell periphery to the basal pole. All three cell types express the three known inositol 1,4,5-trisphosphate receptors (IP3Rs). Expression of IP3Rs was confined to the area just underneath the luminal and lateral membranes, with no detectable receptors in the basal pole or other regions of the cells. In pancreatic acini and SMG ducts IP3R3 was also found in the nuclear envelope. Expression of ryanodine receptor was detected in submandibular salivary gland cells but not pancreatic acini. Accordingly, cyclic ADP ribose was very effective in mobilizing Ca2+ from internal stores of submandibular salivary gland but not pancreatic acinar cells. Measurement of [Ca2+] i and localization of IP3Rs in the same cells suggests that only a small part of IP3Rs participate in the initiation of the Ca2+ wave, whereas most receptors in the cell periphery probably facilitate the propagation of the Ca2+ wave. The combined results together with our previous studies on this subject lead us to conclude that the internal Ca2+ pool is highly compartmentalized and that compartmentalization is achieved in part by polarized expression of Ca2+ channels.

Ca 2ϩ mobilizing agonists trigger several forms of [Ca 2ϩ ] i signals, which include [Ca 2ϩ ] i oscillations and Ca 2ϩ waves (1). It has now been recognized in many cell types that Ca 2ϩ signaling is highly organized and compartmentalized. This issue has been extensively discussed in a series of recent reviews in a special issue of Cell Calcium (2). In the case of pancreatic acini compartmentalization of signaling and intracellular Ca 2ϩ pools was suggested when the quantal nature of Ca 2ϩ release from internal stores (3,4) was described. Subsequently, a series of elegant papers demonstrated the specialization of the luminal pole (LP) 1 of these cells. Ca 2ϩ waves tended to start in this region and spread to the basal pole (BP) (5)(6)(7)(8). When the cells are stimulated to trigger [Ca 2ϩ ] i oscillations, the oscillation tended to occur in and remain confined to the LP (6,7,9). Similar LP oscillations could be initiated with infusion of low concentration of IP 3 into the cells (7,9). More recently we showed a higher order of organization of Ca 2ϩ signaling in this pole, in that the initiation site of Ca 2ϩ waves was agonist specific in the same cell type (10).
The structural features responsible for the compartmentalized organization of Ca 2ϩ signaling in acinar and other cell types are not well understood. Several findings suggest that the properties of IP 3 -mediated Ca 2ϩ release may account for the specialization of the LP. Analysis by reverse transcription polymerase chain reaction and WB showed the presence of all three types of IP 3 R in the pancreas (11,12). However, for the most part, their specific cellular expression and localization is not known. In one study clear localization of type 3 IP 3 R (IP 3 R3) in the LP has been reported (13). Propagation of the Ca 2ϩ wave from the LP to the BP was suggested to occur through a series of Ca 2ϩ -induced Ca 2ϩ release events mediated by the ryanodine receptor (RyR) (5,14). Accordingly, it was reported that cADPR, an activator of the RyR (15), releases Ca 2ϩ from the IS of pancreatic acini (16). More recently Gerasimenko et al. (17) reported that the secretory granules, which are located in the LP, release Ca 2ϩ in response to IP 3 and cADPR.
Compared with pancreatic acini, there is little information on Ca 2ϩ signaling by SMG acinar and duct cells. Several studies reported the response of the two cell types to Ca 2ϩ -mobilizing agonists including carbachol, epinephrine, substance P, and isoproterenol (18 -20). Studying Ca 2ϩ signaling in SMG cells has several advantages. First, it allowed us to determine the generality of polarized arrangement of Ca 2ϩ signaling and expression of Ca 2ϩ transporters in secretory epithelial cells. Second, SMG cells were shown to express mRNA for type 1 RyR (21), which allows a more clear evaluation of the role of RyR in Ca 2ϩ signaling in polarized cells. Third and most important for this study, is that the SMG duct is a single layer epithelium with very well defined LP and BP, which allows an unequivocal localization of Ca 2ϩ transporters, in particular the Ca 2ϩ pumps described in the accompanying manuscript (33).
In the present study we used [Ca 2ϩ ] i imaging at high temporal resolution to show that in all three cell types Ca 2ϩ waves start in the LP and spread through the cell periphery to the BP. This pattern correlated well with the polarized expression of IP 3 Rs in all cells and RyR in SMG cells. The results also highlight the problem of the mechanism of Ca 2ϩ wave propagation in the face of the paucity of Ca 2ϩ release channels from the cell interior.

EXPERIMENTAL PROCEDURES
Materials-cADPR was purchased from Calbiochem or Sigma. Preparation and characterization of the pAb raised against IP 3 R1-3 are described in Ref. 12. IgG2a Clone 2 mAb against IP 3 R3 was from Transduction Laboratories (Lexington, KT). Two anti-ryanodine receptor Abs were used, mAb Clone 34-C from Affinity Bioreagents (22) and mAb anti-RyR from Calbiochem (23).
Preparation of Tissue Slices, Acini, and Duct Fragments-The pancreas and SMG of 100 -150-g rats were removed immediately after sacrifice of animals, immersed in the O.C.T. compound (Miles, Elkhard, IN) and snap frozen in liquid nitrogen. Frozen glands were used to cut 4-m sections. The sections were plated on polylysine-coated slides, dried, and kept at Ϫ20°C until use. A mixture of acini and duct fragments from all glands were prepared by published procedures (4,10,20). The cells were kept suspended in cold PSA until use. The composition of PSA was (in mM) NaCl 140, KCl 5, MgCl 2 1, CaCl 2 1, Hepes 10 (pH 7.4 with NaOH), glucose 10, pyruvate 10, bovine serum albumin 0.1%, and soybean trypsin inhibitor 0.02%.
Preparation of Single Acinar and Duct Cells-After removal of the glands and fine mincing, the minced pancreas was treated with trypsin/ EDTA and then collagenase to liberate single pancreatic acinar cells exactly as described before (24). The minced SMG was washed once with 20 ml of PSA by 10 s of centrifugation at 120 ϫ g. The pellet was resuspended in 5 ml of PSA containing 0.8 mg/ml collagenase type CLS4, 254 units/mg (Worthington) and was digested for 20 min at 37°C. The tissue was washed twice with a PBS solution resuspended in 5 ml of PBS containing 0.05% trypsin/0.02% EDTA and incubated for 8 min at 37°C. The trypsin-treated tissue was washed twice with 30 ml of PSA, resuspended in 6 ml of PSA containing 0.53 mg/ml collagenase CLS4, and incubated for 20 min at 37°C. The cells were washed twice with 20 ml of PSA by a 2-min centrifugation at 150 ϫ g and kept on ice until use. Single duct and acinar cells were distinguishable by microscopy based on their different size, and their identity was verified during electrophysiological recording by measurement of membrane capacitance that averaged 19.74 Ϯ 0.55 picofarad (n ϭ 12) in acinar and 4.33 Ϯ 0.06 picofarad (n ϭ 6) in duct cells.
[Ca 2ϩ ] i Imaging-Loading cells with Fura 2 and image acquisition and analysis was as detailed before (10). Acini and ducts with well defined basal and apical borders were selected for experimentation. To maximize the temporal resolution, fluorescence was measured at a single excitation wavelength of 380 nm, averaging four consecutive images for each time point. Under these conditions using a frame size of 256 ϫ 240 pixels allowed recording the fluorescence of 4 -5 acinar cells at a resolution of up to 7.7 Hz. During perfusion with the control solution and just before the first stimulation, the image of the resting cells was acquired. This was taken as the fluorescence at time 0 (F 0 ). Pixel values of all subsequent images were divided by this image, and the traces are the calculated F 0 /F t , where F t is the fluorescence at time t. For the experiments in Figs. 7 and 8 at the end of fluorescence recording, the perfusion medium was aspirated, and the bright field image at high (experimental) and low magnification was recorded. Then the bottom of the coverslip was tagged with a marker before fixation and extraction of the cells with cold methanol. The extracted cells were processed for IC staining with IP 3 Rs Ab as described below. At the end of Ab staining, coordinates were labeled around the tag, the tag was cleared to allow imaging by confocal microscopy, and the coordinates together with the low magnification bright field image were used to identify the cell cluster from which [Ca 2ϩ ] i was recorded.
Whole Cell Current Recording-The tight seal whole cell current recording of the patch clamp technique (26) was used for measurement of Ca 2ϩ -activated Cl Ϫ current, which directly correlates with changes in [Ca 2ϩ ] i (27). The experiments were performed with single acinar or duct cells perfused with solution A. The standard pipette solution contained (in mM): KCl 140, MgCl 2 1, EGTA 0.5, Na 2 ATP 5, and Hepes 10 (pH 7.3 with KOH) as described in previous studies (25). In some experiments this solution also contained 1-100 M cADPR and 1.5 mg/ml heparin. Seals of 6 -10 gigaohms were produced on the cell membrane, and the whole cell configuration was obtained by gentle suction or voltage pulses of 0.5 V for 0.3-1 ms. The patch clamp output (Axopatch-1B, Axon Instruments) was filtered at 20 Hz. Recording was performed with patch clamp 6 and a Digi-Data 1200 interface (Axon Instruments). In all experiments the Cl Ϫ and cation equilibrium potentials were about 0 mV. All the traces shown were at a holding potential of Ϫ40 mV.
Western Blot-Packed, purified pancreatic acini, and SMG ducts and acini were extracted with a buffer containing 62.5 mM Tris (pH 6.8 with HCl), 8 M urea, 2% SDS, and 100 mM dithiothreitol in a ratio of 1:2. 10-l samples containing about 75 g of protein were separated by an SDS-polyacrylamide gel electrophoresis using a 7.5% polyacrylamide gel. The separated proteins were transferred to 0.2-m polyvinylidene difluoride membranes, and the membranes were blocked by a 1-h incubation at room temperature in 5% nonfat dry milk in a solution containing 20 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 0.05% Tween 20 (TTBS). The IP 3 Rs were detected by a 1-2-h incubation of individual membranes with the respective Ab diluted in TTBS.
Immunocytochemistry-Tissue slices and the cells attached to glass coverslips used for [Ca 2ϩ ] i measurements were fixed and permeabilized with 0.5 ml of cold methanol for 10 min at Ϫ20°C. After removal of methanol, the cells were washed with PBS and incubated in 0.5 ml of PBS containing 50 mM glycine for 10 min at room temperature. This buffer was aspirated and the nonspecific sites were blocked by a 1 h incubation at room temperature with 0.25 ml of a PBS solution con- taining 5% goat serum, 1% bovine serum albumin, and 0.1% gelatin (blocking medium). The medium was aspirated and replaced with 50 l of blocking medium containing control serum or a 1:50 dilution of pAb against IP 3 R1 and IP 3 R2, 1:25 dilution of pAb and 1:2000 dilution of mAb against IP 3 R3, and 1:250 dilution of mAb against the RyR. In several control experiments, the primary pAbs against the different IP 3 Rs were preincubated with the respective peptides used to raise them (see Ref. 12). Such preincubation completely blocked detection of the respective IP 3 Rs by WB or IC. After incubation with the primary Ab for 1.5 h at room temperature and three washes with blocking medium, the Ab were detected with goat IgG tagged with fluorescein. Images were collected with a Bio-Rad MRC 1000 confocal microscope.

Properties of [Ca 2ϩ ] i Waves in SMG Cells-In a previous
study we showed that in maximally stimulated pancreatic acini, Ca 2ϩ waves tend to initiate in the apical pole and spread through the cell periphery to the basal pole (10). Fig. 1 and Table I show that this is not a unique phenomenon to pancreatic cells, because it can be demonstrated in SMG acinar and duct cells. With all agonists the Ca 2ϩ wave tended to initiate in the LP or the lateral region adjacent to the LP. However, differences among agonists were noted in terms of wave speed and the time required to cause maximal increase in [Ca 2ϩ ] i . Epinephrine induced the fastest wave in duct cells; the [Ca 2ϩ ] i wave initiated by isoproterenol in duct cells was significantly slower than the others as well as the time required to achieve peak Ca 2ϩ (Table I). It is important to note that acinar cells are about 2.5 times bigger than duct cells, which indicates that the wave travels about 2-3 times faster in SMG acinar compared with duct cells.
WB Analysis of IP 3 R-To study whether and how the expression of different IP 3 Rs in specific parts of the cells correlate with the [Ca 2ϩ ] i wave, we analyzed the expression of type 1-3 IP 3 R by WB and determined their localization by IC in pancre-atic and SMG cells. Fig. 2 shows WB analysis of the different IP 3 R in pancreatic acini and SMG duct and acinar cells. For this we took advantage of our technique for separation between SMG acinar and duct cells, which yield better than 95% pure duct and acinar preparations (20). IP 3 Rs in all cells were sensitive to proteolysis. Including 100 mM of the potent protease inhibitor benzamidine and 0.3% soybean trypsin inhibitor reduced the intensity of minor bands and increased the level of intact receptors. However, even under these conditions proteolysis could not be completely prevented. Because of this problem it is not possible to reliably determine the relative abundance of the different types of IP 3 R. In the case of IP 3 R3 we used two separate Ab to verify the expression of this receptor because the two Ab gave somewhat different results in IC (see below). Both Abs showed that SMG acinar cells express the highest level of IP 3 R3. The pAb, but not the mAb, detected expression of IP 3 R3 in SMG ducts. The relative intensity of labeling of SMG duct and acini by the pAb (about 1:3) excludes contamination of the duct preparation with acini as the source of labeling.
Immunolocalization of IP 3 R-Previously only the localization of IP 3 R3 in the LP of pancreatic acini has been reported (13). Fig. 3 demonstrates the localization of all IP 3 Rs in the LP of pancreatic acini and SMG acinar and duct cells. Remarkably all three epithelial cell types expressed high levels of IP 3 R1 and IP 3 R2 only close to the luminal and lateral membranes. In additional experiments when up to 2-3 times higher concentrations of Ab were used, still no IP 3 R expression could be observed in any region of the cells beyond that shown in Fig. 3.
Cell-specific expression and localization was observed with IP 3 R3 using two different Ab, a pAb raised against a 13-amino acid sequence of the C-terminal of IP 3 R3 (12) and a mAb raised against amino acids 22-230 of IP 3 R3 (28). In pancreatic acini FIG. 2. WB analysis of IP 3 Rs. Isolated SMG duct (SMD) and acinar cells (SMA) and pancreatic acini (PnA) were packed and dissolved in sample buffer containing protease inhibitors and with (ϩ) or without (Ϫ) 100 mM benzamidine (Benz) and 0.3% soybean trypsin inhibitor (STI). About 75 g of proteins from each cell type were separated by SDS-polyacrylamide gel electrophoresis and analyzed for the level of IP 3 R1, IP 3 R2, and IP 3 R3 using the pAb for the corresponding IP 3 R or the mAb for IP 3 R3 (far right blot).  Fig. 1 were used to identify the initiation sites of the [Ca 2ϩ ] i waves. Time to spread is defined as the time between traces 1 and 5 in each experiment, and time to peak is the time needed to achieve maximal [Ca 2ϩ ] i increase in region 1 in each cell. The results are given as means Ϯ S.E. Significant differences (p Ͻ 0.05) were measured in time to spread and time to peak in duct cells when stimulation with epinephrine is compared to that of other agonists. both Ab detected expression of IP 3 R3 in the LP and next to the lateral membrane, which closely resembled that of IP 3 R1 and IP 3 R2. However, only the pAb detected expression around the nuclear envelope. The ability of the pAb but not the mAb to detect the IP 3 R3 in the nuclear envelope is further highlighted when the pattern of expression of IP 3 R3 in the SMG is examined. In SMG acini, expression of IP 3 R3 was confined to the LP and lateral membrane. With the pAb IP 3 R3 was detected only around the nuclear envelope of duct cells. The mAb did not stain the SMG duct. Unlike both types of acini, the SM duct did not express any IP 3 R3 in the LP or lateral membrane. After submission of the present study similar localization of IP 3 Rs in pancreatic acini using the same pAb was reported (29). RyR in SM and Pancreatic Cells-The confinement of IP 3 R expression to just underneath the luminal and lateral membranes in the face of the propagated Ca 2ϩ waves raised the possibility that RyR may mediate the propagation of the waves. We expected all cell types to express the RyR because it was reported that cADPR releases Ca 2ϩ from IS (16) and isolated zymogen granules (17) of pancreatic acini and the SMG express type 1 RyR (21). The use of mAb that recognize all isoforms of known RyR and functional characterization of Ca 2ϩ release showed that this may not be the case. Fig. 4 shows that we could not detect any RyR in pancreatic acini by IC. On the other hand both SMG acinar and duct cells stained by the same Ab. Also in this case expression of RyR was confined to the LP and lateral membrane.
Considering the reported effect of cADPR on Ca 2ϩ release in pancreatic acini (16,17), we examined the effect of this compound on Ca 2ϩ release from IS of pancreatic acini and SM cells using two different experimental protocols. In Fig. 5A it is shown that 10 M cADPR infused through a patch pipette into pancreatic acinar cells caused Ca 2ϩ release from IS. However, with 10 M cADPR Ca 2ϩ release was observed in only four of nine cells, was much smaller than that caused by agonist (Fig.  5A), and was blocked by heparin (Fig. 5B) to the extent that in five of five experiments with heparin no Ca 2ϩ release from IS was observed. These findings confirm all aspects of the data reported by Thorn et al. (16). Unlike the case in pancreatic acini, 10 M cADPR potently released Ca 2ϩ from IS of SMG acinar and duct cells (Fig. 5, C and E), the release was observed in eight of nine acinar and four of four duct cells, and it was not blocked by heparin (Fig. 5, D and F) in three of three acinar and three of three duct cells.
A different protocol comparing the effect of cADPR in the two cell types is shown in Fig. 6. In SLO permeabilized cells cADPR up to 100 M failed to release Ca 2ϩ from IS of pancreatic acini (Fig. 6, trace a). On the other hand 10, 100 (not shown) and 25 M cADPR released Ca 2ϩ from IS of SMG cells (Fig. 6, trace b), and this release was not blocked by heparin (Fig. 6, trace c). Relatively high concentrations of cADPR were needed, probably because of the presence of an ATP regeneration system and high concentration of Mg 2ϩ in the uptake medium. Additional evidence for the differences between SMG and pancreatic cells was obtained with caffeine. At 2 and 5 mM, caffeine failed to increase [Ca 2ϩ ] i in pancreatic acinar cells, but it inhibited the effect of agonist stimulation, as reported before (30). On the other hand 5 mM caffeine increased [Ca 2ϩ ] i in SMG acinar and duct cells by about 220 Ϯ 36 nM (n ϭ 4 for each cell type) (results not shown).

Correlation between IP 3 R Localization and [Ca 2ϩ ] i Initiation
Sites-Considering the quantal behavior of Ca 2ϩ release (3,4), it was of interest to determine whether initiation sites of Ca 2ϩ release correlated with the entire or part of the IP 3 R pool. To achieve that we recorded [Ca 2ϩ ] i at high temporal resolution from small acinar clusters, fixed the cells, and stained them with Ab raised against IP 3 Rs. An example for each IP 3 R using SMG acinar cells is shown in Fig. 7. Despite the low efficiency of labeling with IP 3 R1 in isolated SM acinar cells (Fig. 7A, c), it is clear that area 1 did not include the entire IP 3 R1 pool in the LP of cell 1. This is even more clear for IP 3 R2 (Fig. 7B) and IP 3 R3 (Fig. 7C). In each case IP 3 R is apparent in the entire LP and most of the lateral membrane, but the initial [Ca 2ϩ ] i increase occurred in a limited region of the cells.
A better resolution of the initiation sites and their correlation with IP 3 R expression was obtained in three cells clusters of pancreatic acini (Fig. 8). In this case we also attempted to correlate the [Ca 2ϩ ] i initiation sites with the special distribution of IP 3 Rs. High background precluded such studies with IP 3 R1 Ab in isolated pancreatic acini. In the cases of IP 3 R2 and IP 3 R3, the sharp [Ca 2ϩ ] i initiation sites correlated with only part of the IP 3 R pool. Interestingly, it appears that the [Ca 2ϩ ] i wave initiation sites correlated best with expression of IP 3 Rs in specific parts of the cells. For example the initiation sites correlated best with expression of IP 3 R2 in the bottom part of cells 1 and 2. Similar correlations were observed in several experiments with each IP 3 R. Whether this indicates the spacial location of the initiation site or it represents cell fixation artifact remains to be determined.

DISCUSSION
Ca 2ϩ signaling is governed by the action of Ca 2ϩ pumps and Ca 2ϩ channels (31). The pumps remove Ca 2ϩ , whereas the channels release Ca 2ϩ into the cytosol. The coordinated activation and inactivation of these transporters yield [Ca 2ϩ ] i signals in the form of Ca 2ϩ oscillations and/or [Ca 2ϩ ] i waves. The localization and participation of the Ca 2ϩ pumps in the [Ca 2ϩ ] i wave initiated by agonists is described in the accompanying manuscript (33). In the present studies we show the polarized expression of Ca 2ϩ channels in three secretory epithelial cells and provide evidence for their compartmentalization to initiate and propagate the Ca 2ϩ waves.
WB analysis and IC revealed that all three cell types express the three known IP 3 Rs. In agreement with the reported high levels of type 1 RyR mRNA in SMG tissue (21), IC showed the presence of RyR in SMG cells. This is further supported by the ability of cADPR to release Ca 2ϩ from IS of SMG cells. As was shown for other cells (15,32), the effect of cADPR in SMG cells was not inhibited by a high concentration of heparin. cADPR was particularly effective in SMG cells because stimulation with maximal concentration of agonist released only a small amount of Ca 2ϩ from IS of cells infused with 10 M cADPR. Finally, caffeine caused a small Ca 2ϩ release form IS of SMG acinar and duct cells. Hence, we believe that SMG acinar and duct cells indeed express an active RyR.
There is less certainty as to the expression of RyR in pancreatic acini. Thus, Ryr was not detected by IC, cADPR was ineffective in releasing Ca 2ϩ from IS of permeabilized cells, and caffeine did not increase [Ca 2ϩ ] i in intact pancreatic acinar cells. The only exception was that infusion of 10 M cADPR through a patch pipette into pancreatic acini causes a small Ca 2ϩ release from internal stores. However, this observation is weakened by the finding that the effect of cADPR was inhibited by heparin (see also Ref. 16), an inhibitor of the IP 3 channel (4,25). A possible explanation for all the findings is that pancreatic acini express very low levels of RyR, below the detection threshold of the Ab, which is not sufficient to observe Ca 2ϩ release in permeable cells. This may be inferred from the observation that 10 M cADPR induced a much smaller Ca 2ϩ release in pancreatic compared with SMG cells. The inhibition by heparin may suggest functional interaction between the IP 3 R and RyR. In SMG cells the high abundance of RyR allowed Ca 2ϩ release even when IP 3 R were inhibited by heparin. Obviously further work is needed to completely resolve the question of RyR in pancreatic acinar cells.
With the exception of IP 3 R3 all Ca 2ϩ channels were confined to the LP and lateral membrane. The pAb raised against IP 3 R3 detected clear expression of this receptor in the nuclear region of pancreatic acinar and SMG duct cells. On the other hand, the mAb against the same receptor failed to detect any nuclear staining, although they detected the IP 3 R3 in the LP of pancreatic and SM acinar cells. Because the mAb and pAb recognize different sequences of IP 3 R3, WB analysis with the pAb clearly shows the expression of IP 3 R3 in duct cells, and the only staining of IP 3 R3 in duct cells is found in the nuclear region, we can conclude that IP 3 R3 is indeed expressed in the nuclear envelope. The differential reactivity with the mAb and pAb may indicate expression of two different isoforms of IP 3 R3 in the cell periphery and the nuclear envelope.
In all three cell types and with all Ab we could not detect any Ca 2ϩ channels in the granular region. We paid particular attention to this region because it was reported that both IP 3 and cADPR released Ca 2ϩ from single isolated zymogen granules (17). After submission of the present work, IP 3 Rs localization in pancreatic acini and analysis of isolated granules also failed to detect IP 3 Rs associated with the granular region (29). However, it is possible that the level of IP 3 Rs in the granules and other parts of the cells are too low to be detected by IC. A more interesting possibility is that only subset of the granules respond to IP 3 and may have originated from the so called "fusion competent" pool next to the luminar membrane, the site of high levels of IP 3 Rs. The most popular model for propagation of Ca 2ϩ waves is sequential Ca 2ϩ -induced Ca 2ϩ release events mediated by the effect of Ca 2ϩ on IP 3 R and/or RyR (1,27). This and all other models assume and require expression of Ca 2ϩ release channels throughout the cytoplasm. This was not found to be the case for all the cells studied. The Ca 2ϩ release channels were concentrated next to the luminal and lateral membranes. Such a localization can account well for the finding that in most pancreatic acinar (see Ref. 10) and SMG cells ( Fig. 1 and Table  I) the Ca 2ϩ waves were initiated in the LP and tended to propagate along the cell periphery. However, in all cells [Ca 2ϩ ] i eventually increased in all parts of the cytoplasm. How the Ca 2ϩ propagated from sites of high Ca 2ϩ release channel expression to the center of the cells is not clear. One possibility is that peripheral to central propagation occurs by diffusion. This is not very likely because, as shown in the following manuscript (33), wave propagation is a controlled process dependent on the activity of sarco/endoplasmic reticulum Ca 2ϩ ATPase pumps. Another possibility is that Ca 2ϩ release events in the cell periphery are sufficiently large to reach the nuclear membrane and activate IP 3 R3. In the space between these two sites of high levels of Ca 2ϩ release channels, expression of the channels may be below the detection level of all Ab used but sufficient to allow Ca 2ϩ release when activated by IP 3 and/or the Ca 2ϩ released in the initial event.
Polarized expression of Ca 2ϩ release channels and propagation of the Ca 2ϩ wave through the cell periphery highlight the compartmentalized arrangement of Ca 2ϩ signaling complexes. In previous studies we (4,10,25) and others (5-9) provided multiple functional evidence for compartmentalization of Ca 2ϩ signaling. The results in Figs. 7 and 8 provide the first correlation between pattern of IP 3 R expression and initiation sites of the Ca 2ϩ wave. Multiple and agonist specific initiation sites (10) would suggest that specificity of initiation sites is determined by the agonist but that initiation sites always reside at a region of high expression level of IP 3 Rs. Other Ca 2ϩ transporters that control the initiation sites and propagation of Ca 2ϩ waves are the Ca 2ϩ pumps. The role of the pumps in these events are described in the accompanying manuscript (33).