The Plasma Membrane Q-SNARE Syntaxin 2 Enters the Zymogen Granule Membrane during Exocytosis in the Pancreatic Acinar Cell*

During exocytosis in the pancreatic acinar cell, zymogen granules fuse directly with the apical plasma membrane and also with granules that have themselves fused with the plasma membrane. Together, these primary and secondary fusion events constitute the process of compound exocytosis. It has been suggested that the sequential nature of primary and secondary fusion is a consequence of the requirement for plasma membrane soluble N-ethylmaleimide-sensitive fusion protein attachment protein receptors, such as syntaxin 2, to enter the membrane of the primary fused granule. We have tested this possibility by determining the location of syntaxin 2 in unstimulated and stimulated pancreatic acini. Syntaxin 2 was imaged by confocal immunofluorescence microscopy. Fused granules were detected both through their filling with the aqueous dye lysine-fixable Texas Red-dextran and through the decoration of their cytoplasmic surfaces with filamentous actin. In unstimulated cells, syntaxin 2 was exclusively present on the apical plasma membrane. In contrast, after stimulation, syntaxin 2 had moved into the membranes of fused granules, as judged by its location around dye-filled structures of 1-μm diameter that were coated with filamentous actin. At long times of stimulation (5 min), the majority (85%) of dye-filled granules were also positive for syntaxin 2. In contrast, at shorter times (1 min), more dye-filled granules (29%) were syntaxin 2-negative. We conclude that syntaxin 2 enters the membrane of a fused zymogen granule after the opening of the fusion pore, and we suggest that this movement might permit the onset of secondary fusion.

Regulated exocytosis in the pancreatic acinar cell, in response to a rise in intracellular Ca 2ϩ concentration (1), involves the fusion of the membranes of zymogen granules both with the apical plasma membrane and with each other (2)(3)(4)(5). This latter process, known as sequential or compound exocytosis, occurs in only a few cell types and, in the case of the acinar cell, is likely to be an adaptation to increase the efficiency of digestive enzyme secretion at the spatially restricted apical pole of the cell.
It is now accepted that membrane fusion events occurring along the secretory pathway are mediated by trans-membrane complexes of soluble N-ethylmaleimide-sensitive fusion protein attachment protein receptor (SNARE) proteins (6). The SNARE 1 complex consists of four intertwined amphipathic ␣-helices (7). The SNAREs of one membrane contribute three helices, each of which contains a strategically located glutamine residue; hence, such SNAREs have been termed Q-SNAREs (e.g. syntaxin and 25-kDa synaptosome-associated protein (SNAP-25)). The fourth helix is provided by the SNARE present on the apposing membrane and contains a similarly placed arginine residue, hence the name R-SNARE (e.g. synaptobrevin). Exocytotic membrane fusion in the acinar cell is known to be SNARE-dependent (8 -10). Syntaxin 2, the Q-SNARE present on the apical plasma membrane (8,9), is required for granule-plasma membrane fusion in vitro (9). In contrast, syntaxin 4, located on the basolateral plasma membrane (8), is likely to be involved in basolaterally directed membrane traffic. Syntaxins 3, 7, and 8 are all present on the granule membrane (8 -10), although their roles are still unclear. Surprisingly, SNAP-23, the non-neuronal isoform of SNAP-25, seems to be located on the basolateral plasma membrane (11) and on the granule membrane (10). Both endobrevin (VAMP 8) and synaptobrevin 2 are present on the granule membrane (9,10). The phenotype of an endobrevin knockout mouse indicates that this protein acts as the major R-SNARE for zymogen granule exocytosis (10). In contrast, synaptobrevin 2 seems to play only a minor role in granule exocytosis (9).
The exocytotic process in the mouse exocrine pancreas has recently been studied in real time in isolated living acini using two-photon excitation imaging (4,5). In these experiments, a membrane-impermeant fluorescent dye is added to the bathing medium, and the cells are imaged during stimulation with an appropriate secretagogue, such as acetylcholine or carbachol. Shortly after agonist application, ⍀-shaped fluorescent spots begin to appear in the subapical region, which have an approximate diameter of 1 m, consistent with the filling of zymogen granules with dye through the open fusion pore (4,5). These dye-filled granules are visible at the apical membrane for several minutes, indicating that the fusion pore remains open for extended periods (5). Two types of fusion events can be distinguished: those in which the granule fuses directly with the plasma membrane (primary fusion event) and those in which a granule lying deeper within the cell fuses with a granule that has itself already opened to the exterior (secondary fusion event). Analysis reveals that there are approximately equal numbers of primary and secondary fusion events (4). During the exocytotic response of a pancreatic acinar cell, granules apparently fuse together only when the first granule has made contact with the plasma membrane (4,5), with a mean time delay between primary and secondary events of 10 -20 s (4, 5). To account for this delay, it has been suggested that secondary fusion requires the recruitment of components of the fusion machinery from the plasma membrane (4). This might, for example, involve the migration of syntaxin 2 into the membrane of the fused granule, although this would raise the issue of the role of syntaxins 3, 7, and 8, which are already present on the granule membrane (8 -10).
In the present study, we sought to determine whether the Q-SNARE syntaxin 2, initially present on the apical plasma membrane, migrates into the membranes of the zymogen granules as they undergo exocytotic membrane fusion. The answer to this question will provide important information about the nature of the fusion pore formed between the granule and plasma membranes and also about the mechanism underlying compound exocytosis.
Preparation of Recombinant Syntaxins-Full-length syntaxin 2 and the cytoplasmic domains of syntaxins 7 and 8 were expressed in the vector pGEX-4T (Amersham Biosciences). Full-length syntaxins 1, 3, and 4 were expressed in pGEX-2T. Syntaxins tagged with glutathione S-transferase (GST) were purified from bacterial lysates by binding to glutathione-Sepharose (Pharmacia). Bead slurries were suspended in sample buffer and heated to 95°C to release the GST-syntaxins for analysis by SDS-PAGE.
Gel Electrophoresis and Immunoblotting-Proteins were separated by SDS-PAGE and then electrophoretically transferred to nitrocellulose (Schleicher and Schü ll) by semidry blotting. Blots were probed with primary antibodies at a dilution of 1:1000. Immunoreactive bands were visualized using horseradish peroxidase-conjugated secondary antibodies (1:1000) and enhanced chemiluminescence (Pierce and Warriner).
Preparation of Mouse Pancreatic Acini-Acini were prepared essentially as described previously (13). A 5-8-week-old mouse was killed by exposure to a rising concentration of CO 2 , followed by cervical dislocation. The pancreas was excised and immediately transferred to ice-cold extracellular medium (135 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 10 mM glucose, 2 mM CaCl 2 , and 10 mM HEPES, pH 7.7). Collagenase (1000 units, type 1A, in 5 ml of extracellular medium; Sigma) was injected into the pancreas and the tissue was incubated at 37°C for 7 min. The supernatant was discarded and the pancreas was washed twice in extracellular medium, followed by trituration with a large-bore plastic pipette. Once the cell suspension flowed freely, trituration was carried out twice more with pipettes with progressively narrower bores. Cells were pelleted from the suspension by centrifugation at 1000 ϫ g for 1 min. The pellet was resuspended in 10 ml of extracellular medium and the acini were gently washed twice in the same medium. Cells were kept at room temperature for 10 min before 100-l samples were plated onto poly-D-lysine (0.1%) pre-treated coverslips.
Confocal Immunofluorescence Microscopy-All steps were carried out at room temperature, except where otherwise indicated. Acini attached to coverslips were used either unstimulated or after stimulation with 2 M acetylcholine, usually for 5 min. Where appropriate, the fluid phase marker LF Texas Red-dextran (Molecular Probes; 1 mg/ml) was added 5 min before acetylcholine stimulation. After incubation with acetylcholine, the cells were washed twice with 2 ml of phosphate-buffered saline (PBS; 10 mM sodium phosphate, pH 7.4, and 150 mM NaCl) for 10 min each time and then fixed in 4% (w/v) paraformaldehyde in sucrose/ phosphate buffer (210 mM sucrose and 30 mM sodium phosphate, pH 6.8), for 30 min. For studies using LF Texas Red-dextran, the initial washes were omitted, and the cells were washed three times in 4% paraformaldehyde over 30 min. After fixation, the cells were washed twice for 5 min in PBS, twice for 5 min in high-salt PBS (20 mM sodium phosphate, pH 7.4, and 500 mM NaCl), and then incubated for 15 min in permeabilizing buffer (PBS containing 0.1% (w/v) Triton X-100). Cells were washed three times for 5 min in permeabilizing buffer containing 5% (w/v) goat serum (blocking buffer) and then incubated with primary antibody overnight at 4°C on Parafilm. The cells were washed twice with high-salt PBS for 5 min and once for 5 min in blocker, before incubation with secondary antibody for 90 min. Cells were washed twice for 5 min with high-salt PBS, once with PBS, and finally once with low-salt PBS (10 mM sodium phosphate, pH 7.4, and 5 mM NaCl). The coverslips were mounted on glass slides in Vectashield mounting medium (Vector Laboratories, Burlingame, CA), sealed with nail varnish, and stored in the dark at 4°C.
Specimens were imaged using a Zeiss 100 M Axioscope confocal laser scanning microscope, with a 63ϫ objective lens of numeric aperture 1.3, capturing an optical slice of ϳ1 m. Images were collected with the appropriate filters: fluorescein isothiocyanate and Alexa Fluor 488phalloidin were excited using the 488-nm line of a krypton/argon laser and imaged with a 505-550 band-pass filter. Cy3 and LF Texas Reddextran were excited with a 543 nm line of a Helium/Neon laser and collected with a long-pass 560 nm filter. All images were captured using the multitrack mode of the microscope to reduce crosstalk of fluorescent signals. Using this mode, cross-talk was estimated to be less than 2%. Analysis was performed using LSM 510 and ImageJ software. The statistical significance of differences between means was assessed using Student's t test for unpaired data.

RESULTS
Because the pancreatic acinar cell is known to contain syntaxin isoforms 2, 3, 4, 7, and 8 (8 -10), it was important to establish that the antibody we intended to use to detect syntaxin 2 was specific for that isoform. To this end, we prepared these isoforms (along with syntaxin 1, which is not expressed in the acinar cell) as GST fusion proteins, and tested the ability of the anti-syntaxin 2 antibody to discriminate between them. Syntaxins 1-4 were full-length proteins, whereas syntaxins 7 and 8 were cytoplasmic domains. The Coomassie blue-stained gels shown in Fig. 1A demonstrate the approximately equal loading of the various syntaxin isoforms. The major bands seen on the gels migrate at positions appropriate to the combined molecular masses of syntaxin (28 -35 kDa) and GST (25 kDa). The anti-syntaxin 2 antibody was a rabbit polyclonal raised against the entire cytoplasmic domain and has been used previously (9). As shown in Fig. 1A, the antibody gave a strong signal when blotted against GST-syntaxin 2 but did not detect syntaxins 1, 3, 4, 7, or 8. Fig. 1B shows a Coomassie bluestained gel of plasma membrane and zymogen granule membrane fractions isolated from rat pancreas. As expected from previous results (9), the anti-syntaxin 2 antibody detected a 35-kDa band in the plasma membrane fraction but not in the zymogen granule membrane fraction.
We used confocal immunofluorescence microscopy with the anti-syntaxin 2 antibody to locate syntaxin 2 within mouse pancreatic acini. As shown in Fig. 2A, syntaxin 2 was exclusively localized to the apical membranes of the acinar cells (red staining). The figure also shows the location of filamentous actin (F-actin), as detected using Alexa 488-phalloidin (green staining). F-actin is known to be concentrated at the apical pole of the acinar cell, just beneath the apical membrane (14 -16). In the set of magnified images shown in Fig. 2B syntaxin 2 is clearly concentrated inside the "tram-line" arrangement of Factin, consistent with its location on the plasma membrane. Close inspection of the red image also reveals that the syntaxin 2 signal appears as two parallel lines (arrowheads), indicating that it is present on two closely apposed apical membranes.
To follow exocytosis in the acini, we used LF Texas Reddextran (3,000 Da), an extracellular dye. As explained above, the dye will enter a fusing granule through the fusion pore, enabling the detection of single exocytotic fusion events. As shown in Fig. 3A, unstimulated acini showed little staining for LF Texas Red-dextran, although some dye entered the spaces between the cells (compare the red signal with the typical tramline appearance of F-actin beneath the apical membrane). In contrast, after a 5-min stimulation of the acini with acetylcholine (2 M), the LF Texas Red-dextran could be seen in clusters of granular structures close to the apical membranes (arrowheads, Fig. 3B). The diameter of these features (ϳ1 m) is consistent with the filling of fused zymogen granules with the extracellular dye. Stimulation of the acini also caused a dramatic remodeling of the actin-based cytoskeleton, so that the fused granules became surrounded by F-actin (Fig. 3B,  arrows). This actin remodeling has been described previously (14 -16). Although it was initially reported that the coating of the zymogen granules with F-actin precedes exocytotic membrane fusion (14), more recent evidence indicates that membrane fusion occurs first (15,16), and it has been suggested that the actin coat stabilizes the fused granule and prevents the abnormal progression of fusion that occurs during acute pancreatitis (16). The relationship between the F-actin and the LF Texas Red-dextran signals is more obvious in the higher power images shown in Fig. 3C. Note that some of the fused granules are adjacent to the apical membrane (arrowhead), although others are much deeper within the cell (arrow) and   C). For both conditions, LF Texas Red-dextran (1 mg/ml) was added to the extracellular medium before incubation. Acini were fixed with 4% paraformaldehyde, permeabilized, and incubated with Alexa 488-phalloidin (1:100). In unstimulated cells (A), F-actin was localized beneath the apical plasma membrane, and there was little LF Texas Red-dextran labeling. In stimulated cells (B), LF Texas Red-dextran labeled granules that had fused with the apical plasma membrane (arrowheads). These fused granules were also surrounded by actin cytoskeleton (arrows). Two distinct types of fusion event can be observed (C): granule-plasma membrane fusion (arrowhead) and granule-granule fusion (arrow). Scale bars, A and B, 10 m; C, 5 m. seem to be linked to the membrane via other fused granules. These two types of feature are consistent with primary and secondary fusion events, respectively.
We next examined the location of syntaxin 2 before and after stimulation of exocytosis. As shown in Fig. 4, syntaxin 2 lined the apical membrane of unstimulated acinar cells. (Note the "tram-line" appearance of the staining, indicated by the arrowhead). The LF Texas Red-dextran had moved into the narrow spaces between the cells and gave a very similar signal to that of syntaxin 2. After stimulation (Fig. 5), the pattern of staining was dramatically different. Granules stained with LF Texas Red-dextran were seen as before, with the majority of granules now also stained for syntaxin 2 (Fig. 5A, top, arrowhead). Quantitation of LF Texas Red-dextran and syntaxin 2 staining in 11 images showed that after a 5-min stimulation there were 9.6 Ϯ 0.7 (n ϭ 38 cells) LF Texas Red-dextran positive exocytotic events per acinar cell, of which 8.3 Ϯ 0.6 (i.e. 85 Ϯ 2%) also stained for syntaxin 2. In many instances, the syntaxin 2 staining patterns were clearly circular, and surrounded the central red staining of the LF Texas Red-dextran (Fig. 5A, bottom). In a control experiment (data not shown), we found that the migration of syntaxin 2 into the granule membranes did not require the presence of LF Texas Red-dextran. Further, a preimmune serum, taken from the same rabbit used to generate the anti-syntaxin 2 antibodies, gave no specific staining (data not shown).
When acini were stimulated with acetylcholine for 5 min, 15% of the granules that were filled with LF Texas Red-dextran were not stained for syntaxin 2 (e.g., Fig. 5A, arrow), and these granules were always terminal events. Because a finite time will be required for syntaxin 2 to diffuse around the open fusion pore into the granule membrane, these features may represent cases where fixation has "frozen" a granule after fusion pore opening but before significant syntaxin 2 migration. After 5 min of stimulation, the peak of the exocytotic burst is virtually complete (17). To study an earlier phase of the exocytotic response, we imaged cells that had been fixed after a 1-min stimulation, when exocytosis is occurring robustly (17). Analysis of 10 images taken at this time showed that there were 5.1 Ϯ 0.5 (n ϭ 33 cells) LF Texas Red/dextran-positive exocytotic events per acinar cell, of which 3.7 Ϯ 0.5 (i.e. 71 Ϯ 3%) also stained for syntaxin 2. Hence, the number of granules that had fused with the plasma membrane was significantly smaller after 1 min of stimulation than at 5 min (p Ͻ 0.001); of these fused granules, a significantly smaller proportion had taken up syntaxin 2 (p Ͻ 0.001). Fig. 5B shows an image taken after a 1-min stimulation. It can be seen that, compared with Fig. 5A, the number of granules filled with LF Texas Red-dextran is small; the fused granules line the apical membrane, and there are no detectable secondary fusion events. Further, although some of the fused granules stain positively for syntaxin 2 (arrowhead), others do not (arrows). This result supports the suggestion made above that the movement of syntaxin 2 into the membrane of fused granules lags behind the opening of the fusion pore; this entry of syntaxin 2 into that granule membrane may permit the secondary fusion seen so clearly at later times during the response (Fig. 5A).
In a final experiment, we compared the staining pattern of syntaxin 2 in stimulated acini with that of F-actin, visualized using Alexa 488-phalloidin. As shown in Fig. 6A, circular features were visible near the apical plasma membrane in both the red channel (syntaxin 2) and the green channel (F-actin). At higher magnification, it is clear that the red labeling is inside the green labeling (Fig. 6B, arrowheads). This relationship between the two staining patterns is consistent with the presence of syntaxin 2 on the cytoplasmic surface of the granules, and the coating of the surfaces of fused granules with F-actin. DISCUSSION Herein, we present evidence that in mouse pancreatic acinar cells the plasma membrane Q-SNARE syntaxin 2 moves into the granule membrane on cell stimulation. At rest, syntaxin 2 is present in the apical domain of the acinar cells in a band peripheral to F-actin, consistent with its localization on the apical plasma membrane. After cell stimulation, syntaxin 2 is observed in regions around zymogen granules that have undergone exocytosis, as marked with the aqueous dye LF Texas Red-dextran. We suggest that this movement of syntaxin 2 may play a role in the control of secondary exocytotic events during compound exocytosis.
Previous work in pancreatic acinar cells provides abundant evidence for the existence of compound exocytosis (2)(3)(4)(5). Recent imaging experiments show that this process occurs in a sequential manner, such that secondary granules fuse with a primary granule with a consistent delay of 10 -20 s (4,5). It has been proposed (4) that this delay is indicative of movement of SNAREs from the plasma membrane to the vesicle membrane, effectively providing the secondary vesicles with a cognate Q-SNARE. Evidence has been provided that in pancreatic ␤ cells, a cell that also undergoes compound exocytosis, SNAP-25 may move into the primary granule membrane after exocytosis (18). We now show that in pancreatic acinar cells, syntaxin 2 may move and may therefore fulfil a role as a target SNARE for compound exocytosis.
Our present results contrast with two previous lines of evidence suggesting that in secretory epithelial cells, the fusion pore may operate as a barrier to diffusion of proteins and lipids. First, an early freeze-fracture electron microscopic study of rat parotid acinar cells showed that the low density of intramembrane particles, presumed to be proteins, in the granule membrane does not change on exocytosis (19), suggesting that granule membrane integrity is maintained and that there is no mixing with the plasma membrane. It should be mentioned, however, that in the guinea pig pancreas the density of intramembrane particles in the granule membrane and the plasma membrane are very similar, making it impossible to determine whether or not the membranes in this cell type mix upon fusion (20). More recently, Thorn et al. (5) have shown that over the lifetime of a fused granule its lipid identity is apparently maintained and that lipophilic dye present in the apical plasma membrane does not enter the granule membrane. These findings have led to the suggestion that granule membranes might be recycled back through the Golgi complex to be reused in further rounds of exocytosis.
There is no easy way to reconcile the previous findings with those presented here, especially because Thorn et al. (5) used the same mouse pancreatic acinar cell preparation. It could be that lipid movement is effectively suppressed in the pancreatic acinar cells, but that protein movement can still take place. It is known that granule lipid content is distinct from that of the apical plasma membrane (21); hence, the fusion pore may act as a barrier to lipid movement. Another possibility is that syntaxin 2 may be actively translocated into the granule membrane. Such a mechanism has been proposed in mast cells, another cell type that manifests compound exocytosis, where movement of the Q-SNARE SNAP-23 (22) into the secretory granule apparently precedes exocytosis. SNAP-23 is therefore not simply diffusing from the plasma membrane into the granule membrane but rather is actively transported.
The nature of the fusion pore in pancreatic acinar cells is unknown. Because granule contents consist of digestive enzymes (i.e. proteins), the fusion pore must have a diameter of several nanometers, and there is evidence from AFM imaging of live acinar cells that the pores are in fact as large as 100 -180 nm (23). In contrast, the fusion pores in neurons and neuroendocrine cells have diameters of less than 1 nm (24). Recently, it has been reported that the fusion pore in PC-12 cells is lined by the transmembrane segments of five to eight syntaxin molecules (25). However, it seems unlikely that the much larger fusion pores in the acinar cell are similarly lined by a ring of proteins, which would act as a physical barrier to mixing between the granule and plasma membranes and also prevent the collapse of the granule membrane into the target membrane after fusion. If movement of lipids and/or proteins between the two membranes is in some way controlled, this might instead depend on features of the membranes themselves that render them immiscible or on the operation of a scaffold on the cytoplasmic surface of the granule (possibly involving F-actin) that acts as a molecular "picket fence" (26).
Perhaps the most intriguing possibility is that the differences between our results and those of Thorn et al. (5) arise from the use different types of stimulation. In this study, we intentionally used maximal agonist concentrations, bath-applied for long periods, to elicit a large response. In contrast, Thorn et al. (5) examined single-granule exocytosis, which was either spontaneous or induced by the focal uncaging of caged carbachol, a stimulus that is small and transient. Under these conditions, compound exocytosis was observed only rarely. It is therefore possible that single vesicle exocytotic events preserve granule integrity, whereas loss of granule integrity and movement of SNAREs is a requirement to trigger compound exocytosis. To test this possibility, we would need to extend the present findings and follow the movement of SNAREs in living cells, where we can track the progress of single granules and determine whether or not compound exocytosis occurs. To do this, however, we would have to generate a syntaxin 2 construct labeled (e.g. with green fluorescent protein) so as to minimize effects on the properties of the protein and to express this construct efficiently in acinar cells, which are notoriously difficult to transduce, even using adenoviral expression systems (27). The "ideal" experiment, therefore, presents enormous technical obstacles.
In conclusion, we have demonstrated the movement of syntaxin 2 into the membrane of the fusing zymogen granule during exocytosis. Although we have not shown that this movement is necessary for the triggering of secondary granule fusion, the apparent lag between the opening of the fusion pore (allowing the entry of LF Texas Red-dextran) and the redistribution of syntaxin 2 is certainly consistent with this notion.