Presence of a Complex Containing Vesicle-associated Membrane Protein 2 in Rat Parotid Acinar Cells and Its Disassembly upon Activation of cAMP-dependent Protein Kinase*

Amylase release from parotid acinar cells is mainly induced by the accumulation of intracellular cAMP, presumably through the phosphorylation of substrates by cAMP-dependent protein kinase (PKA). However, the molecular mechanisms of this process are not clear. In a previous study (Fujita-Yoshigaki, J., Dohke, Y., Hara-Yokoyama, M., Kamata, Y., Kozaki, S., Furuyama, S., and Sugiya, H. (1996) J. Biol. Chem. 271, 13130–13134), we reported that vesicle-associated membrane protein 2 (VAMP2) is localized at the secretory granule membrane and is involved in cAMP-induced amylase secretion. To study the formation of the solubleN-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) complex containing VAMP2 in parotid acinar cells, we prepared rabbit polyclonal antibody against the peptide corresponding to Arg47-Asp64 of VAMP2 (anti-SER4256). The recognition site of anti-SER4256 overlaps the domain involved in binding target membrane SNAREs (t-SNARES). Then we examined the condition of VAMP2 by immunoprecipitation with anti-SER4256. VAMP2 was not included in the immunoprecipitate from solubilized granule membrane fraction under the control conditions, but incubation with cytosolic fraction and cAMP caused immunoprecipitation of VAMP2. The effect of cytosolic fraction and cAMP was reduced by addition of PKA inhibitor H89. Addition of both the catalytic subunit of PKA and the cytosolic fraction allowed immunoprecipitation of VAMP2, whereas the PKA catalytic subunit alone did not. These results suggest that (1) the t-SNARE binding region of VAMP2 is masked by some protein Xand activation of PKA caused the dissociation of X from VAMP2; and (2) the effect of PKA is not direct phosphorylation ofX, but works through phosphorylation of some other cytosolic protein.

In rat parotid acinar cells, stimulation of ␤-adrenergic receptors and the subsequent accumulation of cAMP induces the exocytosis of amylase (1). Although most regulated exocytosis systems are mediated by elevation of intracellular calcium ions, there are some exocytosis systems that are also regulated by intracellular cAMP. For example, glucagon-induced insulin release from pancreatic beta cells and luteinizing-hormone secretion from pituitary cells are mediated by cAMP (2,3). In these systems, however, the amount of secretion induced by cAMP alone is less than that induced by calcium. In contrast, amylase secretion from parotid acinar cells is mainly regulated by intracellular cAMP without elevation of calcium. Therefore, this is an appropriate system in which to study the mechanism of cAMP-dependent exocytosis.
In calcium-dependent systems, several calcium-binding proteins such as synaptotagmin, syncollin, and calcium-dependent activator protein for secretion, were identified as candidates for the calcium sensor(s) that suppress exocytosis at low concentrations of cytosolic calcium ions and enhance by binding with calcium (4 -7). In cAMP-regulated exocytotic systems, activation of cAMP-dependent protein kinase (PKA) 1 is thought to play an essential role in the process. During amylase secretion from parotid acinar cells, the activity of PKA was enhanced (1), and the catalytic subunit of PKA was shown to be sufficient to cause amylase secretion in permeabilized acinar cells (8). Therefore, it is likely that PKA phosphorylates a protein involved in exocytosis. Several proteins were reported to be phosphorylated upon ␤-adrenergic stimulation (9 -11). However, it is not clear which phosphorylation is crucial for triggering cAMP-dependent exocytosis.
We previously reported (12) that vesicle-associated membrane protein 2 (VAMP2) is localized at the secretory granule membrane of rat parotid acinar cells and that it plays an important role in cAMP-dependent amylase secretion. However, target membrane SNAREs (t-SNAREs) such as syntaxin1 or SNAP-25 were not detected in rat parotid acinar cells. In adipocytes, VAMP2 is also localized to vesicles that store the glucose transporter GLUT4, and functions cooperatively with syntaxin4 in the translocation of GLUT4 to the cell surface (13,14). Neutrophils and mast cells also express syntaxin4 and VAMP2, and these are possibly involved in the exocytosis of secretory granules (15,16). Aquaporin-2 is translocated from intracellular vesicles to the apical membrane upon production of cAMP induced by vasopressin in kidney collecting duct cells (17)(18)(19), in which VAMP2 and syntaxin4 were detected at intracellular vesicles and the apical membrane, respectively (20,21). Although many syntaxin isotypes were identified, only syntaxin1 and 4 can associate with VAMP2 in vitro (22). The * This study was supported in part by a grant-in-aid for scientific research from the Ministry of Education, Science, and Culture of Japan in 1997-1998, a Suzuki Memorial Grant of Nihon University School of Dentistry at Matsudo (research grant for Assistant) in 1997, and Research for the Frontier Science grant (the Ministry of Education, Science, Sports and Culture). 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 U.S.C. Section 1734 solely to indicate this fact.
combination of VAMP2 and syntaxin4 is possibly a common pairing in regulated exocytosis and vesicular trafficking outside the neuron.
If these SNARE proteins are involved in cAMP-dependent amylase secretion, the formation of the SNARE complex may be regulated by phosphorylation by PKA. SNARE proteins are thought to function in the recognition and docking between vesicles and target membrane by interacting with each other. It has been reported that the interaction of recombinant syn-taxin4, and SNAP-23 was modulated by phosphorylation of syntaxin4 by exogenous PKA (23). However, Foster et al. (23) showed that VAMP2 binding to syntaxin4 was not affected by PKA phosphorylation of syntaxin4.
In this study, we tried to identify the t-SNAREs that interact with VAMP2 in parotid acinar cells and to clarify the regulatory mechanism of VAMP2. To study the condition of VAMP2 in rat parotid acinar cells, we prepared antibodies against VAMP2, and immunoprecipitated VAMP2 using the antibody. Whether VAMP2 is immunoprecipitated or not may depend on the proteins bound to VAMP2.

EXPERIMENTAL PROCEDURES
Materials and Antibodies-Collagenase and the catalytic subunit of cAMP-dependent protein kinase were purchased from Roche Molecular Biochemicals GmbH. Trypsin, trypsin inhibitor, and control rabbit IgG were purchased from Sigma. Protein A-Sepharose 4FF was from Amersham Pharmacia Biotech.
Peptides corresponding to the amino acid sequence of Ser 2 -Pro 20 , Arg 47 -Asp 64 of VAMP2 were synthesized and conjugated to keyhole limpet hemocyanin. Anti-SER4253 and anti-SER4256 were raised against each conjugated peptide and purified using peptide-coupled affinity columns ( Fig. 2A). Affinity-purified anti-syntaxin4 antibody was a kind gift from Dr. Amira Klip (The Hospital for Sick Children, Toronto).
Preparation of Cytosolic and Total Membrane Fractions-Parotid glands were minced and dispersed by trypsin and collagenase. Dispersed acinar cells were homogenized with 320 mM sucrose, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 1 mM PMSF, and 20 mM Hepes-NaOH, pH 7.5. The nuclear fraction was removed from the homogenate by centrifugation at 750 ϫ g for 10 min, and the post-nuclear supernatant was centrifuged at 100,000 ϫ g for 60 min. The resultant supernatant and pellet were used as the cytosolic and total membrane fractions, respectively.
Preparation of Secretory Granule Membranes-Secretory granules of the rat parotid glands were isolated by Percoll gradient centrifugation as described previously (24), with some modification. Homogenization was performed with 300 mM sucrose, 1 mM MgCl 2 , 1 mM DTT, 1 mM benzamidine, 0.4 mM PMSF, and 20 mM Hepes-NaOH, pH 7.5. After centrifugation, the Percoll suspension was fractionated from the bottom into 20 tubes. Secretory granules were recovered in the densest fraction. The specific activity of amylase in the final granule fraction was usually 4.5-fold higher than that in the homogenate. Purified whole granules were suspended in buffer A (1 mM MgCl 2 , 1 mM DTT, 1 mM PMSF, and 20 mM Hepes-NaOH, pH 7.5) containing leupeptin (10 g/ml) and antipain (10 g/ml), homogenized and centrifuged at 100,000 ϫ g for 60 min. The pellet was suspended in buffer A containing 150 mM NaCl and used as the granule membrane fraction.
Solubilization of Total Membrane and Secretory Granule Membrane-Total membrane and secretory granule membrane fractions were suspended in solubilizing buffer B (150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 1 mM PMSF, and 20 mM Hepes-NaOH, pH 7.5) containing 4% Triton X-100, and incubated for 45 min. Unsolubilized materials were removed by centrifugation (20,000 ϫ g for 20 min). The supernatants were dialyzed against buffer B containing 1% Triton X-100 overnight. After centrifugation at 20,000 ϫ g for 20 min, the supernatants were obtained as solubilized membrane fractions.
Immunoprecipitation of VAMP2-Solubilized total membrane fraction (200 g) and solubilized granule membrane fraction (6 g) were suspended in reaction buffer (150 mM NaCl, 20 mM NaF, 10 mM MgCl 2 , 2 mM ATP, 10 g/ml leupeptin, 10 g/ml antipain, 1 mM PMSF, 1 mM DTT, 1 mM EGTA, 1% Triton X-100, and 20 mM Hepes-NaOH, pH 7.5) in the absence or presence of cytosolic fraction (50 g), and/or 0.2 mM cAMP, a catalytic subunit of PKA (6 units), and then incubated at 30°C for 10 min. Samples were added to protein A-Sepharose 4FF conjugated with normal rabbit IgG, and were incubated at 4°C for 60 min. After centrifugation, the supernatants were recovered and incubated with anti-SER4256 conjugated with protein A-Sepharose 4FF at 4°C for 2 h. Then protein A-Sepharose was collected by centrifugation and boiled in sample buffer for SDS-polyacrylamide gel electrophoresis.

Syntaxin 4 Was Detected in the Total Membrane Fractions but not in the Granule Membrane Fraction of Rat Parotid
Acinar Cells-In our previous study (12), neither syntaxin 1 nor SNAP-25, which interact with VAMP2 in neuronal cells, were detected in rat parotid acinar cells. A candidate for the t-SNARE target of VAMP2 is syntaxin4 because it binds to VAMP2 in vitro (22) and is reported to interact with VAMP2 in adipocytes (25). To examine the expression of syntaxin4 in parotid acinar cells, cytosolic and total membrane fractions of rat parotid acinar cells were prepared by centrifugation. Similar fractions were prepared from skeletal muscle to serve as positive controls for immunoblotting (22). Western blotting analysis of these samples showed that 35-kDa protein, which reacts with anti-syntaxin4 antibody, is detected in pellets of both parotid acinar cells and skeletal muscle (Fig. 1A). Next, the intracellular localization was analyzed. Cytosol, total membrane, and granule membrane fractions from parotid acinar cells were prepared and used for Western blotting analysis. Syntaxin 4 was detected in the total membrane fraction, but not in the granule membrane fractions where the majority of the VAMP2 was detected (Fig. 1B). The difference in localization suggests that syntaxin4 and VAMP2 function as classical t-SNAREs and vesicle SNAREs, respectively, in rat parotid acinar cells. This is consistent with the previously reported result that botulinum neurotoxin C1 did not inhibit cAMP-dependent amylase secretion (12), because syntaxin4 does not have a cleavage site for botulinum neurotoxin C1.
Immunoprecipitation of VAMP2 from Solubilized Membrane Fractions-The total membrane fractions were solubilized with Triton X-100, and the extracts were used for immunoprecipitation using anti-SER4256. The immunoprecipitates were subjected to Western blotting analysis using anti-SER4253. VAMP2 was not detected in the precipitates from untreated extracts (Fig.  2B, lane 1). Because VAMP2 was present in the solubilized total membrane fraction, it seems likely that the epitope was masked by another protein bound to VAMP2. We attempted to find conditions under which VAMP2 could be made accessible to the antibody. After incubation with the cytosolic fraction, a small amount of VAMP2 was precipitated (Fig. 2B, lane 2). Addition of 0.2 mM cAMP increased the amount of precipitated VAMP2 (Fig.  2B, lane 4), whereas cAMP alone had no effect (Fig. 2B, lane 3). Because PKA is present in the cytosolic fraction, the effect of cAMP is probably to activate PKA. This result suggests that VAMP2 is masked by some other proteins that dissociate upon the activation of PKA.
Immunoprecipitation of VAMP2 from the Solubilized Granule Membrane Fraction-We performed immunoprecipitation from Triton X-100-solubilized granule membrane fractions. Again, VAMP2 was not immunoprecipitated from the extract (Fig. 3A, lane 1). Therefore, syntaxin4 is probably not the protein that binds to VAMP2 to inhibit the access of antibody, because syntaxin4 was not detected in the granule membrane fraction. In the same way as the solubilized total membrane fraction, the addition of cytosol allowed a small amount of immunoprecipitation of VAMP2 from the solubilized granule membrane fraction (Fig. 3A, lane 2), and this was enhanced by cAMP (Fig. 3A, lane 4). PKA inhibitor H89 decreased the amount of VAMP2 in the precipitate, supporting the hypothesis that the effect of cAMP and the cytosolic fraction is due to the activation of PKA (Fig. 3A, lane 5). H85, which does not have inhibitory activity for PKA, did not affect the precipitation (Fig.  3A, lane 6).
Comparison between the Effects of cAMP and Calcium Ion on the Immunoprecipitation of VAMP2-High concentrations of cytosolic calcium ion also induce a small amount of amylase secretion, although cAMP is the main regulator of amylase exocytosis (26). We compared the effects of cAMP and calcium on immunoprecipitation. Reactions 1-3 contained 1 mM EGTA, eliminating free calcium ions, whereas the concentration of calcium ion in reactions 4 and 5 is buffered at 10 M. Calcium ion neither allowed VAMP2 precipitation itself (Fig. 3B, lane  4), nor enhanced the effect of cAMP (Fig. 3B, lane 5). This is consistent with our previous finding that VAMP2 is involved in cAMP-dependent, but not in calcium-dependent amylase secretion (12).
The Role of PKA in the Immunoprecipitation of VAMP2-If the effect of cytosolic fraction and cAMP is due to the activation of PKA, exogenous PKA should also induce the precipitation. We used purified catalytic subunit of PKA in the experiment. Although all reactions contained 2 mM ATP and 10 mM Mg 2ϩ , amounts that are sufficient for phosphorylation, PKA subunit alone did not induce the precipitation of VAMP2 (Fig. 4A, lane  4). In contrast, incubation with both cytosolic fraction and PKA subunit efficiently induced precipitation similar in extent to cytosolic fraction and cAMP (Fig. 4A, lanes 3 and 5). It was confirmed that the effect of cytosolic fraction and cAMP is at least partly due to the activity of PKA. Because addition of cytosolic fraction was necessary, a cytosolic factor other than PKA is also necessary for this reaction.
To determine whether this cytosolic factor is a protein or not, we treated cytosolic fraction with trypsin. Treatment with trypsin significantly decreased the ability of cytosol to allow immunoprecipitation (Fig. 4B, lane 3). Cytosolic fraction protected by trypsin inhibitor before incubation with trypsin retained the activity (Fig. 4B, lane 4). Therefore, the activity is attributable to some protein.
Another issue is the substrate of PKA. To determine whether the substrate is cytosolic factor(s) or granular membrane protein(s), we performed the phosphorylation reaction separately and mixed cytosolic and solubilized granule membrane fractions. After incubation of the cytosolic fraction with PKA at 30°C for 10 min, H89 (final concentration, 10 M) and the solubilized granule membrane fraction were added to the mixture and incubated at 30°C for an additional 10 min. VAMP2 was precipitated under these conditions as efficiently (Fig. 4B, lane 5) as seen with the cytosolic and solubilized granule membrane fractions incubated with PKA in the same reaction (Fig. 4B, lane 2). Next, the solubilized granule membrane fraction was incubated with PKA at 30°C for 10 min, following which H89 and the cytosolic fraction were added and incubated for an additional 10 min. In this case, the efficiency of precipitation was decreased (Fig. 4B, lane  6). Therefore, phosphorylation of some protein in the cytosolic fraction, but not in the granule membrane fraction, is sufficient for the immunoprecipitation of VAMP2.
However, the immunoprecipitation occurring due to the phosphorylation of the solubilized granule membrane fraction is not negligible (Fig. 4B, lane 6). This result may indicate that the inhibitory activity of H89 was incomplete. H89 did not completely inhibit the immunoprecipitation caused by incubation with cytosolic and solubilized granule membrane fractions and cAMP either (Fig. 3A, lane 5). Another possible explanation is that a small amount of the protein phosphorylated by PKA is also present in the granule membrane fraction. In that case, another cytosol-specific protein would also be required  lanes 1 and 3) and the presence (lanes 2, 4 -6) of cytosolic fraction (50 g), and in the absence (lanes 1 and 2) and the presence (lanes 3-6) of cAMP (0.2 mM). In lanes 5 and 6, H89 and H85 were added to samples, respectively. After incubation at 30°C for 10 min, immunoprecipitation with anti-SER4256 was performed. The results are representative of three independent experiments. B, comparison of the effect of cAMP and calcium ion. Solubilized granule membranes were suspended in the aliquots with 1 mM EGTA (lanes 1-3) or that buffered at 10 M calcium ions (lanes 4 and 5). Cytosolic fraction was added (lanes 2-5), and the samples were incubated without (lanes 1, 2, and 4) and with 0.2 mM cAMP (lanes 3 and 5). The results are representative of two independent experiments.
because incubation of the solubilized granule membrane and catalytic subunit of PKA without the cytosolic fraction induced no immunoprecipitation (Fig. 4A, lane 4). DISCUSSION We demonstrated here that the putative t-SNARE binding region of VAMP2 is exposed in a cAMP-dependent manner, which causes immunoprecipitation of VAMP2 with antibody that binds to Arg 47 -Asp 64 of VAMP2 (anti-SER4256). The ability of anti-SER4256 to immunoprecipitate VAMP2 probably reflects the state of VAMP2 and its binding protein in the solubilized granule membrane fractions. When anti-SER4256 cannot immunoprecipitate VAMP2, it suggests that the epitope on VAMP2 is masked by some unidentified protein(s). Disruption of this binding allows VAMP2 to be immunoprecipitated. The epitope of the antibody is included within the region that is necessary for binding to t-SNAREs (27). Therefore, through the exposure of this region, VAMP2 gains the ability to interact with t-SNAREs. In our previous study, we have shown that VAMP2 plays an essential role in cAMP-dependent amylase secretion (12). However, it has not been clarified how cAMP regulates the function of VAMP2. The main function of cAMP is thought to be to activate PKA, but it is unknown what protein mediates between the activation of PKA and exocytosis. The exposure of the t-SNARE binding region of VAMP2 also took place through the activation of PKA. This is the first report of a biochemical function of PKA in cAMP-mediated exocytosis.
A postulated model of this regulatory mechanism is shown in Fig. 5. VAMP2 is possibly masked by some protein X under the control conditions. Upon stimulation, activated PKA phosphorylates some cytosolic protein Y, which subsequently removes X from VAMP2 directly or indirectly. Consequently, VAMP2 interacts with t-SNAREs. Although the t-SNARE proteins involved in amylase secretion have not been identified, syntaxin4 and SNAP-23 are promising candidates because of their VAMP2 binding activity (22,28). The disassembly of the complex containing VAMP2 is probably the first step of cAMP-dependent exocytosis of amylase.
What proteins are X and Y? One candidate for X is t-SNARE.
However, X is distinct from syntaxin4, because syntaxin4 is not present in the granule membrane fractions. If X is t-SNARE, the removal of X from VAMP2 probably implies the disassembly of pre-existing SNARE complex on the granule membrane, which corresponds to the priming step of the exocytotic process. In this case, a candidate for the cytosolic factor Y is ␣ soluble N-ethylmaleimide-sensitive factor attachment protein (␣-SNAP). ␣-SNAP was reported to be phosphorylated by PKA, and its binding affinity to the SNARE complex was reduced by the phosphorylation (29). The meaning of the reduced binding affinity is not clear. However, the possibility was proposed that dissociation of ␣-SNAP is necessary for efficient disassembly of the SNARE complex (29). If this is the case, ␣-SNAP phosphorylation by PKA promotes the disassembly of the pre-existing SNARE complex. There is another possibility that X is not t-SNARE, but some inhibitor of secretion. In this case, the cytosolic factor Y removes X from VAMP2 to promote the exocytotic process, just as Rab-like small GTP-binding protein displaces a member of the Sec1 protein family from t-SNARE to initiate SNARE complex formation in yeast (30).
In either case, cAMP-dependent dissociation of X from VAMP2 induces the formation of the SNARE complex that is necessary for exocytosis. In parotid acinar cells, cAMP is a main inducer of exocytosis. Cytosolic free calcium, which alone causes only a small amount of amylase release, significantly augments cAMP-dependent secretion (26,31). Therefore, we predicted that the exocytotic process in amylase secretion can be separated into cAMP-dependent docking and sequential calcium-enhanced fusion steps (32). In chromaffin and neuronal cells, most vesicles are already docked to the target membrane and await only calcium ions to trigger the calcium sensor and allow membrane fusion (33,34). In contrast, it seems that the exocytotic process of amylase secretion is arrested before docking and that cAMP triggers the docking step. In this study, we found that the exposure of the t-SNARE binding region of VAMP2 occurred upon stimulation by cAMP but not by calcium elevation. These results support our hypothesis that the docking step of amylase secretion is initiated in a cAMP-dependent manner by activation of VAMP2. This regulatory mechanism is expected to be common to cAMP-dependent exocytotic systems.
Acknowledgments-We thank Dr. Amira Klip and Leonard Foster for the anti-syntaxin4 antibody and for helpful comments on the manuscript.  4) and with (lanes 2, 3, and 5) cytosolic fraction (50 g). cAMP was added to lane 3, and catalytic subunit of PKA (6 units) was added to lanes 4 and 5. The results are representative of three independent experiments. B, phosphorylation of cytosolic protein is necessary and sufficient for the immunoprecipitation of VAMP2. Solubilized granule membrane fractions were incubated without (lane 1) or with (lane 2) cytosolic fraction and catalytic subunit of PKA. Cytosolic fraction was incubated with trypsin, and after incubation, trypsin inhibitor was added (lane 3). Lane 4, cytosolic fraction was added with trypsin inhibitor first and then incubated with trypsin at 30°C for 10 min. Trypsintreated, and trypsin inhibitor and trypsin-treated cytosolic fraction were incubated with solubilized granule membrane fractions and catalytic subunit of PKA. Lane 5, cytosolic fraction was incubated with catalytic subunit of PKA at 30°C for 10 min. After that, 10 M H89 and solubilized granule membrane fraction were added, and it was incubated a second time at 30°C for 10 min. Lane 6, solubilized granule membrane fraction was incubated with catalytic subunit of PKA at 30°C for 10 min, and incubated a second time with H89 and cytosolic fraction at 30°C for 10 min. The results are representative of four independent experiments.
FIG. 5. A hypothesis of the regulatory mechanism of the interaction between v-and t-SNARE during the stimulation that causes exocytosis.