INTRODUCTION

Regulated exocytosis involves the highly controlled targeting, docking, and fusion of secretory vesicles to the plasma membrane. Studies using cell permeabilization and non-hydrolyzable GTP analogues such as GTPS¹ have indicated that G-proteins play key roles in regulated exocytosis (1-3). Rab proteins, members of the Ras-related small G-protein family, are well known to participate in various steps of intracellular vesicle trafficking including exocytosis (4-7). Isoforms of Rab3 have been localized to secretory granules of neuronal and non-neuronal cells, although their role in exocytosis is still not clear (7-9). Recently, we and others have reported that Rab3D (10) is localized to secretory granules of various tissues including exocrine pancreas and its localization implies that it may be involved in regulated exocytosis (11-13). In addition to small G-proteins, recent evidence suggest that heterotrimeric G-proteins or their isolated -subunits also play important roles in intracellular vesicle trafficking and vesicle formation in addition to their classical functions in receptor-coupled signal transduction. For example, a heterotrimeric G-protein(s) was shown to be required in endosome fusion in a cell-free system (14). In LLC-PK1 epithelial cells, when G_i3 was overexpressed on Golgi membranes, constitutive secretion was retarded (15). By using an ADP-ribosylation method, G_s as well as G_i/o on Golgi membranes were also demonstrated to participate in the regulation of vesicle formation in PC12 cells (16). In addition, evidence suggests heterotrimeric G-proteins may have a role in regulated exocytosis since G_o on secretory granules was recently shown to play an inhibitory role in calcium-stimulated norepinephrine release from chromaffin cells (17, 18). In insulin secreting B-cells, G_i on secretory vesicles has been demonstrated to be involved in mastoparan-induced insulin secretion (19). Although G_q/11 has recently been localized to Golgi membrane (20), its function in vesicle trafficking is still unknown.

The exocrine pancreas is a model system widely used for studying regulated exocytosis. Several small GTP-binding proteins have been shown to be present on zymogen granules, and two have recently been identified as Rab3D and Rab5 (12, 13, 21). However, little is known about the presence or function of heterotrimeric G-proteins on zymogen granules. Although the existence of a pertussis toxin-sensitive G-protein on pancreatic zymogen granules was previously suggested (22) its identity is still uncertain. In the current work, we have identified a heterotrimeric G-protein(s), which localizes to zymogen granules and appears to participate in regulated exocytosis. We found that both _q/11 and G-protein -subunit exist on zymogen granules. We also examined the effect of substance-P-related peptides on calcium-induced amylase release from streptolysin-O permeabilized acini. Substance-P-related peptide GPAnt-2a and GPAnt-1 were reported to antagonize G_q/11 and G_o/i, respectively (23) and thus these peptides are widely used for analyzing the function of heterotrimeric G-proteins on both plasma membranes and secretory vesicles (17, 18, 24). We found that GPAnt-2a, but not GPAnt-1, enhanced calcium-stimulated amylase secretion from streptolysin-O permeabilized pancreatic acini and that GPAnt-2a, but not GPAnt-1, inhibited GTPase activity on zymogen granules. These results suggest that G_q/11 on zymogen granules plays a tonic inhibitory role in calcium-regulated amylase release from pancreatic acini.

MATERIALS AND METHODS

Streptolysin-O was purchased from Welcome Diagnostic (Greenville, NC), chromatographically purified collagenase from Worthington Biochemicals, bovine serum albumin (Fraction V) from ICN Immunobiologicals (Lisle, IL), and protein A-agarose from Pierce. The following compounds were purchased from Sigma: Mg-ATP, creatine kinase, creatine phosphate, GTP, and activated charcoal. [-³²P]GTP (30 Ci/mmol) and ¹²⁵I-Bolton-Hunter CCK-8 (2200 Ci/mmol) were purchased from DuPont NEN (Boston, MA). GPAnt-1 and GPAnt-2A peptides were purchased from Sigma and Bio-Mol (Plymouth Meeting, PA), respectively. Affinity-purified polyclonal anti-_q/11 antibody (C-19L), anti-_o antibody (K-20), anti-_i (C-10L), anti-G-protein--subunit antibody (T-20), and anti-phospholipase C1 antibody (G-12) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-_s antiserum (RM-1) was purchased from DuPont NEN. Monoclonal antibody against rat pancreatic GP-2 was a gift from Dr. Anson Lowe (Stanford University). Polyclonal antibody (31B) to the  subunit of the sodium pump has been described previously (25). Secondary antibodies included peroxidase-coupled goat anti-rabbit immunogloblin G (IgG, Amersham) and fluorescein isothiocyanate-labeled goat anti-rabbit IgG (Sigma).

Purified zymogen granules were prepared by Percoll gradient separation by previously published procedures (13, 26-28). Their purity has been determined both by analysis of several subcellular enzymatic organelle markers and by electron microscopic analysis and shown to be enriched 100-400-fold compared with other organelles (26). Granule membranes were isolated by ultracentrifugation after lysis with nigericin as described previously (28). Plasma membrane-rich fractions were prepared by sucrose gradient centrifugation by previously published procedures (29). ¹²⁵I-CCK binding to plasma membrane and zymogen granule membrane was carried out as described previously (29) except that membranes were collected on Whatman GFF filters and washed three times with ice-cold buffer to terminate binding. Protein content of membranes was determined with the Bio-Rad protein assay kit, using bovine serum albumin as standard.

Electrophoresis was performed as described (28). Variable amounts of protein from each sample were loaded per lane onto 7.5% or 10% SDS-polyacrylamide electrophoresis gels and run at 200 V. After gel electrophoresis, proteins were transferred to nitrocellulose membranes at 30 volts overnight. Western blotting was carried out as described previously, using the enhanced chemiluminescence reagent to visualize the secondary antibody.

Freshly prepared pancreatic lobules were fixed for 2 h at 4 °C with a mixture of 2% formaldehyde (prepared from paraformaldehyde) and 0.25% glutaraldehyde in phosphate-buffered saline. Fixed tissue was rinsed in phosphate-buffered saline, cryoprotected with sucrose, and frozen as described previously (13, 30). Immunofluorescence localization of _q/11 in 5-µm thick cryostat sections followed the procedures described previously in detail (13, 31). Polyclonal anti-_q/11 was used at 10 µg/ml. Specificity of staining was assessed by preincubation of primary antibody with 10-fold excess by weight of peptide used to generate the antibody. For comparative purpose, some cryostat sections were exposed to monoclonal antibody to the zymogen granule membrane marker, GP-2. Sections were examined by epifluorescence microscopy and confocal fluorescence microscopy (Bio-Rad MRC-600). Digitized images were processed using Photoshop 3.0 software.

Acini were prepared by collagenase digestion as described previously (32). After isolation, acini were suspended in permeabilization buffer consisting of 20 mM PIPES (pH 7.0), 140 mM potassium glutamate, 0.91 mM MgCl₂, 5 mM EGTA, 1 mg/ml bovine serum albumin, 0.1 mg soybean trypsin inhibitor, 1 mM ATP (Mg salt), and 0.5 IU/ml streptolysin-O. For permeabilizing acini and introducing G-protein antagonist peptides into acini, aliquots of the acinar suspension were incubated at 37 °C for 1.5 min in the presence of various amounts of the indicated peptide, then placed in ice-cold water bath for 5 min and aliquoted into 200-µl samples. Amylase release was initiated by adding 200 µl of permeabilization buffer supplemented with CaCl₂ to give a final concentration of 1 mM free Mg²⁺ and the specified concentrations of free Ca²⁺ which were calculated using a computer program as described previously (32). After incubation for 5 min at 30 °C, samples were centrifuged for 10 s in an microcentrifuge. Amylase released into the supernatant during the incubation was quantified using the Phadebas Amylase Test (Pharmacia, Columbus, OH) and expressed as a percent of total amylase in the acini at the beginning of the incubation.

For immunoprecipitation of _q/11, purified zymogen granules were sonicated for 15 s in lysis buffer containing 25 mM Tris (pH 7.5), 1.5 mM dithiothreitol, 3 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride, 2 mM EGTA, 10 µg/ml leupeptin, and 0.5% Triton X-100. The insoluble fraction was removed by centrifugation at 10,000 × g for 10 min at 4 °C. The supernatant was incubated with 2 µg of anti-G_q/11 antibody at 4 °C for 2 h followed by incubation with protein A-agarose beads for 1 h. The G_q/11-bound beads were washed with the same buffer and used for Western blotting or assay of GTPase activity.

Purified zymogen granules (200 µg of protein) or immunoprecipitated _q/11 protein from purified zymogen granules (2 mg) were suspended in 60 µl of HEPES-buffered solution (50 mM HEPES, pH 7.2, 1 mM dithiothreitol). The reaction was initiated by adding 40 µl containing of 50 mM HEPES (pH 7.2), 1 mM dithiothreitol, 2.5 mM EGTA, 250 mM NaCl, 0.5 mg/ml creatine kinase, 5.5 mM creatine phosphate, 2.5 mM ATP, 2.5 µM GTP, and 3 µCi of [-³²P]GTP followed by incubation for 15 min at 30 °C. The reaction was terminated by the addition of 900 µl of cold phosphoric acid (pH 2.3) supplemented with 25% activated charcoal by weight. The samples were centrifuged at 10,000 × g for 15 min at 4 °C and the [³²P]P_i in the supernatant (300 µl) was determined by liquid scintillation counting. Nonenzymatic hydrolysis of [-³²P]GTP was subtracted from all data.

RESULTS

To determine the heterotrimeric G-protein -subunit(s) localized to zymogen granules, we performed Western blotting of purified zymogen granule membranes and a plasma membrane-rich fraction using antibodies specific to _q/11, _s, _o, and _i. Fig. 1 shows that _q/11, _s, and _o were all present in a plasma membrane-rich fraction; _i was only faintly visualized in plasma membranes and was therefore not studied further. Of the four G proteins, only _q/11 was detected on zymogen granule membranes in Western blotting. In addition, the Na⁺ pump  subunit was found on the plasma membrane but not granule membrane (Fig. 1). To quantitate the relative abundance of _q/11 on plasma membrane and zymogen granule membrane we carried out a dilution series of plasma membrane from 30 to 0.1 µg and compared the immunoblot signal to 30 µg of granule membranes. In two separate preparations 30 µg of granule membranes were comparable to 1 µg of plasma membranes. Based on the method of preparation and relative purity of granules (26) this is not likely due to contamination. To confirm this quantitatively, we evaluated CCK receptor binding as a well established pancreatic plasma membrane marker. Specific binding was observed on plasma membranes (total binding 20,173 ± 211 cpm, nonspecific 318 ± 29, mean ± S.E., n = 4). By contrast, no specific binding was present on zymogen granule membranes (total binding 306 ± 24 cpm, nonspecific 340 ± 42).

Western blotting of G_q/11, G_s, and G_o -subunits in zymogen granule membranes and plasma membrane-rich fraction.

Recently, multiple heterotrimeric G-protein -subunits were shown to be localized to the Golgi apparatus of pancreatic acinar cells using immunohistochemistry, but no -subunits were detected in the Golgi apparatus (33). These investigators suggested that G-protein -subunits are localized to the Golgi apparatus independent of -subunits. To determine whether _q/11 exists alone or as part of a complete G-protein complex, zymogen granule membranes were probed with a -subunit antibody directed against a common region in the COOH-terminal of all -subunits. A 37-kDa band was found to be present in both zymogen granule membranes and plasma membrane-rich fraction (Fig. 2A). These data suggest that _q/11 protein on zymogen granules maintains an interaction with -subunit as an intact heterotrimeric G-protein as is the case for the plasma membrane. We also carried out Western blotting using antiphospholipase C1 antibody; phospholipase C1 was detected in the plasma membrane-rich fraction, but not in purified zymogen granule membranes (Fig. 2B). These observations in total, confirm the purity of our zymogen granule and indicate that _q/11 in granule membrane preparations is not due to contamination with plasma membranes.

Western blotting of G-protein -subunits and phospholipase C1 in zymogen granule membranes and plasma membrane-rich fraction.

To further address the localization of _q/11, we analyzed the distribution of _q/11 protein in cryostat sections of pancreatic lobules by immunofluorescence and confocal fluorescence microscopy. As shown in Fig. 3A, _q/11 staining was pronounced along the basolateral membranes of acinar cells, and more weakly localized to the granule region in the apical cytoplasm. Although not quantitative, the weaker immunofluorescence signal over the granule area compared with the plasma membrane is consistent with the similar pattern resolved in Western blots (Fig. 1). When anti-_q/11 antiserum was preincubated with the peptide used to generate the antibody, the specific staining associated with both the basolateral membrane and the granules was abolished (Fig. 3B). Immunofluorescence staining of the apical cytoplasm with anti-_q/11 appeared at higher magnification to be punctate and to be associated specifically with the zymogen granules (Fig. 3C). As shown in this figure, no staining was apparent at the apical membrane of the acinar cells.

Immunofluorescence localization of _q/11 in pancreatic lobules by confocal fluorescence microscopy.

To study the function of G_q/11 on zymogen granule membranes, we utilized the SP-related peptides GPAnt-1 and GPAnt-2a which are known to be antagonists of G_i/o and G_q/11, respectively (23). When these peptides are introduced into permeabilized acini and amylase secretion is stimulated by 10 µM free calcium, an effect on heterotrimeric G-proteins at the plasma membrane to mobilize intracellular calcium can be excluded and the function of the G-proteins at steps distal to intracellular calcium mobilization can be evaluated. GPAnt-2a peptide potentiated amylase secretion induced by 10 µM free calcium from streptolysin-O permeabilized acini in a dose-dependent manner (Fig. 4). In contrast, GPAnt-1 did not alter calcium-stimulated amylase secretion. We also evaluated the peptide from the carboxyl-terminal of _q/11 used to generate the antibody and it had no effect on secretion (not shown). None of the peptides affected basal amylase secretion even at 100 µM. These data demonstrated that GPAnt-2a peptide specifically potentiated calcium-stimulated amylase release at a late step of exocytosis.

To confirm that GPAnt-2a peptide augments calcium-stimulated amylase secretion from permeabilized acini by antagonizing G-proteins on zymogen granules, we evaluated the GTPase activity of zymogen granules. As shown in Fig. 5A, 100 µM GPAnt-2a peptide inhibited GTPase activity on zymogen granules approximately 25%. GPAnt-1 peptide, however, had no effect on the GTPase activity on zymogen granules. Since multiple small G-proteins are also known to be localized on zymogen granules (27, 34), total GTPase activity of zymogen granules most likely represents the activity of multiple G-proteins. Therefore, the fact that GPAnt2a inhibited only 25% of the GTPase activity on zymogen granules is not surprising.

To further determine that GPAnt-2a peptide specifically inhibits G_q/11 protein function on zymogen granules, we examined the inhibitory effect of G-protein antagonist peptides on GTPase activity of G_q/11 immunoprecipitated from purified zymogen granules. Fig. 5B shows immunoprecipitated G_q/11 protein from zymogen granules visualized by Western blotting. Using this immunoprecipitant, the effect of G-protein antagonist peptides on G_q/11 GTPase activity was assayed. As shown in Fig. 5C, GPAnt-2a peptide markedly inhibited immunoprecipitated G_q/11 GTPase activity by about 60%. GPAnt-1 peptide, however, had no effect on GTPase activity of G_q/11. Taken together, these data suggest that the inhibition of GTPase activity of zymogen granules by GPAnt-2A peptide results in the potentiation of calcium-stimulated amylase secretion from permeabilized acini.

DISCUSSION

Transduction and amplification of receptor-mediated signals to phospholipase C at the level of the plasma membrane is the only currently established function of G_q/11 (35, 36). Although some studies have predicted the participation of G_q/11 in intracellular vesicle trafficking (20, 37), such a role has not been conclusively demonstrated. In the present study, we have described a role for G_q/11 on secretory vesicles in exocytosis. We showed the localization of G_q/11 on zymogen granules by Western blotting, immunoprecipitation, and immunohistochemistry. We next demonstrated that G_q/11 antagonist peptide GPAnt-2a augmented calcium-stimulated amylase release from permeabilized acini. Furthermore, we observed that GPAnt-2a inhibited GTPase activity of both zymogen granules and immunopurified G_q/11 from zymogen granules. Accordingly, it is suggested that GPAnt-2a enhanced calcium-induced amylase secretion from permeabilized acini by inhibiting GTPase activity of G_q/11 on zymogen granules.

Of central importance to our evaluation of the role of _q/11 on zymogen granule membranes is the purity of the granules and the certainty that the observed signal is not due to contamination with G_q/11-rich plasma membrane. The preparation of granules makes use of the high density of intact granules which band in Percoll gradients separate from other organelles and are enriched 250-500-fold relative to DNA (nuclei), glutamate dehydrogenase (mitochondria), and NADH dehydrogenase (endoplasmic reticulum) (26). In a recent study the purified granule preparation was used to show the absence of inositol 1,4,5-trisphosphate receptors or inositol 1,4,5-trisphosphate-induced Ca²⁺ release in purified granules (38) although both are present in contaminating membranes present in crude granules prepared by simple differential centrifugation. Finally, in the present work _o, _s,  subunits of the sodium pump, phospholipase C1, and CCKA receptors were all identified in the plasma membrane but not zymogen granule membranes; if the granule membranes were contaminated with as little as 1% of plasma membranes we would have detected saturable CCK binding in the granule membranes. The results with CCK binding are especially useful since there is a large and quantifiable signal in plasma membranes. Finally, the immunohistochemistry reveals specific labeling of the zymogen granules that could be completed by preincubating antisera with the immunizing peptide. Thus, the data as a whole makes a convincing case that _q/11 is present on the zymogen granules.

G-proteins have been suggested to be involved in a late step of exocytosis as a result of observations obtained using permeabilized cell systems. When a non-hydrolyzable GTP analogue, GTPS was introduced into neutrophils (1), mast cells (2), insulin secreting RIN cells (3), and chromaffin cells (39), in the presence of free calcium, exocytosis was markedly enhanced. These findings strongly supported the involvement of G-proteins in the late steps of exocytosis. However, since GTPS can activate both heterotrimeric G-proteins and small G-proteins, other reagents specific to trimeric G-proteins are required to analyze their function in exocytosis. AlF₄ and mastoparan, both of which are known to be specific activators of heterotrimeric G-proteins, have been intensively used in permeabilized cell systems. When AlF₄ was introduced into permeabilized secretory cells, their calcium-stimulated exocytosis was enhanced (40, 41) or attenuated (17). These findings strengthen the idea that heterotrimeric G-proteins are involved in the late steps of exocytosis. Although AlF₄ is a powerful tool for studying heterotrimeric G-proteins, it activates all isoforms. By contrast, mastoparan, an amphiphilic tetradecapeptide purified from wasp venom, is more specific to inhibitory heterotrimeric G-proteins G_i and G_o (42, 43). Application of mastoparan to permeabilized cell systems suggest the direct linkage of G_i or G_o to the exocytotic machinery (44-46). In addition to mastoparan, pertussis toxin is also a specific tool to analyze G_i and G_o. It catalyzes their ADP-ribosylation, and results in their non-responsiveness to activating signals. Thus pertussis toxin pretreatment affected the exocytosis induced by exogenously introduced free calcium into permeabilized chromaffin cells (47, 48). Therefore, the participation of mastoparan and pertussis toxin sensitive G-proteins, G_i and G_o, in the late steps of exocytosis have been intensively studied in these cells. G_i and G_o were recently found on small synaptic vesicles of neuronal cells by immunocytochemistry, and G_o has also been found on large dense core vesicles of adrenal chromaffin cells (49). In exocrine tissues, both G_i and G_o have been reported on parotid secretory granule membranes by Western blotting (50). Since various proteins on secretory vesicles play important roles at the step of vesicle docking and fusion with plasma membrane (51, 52), heterotrimeric G-proteins on secretory vesicles are now assumed to be one of the G-proteins involved in the late steps of exocytosis. Recently, Vitale et al. (17, 18) have shown that G_o on secretory granules has an inhibitory effect on the final stages of exocytosis in chromaffin cells. In insulin-secreting -cells, Konrad et al. (19) have described that G_i on insulin containing vesicles plays a stimulatory role in the mastoparan-induced insulin secretion. These data suggest that different heterotrimeric G-proteins on secretory granules play distinct roles in regulated exocytosis. Concerning other heterotrimeric G-proteins, G_s is shown to be localized to secretory granule membranes of neuroendocrine tissues and parotid glands (20, 48). However, little is known about its function.

G_q/11 has not previously been reported on secretory granules although it was shown to exist in the Golgi apparatus in some neuroendocrine cells (20) and pancreatic acini (33). Its function in the Golgi is still unclear. In the current study, we have shown that G_q/11 is present on pancreatic zymogen granules and acts to inhibit calcium-triggered exocytosis. These findings are consistent with the previous evidence that G-proteins interact with intracellular calcium to bring about exocytosis (2, 53). Moreover, our findings that G_q/11 antagonist GPAnt2a potently augmented calcium-induced amylase secretion, but not basal secretion, reinforces the synergistic role of G_q/11 with intracellular free calcium in regulated exocytosis.

Currently, the process of exocytosis is thought to occur in two phases. The first step of docking and fusion of primed secretory vesicles to plasma membranes is completed in the first 5 min. The second step is the sequential release of vesicles from a reserve pool (54). According to this theory, our findings that GPAnt-2a potently enhanced the initial amylase release over a 5-min incubation suggests that the target protein of GPAnt-2a, G_q/11, acts on the docking and fusion steps of exocytosis in acinar cells. Taken together, it is reasonable to speculate that the G_q/11 target of GPAnt-2a is on the zymogen granules. To confirm our hypothesis, we examined the GPAnt-2a effect on both zymogen granules and G_q/11 isolated from zymogen granules. GPAnt-2a inhibited both GTPase activities. Thus we have concluded that G_q/11 on zymogen granules tonically inhibits calcium-induced amylase release from pancreatic acini.

Heterotrimeric G-proteins exist on plasma membranes as a complex of , , and  subunits at resting state, and separates into free  subunit and a complex when activated; , subunits are assumed always to exist as a complex (55, 56). We have shown that -subunit as well as _q/11 is localized to zymogen granule membranes. Although we did not evaluate the presence of  subunits, it is reasonable to conclude that G_q/11 exists as a complete complex of all three subunits on zymogen granules.

In the classical signal transduction pathway involving G_q/11 on the plasma membrane, phospholipase C is the sole effector protein known thus far. In our Western blotting studies, we could detect phospholipase C1 in the plasma membrane-rich fraction but not in zymogen granule membranes (Fig. 2B). Although this observation reinforces the purity of the zymogen granule membrane preparation, it concomitantly implies that the effector of the G_q/11 on zymogen granules still remains uncertain. F-actin has been assumed to play inhibitory roles in exocytosis by blocking secretory vesicle movement toward the plasma membrane (57). Vitale et al. speculated that the G-protein which is involved in the negative control of exocytosis may be coupled with some component controlling the subplasmalemmal cytoskeleton and the movement of secretory granules to the exocytotic sites (17). Their speculation is based on results concerning G_o of chromaffin secretory granules. However, the interaction between G_o and F-actin is not yet elucidated. On the other hand, Ibarrondo et al. (37) have recently reported the close association of G_q/11 with F-actin filaments in mammary tumor cell line. Moreover, it was shown that assembly and disassembly of F-actin are essential to the final steps of exocrine pancreatic exocytosis (54). Taken together, it is more likely for G_q/11 that some protein(s) which regulates the actin network organization might be the effector of the G-proteins on secretory granules. Alternatively, the possibility still remains that undefined isoforms of phospholipase C might exist on zymogen granules as an effector of G_q/11. In any case, further studies are needed to elucidate the effector of G_q/11 on zymogen granules.

In conclusion, we have demonstrated a tonic inhibitory action of G_q/11 which localizes to zymogen granules in pancreatic acinar exocytosis. These observations provide new insights for understanding the synergistic regulation of exocytosis by intracellular calcium and G-proteins.