Trimeric G Proteins Control Exocytosis in Chromaffin Cells G o REGULATES THE PERIPHERAL ACTIN NETWORK AND CATECHOLAMINE SECRETION BY A MECHANISM INVOLVING THE SMALL GTP-BINDING PROTEIN Rho*

Besides having a role in signal transduction, hetero- trimeric G proteins may be involved in membrane trafficking events. In chromaffin cells, G o is associated with secretory organelles and its activation by mastoparan inhibits the ATP-dependent priming of exocytosis. The effectors by which G o controls exocytosis are currently unknown. The subplasmalemmal actin network is one candidate, since it modulates secretion by controlling the movement of secretory granules to the plasma mem- brane. In streptolysin-O-permeabilized chromaffin cells, activation of exocytosis produces disassembly of corti- cal actin filaments. Mastoparan blocks the calcium-evoked disruption of cortical actin, and this effect is specifically inhibited by antibodies against G (cid:97) o and by a synthetic peptide corresponding to the COOH-terminal domain of G (cid:97) o . Disruption of actin filaments with cytochalasin E and Clostridium perfringens iota toxin par- tially reverses the mastoparan-induced inhibition of secretion. Furthermore, the effects of mastoparan on cortical actin and exocytosis are greatly reduced in cells treated with Clostridium botulinum C3 exoenzyme, which specifically inactivates the small G protein Rho. We propose exocytotic illustrates a quantitative analysis obtained by cells displaying intact cortical actin ring but no surface DBH patches (phalloidin-positive exocytotic DBH patches, but a fragmented cortical actin network (DBH-positive 200 single-rounded were examined per coverslips. represents typical obtained in the rhodamine (phalloidin) fluorescein focal by a

Studies on diverse secretory cell types have highlighted the potential roles of heterotrimeric G proteins in intracellular membrane trafficking events (1)(2)(3). ␣ and ␤␥ subunits of G i and G o proteins have been found associated with the membrane of secretory granules in various neuroendocrine cells (4 -6), suggesting a role in Ca 2ϩ -regulated exocytosis. Accordingly, the participation of a plasma membrane-bound G i3 protein in the late stages of exocytosis in mast cells has been demonstrated (7). Direct control of exocytosis by G i and G o proteins has also been described in insulin-secreting cells (8) and in chromaffin cells (6,9,10). Thus regulated exocytosis may represent a possible effector pathway for trimeric G proteins, although the mechanism by which this class of G proteins relates to the exocytotic machinery remains to be elucidated.
In chromaffin cells, stimulation of the secretory granuleassociated G o by mastoparan and compounds known to stimu-late G␣ subunits inhibits catecholamine secretion by selectively interfering with the ATP-dependent priming step of exocytosis (6,9). Although the molecular machinery underlying the ATP-dependent reaction is not clearly understood, this finding predicts that the granule-bound G o may control an effector related to the first stages of the exocytotic pathway, presumably the recruitment of secretory granules and/or the preparation of the docking/fusion machinery. Many secretory cells display a cortical network of actin filaments that forms a physical barrier to exocytosis for the majority of secretory granules, since they are excluded from the subplasmalemmal zone (11)(12)(13)(14). Activation of exocytosis produces disassembly of the actin network in several secretory cell types, including chromaffin cells (15,16), mast cells (17), pancreatic acinar cells (18), and synaptosomes (19). A close correlation between the disassembly of the actin cytoskeleton, the number of secretory granules in the cortical areas and the initial rate in secretion has been also reported (14). Actin filament disassembly is not by itself a sufficient trigger to allow exocytosis to occur (15). However, rises in intracellular calcium are not capable of stimulating catecholamine release if the peripheral actin barrier has not previously been removed (20). Thus, the cortical actin network represents a dominant negative clamp, which blocks the exocytotic process and its disassembly may form part of the priming reaction.
Previous reports by several investigators have suggested possible interactions between trimeric G proteins and the actin cytoskeleton. In neutrophils, studies based on the use of mastoparan, aluminium fluoride, and pertussis toxin, which specifically ADP-ribosylates G i and G o proteins, suggest that trimeric G proteins are closely linked to the actin organization (21)(22)(23). In mast cells, a trimeric G protein seems to participate in the reorganization of the actin cytoskeleton in response to cell activation (24). The specific association of trimeric G proteins with the actin cytoskeleton during thrombin receptormediated platelet activation has also been reported (25). The aim of the present work was to assess whether the cortical actin network represents a possible effector by which the granule-bound G o controls the exocytotic pathway in chromaffin cells. Using streptolysin-O (SLO) 1 -permeabilized cells, we show that the introduction of mastoparan into the cytosol inhibits the disruption of the subplasmalemmal actin network in calcium-stimulated cells. This effect can be selectively reversed by affinity-purified antibodies prepared against G␣ o and by a syn-thetic peptide corresponding to the COOH-terminal sequence of G␣ o . Furthermore, the mastoparan-induced inhibition of secretion can be partially reversed by agents known to affect the assembly of actin and by the Clostridium botulinum C3 ADPribosyltransferase, which specifically inactivates the small GTPase Rho by ADP-ribosylation. Our results raise the intriguing possibility that the secretory granule-associated G o protein controls the priming step of exocytosis by modifying the actin cytoskeleton underlying the plasma membrane through a sequence of events possibly implicating Rho.

MATERIALS AND METHODS
Culture of Chromaffin Cells-Chromaffin cells were isolated from fresh bovine adrenal glands by retrograde perfusion with collagenase and purified on self-generating Percoll gradients (26). Cells were suspended in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, and containing cytosine arabinoside (10 M), fluorodeoxyuridine (10 M), streptomycin (50 g/ml), and penicillin (50 units/ ml). Cells were cultured as monolayers either on 24 multiple 16-mm Costar plates (Costar, Cambridge, MA) at a density of 2.5 ϫ 10 5 cells/ well or on fibronectin-coated glass coverslips at a density of 2 ϫ 10 5 cells. Experiments were performed 3-7 days after plating. 2) and twice with Ca 2ϩ -free Locke's solution (containing 1 mM EGTA). Cells were subsequently permeabilized for 2 min with 15 units/ml SLO (Institut Pasteur, Paris, France) in 200 l/well Ca 2ϩfree KG medium (150 mM potassium glutamate, 10 mM PIPES, pH 7.0, 5 mM nitrilotriacetic acid, 0.5 mM EGTA, 5 mM MgATP, 4.5 mM magnesium acetate, 0.2% bovine serum albumin). Extracellular fluids were then removed, and cells were incubated 10 min in 200 l/well Ca 2ϩ -free KG medium in the presence of either mastoparan or GAP-43 and when indicated G␣ peptides, anti-G␣ antibodies, C3 transferase, or iota toxin (Ia component). Cells were subsequently stimulated for 10 min with KG medium containing CaCl 2 . The free Ca 2ϩ concentration in the KG medium was calculated by a computer program according to Flodgaard and Fleron (27), kindly provided by T. Saermark, University of Copenhagen, Denmark, using the stability constants given by Sillen and Martell (28).
[ 3 H]Noradrenaline Release-Catecholamine stores were labeled by incubating chromaffin cells with [ 3 H]noradrenaline (13.3 Ci/mmol; Amersham Corp., Les Ulis, France) for 45 min in Locke's solution. Cells were then washed, permeabilized with SLO, and stimulated with calcium as described above. [ 3 H]Noradrenaline release after stimulation was determined by measuring the radioactivity present in the incubation medium and in cells after precipitation with 10% (w/v) trichloroacetic acid. The amount of released [ 3 H]noradrenaline is expressed as a percentage of total radioactivity present in the cells before Ca 2ϩ -induced stimulation. When indicated, data are given as the net secretory values obtained by subtracting the basal release established in Ca 2ϩfree KG medium from the total release measured in the KG medium containing 20 M free calcium. Release experiments were performed in triplicate on at least two different cell preparations. In the figures that are representative of a typical experiment, data are given as the mean of triplicate determinations on the same cell preparation Ϯ S.E.
Peptides and Proteins-Mastoparan was obtained from Sigma. Synthetic peptides were obtained from Neosystem (Strasbourg, France): COOH-terminal G␣ o (ANNLRGCGLY) and COOH-terminal G␣ i3 (KNNLKECGLY) were further purified by high performance liquid chromatography and dissolved in Ca 2ϩ -free KG medium at 10 mM. GAP-43 was purified from a cytosolic fraction obtained from the bovine brain according to a previously published method (31).
Immunocytochemistry and Confocal Laser Scanning Microscopy-Chromaffin cells grown on fibronectin-coated glass coverslips were washed with Locke's solution, permeabilized, and incubated for 10 min in Ca 2ϩ -free KG buffer (resting cells) or in KG buffer containing 20 M free Ca 2ϩ (stimulated cells). Cells were subsequently fixed for 15 min in 4% paraformaldehyde in 0.12 M sodium/potassium phosphate, pH 7.0, and for a further 10 min in fixative containing 0.1% Triton X-100. Following several rinses with phosphate-buffered saline (PBS), cells were pretreated with 3% bovine serum albumin (BSA), 10% normal goat serum in PBS to reduce nonspecific staining.
To identify chromaffin cells, cells were incubated for 1 h at 37°C with antibodies against DBH diluted to 1:1200 in PBS containing 3% BSA in a moist chamber. Cells were then washed with PBS and subsequently incubated for 1 h at 37°C with goat anti-rat IgG conjugated to DTAF diluted to 1:200 in PBS containing 3% BSA. The transient accessibility of DBH on the plasma membrane of stimulated chromaffin cells (30) was tested by incubating SLO-permeabilized cells for 10 min in KG medium containing 20 M free Ca 2ϩ in the presence of anti-DBH antibodies diluted to 1:50. Cells were then fixed, washed, and processed for immunofluorescence labeling.
Actin filaments (F-actin) were stained by incubation with rhodamine (TRITC)-conjugated phalloidin (Sigma) at a concentration of 0.5 g/ml in PBS for 15 min at room temperature. Coverslips were then extensively washed with PBS, rinsed with water, and mounted in Moviol 4 -88 (Hoechst). The percentage of chromaffin cells displaying an intact cortical actin network was estimated by double labeling with rhodamine-conjugated phalloidin and anti-DBH antibodies and counting 200 single-rounded DBH-positive cells per coverslip in randomly selected areas of the coverslips. Each DBH-labeled cell was classified as having either a continuous and homogeneous cortical rhodamine fluorescent ring or a disrupted one. To avoid personal bias, a single-blind method was used: the cells were examined and classified without knowing they were from control or treated preparations.
Immunofluorescence staining was monitored with a Zeiss laser scanning microscope (LSM 410 invert) equipped with a planapo oil (63ϫ) immersion lens (numerical aperture ϭ 1.4). DTAF emission was excited using the argon laser 488-nm line, whereas TRITC was excited using the He/Ne laser 543-nm line. The emission signals were filtered with a Zeiss 515-565-nm filter (DTAF emission) or with a long pass 595-nm filter (TRITC signal). Cells were subjected to optical serial sectioning to produce images in the X-Y plane. Each optical section was scanned eight times to obtain an averaged image. Images were recorded digitally in a 768 ϫ 576-pixel format. Nonspecific fluorescence was assessed by incubating cells with the secondary fluorescent antibodies and measuring the average intensity value for each fluorochrome. This value was then subtracted from all specific images.

Effect of Mastoparan on the Cortical Actin Network in SLO-
permeabilized Chromaffin Cells-The effect of mastoparan on the peripheral actin cytoskeleton was analyzed in SLO-permeabilized chromaffin cells by confocal microscopy using rhodamine-conjugated phalloidin, which binds to filamentous but not to monomeric actin. In permeabilized cells incubated in Ca 2ϩfree medium (Fig. 1A), rhodamine-phalloidin fluorescence was most intense at the cell periphery forming a continuous and homogeneous cortical ring, in agreement with the fact that in chromaffin cells the majority of actin filaments are concentrated in the subplasmalemmal region (14 -16). Stimulation with 20 M free calcium strongly reduced the binding of rhodamine-phalloidin in the cell periphery, revealing the disruption of the cortical actin filaments (Fig. 1B). The introduction of 20 M mastoparan into the incubation medium of permeabilized cells had no detectable effect on actin filaments in resting cells ( Fig. 1C) but totally abolished the disassembly of cortical actin observed in response to a rise in cytosolic calcium (Fig.  1D). In contrast, preincubation with mastoparan before permeabilization did not affect the Ca 2ϩ -evoked actin disassembly ( Fig. 1, E and F), indicating that the ability of mastoparan to stabilize the actin network was related to its direct introduction into the cytoplasm through the pores created in the plasma membrane. Fig. 1G illustrates a quantitative analysis of the chromaffin cell population displaying an intact cortical actin network under resting and stimulating conditions. Stimulation with 20 M free Ca 2ϩ strongly reduced the percentage of SLOpermeabilized cells presenting an intact fluorescent actin ring. Mastoparan inhibited the Ca 2ϩ -evoked disruption of cortical actin, but only when added to the incubation medium after SLO permeabilization. Thus, the presence of mastoparan into the cytosol of permeabilized chromaffin cells stabilized the peripheral actin network and thereby prevented its disassembly upon Ca 2ϩ -induced stimulation.

Mastoparan Prevents the Ca 2ϩ -evoked Disruption of Cortical Actin by Stimulating an Endogenous Trimeric G o Protein-
Mastoparan is a tetradecapeptide that selectively activates G i and G o proteins by inserting into membranes and forming an ␣-helix which resembles the trimeric G protein interacting domain of G protein-coupled receptors (34). On the other hand, mastoparan is an amphiphilic peptide, and its introduction into permeabilized cells has been reported to nonspecifically perturb intracellular membranes (35). To assess whether mastoparan blocked the actin network by stimulating an endogenous trimeric G protein, we attempted to antagonize the effect of mastoparan with affinity-purified antibodies prepared against G␣ o and G␣ i3 , and with synthetic peptides corresponding to the carboxyl terminus of G␣ o and G␣ i3 .
The effect of the carboxyl terminus G␣ o and G␣ i3 peptides was examined because such peptides prevent the stimulation of G o and G i proteins by their respective receptors or by mastoparan (36,37). Fig. 2A illustrates the effect of the synthetic G␣ o and G␣ i3 peptides on the mastoparan-induced inhibition of actin disassembly. Permeabilized chromaffin cells were exposed to mastoparan in the presence or absence of G␣ o and G␣ i3 peptides and subsequently stimulated with 20 M free Ca 2ϩ . The number of chromaffin cells displaying an intact cortical actin ring was estimated following rhodamine-phalloidin staining. In resting cells, neither mastoparan nor G␣ peptides modified the percentage of cells having an intact cortical actin network ( Fig. 2A). Mastoparan inhibited the disassembly of peripheral actin filaments in Ca 2ϩ -stimulated cells, and this inhibition was unaffected by the presence of G␣ i3 peptide. How-ever, the stabilizing effect of mastoparan was strongly reduced in the presence of G␣ o peptide.
Mastoparan has been reported to activate trimeric G i and G o proteins by interacting with the carboxyl terminus of the ␣ subunit (38). Thus, the effect of affinity-purified antibodies raised against the carboxyl terminus of G␣ o and G␣ i3 was examined. In the absence of Ca 2ϩ , the percentage of cells with an intact cortical actin network was not modified by the introduction of mastoparan and anti-G␣ antibodies into the cytosol of permeabilized cells (Fig. 2B). However, the anti-G␣ o antibodies selectively blocked the stabilizing effect of mastoparan on peripheral actin in Ca 2ϩ -stimulated cells (Fig. 2B), in agreement with the results obtained with the synthetic G␣ o peptide. In contrast, anti-G␣ i3 antibodies did not modify the inhibitory effect of mastoparan on Ca 2ϩ -evoked disassembly of cortical actin (Fig. 2B).
These findings support the idea that mastoparan blocks the Ca 2ϩ -evoked disruption of cortical actin network by stimulating an intracellular trimeric G o protein. It is noteworthy that the G␣ o peptide and anti-G␣ o antibodies reverse at similar concentrations both the inhibitory effect of mastoparan on the ATP-dependent priming step of secretion (6) and the stabilizing effect of mastoparan on the peripheral actin cytoskeleton, suggesting a close link between these two events.
We reported previously that the "growth-associated protein" GAP-43 (neuromodulin), a neuronal protein enriched in presynaptic terminals, specifically stimulates the secretory granule-associated G o when introduced into the cytosol of chromaffin cells and thereby inhibits the ATP-dependent priming step of Ca 2ϩ -regulated secretion (31). To confirm that mastoparan inhibits the Ca 2ϩ -evoked disassembly of cortical actin by activating the granule-bound G o protein, we examined whether bovine brain GAP-43 affected actin organization like mastoparan. Permeabilized chromaffin cells were incubated in the presence of 1 M cytosolic GAP-43 or 20 M mastoparan and subsequently stimulated with calcium. Cells were then fixed and labeled with rhodamine-phalloidin and anti-DBH antibodies to visualize actin filaments in chromaffin cells. Table I shows that GAP-43 did not affect the peripheral actin cytoskeleton in resting cells since the percentage of cells displaying an intact actin ring was similar in control cells and in cells incubated with either mastoparan of GAP-43. However, we found that GAP-43 mimicked the inhibitory effect of mastoparan on Ca 2ϩevoked actin disassembly (Table I). In the presence of 1 M cytosolic GAP-43, approximately 70% of the Ca 2ϩ -stimulated cells still had an intact cortical actin ring. These results strengthen the idea that activation of the secretory granuleassociated G o protein can stabilize the cortical actin network and prevent its Ca 2ϩ -induced dissociation upon cell stimulation.
Correlation between the Mastoparan-induced Inhibition of Secretion and the Stabilization of the Cortical Actin Network-Exocytosis can be visualized by immunofluorescence in living cells with anti-DBH antibodies present in the cell incubation medium (16,30,39). DBH, which is exclusively located on the inner face of secretory granule membranes, is exposed on the cell surface during exocytosis. Thus, the secretory activity can be evaluated by the appearance of fluorescent patches at the cell surface corresponding to DBH immunoreactivity. To correlate the exocytotic activity with the subplasmalemmal actin network organization, SLO-permeabilized cells were stimulated with 20 M free calcium in the presence of anti-DBH antibodies. Cells were then washed rapidly, fixed, and stained with rhodamine-conjugated phalloidin. DBH immunoreactivity was detected with fluorescein-conjugated secondary antibodies. Confocal analysis indicated that fluorescent surface patches corresponding to DBH were generally observed in cells having a disrupted peripheral actin network (Fig. 3). Conversely, cells with a strong fluorescent actin ring were not labeled with DBH antibodies (Fig. 3). To probe the effect of mastoparan, the number of cells displaying a fluorescent actin ring and the number of cells labeled with anti-DBH antibodies were counted in randomly select areas of the coverslips. We found that mastoparan strongly reduced the Ca 2ϩ -evoked exocytotic activity visualized by the appearance of fluorescent DBH surface patches (Fig. 3), in agreement with our previous results, indicating that mastoparan inhibits the secretion of catecholamines in both intact and permeabilized chromaffin cells (9,40). Furthermore, the mastoparan-induced decrease in DBH labeling was accompanied by a parallel increase in phalloidinstaining in Ca 2ϩ -stimulated cells (Fig. 3), a result indicative of a close relationship between the inhibitory effect of mastoparan on secretion and actin disassembly.
To further characterize the correlation between the effect of mastoparan on the actin cytoskeleton and the exocytotic process, we used two distinct actin filament-disrupting molecules, namely cytochalasin E and C. perfringens iota toxin, and examined whether these molecules interfere with the mastoparan-induced inhibition of Ca 2ϩ -evoked secretion. At 20 M, mastoparan inhibited Ca 2ϩ -stimulated [ 3 H]noradrenaline release in SLO-permeabilized chromaffin cells by approximately 75% (Fig. 4). Treatment with cytochalasin E or iota toxin partially reversed the mastoparan-induced inhibition of secretion. Preincubation of cells with 20 M cytochalasin E reduced the maximal inhibitory effect of mastoparan to 42% (Fig. 4). Higher concentrations of cytochalasin E did not further reduce the inhibitory effect of mastoparan (data not shown). Incubation of SLO-permeabilized cells with 10 g/ml iota toxin inhibited to a similar extent the effect of mastoparan on Ca 2ϩ -evoked secretion (Fig. 4). In parallel experiments, we verified that both  of the peripheral actin network SLO-permeabilized chromaffin cells were incubated for 10 min with calcium-free KG medium alone (control) or containing either 20 M mastoparan or 1 M GAP-43. Cells were subsequently incubated with calcium-free KG medium or stimulated with KG medium containing 20 M free calcium. Cells were then fixed and labeled with rat anti-DBH antibodies detected with fluorescein anti-rat antibodies in combination with rhodamine-conjugated phalloidin. 200 DBH-positive cells were counted per coverslips and classified as having an intact or disrupted cortical rhodamine fluorescent ring. The proportion of cells displaying an intact actin network was expressed as the percentage of total counted cells. GAP-43 mimicked the stabilizing effect of mastoparan on cortical actin. [ cytochalasin E and iota toxin disrupted the cortical actin network visualized by rhodamine-phalloidin staining in control and mastoparan-treated cells (data not shown). These experiments indicate that the inhibition of secretion by mastoparan is at least partially linked to the stabilization of the cortical actin network. However, mastoparan may also interfere with some other important step of the exocytotic pathway to account for the residual inhibition of secretion observed in cells having their peripheral cytoskeleton disrupted by cytochalasin E or iota toxin.

Effect of C. botulinum C3 Transferase on Mastoparan-induced Stabilization of Cortical Actin and Inhibition of Secretion-
The small GTP-binding protein Rho is known to regulate the actin cytoskeleton organization (41,42). To investigate the possible implication of Rho in the mastoparan-induced actin stabilization, we used the bacterial exoenzyme C. botulinum C3 ADP-ribosyltransferase, which specifically inactivates Rho. Chromaffin cells were permeabilized with SLO, incubated with mastoparan in the presence or absence of C3 transferase, and subsequently stimulated with calcium. Cells were then fixed and processed to visualize actin filaments. Confocal fluorescent images of resting and stimulated control cells and C3 transferase-treated cells are shown in Fig. 5A. Under resting conditions, incubation with 10 g/ml C3 transferase generally preserved the cortical actin network, although the peripheral rhodamine-phalloidin fluorescence often appeared less dense, suggesting some fine modifications in the organization of the actin cytoskeleton. The proportion of cells displaying an intact cortical actin ring under each experimental conditions is illustrated in Fig. 5B. Treatment with C3 transferase did not modify the proportion of cells having a disrupted cortical actin ring upon Ca 2ϩ -induced stimulation. As expected, mastoparan blocked the Ca 2ϩ -evoked actin disassembly. This stabilizing effect of mastoparan was completely abolished in cells incubated with C3 transferase (Fig. 5).
We also examined the effect of C3 transferase on the mastoparan-induced inhibition of secretion. SLO-permeabilized chromaffin cells were incubated with increasing concentrations of C3 transferase in the presence or absence of mastoparan and subsequently stimulated with calcium. Treatment with C3 transferase did not significantly modify the Ca 2ϩ -evoked catecholamine release, but abolished the mastoparan-induced inhibition in a dose-dependent manner. As illustrated in Fig. 6, mastoparan inhibited secretion by 63% in control cells, and this inhibition was progressively reduced to 16% in the presence of increasing concentrations of C3 transferase. Thus, the inactivation of Rho by C3 transferase prevented mastoparan from stabilizing the actin cytoskeleton and inhibiting secretion, indicating that Rho might be involved in the pathway by which the granule-associated G o protein controls the organization of the cortical cytoskeleton in chromaffin cells. DISCUSSION We previously investigated the function(s) of trimeric G proteins in regulated exocytosis in chromaffin cells using mastoparan (6,9,10,40). Mastoparan is a peptide from wasp venom that stimulates the GTPase of G i and G o proteins by a mechanism that is virtually identical with that of agonist-bound receptors (34,38). We found that mastoparan inhibited catecholamine secretion in intact (40) and permeabilized chromaffin cells (6,9,10) by interfering with the ATP-dependent priming step of calcium-evoked exocytosis. Specific antibodies raised against G␣ o and a synthetic peptide corresponding in sequence to the COOH terminus of G␣ o reversed this inhibitory effect of mastoparan (6,9), indicating that a trimeric G o protein acts as a negative control on the exocytotic machinery in chromaffin cells. Confocal immunocytochemical data and functional studies on permeabilized cells suggested the association of G o with the membrane of secretory granules (6). This observation predicted the existence of a putative intracellular pseudoreceptor, which may affect exocytosis by controlling the granule-bound G o . Indeed, we found that cytosolic GAP-43 can stimulate guanine nucleotide binding and exchange activity in chromaffin granule membranes and block calcium-evoked exocytosis in permeabilized chromaffin cells (31). The latter effect was completely inhibited by anti-G o antibodies, suggesting that GAP-43 may represent a possible endogenous upstream regulator for the granule-associated G o , thereby controlling calcium-regulated secretion in chromaffin cells (31).
The aim of the present study was to identify the putative effector(s) by which the granule-associated G o protein inhibits the exocytotic response. Since the cortical actin network acts as a physical barrier to prevent granules from undergoing exocytosis (11,12,43), we thought that the subplasmalemmal cytoskeleton might be a possible candidate. The experiments presented here show that the introduction of mastoparan into the cytosol of permeabilized chromaffin cells prevents the calcium-evoked disruption of peripheral actin filaments. This effect was specifically reversed by the COOH-terminal peptide of G␣ o and affinity-purified antibodies raised against G␣ o , indicating that mastoparan stabilized the cortical actin cytoskeleton by activating G o . Mastoparan is an amphiphilic peptide that penetrates the plasma membrane and activates associated G proteins. The fact that mastoparan stabilized actin filaments providing that the peptide penetrated into the cytoplasm implies that mastoparan stabilizes the actin cytoskeleton most likely by stimulating the secretory granule-associated G o accessible only in permeabilized cells.
We found a close correlation between the inhibitory effect of mastoparan on exocytosis evaluated by the appearance of DBH immunoreactivity at the cell surface and the action of mastoparan on cortical actin filaments visualized with rhodamineconjugated phalloidin. This observation is in line with the idea that mastoparan inhibits secretion by stabilizing the cortical actin network. To determine further the relationship between the effects of mastoparan on peripheral actin and exocytosis, we examined the effect of mastoparan in cells having their cortical actin filaments disorganized by cytochalasin E or iota toxin. The mechanism of disruption of actin filaments by cytochalasins differs from that of clostridial toxins. Cytochalasins block actin polymerization by binding to the barbed end of the actin filament, but these agents also increase the nucleation of G-actin and enhance actin ATPase activity (44). C2 and iota toxins not only block polymerization at the barbed end of actin by ADP-ribosylating G-actin but also inhibit the ATPase activity of actin (45). However, despite their distinct effects on actin filaments, neither cytochalasin E nor iota toxin were able to reverse completely the inhibitory effect of mastoparan on secretion, although both agents disrupted completely the cortical actin network in chromaffin cells. This observation suggests that exocytosis may not simply require the depolymerization of cortical actin filaments but rather a subtle reorganization of the peripheral actin that agents like cytochalasins or clostridial toxins are unable to mimick. Evidence in favor of a role for actin filament disassembly in secretion has been obtained in PC12 cells with C2 toxin, but the toxin produced a biphasic action on noradrenaline release indicating that actin may play more than one role in exocytosis (46). In mast cells, activation of secretion triggers not only the disassembly of cortical actin but also the appearance of actin filaments that seem to provide a structural support for degranulation (24). To what extent secretion in chromaffin cells requires the presence of short actin filaments and/or de novo actin polymerization is currently unknown, but certainly merits further investigation. On the other hand, by stimulating the granule-bound G o protein, mastoparan may affect another important step in the exocytotic pathway. We recently described the possible involvement of the GTP-binding ADP-ribosylation factor 6 (ARF6) in calciumevoked secretion in chromaffin cells. ARF6 was found associated to the membrane of secretory chromaffin granules through an interaction with G␤␥ subunits (47). Stimulation of chromaffin cells triggered the rapid dissociation of ARF6 from secretory granules by a mechanism sensitive to aluminium fluoride (47). Mastoparan as well as aluminium fluoride is likely to prevent the translocation of ARF6 by maintaining the granule-bound G o in an activated state, an event which may also contribute to the inhibitory effect exerted by mastoparan on the exocytotic pathway.
We investigated the possible sequence of events leading to the stabilization of the cortical cytoskeleton in response to G␣ o activation. Although it has long been known that Ca 2ϩ is a key regulator of the cytoskeleton, evidence is now accumulating that Rho GTPases, a subgroup of the Ras superfamily of small GTP-binding proteins, represent other important modulators of actin cytoskeleton (41,42,48). In fibroblasts, polymerized actin is assembled into a variety of distinct structures, which have now all been shown to be controlled by members of the Rho family (49). In mast cells, Rho and Rac have been implicated in the signaling pathways that lead both to cytoskeleton reorganization and to secretion (50 -52). The exoenzyme C3 from C. botulinum is a useful tool for examining the cellular function of Rho, because it specifically ADP-ribosylates the protein at an asparagine residue in the putative effector domain (53,54). We show here that the introduction of C3 transferase in permeabilized chromaffin cells prevented the mastoparan-induced stabilization of cortical actin and strongly reversed the inhibitory effect of mastoparan on secretion. These results suggest the possible involvement of Rho in the pathway by which G o controls the peripheral actin cytoskeleton and the ATP-dependent priming step of exocytosis. Interestingly, the control of Rho-dependent actin polymerization by the ␣ subunit of trimeric G proteins has been recently demonstrated in Swiss 3T3 fibroblasts (55). In mast cells, Norman et al. (24) have also suggested the occurrence of a plasma mem-brane-associated trimeric G protein that might be able to transduce signals to Rho and Rac via a putative cytosolic factor. Hence, the direct effector coupling the ␣ subunit of trimeric G proteins to the regulation of Rho activity remains unclear. RhoGDI might represent an attractive candidate, since its introduction into the cytosol inhibits exocytosis in mast cells (51,52).
The precise mechanism by which Rho controls the organization of actin is not yet fully understood. However, use of cellfree assays and intact cell systems has shown that Rho regulates several enzymes, including phosphatidylinositol-4phosphate 5-kinase and phosphoinositide 3-kinase (56 -58), implying that Rho regulates the actin cytoskeleton through the formation of polyphosphoinositides, which are known to modulate the activity of various actin-binding proteins (59). Phosphatidylinositol (4,5)-bisphosphate decreases the actin filament severing activities of gelsolin and scinderin (60,61), two proteins that have been found associated to the subplasmalemmal cytoskeleton in chromaffin cells (61,62). Furthermore, recombinant scinderin facilitates exocytosis in permeabilized chromaffin cells, an effect that can be blocked by phosphatidylinositol (4,5)-bisphosphate (63). These observations suggest that Rho may stabilize the cortical actin network in chromaffin cells by controlling the level of phosphatidylinositol (4,5)bisphosphate and thereby modulating the actin-severing activity of scinderin and/or gelsolin. Since phosphatidylinositol-4phosphate 5-kinase has been identified among the cytosolic proteins involved in the ATP-dependent priming reaction of exocytosis (64), it is tempting to speculate that the Rho-dependent synthesis of phosphatidylinositol (4,5)-bisphosphate is the link integrating the granule-bound G o with the regulation of the cortical actin network, a scheme that might represent some of the biochemical reactions underlying the priming of exocytosis in neuroendocrine cells.
The intracellular regulatory mechanism of G o activation and inactivation in resting and stimulated chromaffin cells remains elusive. Cytosolic GAP-43 is an attractive candidate, since the protein is a major substrate for protein kinase C, binds calmodulin, and is therefore sensitive to variations in cytosolic calcium. Although the introduction of GAP-43 modulates the exocytotic response in both adrenergic and noradrenergic permeabilized cells (31), its apparent absence in adrenergic cells (65) raises the question of its general function in neuroendocrine cells. Alternatively, novel putative endogenous regulators of trimeric G proteins have been discovered recently. These newly identified proteins interacting either at the level of the ␣ subunit (66) or with the ␤␥ complex (67) may well represent potential partners for the secretory granule-associated G o during the exocytotic process. We previously proposed that G o acts essentially as an inhibitory clamp preventing the priming of secretory granules and the ongoing of exocytosis in resting cells. The present data indicating that stimulation of the granule-bound G o results in actin filament stabilization suggest that G o could also play an active function in the terminal phase of exocytosis by facilitating the rapid re-assembly of cortical actin filaments at the site of fusion between granule and plasma membranes. In line with this hypothesis, it is interesting to note that active Rho has recently been described as a key regulator of the association of actin filaments with the plasma membrane (68). Further characterization of the identity and calcium sensitivity of the upstream regulators controlling the activation/inactivation cycle of G o in stimulated chromaffin cells is now required to provide a more detailed picture of the exocytotic stages involving the participation of the secretory granule-bound G o protein.