Identification of a potential effector pathway for the trimeric Go protein associated with secretory granules. Go stimulates a granule-bound phosphatidylinositol 4-kinase by activating RhoA in chromaffin cells.

Besides having a role in signal transduction, heterotrimeric G proteins may be involved in membrane trafficking events. In chromaffin cells, Go is associated with secretory organelles, and its activation inhibits the ATP-dependent priming of exocytosis. By using permeabilized cells, we previously described that the control exerted by the granule-bound Go on exocytosis may be related to effects on the cortical actin network through a sequence possibly involving Rho. To provide further insight into the function of Rho in exocytosis, we focus here on its intracellular localization in chromaffin cells. By immunoreplica analysis, immunoprecipitation, and confocal immunofluorescence, we found that RhoA is specifically associated with the membrane of secretory chromaffin granules. Parallel subcellular fractionation experiments revealed the occurrence of a mastoparan-stimulated phosphatidylinositol 4-kinase activity in purified chromaffin granule membranes. This stimulatory effect of mastoparan was mimicked by GAP-43, an activator of the granule-associated Go, and specifically inhibited by antibodies against Galphao. In addition, Clostridium botulinum C3 exoenzyme completely blocked the activation of phosphatidylinositol 4-kinase by mastoparan. We propose that the control exerted by Go on peripheral actin and exocytosis is related to the activation of a downstream RhoA-dependent phosphatidylinositol 4-kinase associated with the membrane of secretory granules.

Besides having a role in signal transduction, heterotrimeric G proteins may be involved in membrane trafficking events. In chromaffin cells, G o is associated with secretory organelles, and its activation inhibits the ATPdependent priming of exocytosis. By using permeabilized cells, we previously described that the control exerted by the granule-bound G o on exocytosis may be related to effects on the cortical actin network through a sequence possibly involving Rho. To provide further insight into the function of Rho in exocytosis, we focus here on its intracellular localization in chromaffin cells. By immunoreplica analysis, immunoprecipitation, and confocal immunofluorescence, we found that RhoA is specifically associated with the membrane of secretory chromaffin granules. Parallel subcellular fractionation experiments revealed the occurrence of a mastoparanstimulated phosphatidylinositol 4-kinase activity in purified chromaffin granule membranes. This stimulatory effect of mastoparan was mimicked by GAP-43, an activator of the granule-associated G o , and specifically inhibited by antibodies against G␣ o . In addition, Clostridium botulinum C3 exoenzyme completely blocked the activation of phosphatidylinositol 4-kinase by mastoparan. We propose that the control exerted by G o on peripheral actin and exocytosis is related to the activation of a downstream RhoA-dependent phosphatidylinositol 4-kinase associated with the membrane of secretory granules.
Studies on diverse secretory cell types have established a crucial role for GTP-binding proteins in the regulation of calcium-dependent exocytosis. Trimeric G proteins have been found associated with the membrane of various secretory granules (1)(2)(3), suggesting their participation in the exocytotic reaction. Accordingly, the involvement of a plasma membranebound G i3 protein in the late stages of exocytosis in mast cells has been demonstrated (4). Direct control of exocytosis by G i and G o proteins has also been described in insulin-secreting cells (5) and in chromaffin cells (3, 6 -9). Besides heterotrimeric G proteins, a subset of small GTPases appears to control the exocytotic reaction. Rab3 and the ADP-ribosylation factor 6 which are specifically localized on secretory vesicles (10,11) have been proposed to serve as a regulator of the exocytotic fusion in chromaffin cells (12,13), anterior pituitary cells (14), and melanotrophs (15). In addition, recent investigations led to the idea that Rho may be a component of the molecular machinery underlying regulated secretion (16 -20).
Rho belongs to the GTPase family consisting of Rho, Rac, and Cdc 42 proteins. This family has been implicated in a number of cellular functions requiring the reorganization of actin-based structures (21)(22)(23). In secretory cells, cytoskeletal rearrangements are a prerequisite for exocytosis since actin forms a cortical barrier that must be reorganized to enable docking and/or fusion of the secretory granules with the plasma membrane (24 -26). Rho together with a trimeric G protein regulates the changes in the actin cytoskeleton observed in activated mast cells (17,27). In chromaffin cells, we recently described that the secretory granule-associated G o controls the priming of exocytosis by modifying the cortical actin network through a sequence of events that eventually involves Rho (20). Thus, Rho seems to be an integral component of the signaling pathway leading to the cytoskeletal redistribution necessary for secretion, although the mechanism by which Rho relates to the actin organization remains to be elucidated.
To identify the partners of Rho signaling in calcium-evoked secretion, we examine here the intracellular distribution of Rho in chromaffin cells. By immunoreplica analysis and confocal immunofluorescence, we demonstrate that RhoA is a specific component of the membrane of secretory chromaffin granules. Furthermore, our data reveal that the phosphatidylinositol (PtdIns) 1 4-kinase present on chromaffin granules (28,29) can be stimulated by specific activators of G␣ o through a mechanism sensitive to Clostridium botulinum C3 ADP-ribosyltransferase. We propose that the granule-bound G o controls the peripheral actin cytoskeleton and exocytosis through an effector pathway involving the sequential participation of RhoA and PtdIns 4-kinase both located on the membrane of secretory chromaffin granules.

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 (30). 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 on either 24 multiple 16-mm Costar plates (Cambridge, MA) at a density of 2.5 ϫ 10 5 cells/well or fibronectin-coated glass coverslips at a density of 2 ϫ 10 5 cells or 100-mm Costar plates at a density of 5 ϫ 10 6 cells/plate. Experiments were performed 3-7 days after plating.
Subcellular Fractionation of Cultured Chromaffin Cells-Cultured chromaffin cells (50 ϫ 10 6 cells) were washed with Locke's solution, rapidly frozen and thawed, and collected in 3.7 ml of buffer A containing 20 mM HEPES, pH 7.5, 1 mM EDTA, 1 mM DTT, 20 g/ml pepstatin, 20 g/ml aprotinin, 20 g/ml trypsin inhibitor, 20 g/ml RNase, 0.1 mM phenylmethylsulfonyl fluoride (total homogenate). Following 5 min incubation on ice, 2.5 ml of the homogenate was centrifuged for 45 min at 100,000 ϫ g. The supernatant was saved (cytosol) and the pellet homogenized in 500 l of buffer A (membrane-bound fraction).
Subcellular Fractionation of Bovine Adrenal Medulla-Plasma and chromaffin granule membranes were purified from bovine adrenal medulla as described previously (3). Briefly, adrenal medullary glands were homogenized in 0.32 M sucrose (10 mM Tris-HCl, pH 7.4) and then centrifuged at 800 ϫ g for 15 min. After centrifugation at 20,000 ϫ g for 20 min, the pellet was resuspended in 0.32 M sucrose (10 mM Tris-HCl, pH 7.4), layered on a continuous sucrose density gradient (1-2.2 M sucrose, 10 mM Tris-HCl, pH 7.4), and centrifuged for 1 h at 100,000 ϫ g. Twelve 1-ml fractions were collected from top to bottom and analyzed for protein by the Bradford procedure. The distribution of dopamine-␤hydroxylase (D␤H) and Na ϩ /K ϩ ATPase was estimated as described previously (31,32). Plasma membranes were purified from fractions 2 and 3, which contained the highest Na ϩ /K ϩ ATPase activity. Chromaffin granule membranes were recovered from fractions 11 and 12, which contained the highest D␤H activity. Fractions were diluted 10 times in TED buffer (20 mM Tris-HCl, pH 8.0, 1 mM EDTA, 1 mM DTT), and membranes were collected by centrifugation for 30 min at 100,000 ϫ g. Membranes were suspended in TED buffer and stored at Ϫ20°C.
Phosphatidylinositol Kinase Assay-PtdIns kinase activity was detected as described by Husebye and Flatmark (28). Chromaffin granule membrane proteins (50 g) were preincubated for 10 min in the presence or absence of the indicated concentrations of mastoparan in assay buffer containing 30 mM HEPES, pH 7.0, 0.1 mM EGTA, 5 mM MgCl 2 in a final volume of 300 l. The reaction was started by addition of 30 l of the ATP solution containing 2.5 Ci of [␥-32 P]ATP (3000 Ci/mmol, Amersham Pharmacia Biotech) and 5 mM MgATP. After 2 min, the reaction was stopped by addition of 800 l of chloroform, methanol, 12 M HCl (20/40/1, v/v/v). The chloroform phase was extracted, evaporated under vacuum, and then redissolved in 20 l of chloroform/methanol (2/1, v/v). Phospholipids were separated on one-dimensional TLC oxalate-coated silica gel plates in chloroform, methanol, 4 M NH 4 OH (9/7/2, v/v/v). Labeled phospholipids were visualized with 32 P-imaging plates using the Bio-Imaging Analyzer FUJIX BAS1000. Labeled phospholipids were compared with standard lipids stained with iodine vapor.
PtdIns kinase assays in the presence of LY294002 (Lilly) or quercetin (Sigma) were performed with a reduced concentration of MgATP (20 M). Membranes were preincubated for 15 min in assay buffer containing either LY294002 (100 M) or quercetin (100 M) before incubation for 10 min in the presence or absence of mastoparan (20 M). The reaction was subsequently started with [␥-32 P]ATP.
Deacylation and HPLC Analysis-For definitive identification of the phosphatidylinositol monophosphates detected on the TLC plates, the radioactive spots were excised, deacylated (33), and subjected to HPLC analysis using a 10-m Partisil SAX ion-exchange column (Interchim, Montluçon, France). After sample injection, the column was equilibrated with water for 5 min, and 32 P-labeled glycerophosphatidylinositol monophosphates were separated using a 60-min linear gradient of 0 -0.18 M ammonium phosphate, pH 3.8 (33). A mixture of AMP and ADP was applied with the deacylated samples, and their 260-nm absorbance served to standardize elution times that were similar from one run to the next. Radioactivity was detected by assaying 200-l eluate fractions. 32 P-Labeled glycerophosphatidylinositol 4-phosphate (gPt-dIns-4-P) standard was prepared by deacylation of [ 32 P]PtdIns-4-P obtained by lipid extraction and TLC separation of 32 P-labeled quiescent CHO cells as described (33). The 32 P-labeled glycerophosphatidylinositol 3-phosphate (gPtdIns-3-P) standard was prepared by deacylation of [ 32 P]PtdIns-3-P produced in vitro by phosphatidylinositol 3-kinase (p85) immunoprecipitated from CHO cells and incubated in the presence of PtdIns and [␥-32 P]ATP (33).
GTPase Activity-The GTPase activity of purified chromaffin granule membranes was estimated according to the procedure previously described (3,6). Briefly, granule membrane proteins (10 g) were incubated for 10 min at 30°C in 100 l of 50 mM HEPES, pH 8.0, 1 mM EDTA, 1 mM DTT, 100 mM NaCl, and 1 mM MgCl 2 in the presence of the indicated concentrations of mastoparan. The GTPase reaction was started by the addition of GTP at a final concentration of 0.8 M with 1 nCi [␥-32 P]GTP (NEN Life Science Products, 30 Ci/mmol) and stopped by the addition of 250 l of cold 5% (w/v) Norit in phosphate buffer. After centrifugation, the 32 P i in the supernatant was determined by 32 P Cerenkov counting. Non-enzymatic hydrolysis of [␥-32 P]GTP during the assay was subtracted from all data. Assays were performed in triplicate.
Immunocytochemistry and Confocal Laser Scanning Microscopy-Chromaffin cells grown on fibronectin-coated glass coverslips were washed with Locke's solution and 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, 10% normal goat serum in PBS to reduce nonspecific staining. Cells were incubated for 2 h at 37°C with the primary antibodies in PBS containing 3% bovine serum albumin in a moist chamber. Cells were then washed with PBS and subsequently incubated for 1 h at 37°C with the respective secondary antibodies diluted to 1:200 in PBS containing 3% bovine serum albumin. Finally, coverslips were extensively washed with PBS, rinsed with water, and mounted in Moviol 4 -88 (Hoechst).
Immunofluorescence staining was monitored with a Zeiss laser scanning microscope (LSM 410 invert) equipped with a planapo oil (63 ϫ) immersion lens (numerical aperture ϭ1.4). DTAF emission was excited using the argon laser 488-nm line, whereas TRITC was excited using the helium/neon 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). Nonspecific fluorescence was assessed by incubating cells with the secondary fluorescent antibodies and measuring the average intensity value for each fluorochrome. This value was subtracted from all images.
Immunoprecipitation and Immunoblotting-Chromaffin granule membrane proteins (250 g) were subjected to ADP-ribosylation with recombinant C3 transferase. Granule membranes were subsequently immunoprecipitated with 30 l of anti-D␤H antiserum and 100 l of ImmunoPure Immobilized protein A (Pierce) (10). The beads were collected by centrifugation, and washed and bound proteins were analyzed by SDS-gel electrophoresis and autoradiography.
Antibodies-The rabbit polyclonal anti-D␤H antiserum used for chromaffin granule membrane immunoprecipitation was prepared in our laboratory and its specificity demonstrated (24). Rabbit polyclonal anti-bovine chromogranin A (CGA) antibodies (34) were used at 1:5000 dilution. Mouse monoclonal antibodies against RhoA and rabbit polyclonal antibodies against RhoB (Santa Cruz Biotechnology, INC) were used at 1:10 dilution for immunocytochemical experiments and at 1:100 for immunodetection on nitrocellulose sheets. Donkey anti-rabbit IgGs conjugated to dichlorotriazinyl aminofluorescein (DTAF) were used at 1:200 dilution (Chemicon International Inc.). Goat anti-mouse IgG conjugated to tetramethylrhodamine (TRITC) were used at 1:200 dilution (Chemicon International Inc.). Affinity purified antibodies against the COOH-terminal G␣ o peptide (ANNLRGCGLY) or G␣ i peptide (KNNLK-DCGLF) were prepared as described previously (6), and their specificity against non-denaturated G␣ protein was demonstrated. Mouse monoclonal antibodies against the p85 subunit of phosphatidylinositol 3-kinase were purchased from Upstate Biotechnology (Lake Placid, NY).
Peptides, Proteins, and Toxins-Mastoparan was obtained from Sigma. Cytosolic and membrane-bound forms of GAP-43 were purified from bovine brain according to a previously published procedure (8). C. botulinum exoenzyme C3 ADP-ribosyltransferase (C3 transferase) prepared and purified as described (35) was a generous gift of Dr. M. R. Popoff (Institut Pasteur, Paris).

Subcellular Localization of C3-catalyzed ADP-ribosylated
Proteins in Chromaffin Cells-To investigate the intracellular distribution of Rho in chromaffin cells, we made use of the specific ADP-ribosylation reaction with C. botulinum C3 ADPribosyltransferase, in which [ 32 P]ADP-ribose is incorporated on residue asparagine 41 of Rho (36,37). Plasma membranes, chromaffin granule membranes, and cytosolic proteins were prepared from bovine adrenal medulla by subcellular fractionation on sucrose density gradients, and the presence of substrates for C3 transferase was analyzed by gel electrophoresis and autoradiography. We found that C3 transferase consistently labeled a 21-kDa protein present in chromaffin granule membranes and to a smaller extent in the cytosol (Fig. 1A). In contrast, the plasma membrane contained virtually no detectable ADP-ribosylated product when incubated with C3 transferase (Fig. 1A). Control experiments performed in the absence of C3 transferase confirmed that the granule-associated 21-kDa protein was specifically labeled in a toxin-dependent fashion (Fig. 1A).
The specific association of the C3 substrate with chromaffin granule membranes was further substantiated by analyzing the distribution pattern of [ 32 P]ADP-ribosylated proteins in fractions collected from a sucrose density gradient layered with a crude membrane preparation. As illustrated in Fig. 1B, the radiolabeled 21-kDa protein was not evenly distributed throughout the sucrose gradient but showed a major peak in fractions 10 -12. These fractions contained the chromaffin granule membranes as revealed by immunodetection with anti-D␤H antibodies, a specific marker for chromaffin granules. We also used an immunoadsorption procedure to exclude the association of the 21-kDa C3 substrate with a contaminating organelle that may comigrate with chromaffin granules in the sucrose density gradients. Fig. 1C shows that preincubation of purified granule membranes with anti-D␤H antibodies followed by protein A-Sepharose addition resulted in the co-adsorption of the 21-kDa [ 32 P]ADP-ribosylated protein, in agreement with the idea that secretory granule membranes do contain a specific protein substrate for C3 transferase.
RhoA Is a Specific Component of the Secretory Chromaffin Granule Membrane-Since Rho and Rac are potential substrates of C3 transferase (37), we characterized further the granule-associated 21-kDa protein by immunodetection on nitrocellulose sheets. Fig. 2 illustrates a Western blot analysis of adrenal medullary plasma membranes, chromaffin granule membranes, and cytosolic proteins using anti-RhoA and anti-RhoB antibodies. RhoA was detected in the granule membrane fraction, displaying an apparent molecular mass of 21 kDa ( Fig. 2A). We also observed the presence of a substantial amount of RhoA in the cytosolic fraction but not in the plasma membrane ( Fig. 2A), in agreement with the results obtained by C3-induced ADP-ribosylation. In contrast, no specific immunosignal for RhoB was detectable among the three subcellular fractions (Fig. 2B). To evaluate the portion of RhoA present in the cytosol, cultured chromaffin cells were collected, and the content of RhoA was estimated in three fractions defined as the total homogenate, the cytosol, and the membrane-bound compartment. By C3-catalyzed ADP-ribosylation and immunoreplica analysis using the anti-RhoA antibodies, we found that the cytoplasmic pool represented approximately 15-20% of the total RhoA present in chromaffin cells (Table I).
Two-dimensional gel electrophoresis of purified chromaffin granule membranes revealed the presence of two spots labeled with the anti-RhoA antibody. These two spots displayed only a different isoelectric point value (respectively, 6.2 and 6.5) but no difference in the apparent molecular mass (data not shown).
Occasionally, the RhoA-labeled band present in chromaffin granule membranes could also be resolved into two components on monodimensional gels ( Fig. 2A). This observation may reflect some post-translational modification, such as phosphorylation. Consistent with this idea, studies in human lymphocytes (38) indicate that membrane-bound RhoA is a target for protein kinase A-mediated phosphorylation, whereas cytosolic RhoA is protected from phosphorylation by its binding to the GDP dissociation inhibitor. . Labeled proteins were then resolved by SDS-gel electrophoresis and detected by autoradiography. B, 12 fractions (15 g of protein/ fraction) collected from a continuous sucrose density gradient layered with the crude chromaffin membrane pellet were used as substrates for C3-catalyzed [ 32 P]ADP-ribosylation. Proteins were then separated by electrophoresis and analyzed by autoradiography to detect the incorporation of [ 32 P]ADP-ribose or transferred to nitrocellulose sheets to detect chromaffin granule membranes using anti-D␤H antibodies. Chromaffin granules were recovered in fractions 10 -12 which also contained the highest amount of C3 toxin substrate. C, chromaffin granule membranes were subjected to C3-catalyzed [ 32 P]ADP-ribosylation and subsequently immunoprecipitated with anti-D␤H antibodies and protein A-Sepharose. Control experiments were performed with protein A-Sepharose and granule membranes in the absence of anti-D␤H antibodies or with protein A-Sepharose and anti-D␤H antibodies in the absence of granule membranes. Precipitated proteins were separated by gel electrophoresis, and labeled proteins were analyzed by autoradiography. A 21-kDa substrate for C3 transferase is specifically associated with the membrane of purified chromaffin granule.
Next, we examined the intracellular distribution of RhoA in cultured chromaffin cells by immunofluorescence and confocal microscopy. To confirm the association of RhoA with secretory granules, double-labeling experiments were performed with antibodies against RhoA and against chromogranin A (CGA), a specific marker for chromaffin granules. Immunostaining with RhoA antibodies revealed a punctate staining pattern (Fig. 3, A  and D) that was very similar to the labeling obtained with anti-CGA antibodies (Fig. 3, B and E). The cytofluorogram representing the pixels measured in the CGA and RhoA images (data not shown), together with the mask constructed from the dots located diagonally in the cytofluorogram (Fig. 3, C and F), revealed that almost all individual chromaffin granules were labeled with the anti-RhoA antibodies. We could not detect a specific staining of the cytosol with the RhoA antibodies, in agreement with the observation that most of the cellular RhoA is membrane-bound in chromaffin cells (Table I). Experiments performed with anti-RhoB antibodies failed to stain chromaffin cells (data not shown).
Effect of Mastoparan and GAP-43 on the Secretory Granuleassociated Phosphatidylinositol 4-Kinase Activity-We previously reported that the secretory granule-associated G o protein regulates the peripheral cytoskeleton in chromaffin cells by a mechanism involving the small GTP-binding protein Rho (20). Since phosphoinositide kinases are potential effectors for Rho in mediating cytoskeletal rearrangements (21-23), we investigated the possible functional relationship between the granulebound form of G o , RhoA, and the PtdIns 4-kinase activity described in chromaffin granule membranes (28,29). Mastoparan, a selective activator of G i and G o proteins (39), was used in our previous studies of permeabilized chromaffin cells to stimulate the granule-associated G o and its downstream effectors (3,6,20). Hence, we first examined the effect of mastoparan on PtdIns kinase activity in purified chromaffin granule membranes. Fig. 4 shows the PtdIns kinase activity detected in plasma and chromaffin granule membranes using endogenous PtdIns as substrate. PtdIns monophosphate (PtdInsP) was the only 32 P-labeled derivative of PtdIns found in lipid extracts of both plasma and granule membranes labeled with [␥-32 P]ATP (Fig. 4A). Note that the PtdIns kinase activity was approximately 10 times more important in the granule membrane than in the plasma membrane. A quantitative analysis of PtdIns labeling in granule membranes was performed in the presence of either quercetin, an inhibitor of PtdIns 3-and 4-kinases (40), or LY294002, a specific inhibitor of PtdIns 3-kinase (41). As illustrated in Fig. 4B, only quercetin inhibited the formation of radiolabeled PtdInsP, in line with the presence of PtdIns 4-kinase among the components of the chromaffin granule membrane (28,29). Mastoparan strongly stimulated the formation of radiolabeled PtdInsP in purified chromaffin granule membranes (Fig. 4B). Moreover, the mastoparan-stimulated PtdIns kinase activity was not significantly affected by LY294002 but largely inhibited by quercetin (Fig. 4B). To identify further the PtdIns kinase activated by mastoparan, radiolabeled PtdInsP was extracted from the thin layer plate, deacylated, and analyzed by anion-exchange HPLC. Fig. 4C shows that deacylated PtdIns-3-P and PtdIns-4-P are readily separated by this method. The deacylated PtdInsP recovered from granule membranes incubated in the presence of mastoparan eluted as one single peak that comigrated with the 4-isomer (Fig. 4C). The 3-isomer was not detectable in either control or mastoparanstimulated samples (Fig. 4C), thereby confirming the identity of the mastoparan-activated lipid kinase in secretory granule membranes as PtdIns 4-kinase. Fig. 5 shows the dose-response curve for the mastoparaninduced activation of PtdIns 4-kinase in chromaffin granule membranes. Mastoparan produced a dramatic dose-dependent stimulation of the granule-associated PtdIns kinase activity in the 5-40 M range. At 40 M, mastoparan stimulated the PtdIns 4-kinase activity by 540 Ϯ 54% (Fig. 5). Interestingly, mastoparan stimulated the granule-associated G proteins with the same concentration dependence since both GTPase and PtdIns 4-kinase were activated by mastoparan with EC 50 of 12 and 18 M, respectively (Fig. 5).
We previously reported that the neuronal "growth-associated protein" GAP-43 (neuromodulin) inhibits the ATP-dependent priming of secretion and stabilizes the cortical actin network by specifically stimulating the secretory granule-associated G o when introduced into the cytosol of permeabilized chromaffin cells (8,20). To strengthen the idea that mastoparan activated PtdIns 4-kinase in chromaffin granule membranes by activating the granule-bound form of G o , we examined whether GAP-43 was also able to stimulate PtdIns 4-kinase activity. Purified chromaffin granule membranes were exposed to either cytosolic or membrane-extracted bovine brain GAP-43. As shown in Fig. 6, we found that the soluble form of GAP-43 mimicked in a dose-dependent manner the stimulatory effect of mastoparan on PtdIns kinase activity. Maximal effect was observed in the presence of 5 M soluble GAP-43 for which the PtdIns kinase activity was enhanced by 300 Ϯ 10%. Interestingly, the membrane-extracted GAP-43 which is totally ineffective in stimulating the granule-bound G o (8) was also unable to stimulate the granule-associated PtdIns kinase activity (Fig.  6). This observation emphasizes the apparent relationship be-  tween the activation of the secretory granule-associated G o and the increase of PtdIns 4-kinase activity in the granule membrane.
Mastoparan-induced Activation of the Granule-associated Phosphatidylinositol 4-Kinase Requires G o -Mastoparan has been reported to activate G i and G o proteins by interacting with the carboxyl terminus of the ␣ subunit (42). To assess whether mastoparan enhanced the granule-associated PtdIns 4-kinase activity by stimulating the granule-bound G o , we attempted to antagonize the effect of mastoparan with affinity purified antibodies prepared against the carboxyl terminus of G␣ o and G␣ i . Chromaffin granule membranes were incubated overnight with either anti-G␣ o or anti-G␣ i antibodies and then exposed to 20 M mastoparan. Fig. 7 shows that the presence of the anti-G␣ i antibodies did not significantly affect the mastoparanstimulated PtdIns kinase activity. In contrast, the stimulating effect of mastoparan on PtdInsP formation was completely abolished by the anti-G␣ o antibodies (Fig. 7), in agreement with the idea that mastoparan enhanced the granule-associated PtdIns 4-kinase activity by stimulating the granule-bound G o . It is noteworthy that the anti-G␣ o antibodies reversed at the same concentration the effects of mastoparan on secretory granule-associated PtdIns 4-kinase, peripheral actin organization (20), and ATP-dependent catecholamine secretion (3), implying a possible link between these three events in the exocytotic pathway.
Effect of C. botulinum C3 Transferase on Mastoparan-induced Activation of the Granule-associated Phosphatidylinositol 4-Kinase-To probe the possible role of Rho in the mastoparan-induced activation of PtdIns 4-kinase, we used the C3 ADP-ribosyltransferase to specifically inactivate the granuleassociated Rho protein. Purified chromaffin granule membranes were incubated for 10 min with increasing concentrations of C3 toxin in the presence or in the absence of 20 M mastoparan. The reaction was then started by adding [␥-32 P]ATP, and the formation of radiolabeled PtdInsP was analyzed. Treatment with C3 transferase did not modify the basal level of PtdIns kinase activity (Fig. 8). However, C3 transferase induced a dose-dependent inhibition of the mastoparan-induced PtdIns kinase activity (Fig. 8). At 5 g/ml, C3 toxin completely abolished the mastoparan-dependent formation of PtdInsP (Fig. 8), indicating that the inactivation of Rho prevented mastoparan from activating PtdIns 4-kinase in chromaffin granule membranes. These results reinforce the idea that the secretory granule-associated G o controls exocytosis by stabilizing the actin cytoskeleton (20) through a pathway involving RhoA and the granule-associated PtdIns 4-kinase. DISCUSSION Exocytotic fusion of secretory granules with the plasma membrane requires calcium and ATP. By using permeabilized cell models, it has been possible to establish that exocytosis consists of sequential ATP-dependent and ATP-independent phases of release (7,43,44). The ATP-independent phase has been attributed to the fusion of a small number of docked granules (45,46), whereas the ATP-dependent component has been assumed to include docking of granules and priming of the exocytotic machinery (43). We previously investigated the function(s) of trimeric G proteins in chromaffin cell exocytosis by using selective modulators of G␣ proteins (3, 6 -9, 20). Our findings indicate that two trimeric G proteins act in series in the exocytotic pathway as follows: a secretory granule-associated G o controls the ATP-dependent priming reaction, whereas a plasma membrane-bound G i3 is involved in the late ATPindependent fusion step (3, 6 -9). More recently, immunocytochemical and functional studies performed on permeabilized chromaffin cells indicated that the control exerted by the granule-associated G o on the priming of exocytosis is related to effects on the subplasmalemmal actin cytoskeleton (20). Our results suggested that activated G o stabilized the actin network through a sequence of events possibly implicating the small GTPase Rho (20). In line with this hypothesis, Rho-dependent pathways have been previously described to be under the control of trimeric G proteins in diverse cell types, including fibroblasts (47), leukocytes (48), myocytes (49), and mast cells (27).
To investigate the intracellular localization of Rho in secretory cells, we used here the exoenzyme C3 from C. botulinum which specifically ADP-ribosylates the protein at an asparagine residue in the effector domain (36,37). We found that a 21-kDa substrate of C3 toxin is specifically associated with the membrane of purified secretory chromaffin granules. Accordingly, the presence of C3 substrates in granule-enriched fractions has been previously described in neutrophils (50) and anterior pituitary cells (51). In the present study, we demonstrate by confocal immunocytochemistry the colocalization of RhoA with the chromaffin granule marker chromogranin A, further confirming the specific association of RhoA with the lipids were extracted and separated on TLC plates. 32 P-Labeled phospholipids were compared with standard lipids stained with iodine vapor. The PtdIns kinase activity detected in purified chromaffin granule membranes was not affected by LY294002, a specific PtdIns 3-kinase inhibitor. B illustrates a quantitative analysis of radiolabeled PtdInsP using a PhosphorImager program and expressed as arbitrary units membrane of secretory organelles. This granular distribution fits well with the idea that RhoA may act as a downstream effector for the granule-bound G o controlling the subplasmalemmal actin network in chromaffin cells (20).
Although the best established function of Rho relates to the cytoskeletal organization, the pathways involved in Rho-induced actin regulation remain to be identified. To date, several candidates have been proposed to serve as signaling elements mediating the Rho-dependent cytoskeletal features. For instance, protein kinase N (52,53) and Rho kinase (54,55) have been identified as possible downstream effectors of Rho in actin stress fiber formation. In permeabilized fibroblasts, Mackay et al. (56) demonstrated that moesin and its close relatives, ezrin and radixin, can reconstitute stress fiber assembly and cortical actin polymerization in response to Rho activation. Several reports also indicate that Rho controls the activity of various phosphoinositide kinases (21)(22)(23), thereby regulating the synthesis of PtdIns 4,5-bisphosphate (PtdIns-4,5-P 2 ), a known regulator of actin polymerization (57).
In chromaffin cells, the ATP dependence of secretion has been suggested to reside, at least in part, in maintaining cellular phosphoinositides at a defined level (58). The involvement of phosphoinositides in exocytosis is further supported by the recent identification of three proteins required for the priming of exocytosis, as PtdIns transfer protein (59), PtdIns 5-kinase (60), and PtdIns 4-kinase (29). Both PtdIns transfer protein and PtdIns 5-kinase are cytosolic proteins. In contrast, PtdIns 4-kinase has been identified among the components of the chromaffin granule membrane (28,29). These observations led us to investigate whether Rho-dependent activation of PtdIns 4-kinase might be part of the molecular machinery linking the granule-bound G o to the regulation of the cortical actin network. By using purified chromaffin granule membranes, we show here that activation of G␣ o stimulates the granule-associated PtdIns 4-kinase activity. Moreover, incubation of granule membranes with C. botulinum C3 transferase completely abolished the G o -stimulated PtdIns kinase activity, indicating the participation of Rho in the sequence of events coupling G o to the granule-bound PtdIns 4-kinase. These results correlate well with the idea that Rho-dependent synthesis of PtdIns-4P may be the link integrating the granule-bound G o with the regulation of cortical actin network.
Although PtdIns 4-kinase has been identified as a component of the priming reaction (29), its exact function in the exocytotic machinery remains to be determined. PtdIns 4-kinase produces PtdIns-4P by specifically phosphorylating the D-4 position of the inositol ring. PtdIns-4P can be subsequently phosphorylated by PtdIns 5-kinase to generate PtdIns-4,5-P 2 . Indeed, cytosolic PtdIns 5-kinase has been described as an essential factor for the ATP-dependent priming of chromaffin cell secretion (60). PtdIns-4,5-P 2 is an important regulator of the cytoskeleton organization in many cell types (57). In chromaffin cells, PtdIns-4,5-P 2 decreases the actin filament severing activities of gelsolin and scinderin (57, 61), two proteins that have been found associated with the cortical ring of actin filaments (30,61). Moreover, recombinant scinderin facilitates exocytosis in permeabilized chromaffin cells, an effect that can be blocked by PtdIns-4,5-P 2 (62). Thus, by decreasing the actinsevering activity of scinderin and/or gelsolin, G o /Rho-dependent PtdIns 4-kinase may well contribute to the stabilization of peripheral actin filaments that exclude secretory granules from the subplasmalemmal zone in resting chromaffin cells. Alternatively, PtdIns 4,5-P 2 may be required to maintain or regulate the interactions between actin filaments and the plasma membrane in stimulated cells. For instance, it has been recently demonstrated that Rho-dependent synthesis of PtdIns 4,5-P 2 stimulates the association of ezrin, radixin, and moesin proteins with their membrane-binding partner CD44 and thereby contributes to the cross-linking of actin filaments with plasma membranes (56,63). The occurrence of ezrin, radixin, and moesin proteins in neuroendocrine cells is currently unknown. However, it can be predicted that termination of exocytosis and/or endocytosis requires the rapid re-assembly of cortical actin filaments. In view of its granular location, the G o /Rho-dependent PtdIns 4-kinase represents an attractive candidate capable of maintaining or stimulating the formation of actin filaments specifically at the site of fusion between granule and plasma membranes.
To conclude, the present data indicate that stimulation of the secretory granule-associated G o results in the Rho-dependent activation of PtdIns 4-kinase, a pathway that may contribute to the local increase of PtdIns-4,5-P 2 on the granule membrane. Further investigations are now required to correlate the intracellular fluctuations of PtdIns-4,5-P 2 with the activation/inactivation cycle of the granule-bound G o . It is tempting to propose that G o /Rho-dependent activation of PtdIns 4-kinase represents an important basis for the actin cytoskeletal reorganization underlying membrane trafficking at the site of exocytosis.