ARF6 Regulates the Synthesis of Fusogenic Lipids for Calcium-regulated Exocytosis in Neuroendocrine Cells*

An important role for specific lipids in membrane fusion has recently emerged, but regulation of their biosynthesis remains poorly understood. Among fusogenic lipids, phosphatidic acid and phosphoinositol 4,5-bisphosphate (PIP2) have been proposed to act at various steps of neurotransmitter and hormone exocytosis. Using real time FRET (fluorescence resonance energy transfer) measurements, we show here that the GTPase ARF6, potentially involved in the synthesis of these lipids, is activated at the exocytotic sites in PC12 cells stimulated for secretion. Depletion of endogenous ARF6 by siRNA dramatically inhibited secretagogue-evoked exocytosis. ARF6-siRNA greatly reduced secretagogue-evoked phospholipase D (PLD) activation and phosphatidic acid formation at the plasma membrane and moderately reduced constitutive levels of PIP2 present at the plasma membrane in resting cells. Expression of an ARF6 insensitive to short interference RNA (siRNA) fully rescued secretion in ARF6-depleted cells. However, a mutated ARF6 protein specifically impaired in its ability to stimulate PLD had no effect. Finally, we show that the ARF6-siRNA-mediated inhibition of exocytosis could be rescued by an exogenous addition of lysophosphatidylcholine, a lipid that favors negative curvature on the inner leaflet of the plasma membrane. Altogether these data indicate that ARF6 is a critical upstream signaling element in the activation of PLD necessary to produce the fusogenic lipids required for exocytosis.

Release of neurotransmitters and hormones occurs through exocytosis, a highly regulated process that culminates with the fusion of secretory vesicles/granules and the plasma membrane. In the exocytotic process key proteins such as SNAREs 2 have been shown to be fundamental players by providing the energy required to fuse membranes through the formation of high affinity, parallel, four-␣-helix bundles (1). The role of lipids is less well understood, but recent findings indicate that the shape of the lipids (determined by the size of their head group) and their charge might also be important for the fusion process. In vitro cone-shaped lipids that spontaneously form negative membrane curvatures favor the formation of hemifusion intermediates and fusion when present in the contacting leaflets of apposed membranes (2)(3)(4)(5). Accordingly, the lipid-modifying enzyme PLD, which produces cone-shaped PA, emerges as a major actor in various cellular processes that have in common membrane fusion (6 -13). PIP 2 is another acidic phospholipid known to be important for regulated exocytosis in neuroendocrine cells (14) and neurons (15). Interestingly, recent reconstitution in vitro experiments reveal that PA, located on the acceptor membrane (inner leaflet of the plasma membrane), and PIP 2 , located on the donor membrane (outer leaflet of the vesicle), increase the fusion rate. Together, these findings suggest that SNARE-mediated membrane fusion additionally requires specific and optimally distributed fusogenic lipids (5).
The small GTPase ARF6 is known in vitro to directly activate PIP5K type I ␥ (16) and PLD (17) to generate PIP 2 and PA, respectively. ARF6, the most divergent isoform of the six members of the ARF family, has been proposed to play a role in cell motility (18), vesicle recycling at the plasma membrane (19,20), phagocytosis (21), hormone-regulated secretion (22), and Glut-4 translocation (23), all of which involve vesicle transport and fusion with the plasma membrane. Subcellular studies have shown that ARF6 associates with intracellular vesicular structures including large dense core secretory granules in chromaffin cells (24 -26). Overexpression of various mutants of ARF6 affects the secretory response from neuroendocrine cells suggesting that ARF6 is involved in calcium-regulated exocytosis (26,27). On the other hand, overexpression of some of these ARF6 mutants has been reported to induce a profound redistribution of intracellular proteins and lipids (27), a finding that questions the physiological relevance of studies based on the use of mutated ARF6 proteins.
The present work was undertaken to assess the actual role of ARF6 in regulated exocytosis, using FRET measurements and RNA interference to reduce the level of the endogenous protein. We demonstrate that ARF6 is an essential element of the calcium-regulated exocytotic machinery and show that its prime function is to activate PLD1, thereby providing the exocytotic machinery with the fusogenic PA required for the late fusion reaction.
ARF6 was fused to the N-terminal part of ECFP into XhoI and KpnI sites of pECFPN1. MT2 was fused to the N-terminal part of EGFP or EYFP into EcoRI and SalI sites of pEGFPN1 and pEYFPN1. The PA-binding domain (PABD) of the yeast soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) protein Spo20p was fused to EGFP as described previously (12).
For short interference RNA (siRNA) targeting, rat ARF6 cDNA fragments encoding a 19-nucleotide siRNA sequence (siRNA#1, AATCCTCATCTTCGCCAACAA; siRNA#2, gctgcaccgcattatcaat; siRNA#3, GCACCGCATTATCAAT-GACCG), derived from the target transcript and separated from its reverse 19-nucleotide complement by a short spacer, were annealed and cloned in the BglII and HindIII sites in front of the H1 promoter of either the pEGFP-N2-RNAi plasmid or a modified pXGH5 plasmid encoding for GH as described previously (30). Modified pXGH5 vector with no siRNA sequence is referred to as pGHsuper. Similarly, a mutated form of siRNA#1 was generated by inserting the sequence aatcttgatcttcgccaacaa derived from the target transcript (bold sequence indicates the changes made in the siRNA#1 construct). Cells were also transfected with a pXS plasmid tagged with HA containing mutated rescue ARF6 constructs resistant to siRNA degradation by the mutagenesis of the codon CTC encoding Leu-118 to TTG. Sitedirected mutagenesis was carried out using the QuikChange mutagenesis kit (Stratagene) as described previously (31). PLD1 and PLD2 siRNA plasmids have been described recently (12).
For immunofluorescent and FRET experiments, expression vector were introduced into PC12 cells (1 ϫ 10 5 cells) using GenePorter (Gene Therapy System; 0.5 g ADN/well) on adherent cells according to the manufacturer's instructions. For Western blot, pulldown, and PLD assays, 20 g of each plasmid was electroporated into PC12 cells (10 7 cell/reaction) at 290 V and 1200 microfarads and plated (12). Under these conditions, the transfection efficiency ranged from 60 to 75%, and the co-transfection rate was greater than 90%.
Growth Hormone Release-PC12 cells (24-well plates, 80% confluent) were transfected with the various constructs (0.5 g of each plasmid/well) using GenePorter (Gene Therapy Systems). 72 h after transfection, cells were washed four times with Locke's solution and then incubated for 10 min in Locke's solution (basal release) or stimulated for 10 min with a depolarizing concentration of K ϩ . Where indicated, 1 M lysophosphatidylcholine (LPC) was added during the last wash and during the 10-min incubation under resting conditions (Locke's solution) or stimulated conditions (elevated K ϩ solution). Supernatants were collected, and the cells were harvested by scraping in 10 mM phosphate-buffered saline and broken by three freeze-andthaw cycles. The amounts of GH secreted into the medium or retained within the cells were measured using an enzymelinked immunosorbent assay (ELISA; Roche Applied Science). GH secretion was expressed as a percentage of total GH present in the cells before stimulation. We measured secretion from permeabilized cells as described previously (32). First cells were washed twice with Locke's solution and twice with Ca 2ϩ -free Locke's solution and then permeabilized for 10 min in potassium glutamate buffer containing 10 M digitonin without (resting) or with 20 M free calcium (stimulated).
Fluorescence Resonance Energy Transfer Measurement-For FRET experiments, we used ARF6 fused at its C-terminal end with ECFP as a donor and the probe for active ARF6, MT2 fused with EYFP, as an acceptor. PC12 cells were transfected with plasmids encoding either ARF6-ECFP or MT2-EYFP or co-transfected with both plasmids. For FRET recording, cells were placed on an inverted fluorescence microscope (Axiovert 35M, Zeiss; dichroic mirror 470 nm), which was connected by optical fibers with the excitation and emission monochromators of an Alphascan PTI spectrofluorometer. The cells were superfused with Locke's solution (1 ml/min) and stimulated by superfusion of elevated K ϩ solution for the indicated times. The cells were illuminated at 430 nm (excitation wavelength of ECFP), and emission spectra between 480 and 600 nm were recorded at the indicated times. For FRET measurements, the emission spectra were deconvoluted after base-line subtraction to distinguish donor and acceptor fluorescence intensities as described previously (32).
Determination of PLD Activity-72 h after transfection, PC12 cells were washed four times with Locke's solution and then incubated for 10 min in calcium-free Locke's solution (basal PLD activity) or stimulated in Locke's solution containing a depolarizing concentration of K ϩ . Medium was then replaced by 100 l of ice-cold 50 mM Tris, pH 8.0, and the cells were broken by three freeze/thaw cycles. Samples were collected and mixed with an equal amount of Amplex Red reaction buffer (Amplex Red phospholipase D assay kit, Molecular Probes), and the PLD activity was estimated after a 1-h incubation at 37°C with a Mithras fluorometer (Berthold). A standard curve was established with purified PLD from Streptomyces chromofuscus (Sigma).
Immunofluorescence, Confocal Microscopy, and Image Analysis-Transfected PC12 were washed four times with Locke's solution and then incubated for 10 min either in Locke's solution (resting conditions) or in Locke's solution containing a depolarizing concentration of potassium (stimulation) before the fixation step and further processed for immunofluorescence as described previously (33,34) except for PIP 2 labeling. Briefly, for PIP 2 staining cells were fixed on ice for 3 h in 4% paraformaldehyde-Dulbecco's modified Eagle's medium with 2 mM EGTA, washed three times in 50 mM NH 4 Cl-phosphatebuffered saline, and permeabilized for 4 h at 4°C in permeabilization buffer (1 mM MgCl 2 , 0.2% saponin, 1% fetal calf serum, 0.1% bovine serum albumin, 50 mM glycine, 0.05% NaN 3 in phosphate-buffered saline). Incubation with anti-PIP 2 antibody (1/50) in permeabilization buffer was performed overnight at 4°C and then with the secondary antibody for 2 h at 4°C. Stained cells were visualized using a Zeiss LSM 510 confocal microscope. Quantification was performed using Zeiss CLSM instrument software, version 3.2. The percentage of the EGFP-PABD binding probe co-localizing at the plasma membrane with SNAP25 was measured by determining the double-labeled pixels, expressed as average fluorescence intensity, and then converted to a percentage of the total fluorescence obtained from the two labels in each cell.
The amount of PIP 2 staining was measured and expressed as the average fluorescence intensity (mean pixel intensity multiplied by the pixel number) normalized to the corresponding surface area by dividing by the total plasma membrane surface of each cell. We performed a quantitative cell-to-cell comparison in the same field (fluorescence from nontransfected cells compared with that of transfected cells expressing GFP). The PIP 2 level in the nontransfected cells was fixed at 100%. To validate the specificity of the anti-PIP 2 antibody and our analysis, we used the recently described drug-inducible type IV 5-phosphatase to reduce PIP 2 levels in PC12 cells (35). Quantification revealed that PIP 2 staining was almost completely eliminated (supplemental Fig. 1). On the other hand, overexpression of PIP5K induced an increase in PIP 2 staining of up to 6-fold (supplemental Fig. 1). In conclusion, these data confirm that we were able to detect variation in PIP 2 staining over a large scale and that the PIP 2 antibody used was highly specific.
Data Analysis-In all figures, data are given as the mean values Ϯ S.E. obtained from at least three independent experi-ments performed on different cell cultures, where n represents the number of experiments. Data were analyzed with Minitab statistical software. Statistical significance was established using Student's test, and data were considered significantly different when the p value was less than 0.05. Gaussian distributions of the data were verified.

Endogenous ARF6 Is Activated during Calcium-regulated
Exocytosis-To visualize the activated form of ARF6 in cells, we adapted a recently described probe shown to interact specifically with GTP-bound ARF6 in vitro and in a yeast two-hybrid assay (36). MT2 (metallothionein-2), a ubiquitous 6 -7-kDa metal-containing protein, was cloned from PC12 cells and expressed as MT2-GFP fusion protein. In resting PC12 cells, this ARF6-GTP sensor accumulated in the nucleus and also moderately in the cytosol (Fig. 1A). Stimulation with a depolarizing concentration of potassium led to a reduction of cytosolic MT2-GFP and a concomitant recruitment of the probe to the cell periphery where it co-localized with the plasma membrane marker SNAP25 (Fig. 1A). Quantification revealed that more than 10 Ϯ 1.2% of the MT2-GFP signal co-localized with SNAP25 in stimulated PC12 cells, compared with less than 1% under resting conditions. Accumulation of the ARF6-GTP sensor at the cell periphery was transient and rapidly became undetectable when the cells were placed in a resting condition (data not shown). Similar observations were made in chromaffin and PC12 cells stimulated with various secretagogues, such as nicotine, ATP, or barium (data not shown). These data suggest that endogenous ARF6 is activated transiently at the plasma membrane in cells stimulated for exocytosis.
To assess the dynamics of ARF6 activation in living cells, next we decided to use FRET measurements. For this purpose the fluorophore pair ECFP-EYFP, which has been shown previously in several applications to act as a donor/acceptor pair with a Förster radius of ϳ50 Å, was chosen. PC12 cells were cotransfected to express ARF6-ECFP and MT2-EYFP as donor and acceptor fluorophores. In resting cells, excitation at 430 nm (the excitation wavelength of ECFP) induced an emission spectrum displaying two fluorescence intensity peaks at 495 and 525 nm characteristic of ARF6-ECFP and MT2-EYFP, respectively (Fig. 1B). Bleaching of each fluorophore was estimated in monotransfected cells, and these values were used to correct for bleaching in cotransfected cells. Following cell stimulation with high potassium, ARF6-ECFP fluorescence decreased and MT2-EYFP increased concomitantly (Fig. 1B). We quantified this effect after spectral deconvolution in cells stimulated for different periods of time. As illustrated in Fig. 1C, FRET occurred between ARF6-ECFP and MT2-EYFP in stimulated PC12 cells. Accordingly, immunofluorescence images confirmed that ARF6-ECFP and MT2-EYFP co-localized at the cell periphery in stimulated cells (data not shown). An optimal FRET signal was detected between 45 s and 2 min of stimulation, which correlates well with the time course of GH secretion elicited by high potassium stimulation (compare Figs. 1C and 3A). Taken together, these data indicate that MT2 interacts with ARF6 in the cell periphery of secretagogue-stimulated cells and reveal that activation of ARF6 occurs at the plasma membrane prior to exocytosis and hormone secretion.
Characterization of ARF6-siRNA and Effect on Exocytosis in PC12 Cells-To probe the functional importance of ARF6 in regulated exocytosis, we used a siRNA approach to knock down endogenous ARF6 expression in PC12 cells. To accomplish this and simultaneously enable a quantitative assessment of exocytosis, we engineered plasmids that express both full-length human GH and various siRNA targeted against the sequence of ARF6. Because GH is stored in secretory granules and released by exocytosis, they allow a direct measurement of the secretory activity of siRNA-expressing cells (28,30). These plasmids also allow the identification of the subpopulation of cells that transiently express siRNAs through the immunostaining of GH. Western blot analysis revealed that transient expression of different ARF6-siRNAs reduced the level of endogenous ARF6, whereas actin levels were constant ( Fig. 2A). Densitometry scans from three independent experiments indicated that the level of ARF6 was reduced by about 55 Ϯ 5%, whereas a mutated ARF6-siRNA had no effect. When normalized to the transfection efficiency as evaluated by GH immunostaining, the level of ARF6 in PC12 cells expressing the siRNAs was reduced by up to 91 Ϯ 4% (Fig. 2B). In agreement with their specificity, the three different ARF6-siRNAs did not affect the level of expression of the related isoform ARF1 (Fig. 2A). We also verified that the different ARF6-siRNAs did not affect the expression of levels of the guanine nucleotide exchange factor ARNO and GTPaseactivating protein GIT1 for ARF6 in PC12 cells (28,29).
Because the amount of activated ARF6 may be relatively low even in stimulated cells (37), the residual endogenous ARF6 expressed in siRNA-transfected cells may be sufficient to provide levels of active ARF6 similar to that found in control cells.
To check this possibility, the MT2 probe fused to GST was used to pull down the GTP-bound form of HA-tagged ARF6 (expressed at low level) in cells expressing the various ARF6-siRNAs. As observed for the endogenous ARF6 protein, HA-ARF6 was significantly activated in control cells stimulated with high K ϩ (Fig. 2C). Expression of the ARF6-siRNAs strongly reduced the amount of GTP-loaded HA-ARF6 in stimulated cells by up to 90% as revealed by Western blot quantification (Fig. 2D). Thus, ARF6-siRNAs not only reduced the total level of endogenous ARF6 but also the amount of activated ARF6 found in stimulated cells. To further substantiate these findings, we measured the recruitment of the MT2-GFP probe to the cell periphery and used it as an index for ARF6 activation at the plasma membrane. By quantifying the co-localization of MT2-GFP with SNAP25, we found that expression of the ARF6-siRNA#1, but not the mutated siRNA, strongly reduced the translocation of MT2 to the plasma membrane (Fig. 2, E and  F), confirming that ARF6-siRNA was able to prevent the formation of active ARF6 at the plasma membrane in cells stimulated with a secretagogue.
Expression of the ARF6-siRNA#1 silencer in PC12 cells inhibited GH release evoked by increasing periods of stimulation (Fig. 3A). ARF6 silencing did not modify basal GH release, but it resulted in a strong decrease (ϳ75% inhibition) in the amount of GH secreted during a 10-min exposure to 59 mM K ϩ (Fig. 3A). Similar observations were made with the two other ARF6-siRNAs (data not shown), whereas the mutated siRNA sequence that did not affect endogenous ARF6 levels also had no effect on secretion (Fig. 3A). In parallel we verified that the expression of ARF6-siRNAs neither reduced the expression level of GH nor affected the distribution of GH-containing secretory vesicles (supplemental Fig. 2).
The signaling pathway that leads to secretagogue-evoked rise of cytosolic calcium can be bypassed by permeabilizing the plasma membrane and directly controlling calcium concentration (38). GH secretion was triggered by the presence of 20 M free calcium in the incubation medium of digitonin-permeabilized PC12 cells (Fig. 3B). The three ARF6-siRNAs potently inhibited GH release from permeabilized cells (Fig. 3B), indicating that their inhibitory effect on secretion was not the consequence of a reduction in cytosolic calcium rise but was linked directly to a defect in the exocytotic machinery. These results are in line with the idea that ARF6 is a major positive regulator of calcium-regulated exocytosis. FEBRUARY 20, 2009 • VOLUME 284 • NUMBER 8

JOURNAL OF BIOLOGICAL CHEMISTRY 4839
Depletion of Endogenous ARF6 Moderately Affects PIP 2 Level but Strongly Inhibits PLD Activation and PA Production in PC12 Cells Undergoing Exocytosis-In light of the known regulation of PI5PK activity by ARF6, we examined whether PIP 2 levels were affected in PC12 cells expressing ARF6-siRNA using either the GFP-PH-PLC␦ probe (data not shown) or specific anti-PIP 2 antibodies to detect endogenous PIP 2 . In line with previous reports, PIP 2 essentially was detected at the plasma membrane in PC12 cells (Fig. 4A), and stimulation with a secretagogue apparently did not modify its cellular level and distribution (data not shown). We found that depletion of ARF6 resulted in a moderate ϳ10 -30% decrease of PIP 2 levels present at the plasma membrane in resting PC12 cells (Fig. 4). However, only the effect of the ARF6-siRNA#3 was found to be statistically significant (Fig. 4B).
ARF6 is also a well known activator of PLD1 (17,18), an essential lipid-modifying enzyme for the exocytotic fusion reaction (6,12). On the basis of the results obtained using mutated ARF6 proteins impaired in their ability to activate selected effectors, we had proposed previously that the main function of ARF6 during exocytosis is linked to the regulation of PLD1 activity (26). This hypothesis predicts that a reduction of endogenous ARF6 levels should affect secretagogue-evoked PLD activation. To probe this possibility, PLD activity was measured in homogenates prepared from resting and K ϩ -stimulated PC12 cells expressing the various ARF6-siRNAs. In line with our previous reports (12), stimulation with high K ϩ triggered a marked activation of PLD (Fig. 5A). Down-regulation of endogenous ARF6 by siRNA expression resulted in a 60 -70% inhibition of K ϩ -induced PLD activity in cell homogenates. In contrast, the mutated ARF6-siRNA had no effect on PLD activity (Fig. 5A). In support of the specific involvement of ARF6 in PLD activation in stimulated neuroendocrine cells, depletion of ARF1 by siRNA did not significantly modify the secretagogue-induced PLD activity (supplemental Fig. 3).
Activated PLD produces PA that can be visualized by using the PA-binding domain of the yeast homologue of SNAP25, Spo20p, fused to GFP (12). In secretagogue-stimulated PC12 cells, this PA sensor is recruited to the plasma membrane and reveals the PLD1-mediated formation of PA at the sites of exocytosis (12). To further assess the importance of ARF6 in the upstream signaling pathway of PLD1, we examined the distribution of the PABD-GFP PA sensor in cells expressing ARF6-siRNAs. In these experiments, PC12 cells were first transfected with a plasmid driving the expression of GH alone or GH and the ARF6-siRNA and then, 48 h later, transfected with the PABD plasmid. As shown in Fig. 4B, the PA sensor was recruited to the plasma membrane in stimulated cells where it co-localized with SNAP25, in agreement with our previous results (12). Expression of ARF6-siRNAs, revealed by the presence of GH, largely inhibited the recruitment of PABD at the cell periphery (Fig. 5B). We quantified the percentage of PABD co-localizing with SNAP25 under resting and stimulated conditions and found that ARF6-siRNA was almost as efficient as PLD1-siRNA in inhibiting the recruitment of PABD to the plasma membrane in K ϩ -stimulated PC12 cells (Fig. 4C). Note that PLD2-siRNA did not affect the translocation of PABD to the cell periphery (Fig. 4C). These results confirm that PLD1 is the major source of PA synthesis at the plasma membrane in cells stimulated for exocytosis. They also suggest an absolute requirement for ARF6 activation, as PA is not formed at the sites of exocytosis in cells expressing PLD1 but is depleted in ARF6.
A Mutant of ARF6 That Is Unable to Activate PLD Does Not Rescue GH Secretion from ARF6-depleted PC12 Cells-We previously characterized the ARF6(N48I) mutant, which is selectively unable to activate PLD while retaining its GTP-binding and GDP-dissociation characteristics, and its ability to be activated by ARNO and inactivated by GIT1 (26). ARF6(N48I) is still able to activate in vitro any known ARF6 effectors such as cholera toxin, and based on the analogous ARF1 mutant, one  can assume that it also stimulates PIP5K (39,40). Expression of a wobble-mutated form of ARF6 insensitive to the ARF6-siRNA was able to completely rescue K ϩ -evoked GH release in ARF6-depleted cells (Fig. 6). However, expression of a siRNAresistant ARF6(N48I) construct failed to rescue GH secretion resulting from ARF6 depletion (Fig. 6). These results indicate that the ability of ARF6 to activate PLD is essential for secretion rescue in ARF6-depleted cells, reinforcing the idea that the main function of ARF6 in the exocytotic pathway is the activation of PLD1 at the plasma membrane.

Exocytosis in ARF6-depleted Cells Can Be Rescued by the Addition of an Inverted Cone-shaped Lipid to the Outer Plasma
Membrane Leaflet-Lipids have been proposed to play a decisive role in the late post-docking stages of exocytosis. It has been suggested that once a secretory granule is juxtaposed to the plasma membrane through the formation of SNARE complexes, pore formation, and eventually expansion, proceeds through the progressive formation of a granule/plasma mem-brane stalk and a lipid zippering process that generates hemifusion intermediates (4). Cone-shaped lipids like PA at the juxtaposed membrane leaflets reduce the energy requirements for the curvature process and promote the formation of the hemifusion intermediates. Conversely, inverted cone-shaped lipids present in the outer leaflet of the plasma membrane should bend the outer membrane inward and similarly promote hemifusion. In light of this assumption, we attempted to rescue the secretory activity in ARF6-depleted cells, by challenging them with external application of the inverted coneshaped lipid LPC. Control PC12 cells or cells expressing various siRNA were incubated in the presence of 1 M LPC and then stimulated with high K ϩ . Under our experimental conditions, LPC did not modify basal or K ϩ -stimulated GH release from control cells (Fig.  7). In line with our previous report (12), the addition of LPC rescued K ϩ -evoked GH secretion from PLD1-depleted cells, indicating that exogenous LPC was able to compensate for the decreased production of PA on the inner leaflet of the plasma membrane because of PLD1 knockdown. In other words, exogenous LPC is able to mimic the biophysical modifications of the plasma membrane induced by the PLD1-dependent production of PA. The addition of LPC also rescued K ϩ -evoked GH secretion in cells expressing ARF6-siRNA (Fig. 7). Thus, external LPC promotes exocytosis in ARF6-depleted cells, supporting the idea that endogenous ARF6 is an activator for the plasma membrane-bound PLD1 to produce fusogenic PA at the exocytotic sites.

DISCUSSION
Over the last decade, accumulating evidence has suggested the implication of ARF6 in various membrane trafficking events including endocrine and neuroendocrine exocytosis (41). The downstream effector(s) of ARF6 in these processes remain, however, uncertain, albeit linked to the regulation of lipid synthesis. Because most of the previous studies have employed ARF6 mutants, which are now recognized to artificially disturb the distribution of cellular lipids and SNARE proteins (27), we decided to used an RNA interference strategy to investigate the function of ARF6 in calcium-regulated exocytosis. We found that depletion of endogenous ARF6 profoundly reduces the  3). B, PC12 cells cotransfected with pEGFP-PABD and the pGHsuper, mutated siRNA#1, or pGHsuper-ARF6-siRNA (siRNA#1-3) vector were stimulated for 10 min with 59 mM K ϩ and processed for GH and SNAP25 staining. Mask images obtained by selecting double-labeled pixels show the areas of co-localization of PABD-GFP with SNAP25. Bar, 5 m. The arrows highlight cells that do not express ARF6-siRNA#1, as seen by the absence of GH in which PABD is recruited to the plasma membrane. C, histogram presenting a semiquantitative analysis of the percentage of PABD that co-localizes with SNAP25 in resting and stimulated cells expressing ARF6-siRNA, PLD1-siRNA, or PLD2-siRNA. Data are presented as the mean values Ϯ S.E. obtained in three independent experiments. *, p Յ 0.001 (n ϭ 3 with more than 75 cells analyzed per condition). exocytotic activity of PC12 cells at a step distal to calcium entry. This inhibition of secretion appears to be correlated with a strong reduction of PLD activity in stimulated cells and more specifically to the absence of PA synthesis at the plasma membrane. Moreover, we show that external agents that promote membrane bending are able to compensate for the inhibition of secretion resulting from ARF6 deficiency.
The calcium-triggered merger of two apposed membranes is the defining step of regulated exocytosis. To date, fusion of two lipid bilayers has been modeled as a series of membrane intermediates (4,42). Firstly membranes are brought into close proximity to establish a region of dehydrated contact at the initial contact point. A fusion stalk forms and then radial expands to yield a hemifusion diaphragm, while distal membrane leaflets remain separate. Different lipids such as negative curvature-forming lipids greatly reduce the energy required to overcome the hydration layer, and promote stalk formation and hemifusion when present in the contacting leaflets (3). PA under physiological conditions has been shown to generate spontaneous negative membrane curvature (43). In neuroendocrine cells, morphometric ultrastructural analysis reveals that PA accumulates at the plasma membrane granule docking sites upon cell stimulation with a secretagogue (12). Interestingly, SNARE syntaxin 1A recently has been reported to bind PA (44). Neutralization of the juxtamembrane PA-binding region of syntaxin 1A results in a delay in fusion pore expansion and a decrease in fusion pore diameter (44), in line with the idea that PA is an essential lipid for membrane fusion. Additionally, these findings suggest that syntaxin 1A may to some extent contribute to localizing PA at the fusion site (i.e. where the granules dock at the plasma membrane) (44).
We demonstrated previously that PLD1 is the only source for secretagogue-dependent PA formation at the plasma membrane (12). We show here that depletion of endogenous ARF6 prevents normal PLD1 activation and inhibits exocytosis, but the addition of an inverted cone-shaped lipid, such as LPC, on the outer leaflet of the plasma membrane compensates for the absence of ARF6 and restores exocytosis. In other words, knockdown of ARF6 produces a phenotype that is similar to the previously described PLD1 knockdown, which is also rescued by an exogenous supply of inverted-cone shaped lipids (12). These data support the idea that the reduced exocytotic activity observed in ARF6-depleted cells is primarily the consequence of the lack of synthesis of negative curvature-forming lipids at the inner leaflet of the plasma membrane. Accordingly, exocytosis in ARF6-depleted cells is not restored by ARF6(N48I), a mutant that is unable to stimulate PLD. Together, these findings indicate that PLD1 is likely to be the main target of ARF6 in regulated exocytosis in neuroendocrine cells. In this context it is interesting to recall that amperometric analysis of chromaffin cell secretion indicated that expression of ARF6(N48I) and expression of a catalytically inactive mutant of PLD1 similarly affect the kinetic parameters of individual secretory granule exocytosis (6,45). PIP 2 is required for secretion in neuroendorine cells (27,46,47), even though PIP 2 levels at the plasma membrane do not seem to increase upon stimulation with secretagogues. It has been proposed that PIP 2 regulates the size of the releas-  able pool of secretory granules in chromaffin cells, because a reduction in PIP 2 levels affects directly and dynamically secretion (47). However, inhibition of the secretory response was observed only in cells that had nearly completely lost their constitutive levels of PIP 2 at the plasma membrane (47). Whether the mild reduction of PIP 2 observed in ARF6-depleted cells significantly contributes to the inhibition of exocytosis remains difficult to estimate. Effectively, a small reduction in PIP 2 may affect the number of granules docked at the plasma membrane or the distribution of the numerous PIP 2 -binding proteins involved in regulated exocytosis. PLD1 itself requires PIP 2 for its association with the plasma membrane in stimulated cells (48), but we observed no apparent difference in the distribution of PLD1 in cells having reduced levels of ARF6 (data not shown). Finally, the observation that secretion can be almost fully restored in ARF6-depleted PC12 cells by the exogenous addition of inverted cone-shaped lipids is a strong argument for a late post-docking role for ARF6 in exocytosis. PIP 2 has been described recently as an essential lipid that recruits protein-priming factors that facilitate SNARE complex formation and granule docking, but because of its positive curvature-promoting properties, it may also be a restraint that must be alleviated to allow late exocytotic fusion reactions (49). These considerations, together with the fact that PIP5K activity and PIP 2 synthesis are not directly linked to secretagogue stimulation, render PIP5K an unlikely target for secretagogue-activated ARF6 in the exocytotic pathway.
As shown here by FRET analysis, endogenous ARF6 activation occurs at the plasma membrane and is intimately linked to secretagogue-induced stimulation. This fits well with our previous findings that ARNO, a guanine nucleotide exchange factor for ARF6 located at the plasma membrane, is able to activate HA-tagged ARF6 in stimulated cells (26,29). Moreover, ARNO, ARF6, and PLD1 were found to co-localize on plasma membrane lawns (45). Earlier subcellular fractionation experiments and immunocytochemistry indicate that ARF6 is associated with secretory granules in resting conditions (24,26,50). This implies that ARF6 activation occurs after the docking of the secretory granules at the plasma membrane and posits ARF6 as an ideal factor to spatially and temporally turns on PLD1, allowing a finely tuned production of fusogenic PA at the active site of exocytosis.
Finally, PLD1 is a multidomain protein regulated by various additional GTPases and several kinases (51). In addition to the regulatory role of ARF6 described here, we reported previously that the RalA (30) as well as Rac1(52) GTPases contribute to the activation of PLD1 required for exocytosis. Further studies are now required to understand how these multiple regulatory pathways dynamically control formation and expansion of the fusion pore in neuroendocrine cells.