Mammalian exocyst complex is required for the docking step of insulin vesicle exocytosis.

Glucose stimulates insulin secretion from pancreatic beta cells by inducing the recruitment and fusion of insulin vesicles to the plasma membrane. However, little is currently known about the mechanism of the initial docking or tethering of insulin vesicles prior to fusion. Here, we examined the role of the SEC6-SEC8 (exocyst) complex, implicated in trafficking of secretory vesicles to fusion sites in the plasma membrane in yeast and in regulating glucose-stimulated insulin secretion from pancreatic MIN6 beta cells. We show first that SEC6 is concentrated on insulin-positive vesicles, whereas SEC5 and SEC8 are largely confined to the cytoplasm and the plasma membrane, respectively. Overexpression of truncated, dominant-negative SEC8 or SEC10 mutants decreased the number of vesicles at the plasma membrane, whereas expression of truncated SEC6 or SEC8 inhibited overall insulin secretion. When single exocytotic events were imaged by total internal reflection fluorescence microscopy, the fluorescence of the insulin surrogate, neuropeptide Y-monomeric red fluorescent protein brightened, diffused, and then vanished with kinetics that were unaffected by overexpression of truncated SEC8 or SEC10. Together, these data suggest that the exocyst complex serves to selectively regulate the docking of insulin-containing vesicles at sites of release close to the plasma membrane.

Insulin is accumulated in large dense core vesicles within the pancreatic ␤ cell and is released by exocytosis when glucose concentrations rise (1). Triggering of secretion appears to involve increases in intracellular ATP concentrations (or ATP/ ADP ratio), closure of ATP-sensitive K ϩ channels, and Ca 2ϩ influx through voltage-sensitive Ca 2ϩ channels (1). Because secretion in response to nutrient secretagogues is usually biphasic, with a rapid early phase followed by a subsequent second or sustained phase (2), it has been proposed that functionally distinct groups of vesicles may exist and represent "reserve," "readily releasable," and "intermediate" pools (3,4). At present, however, the mechanisms through which vesicles are recruited to the readily releasable pool, proposed to correspond to a subset of the "morphologically docked" (4) vesicles at the cell surface, are still incompletely understood (1).
Soluble N-ethylmaleimide-sensitive fusion protein attachment protein receptors (SNAREs) 1 are believed to be required for the trafficking of insulin and other secretory hormones from the endoplasmic reticulum to the Golgi apparatus, Golgi and trans-Golgi network to the plasma membrane, and finally exocytosis of mature vesicles (5,6). Dense core vesicle trafficking to the plasma membrane can itself be divided into three stages: transport of insulin vesicles to the plasma membrane, their first interaction with the plasma membrane (docking or tethering), and then subsequent fusion at the plasma membrane (exocytosis). Although SNARE proteins are believed to regulate the final fusion step in ␤ cells (7), the machinery involved in the initial docking or tethering steps and possibly in regulating the sustained phase of insulin secretion (1,8) is undefined (9).
The budding yeast Saccharomyces cerevisiae has provided the genetic identification of proteins required for vesicle transport (10,11). Six of the sec gene products, Sec3p, Sec5p, Sec6p, Sec8p, Sec10p, and Sec15p, have been described in the vesicle transport complex, and mammalian homologues have been identified (12)(13)(14). Furthermore, two novel gene products, Exo70p and Exo84p, have been shown to be part of the complex (13,15,16), now known as the SEC6-SEC8 or "exocyst" complex.
The exocyst complex is concentrated at the sites of active vesicle exocytosis in the yeast bud and is essential for secretion (16,17). Similarly, the mammalian exocyst is localized at the plasma membrane of nerve terminals (12) and, in differentiated PC12 cells, displays a punctate distribution at the termini of cell processes at or near sites of exocytosis (13) or on the secretory vesicle membrane (18). Such findings suggest that the exocyst complex may serve as part of vesicledocking machinery at sites of regulated as well as constitu-tive exocytosis (12,19,20). At present, however, functional data supporting or refuting such a role in mammalian cells is unavailable.
In this study, we examined the role of the exocyst complex in regulated insulin secretion. We show that SEC6 is concentrated on insulin-containing dense core vesicles, whereas SEC5 and SEC8 are concentrated in the cytoplasm and on the plasma membrane, respectively. Although dominant-negative suppression of SEC8 and SEC10 function decreased the density of vesicles at the plasma membrane, overexpression of truncated SEC8 or SEC10 did not affect the kinetics of single exocytotic events. These data are compatible with a model in which the exocyst complex (a) forms as dense core vesicles approach the cell surface, (b) is then involved in the docking of vesicles at the plasma membrane by presently undefined mechanisms, but (c) is not involved in the final fusion step.
Generation of Truncated sec6 and sec8 Adenoviruses-cDNAs encoding C-terminal truncation mutants of rat sec6 (residues 1-333, Ad-sec6 CT ) or rat sec8 (residues 1-426, Ad-sec8 CT ) were amplified by PCR, subcloned into pCR2 (Invitrogen), and sequenced on both strands. These mutants corresponded to those identified by Novick and colleagues (17,24,25), which abrogate function (i.e. secretion) in yeast. We therefore postulated that the constructs would function as dominant negatives in mammalian cells. The cDNAs were subcloned into the pShuttle-CMV vector, linearized with PmeI, and cotransformed with pAdEasy into Escherichia coli strain BJ5183 by electroporation. Recombinants were selected and amplified in E. coli DH5␣. The chosen clones were linearized with PacI to expose the inverted terminal repeats and allow viral packaging when transfected in HEK293 cells. Large scale amplification and titer of viral stocks was performed as described previously (26).
To monitor exocytosis of NPY-mRFP constructs at the single vesicle level, we used a TIRF microscope similar to that described previously (22,28,29) with minor modifications. We used an objective lens with a numerical aperture of 1.45 (PlanAPO ϫ100/1.45 TIRFM, Olympus) to monitor single vesicle exocytosis. To observe the pIRES2-EGFP and NPY-mRFP fluorescence images, a 488-nm laser (argon ion laser, 30 milliwatts, Spectra-Physics) was used for total internal reflection illumination with a long pass filter (515 nm for EGFP, 600 nm for mRFP). Images were collected with a cooled charge-coupled device camera (640 ϫ 480 pixels, IMAGO, Till Photonics; operated with TillvisION software, Till Photonics). Images were acquired every 300 ms with 50 ms of exposure. To analyze the data, fusion events were manually selected, and the average fluorescence intensity of individual vesicles in a 1 ϫ 1-m square placed over the vesicle center was calculated.
The critical angle of our 1.45 numerical aperture objective lens is 63.56°, and the widest possible angle is 71.39°(when measured at the rim of the lens). We introduced the laser beam in the middle of these angles. We calculated the decay length of evanescent field (d) theoretically using the formula, where is the angle of incidence, the wavelength of light, n 1 and n 2 the refractive indices of water (1.33) and glass (1.53). For ϭ 67.5°and ϭ 488 nm, decay constant (d) becomes 81.1 nm, and effectively, a layer of 100 -200 nm is the reach of the evanescent light.
Confocal Imaging-To assess the density of the vesicles in exocyst complex mutant-expressing cells, we employed a confocal microscope (Leica TCS-AOBS laser-scanning confocal microscope). Before monitoring vesicle density, the bottom of the cell was located by inspection in confocal mode. The focal plane was then moved up by 1 m using a piezo motor, which drove the objective lens as controlled by the system software.
Generation and Purification of Polyclonal Anti-SEC5 Antibody-Antibodies were raised in sheep against residues 1-100 of the human SEC5 (GenBank TM accession number: NP_060773) sequence, encompassing the Ral-binding domain. The mouse SEC5 protein is identical in this region apart from residue 83, where arginine is replaced by lysine in the murine sequence. The relevant part of the coding region of SEC5 was inserted as a BamHI/EcoRI fragment into the vector pET28a. The resultant fusion protein, bearing an N-terminal hexahistidine tag, was expressed in E. coli and purified by chromatography on nickelnitrilotriacetic acid-agarose. Antibodies were purified from the antiserum by chromatography on a column of the fusion protein, immobilized on CNBr-activated Sepharose CL-4B.
Insulin Secretion Assay-MIN6 cells (5 ϫ 10 5 cells/well) were seeded in 6-well microtiter plates and infected with null (control), truncated sec6, and truncated sec8 adenoviruses. Culture was continued for 24 h in Dulbecco's modified Eagle's medium containing 25 mM glucose and then at 3 mM glucose for another 16 h. Cells were then washed in phosphate-buffered saline and incubated in KRB medium and 3 mM glucose, 30 mM glucose, or 50 mM KCl. Incubations were performed for 30 min at 37°C in a water bath. Total insulin was extracted in acidified ethanol (30). Total and secreted insulin were measured using radioimmunoassay by competition with 125 I-labeled rat insulin (Linco Research, St. Charles, MO) according to the manufacturer's instructions.
Data Analysis-Data are given as means Ϯ S.E. Comparisons between means were performed using one-way analysis of variance with GraphPad Prism TM (GraphPad Software, San Diego, CA) software.

The Exocyst Complex Is Associated with Insulin-containing
Vesicles-The exocyst complex has been shown to be present in high concentrations in developing brain and is believed to play a role in synaptogenesis (12,31). To examine the cellular localization of the complex in pancreatic ␤ cells, formaldehyde-fixed pancreatic MIN6 ␤ cells were probed with antibodies against the three exocyst complex subunits (SEC5, SEC6, and SEC8).
Confocal images revealed that SEC5, SEC6, and SEC8 immunoreactivity displayed a markedly different subcellular localization (Fig. 1). SEC5 immunofluorescence was observed largely as diffuse, cytosolic fluorescence, although some punctate staining, partially colocalized with insulin, was apparent (Fig. 1, A-C). In contrast, SEC6 was found as punctate labeling, closely colocalized with insulin-positive vesicles (Fig. 1, D-F). SEC8 was mainly found on the plasma membrane of the cells (Fig. 1, G-I). These observations are similar to those made previously in PC12 cells (18) and suggest that the exocyst complex might regulate vesicle docking or fusion in pancreatic MIN6 ␤ cells.

Truncated Exocyst Component Decreases the Number of
Docked Vesicles on the Plasma Membrane-To determine the significance of exocyst complex function in insulin vesicle recruitment or docking, MIN6 cells were cotransfected with NPY-mRFP and either pIRES2-EGFP-sec8-⌬N or pIRES2-EGFP-sec10-⌬C. The pIRES2-EGFP plasmid is a bicistronic vector that allows the expression of both EGFP and an exocyst subunit mutant as two separate proteins. NPY-mRFP was used here as a surrogate for insulin (22,29) because the latter targets relatively poorly to the dense core vesicle after fusion with fluorescent proteins 2 (28). Cotransfected MIN6 cells were identified as those containing diffuse cytosolic EGFP fluorescence and punctate NPY-mRFP fluorescence. We then counted the number of intracellular and plasma membrane-docked NPY-mRFP vesicles by TIRF and confocal microscopy, respectively (Fig. 2, A-F). All of the cells examined that overexpressed truncated exocyst components showed a significant 40 -50% decrease in the number of docked vesicles at the plasma membrane (control, 1.36 Ϯ 2.2 vesicles/m 2 , n ϭ 12; sec8-⌬N, 0.86 Ϯ 2.8 vesicles/m 2 , n ϭ 7; sec10-⌬C, 0.65 Ϯ 1.1 vesicles/m 2 , n ϭ 12) (Fig. 2G, open squares). In contrast, cells overexpressing sec8-⌬N or sec10-⌬C showed a non-significant ϳ10% decrease in the number of intracellular vesicles (Fig. 2G, filled squares).
We next measured the impact of overexpression of the exocyst complex on glucose-or high K ϩ (depolarization)-induced insulin secretion. Insulin secretion was stimulated ϳ2.5-fold by 30 mM (versus 3 mM) glucose and 4-fold by high (50 mM) K ϩ in control cells (Fig. 2H). In Ad-sec6 CT -or Ad-sec8 CT -infected MIN6 cells, glucose-or high K ϩ -induced insulin secretion was decreased by ϳ40% (Fig. 2H). Taken together, these data suggest that the exocyst complex inhibits the docking step of insulin vesicle delivery to the plasma membrane, thus reducing the number of vesicles available for immediate release.
Effect of Overexpression of Exocyst Complex on Vesicle Fusion-To determine whether the apparent inhibition of docking observed above might be due to the exocyst complex decreasing the rate at which individual vesicles fused, thus causing a potential "bottleneck" that prevented subsequent fusion events, the dynamics of single vesicle exocytosis were analyzed in single NPY-mRFP-expressing vesicles. Although as expected, exocytotic events were detected less frequently in sec8-⌬N-and sec10-⌬C-expressing cells than in cells transfected with empty vector control (Fig. 2E), the kinetics of individual fusion events were identical in each case (Fig. 3D). Thus, stimulation with 30 mM glucose caused NPY-mRFP-expressing vesicles to brighten and spread suddenly during the release of the fluorescent peptide (22,29,32) with an identical time course in control (Fig. 3A), sec8-⌬N-overexpressing cells (Fig.  3B), or sec10-⌬C-overexpressing cells (Fig. 3C)  was observed in the mean values for the rise time, half-widths, or decay time of fluorescence intensity between control and sec8-⌬N-or sec10-⌬C-overexpressing cells (data not shown). These data thus indicate that the exocyst complex does not alter the kinetics of individual insulin exocytosis events but rather regulates the docking step of insulin vesicles to the plasma membrane.
Effect of Overexpression of Exocyst Complex on Total Number of the Vesicles on the Plasma Membrane-We next measured the dynamic changes in the number of insulin vesicles docked at the plasma membrane during stimulation with glucose. During glucose stimulation of control cells (Fig. 4, A and D, open  circles), the total number of docked vesicles slightly increased by up to 110% of the initial value, consistent with previous observations (33). This apparent recruitment of new vesicles to docking sites might thus contribute to the refilling of the docked vesicle pool (possibly equivalent to a readily releasable pool) during the second phase of glucose-induced insulin secretion. In contrast, in either sec8-⌬N-or sec10-⌬C-overexpressing cells, the total number of docked vesicles during glucose stimulation remained low, with no appreciable appearance of newly recruited vesicles to the plasma membrane (Fig. 4, B and C). As a result, in cells expressing sec8-⌬N or sec10-⌬C, the total number of docked vesicles decreased by 10 -40% of the initial number after 10 min of stimulation with high glucose concentrations (Fig. 4D, closed squares and triangles).
Truncated Exocyst Component Does Not Affect Microtubule Dynamics-To determine whether the effects of the dominant-negative sec constructs on vesicle distribution may be due to alterations in microtubule dynamics, the latter were assessed by monitoring the distribution of ␣-tubulin in nullor sec mutant-expressing cells. No differences were apparent in ␣-tubulin distribution in cells expressing any of the interfering sec constructs as assessed by confocal microscopy (Fig.  5). Thus, microtubule rearrangements would seem unlikely to explain the decreased vesicle recruitment to the cell surface. DISCUSSION We show here that the exocyst complex is present in significant concentrations in pancreatic MIN6 ␤ cells. Unexpectedly, we observed very different patterns of immunofluorescence staining for intracellular SEC5, SEC6, and SEC8. Thus, SEC5 immunofluorescence was predominantly seen diffusely throughout the cytoplasm in MIN6 cells. In contrast, SEC6 and SEC8 immunofluorescence were predominantly localized to insulin vesicles and the plasma membrane, respectively (Fig. 1). Thus, it would appear that the exocyst complex does not exist chiefly as a preformed unit but that instead, formation of the holo-complex may accompany translocation of vesicles to docking (pre-exocytosis) sites at the plasma membrane. Several studies in epithelial cells have shown that a portion of the exocyst complex exists in a partially assembled state with some of the protein being found dissociated from SEC6 (34)(35)(36). Also, studies by Finger et al. (37) and Guo et al. (38) showed that the localization of SEC8 to the plasma membrane at sites of polarized exocytosis is dependent on the accumulation of all other protein subunits of the complex in yeast. It therefore seems possible that SEC8 may be recruited to sites on the plasma membrane, probably in conjunction with a subset of other exocyst components prior to assembly of the whole exocyst complex. This may occur during docking of insulin-containing vesicles bearing SEC6 and possibly other exocyst subunits. Recently, several studies have shown that the exocyst complex interacts with active Ral GTPase (39)(40)(41). Wang et al. (42) showed that point mutations in both RalA and the sec5 subunit, which abolish RalA-SEC5 interaction, also abolish the inhibition of GTP-dependent exocytosis in PC12 cells. In contrast, the cleavage of a SNARE protein by Botulinum neurotoxin blocks both GTP-and Ca 2ϩ -dependent exocytosis. Therefore, the above authors (42) concluded that RalA and the exocyst complex are essential for GTP-dependent exocytosis, whereas the final fusion steps are SNARE-dependent. These data are fully compatible with the present findings.
We also analyzed here the docking and fusion of insulin vesicles in pancreatic MIN6 cells overexpressing truncated exocyst components. Our results showed that: 1) inhibition of the exocyst complex inhibited insulin secretion in response to cell depolarization or to glucose; 2) the kinetics of insulin secretion were identical in either control or truncated exocyst component-overexpressing cells; 3) the supply of vesicles and/or the ability to retain them on plasma membrane were impaired, resulting in a decrease in the number of docked insulin vesicles (to ϳ50% of that in normal MIN6 cells; Fig. 2D).
In conclusion, we present direct evidence that the exocyst complex contributes to the docking step of insulin vesicles on the plasma membrane in pancreatic ␤ cells and may be involved in the recruitment of vesicles from a reserve to a readily releasable pool at the plasma membrane. Changes in the efficiency of the docking process may therefore play a role in the normal control of insulin release under conditions where secretion is strongly stimulated, such as during chronic hyperglycemia or during treatment with sulfonylureas. Because such alterations might conceivably contribute to insulin secretory deficiency in some form of type 2 diabetes, exocyst complex components may provide potential therapeutic targets in this disease.