Munc18-1 plays a crucial role in regulated exocytosis in neurons and neuroendocrine cells through modulation of vesicle docking and membrane fusion. The molecular basis for Munc18 function is still unclear, as are the links with Rabs and SNARE [SNAP (soluble N-ethylmaleimide-sensitive factor-attachment protein) receptor] proteins that are also required. Munc18-1 can bind to SNAREs through at least three modes of interaction, including binding to the closed conformation of syntaxin 1. Using a gain-of-function mutant of Munc18-1 (E466K), which is based on a mutation in the related yeast protein Sly1p, we have identified a direct interaction of Munc18-1 with Rab3A, which is increased by the mutation. Expression of Munc18-1 with the E466K mutation increased exocytosis in adrenal chromaffin cells and PC12 cells (pheochromocytoma cells) and was found to increase the density of secretory granules at the periphery of PC12 cells, suggesting a stimulatory effect on granule recruitment through docking or tethering. Both the increase in exocytosis and changes in granule distribution appear to require Munc18-1 E466K binding to the closed form of syntaxin 1, suggesting a role for this interaction in bridging Rab- and SNARE-mediated events in exocytosis.
Intracellular membrane fusion throughout the exocytic and endocytic pathways is mediated by members of the conserved SNARE [SNAP (soluble N-ethylmaleimide-sensitive factor-attachment protein) receptor] protein family  and is driven by the assembly of a SNARE complex between v- (vesicle) and t- (target) SNAREs [2,3]. Additional conserved protein families are involved in SNARE-dependent vesicle fusion, including the Rab  and SM (Sec1/Munc18-like) protein families. The Rab GTPases are involved in all membrane traffic steps and, along with their specific effectors, play key roles in vesicle tethering and docking prior to fusion [5,6]. The functional link between Rabs and SNARE proteins is unclear, but is mediated in part by Rab effectors [7,8]. There is an essential requirement for an SM protein in every vesicular traffic step in cells [9–12], which has been suggested to be related to their ability to interact with their appropriate syntaxin SNARE, as first observed for Munc18-1 [13,14]. A link between Rabs and SM proteins has been inferred from genetic evidence , but direct interactions have been not observed. Recent work has shown that increased vesicle docking as a result of Rab3 overexpression requires Munc18-1, but the mechanism underlying this is unknown . Indeed, the exact functions of SM proteins and how they fit into the steps between Rab and SNARE functioning are still to be resolved.
The most specialized form of vesicle fusion is in regulated exocytosis in neurons and neuroendocrine cells , which is tightly regulated both temporally and spatially and, in the case of neuronal exocytosis, fusion can occur with sub-millisecond kinetics. In neuronal and neuroendocrine exocytosis, the key SNARE proteins are syntaxin 1 and SNAP-25 on the plasma membrane and VAMP (vesicle-associated membrane protein) 1 or 2 on the vesicle membrane [2,18], with the involvement of Munc18-1  and the Rab proteins Rab3 , or Rab3 and Rab27 working in concert [6,20]. Munc18-1 is an essential protein for neurotransmission [11,12,21], and in neuroendocrine cells it has been suggested to be required for dense-core vesicle docking  or to have a role in regulating the fusion event itself . A late role in fusion is supported by observations revealing a stimulatory effect of Munc18-1 and Sec1 in an in vitro fusion assay [24,25]. It is not yet understood how Munc18-1 stimulates vesicle docking and fusion at a molecular level.
SM proteins have been observed to interact with their syntaxin using one or more of three modes of binding . Mode 1, seen only with Munc18-1, involves binding of Munc18-1 to syntaxin 1 in a closed conformation that precludes its assembly into a SNARE complex [27–29]. Mode 2 binding, described originally for the SM proteins Sly1p and Vps45 (vacuolar protein sorting 45) [30–34] and subsequently for Munc18c , involves binding of the SM proteins to the extreme N-terminus of the relevant syntaxin. Mode 3 binding, described first for yeast Sec1 , involves preferential binding to the assembled SNARE complex [24,37]. It is unclear whether all SM proteins undergo all three modes of interaction and whether all three modes are physiologically relevant. It has recently been observed, however, that Vps45 may participate in two modes of binding . In addition, Munc18-1 may undergo all three modes of interaction [25,38,39], and modes 2 and 3 are required for Munc18-1 to stimulate membrane fusion kinetics in vitro . Mode 1 binding has only been observed with Munc18-1 and its role in exocytosis has been debated. It has been suggested that this mode of interaction might be required to chaperone syntaxin during its transportation to the cell surface . However, a requirement for this mode 1 type interaction for syntaxin 1 expression on the plasma membrane has been ruled out, as syntaxin 1 can reach the plasma membrane in Munc18-1 null organisms [21,40] or if syntaxin 1 is mutated in order to prevent mode 1 binding .
We have previously examined the effect of various mutations in Munc18-1 on mode 1 binding to syntaxin 1 and also on dense-core granule exocytosis following expression of mutated proteins in adrenal chromaffin cells [23,42]. These mutations were based on those that have been discovered to produce an altered phenotype in yeast, Drosophila or Caenorhabditis elegans. One of these, the E466K mutation, was modelled on a mutation found in yeast Sly1p that suppressed the need for the Rab protein Ypt1 in ER (endoplasmic reticulum) to Golgi traffic [15,43]. The E466K mutation in Munc18-1 is sited well away from known or postulated interaction sites for syntaxin 1 , and does not affect mode 1 binding to syntaxin 1 in vitro . Interestingly, E466K acted as a gain-of-function mutation, so that expression of mutated Munc18-1 resulted in an increase in the number of exocytotic events after stimulation. We have set out to characterize the molecular basis of this gain of function and this revealed a direct interaction of Munc18-1 with Rab3A. In addition, expression of Munc18-1 with the E466K gain-of-function mutation stimulated secretory granule recruitment to the cell periphery, possibly through docking or tethering in a manner requiring the interaction of Munc18-1 with closed syntaxin 1. These findings raise the possibility that Munc18-1 acts as a direct link between Rabs and SNARE function.
MATERIALS AND METHODS
Plasmid construction and recombinant protein production
Munc18 was cloned into pcDNA3.1 for expression in mammalian cells and into the pGEX-6p-1 vector for GST (glutathione transferase) fusion protein expression . His6–Rab3A was generated as described previously . The cytoplasmic domain of syntaxin 1 (residues 4–266) was expressed and purified as a His6-tagged protein . GST–complexin II was prepared as described previously .
Binding of proteins to immobilized GST-fusion proteins
Extracts of bovine brain were prepared  and pull-down assays were performed as described previously [41,42]. Glutathione–Sepharose beads (GE Healthcare) were washed three times with binding buffer [20 mM Hepes (pH 7.4), 150 mM potassium acetate, 1 mM MgCl2 and 0.05% (v/v) Tween 20]. Escherichia coli protein extract (140 mg) was incubated per ml of a 50%(v/v) slurry of beads in binding buffer for 1 h, then the beads were washed three times in the binding buffer. All incubations were carried out at 4 °C with rotation. Recombinant GST or GST-tagged proteins were added to the beads to a final concentration of 2 μM in a total volume of 200 μl and incubated for 30 min. Bovine brain extract (200 μl) was added to the beads and incubated for 2 h. Beads were washed using CytoSignal spin filters (CytoSignal, Irvine, CA, U.S.A.), with two initial washes with binding buffer containing 1 mg/ml gelatin, then washed three times with binding buffer containing 5% (v/v) glycerol. Proteins were eluted from the beads by the addition of 100 μl Laemmli buffer (Sigma) for 15 min. Samples were separated by SDS/PAGE (12.5% gels) and transferred to nitrocellulose paper for immunoblotting. Commercial antibodies used for Western blotting: mouse monoclonal anti-Rab3A (1:500 dilution), anti-SNAP-25 (1:400 dilution), anti-Munc18 (1:10000), anti-dynamin (1:1000 dilution) and anti-Sec8 (1:1000 dilution) antibodies (BD Biosciences); mouse monoclonal anti-Rab5 (1:40 dilution) antibody (Santa Cruz Biotechnology); rabbit polyclonal anti-Hsc70 (heat-shock cognate 70 stress protein) (1:5000 dilution) and anti-syntaxin 1 (1:100000 dilution) antibodies (Sigma); rabbit polyclonal anti-complexin (1:1000 dilution), anti-synaptotagmin (1:1000 dilution) and anti-SCAMP (secretory carrier membrane protein) (1:1000 dilution) antibodies (Synaptic Systems, Goettingen, Germany). Anti-Csp antibody was used at 1:1000 dilution [47a] and rabbit anti-Vamp2 antibody (1:500 dilution) was a gift from Professor M. Takahashi (School of Life Sciences, University of Tokyo, Tokyo, Japan).
For analysis of the binding of His6-tagged syntaxin 1A (4–266) to immobilized GST–Munc18-1 proteins, glutathione–Sepharose beads were washed three times with binding buffer [20 mM Hepes (pH 7.4), 150 mM NaCl, 1 mM DTT (dithiothreitol), 2 mM MgCl2 and 0.5% (v/v) Triton X-100] and incubated with E. coli protein extract. Recombinant GST-tagged proteins were incubated with beads at a final concentration of 1 μM and in a final volume of 100 μl for 30 min. His6–syntaxin 1A (4–266) (1 μM final concentration) was added to the GST–protein bound beads for 2 h. Beads were washed using CytoSignal filters and samples eluted, separated by SDS/PAGE (12.5% gels) and transferred for immunoblotting as described above, with a monoclonal anti-syntaxin 1 antibody used for detection. For analysis of binding of His6-tagged Rab3A, this was incubated with glutathione–Sepharose immobilized GST–Munc18-1 at a final concentration of 1 μM and in a final volume of 100 μl for 30 min. The guanyl nucleotide dependency of binding was examined by performing binding studies in the presence of either 50 μM GDP[β-S] (guanosine 5′-[β-thio]diphosphate) or GTP[S] (guanosine 5′-[γ-thio]triphosphate) (Sigma) with 1 mM ATP.
Amperometric recording and analysis
The electrochemical recording conditions were as described previously . Briefly, cells were incubated in bath buffer [139 mM potassium glutamate, 0.2 mM EGTA, 20 mM Pipes, 2 mM ATP and 2 mM MgCl2 (pH 6.5)] for 30 min, and a 5 μm-diameter carbon fibre electrode was positioned in direct contact with a targeted adrenal chromaffin cell. Exocytosis from cells was stimulated by pressure ejection of a permeabilization/stimulation buffer [139 mM potassium glutamate, 20 mM Pipes, 5 mM EGTA, 2 mM ATP, 2 mM MgCl2, 20 μM digitonin and 10 μM calcium (pH 6.5)] from a glass pipette oriented on the opposite side of the cell. Amperometric responses were monitored using a VA-10 amplifier (NPI Electronic, Tamm, Germany) and acquired with Axoscope 8 (Axon Instruments). Individual experiments were conducted in parallel on control (untransfected) and transfected cells from the same cell batch. Successful transfection was identified by expression of EGFP (enhanced green fluorescent protein), as previous studies have established a 95% rate of co-expression. Amperometric data were exported from Axoscope and subsequently analysed using Origin (Microcal Software, Northampton, MA, U.S.A.). Spikes were selected for analysis, provided that the spike amplitude was greater than 40 pA to remove any confounding effects of including those fusion events not occurring directly underneath the carbon fibre end. Results are means±S.E.M., and significance was established using non-parametric Mann–Whitney U-tests.
GH (growth hormone) release assay
The effect of Munc18-1 on exocytosis was analysed using an assay for the release of expressed human GH from PC12 (pheochromocytoma cells) cells as described previously [23,44]. Transient transfections were performed using 3 μl Lipofectamine™ 2000, 1 μg of pXGH5 (GH-expressing vector) and either 1 μg pcDNA3 alone or pcDNA3 containing wild-type Munc18, Munc18 E466K or Munc18 R39C/E466K, which were incubated at room temperature (24 °C) for 20 min before being added to one well of a poly-D-lysine-coated 24-well plate (BD Biocoat; BD Biosciences) containing 0.35×106 cells. After 72 h incubation, cells were washed in Krebs buffer [20 mM Hepes (pH 7.4), 145 mM NaCl, 5mM KCl, 1.2 mM NaH2PO4, 10 mM glucose, 1.3 mM MgCl2 and 3 mM CaCl2] and challenged for 15 min with either Krebs buffer alone or Krebs buffer containing 300 μM ATP. Cells were solubilized using PBS containing 0.5% (v/v) Triton X-100 and both secreted and intracellular GH content was assayed using a human GH ELISA (Roche).
PC12 cells were seeded at ∼4×105 cells per well on poly-D-lysine-coated 24-well glass coverslips (BD Biosciences) and cells were allowed to adhere overnight. Prior to transfection, growth medium  was replaced with 400 μl/well Optimem 1 medium (Invitrogen). Transfection mixes were prepared, containing 3 μl Lipofectamine™ 2000 (Invitrogen) per μg of plasmid in 100 μl Optimem 1, and incubated at room temperature for 20 min before being added dropwise to the cells. The cells were incubated in the transfection mixture for 5 h and the medium was then replaced with normal growth medium.
The transfected cells were washed in PBS and then fixed in 0.5 ml 4% (w/v) paraformaldehyde for 30 min at room temperature. Coverslips were washed twice in PBS, incubated with anti-Munc18 antibody (Calbiochem) at 1:400 dilution in PBT [PBS containing 0.3% (v/v) BSA and 0.1% (v/v) Triton X-100] for 1 h and then washed three times with PBT. The coverslips were incubated with Alexa Fluor® 594-conjugated secondary antibody (Invitrogen) at a 1:100 dilution in PBT for 1 h as appropriate. After a final washing step, coverslips were air-dried prior to mounting on to glass microslips using Mowiol (Calbiochem).
Confocal microscopy and quantification
Confocal microscopy following immunostaining was carried out on a Leica TCS-SP microscope (Leica Microsystems, Heidelberg, Germany) using a ×63 oil immersion objective with a 1.4 numerical aperture and the pinhole set to 1 Airy unit. EGFP-tagged constructs were excited using a 488 nm laser and light collected between 500–550 nm. Alexa Fluor® 594 staining was excited using a 594 nm laser and light collected between 625–675 nm. Regions of interest were drawn around the outside of each cell, immediately beneath the cell periphery and around the nucleus. Peripheral fluorescence was calculated by subtraction of the cytoplasmic and nuclear fluorescence from the outer region of interest. Cytosolic fluorescence was calculated by subtraction of nuclear fluorescence. Ratios of peripheral to cytosolic fluorescence were calculated on the basis of the mean fluorescence per pixel in the respective regions rather than total fluorescence, to avoid problems arising from variations in the size of the selected regions of interest. To determine overall expression levels of overexpressed Munc18-1 and its mutants or EGFP–Rab3A, the mean pixel intensity over the whole cell was determined using ImageJ [NIH (National Institutes of Health), Bethesda, MD, U.S.A.].
The E466K mutation increases the interaction of Munc18-1 with Rab3A
Expression of Munc18-1 bearing the E466K mutation (Figure 1a) in adrenal chromaffin cells increases the number of exocytotic events . The E466K mutation does not, however, affect the expression of Munc18-1 in transfected cells or its binding to the known interacting proteins syntaxin1, granuphilin, Mint1 or Mint2 . In order to investigate the molecular basis of the E466K gain-of-function mutation as seen in chromaffin cells, we set out to search for new or enhanced protein interactions with Munc18-1 E466K. The R39C mutation was also introduced into the Munc18-1 E466K background, as this has been shown to inhibit the binding of Munc18-1 to the closed form of syntaxin 1 [23,42], and allowed us to probe the importance of this mode of interaction with syntaxin. In the crystal structure of the Munc18-1/syntaxin 1A complex , Arg39 is present in a cleft in which the closed form of syntaxin 1A resides, and so makes direct contact with syntaxin 1A (Figure 1a). As expected from the structure, the R39C mutation substantially reduces direct binding to GST–syntaxin 1A constructs in assays where the binding is predominantly to closed syntaxin 1A [23,42]. As seen for the R39C mutation alone, the Munc18-1 R39C/E466K double mutant bound markedly less His6–syntaxin 1A (4–266) in a direct binding assay, with a 4.6-fold reduction in binding compared with wild-type Munc18-1 (Figure 1b).
Analysis of Munc18-1 and Munc18-1 mutant binding to Rab3A and syntaxin 1A
GST-fusion proteins for Munc18-1 wild-type, Munc18-1 E466K and Munc18-1 R39C/E466K were initially used in pull-down assays in which proteins bound from a brain extract were detected using Western blotting to search for new or increased protein interactions with Munc18-1 E466K or the Munc18-1 R39C/E466K double mutant. In initial screens, no binding or differences in interaction between the Munc18-1 mutants compared with wild-type Munc18-1 were seen for N-ethylmaleimide-sensitive factor, amisyn, Csp, Doc2, β-adaptin, sec6, sec8, SNAP-25, synapsin, synaptotagmin 1 or synaptophysin. However, increased binding of Rab3A by Munc18-1 E466K was observed (Figure 1c). In these experiments, equal amounts of the GST-tagged Munc18-1 were bound to the glutathione–Sepharose beads (Figure 1c). An increased level of binding of Rab3A from brain extract was also found with the double mutant Munc18-1 R39C/E466K compared with wild-type Munc18-1 (Figures 1c and 1d), indicating that this interaction was not dependent on Munc18-1 binding to closed syntaxin 1A. In independent pull-down experiments, increased binding of Rab3A to Munc18-1 E466K was consistently observed, with low but detectable binding to wild-type Munc18-1 over GST controls also shown in four out of six experiments. From quantification of Western blotting using recombinant Rab3A as a standard, it was estimated that the concentration of Rab3A in brain extracts was approx. 80 nM and 5–8% of the input Rab3A became bound to immobilized Munc18-1 R39C/E466K in these experiments. Many other proteins tested from brain extracts did not bind to wild-type or mutant Munc18-1, but another novel interaction was identified. Specific binding of Sec8 was observed, but this was not consistentlty different for wild-type compared with mutant Munc18-1 (Figure 1c). The specificity of the Rab3A interaction with Munc18-1 was suggested by the finding that in contrast with Rab3A, Rab5 was only detected at background levels in all pull-down experiments performed (Figure 1d). As an additional control, the binding of the chaperone protein Hsc70 was tested, as this often binds to recombinant proteins. Only low levels of Hsc70 were detected bound to wild-type Munc18-1, and the level of binding was not increased by the presence of the double mutation (Figure 1d), indicating that these mutations did not result in misfolding of the protein.
In order to assess whether Munc18-1, Rab3A and syntaxin 1 associate in the same complex, GST–syntaxin 1A (4–266) was incubated with brain extract in pull-down experiments to see if both native Munc18-1 and Rab3A could be found associated with the syntaxin construct (Figure 2a). Munc18-1 was shown to efficiently interact with GST–syntaxin 1A (4–266). In contrast, although Rab3A was readily detected in the brain extract, no Rab3A was detected bound to GST–syntaxin 1A (4–266). To check if this was the case for native syntaxin in SNARE complexes, we also used GST–complexin II in pull-downs, as this binds efficiently only to the assembled SNARE complex . As a control, we used the complexin II R59H mutant construct . This residue is involved in the direct interaction with VAMP in the SNARE complex , and mutation of Arg59 abolished binding of SNARE proteins in the pull-down assay (Figure 2b). Munc18-1 was found specifically associated along with the SNARE complex in the GST–complexin II pull-down, whereas other control proteins were not detected. In contrast, no Rab3A was detected bound to GST–complexin II (Figure 2b), even though it was readily detected in the input brain extract. These data suggest that Rab3A cannot bind to Munc18-1 when it is bound to closed syntaxin [GST–syntaxin 1A (4–266)] or the SNARE complex.
Rab3A does not associate with Munc18-1 bound to syntaxin or the native SNARE complex
The binding of Rab3A to Munc18-1 could have been the result of direct or indirect interactions with complexes containing other proteins in the brain extract. To assess this possibility, pull-down assays were performed using recombinant Rab3A. To provide a control, pull-downs were also carried out using the known Rab3A effector Noc2, which binds Rab3A in vitro . In this direct-binding assay, both wild-type and R39C/E466K Munc18-1 appeared to bind Rab3A equally well, whereas there was little binding to control GST (Figure 3a). This difference, in comparison with the assay using brain extract, may be the result of Rab3A being present at a higher concentration (1 μM) than in brain extracts. The level of Munc18-1 binding to Rab3A was similar to that seen with GST–Noc2 and, from quantification of Western blots of bound Rab3A compared with input Rab3A, a stoichiometry of approx. 0.7 moles of Rab3 per mole of Munc18-1 was determined. In contrast with the situation with Noc2, where binding to Rab3A was 5.5-fold higher in the presence of GTP[S] compared with GDP[β-S], binding of Munc18-1 wild-type or Munc18-1 R39C/E466K to Rab3A was not dependent on the guanine nucleotide-loaded status of Rab3A (Figure 3b).
Direct binding of Rab3A to Munc18-1
Stimulation of exocytosis by Munc18-1 E466K is dependent on binding to closed syntaxin
Expression of the Munc18-1 E466K mutant in adrenal chromaffin cells increased the rate of exocytotic events measured using carbon-fibre amperometry on single transfected cells . Unlike other studies, in which very high overexpression of wild-type Munc18-1 was observed in immature chromaffin cells using viral constructs , we and others have not observed such a stimulatory effect of the wild-type protein in studies on chromaffin  or PC12 cells [50,51]. We used our well-characterized amperometric assay  in order to test whether the specific stimulation of exocytosis by the E466K mutation required high-affinity binding to closed syntaxin 1. To do this, we examined the effect of the additional R39C mutation. Expression of Munc18-1 R39C was previously found to have no effect on the number of exocytotic events measured by amperometry . Example traces are shown immediately after stimulation and the onset of amperometric spikes, which were largely undetectable in unstimulated cells (Figure 4a). As a result of variablity between individual cells and recording electrodes, data were derived from 17–37 cells per condition, and control recordings were taken in parallel from non-transfected cells in the same dishes with the same carbon-fibre electrodes . The spike frequency was expressed as a percentage of the appropriate controls for each Munc18-1 protein. We have previously established that transfection with empty plasmid and EGFP does not itself affect exocytosis . Expression of Munc18-1 E466K increased the frequency of amperometric spikes in chromaffin cells measured over a 210 s period after stimulation (Figure 4c). In order to test whether the specific stimulation of exocytosis by Munc18-1 E466K required the ability of Munc18-1 to bind to closed syntaxin 1, we examined the effect of the additional R39C mutation. Introduction of this second mutation on the Munc18-1 E466K background abolished the stimulatory effect of the E466K mutation on exocytosis (Figures 4a–4c).
The increase in evoked exocytosis by Munc18-1 E466K in adrenal chromaffin and PC12 cells requires mode 1 binding to closed syntaxin 1
An alternative assay was used with PC12 cells, in which the release of newly synthesized GH expressed from a co-transfected plasmid was examined. Cells were stimulated with ATP and release of GH above basal levels was determined. Overexpression of wild-type Munc18-1 did not affect evoked GH release, but GH release was significantly increased by 20% following expression of Munc18-1 E466K (Figure 4d). Again, this stimulation apparently required syntaxin binding as the double mutant had no effect, despite it being expressed in PC12 cells. The difference in the effect of Munc18-1 E466K compared with Munc18-1 R39C/E466K was not the result of differences in the expression levels of the proteins (see below).
Expression of the Munc18-1 E466K mutant increases secretory granule recruitment to the cell periphery
Rab3A is associated with secretory granules but not other organelles in PC12 cells, and we have shown that Rab3A cycles rapidly between the granule and cytosol under resting conditions . In order to examine whether an interaction between the Munc18-1 E466K mutant and Rab3A would affect Rab3A cycling, PC12 cells were transfected to express both the Munc18-1 mutant and EGFP–Rab3A, and the cycling of EGFP–Rab3A was assessed by analysis of FRAP (fluorescence recovery after photobleaching) as described previously . Following photobleaching of a small region of the cell, recovery of EGFP–Rab3A fluorescence on granules did not differ in cells with or without expression of Munc18-1 E466K (Figure 5a). In these experiments, EGFP–Rab3A-labelled granules appeared to be distributed throughout the control cells, consistent with only a proportion (∼30%) of the granules being docked in the particular PC12 cells used. We noticed, however, that in these experiments, more labelled granules appeared to be present at the cell periphery in cells expressing Munc18-1 E466K (Figure 5b). Previous work has implicated both Munc18-1  and Rab3A [6,55] in granule tethering at the plasma membrane or docking in neuroendocrine cells, and so we investigated whether the E466K gain-of-function mutation caused increasing granule recruitment. Cells were transfected to express a Munc18-1 protein and EGFP–Rab3A and were fixed and immunostained with anti-Munc18 serum. Cells overexpressing Munc18-1 following transfection could be detected by immunostaining, due to a higher fluorescence intensity when compared with non-transfected cells . This experiment was performed using a carefully titrated concentration of anti-Munc18 serum that would not detect endogenous Munc18-1, only overexpressed Munc18-1 (Figure 5b). In the initial analysis of EGFP–Rab3A distribution, cells were selected as Munc18-1 high-expressing cells as determined by immunostaining and compared with cells separately transfected with empty vector. Under standardized imaging conditions on the same day, regions of interest were drawn around the cell periphery and the nucleus (Figure 5c), and the mean pixel intensity was determined both at the cell periphery and in the rest of the cell (minus the nucleus) and expressed as a ratio. We determined mean pixel intensity rather than total fluorescence in the defined areas in order to rule out any error resulting from the size of the compartments within the regions of interest. Overexpression of wild-type Munc18-1, which did not increase exocytosis, did not affect the fluorescence ratio compared with low-expressing cells. In contrast, in Munc18-1 E466K-expressing cells, there was a significant increase in peripheral fluorescence intensity for the granule-associated EGFP–Rab3A, compared with controls (Figure 5d). However, the Munc18-1 R39C/E466K double mutant displayed no effect on granule distribution. The lack of effect of the Munc18-1 R39C/E466K mutant was not the result of lower expression, as the levels of expression of wild-type, E466K and R39C/E466K Munc18-1 were very similar, and the expression of EGFP–Rab3A was also similar in cells expressing each of the Munc18-1 constructs (Figures 5d and 5e). Therefore the increase in peripheral granules following expression of Munc18-1 E466K appeared to require the ability of Munc18-1 to interact with closed syntaxin 1.
Expression of Munc18-1 E466K does not affect the dynamics of Rab3A cycling but redistributes secretory granules to the cell periphery
In a subsequent alternative analysis of an independent experiment, Munc18-1 low-expressing cells, in which fluorescence levels were comparable with untransfected cells, were selected to act as controls to be compared with Munc18-1 high-expressing cells on the same coverslips, in order to rule out variations between coverslips. Very similar findings were obtained in this analysis, with Munc18-1 E466K but not wild-type or R39C/E466K Munc18-1 increasing the proportion of labelled granules at the cell periphery in high-expressing cells (Figures 6a and 6b). Unlike Rab3A, Rab27A does not cycle on and off from secretory granules . In an additional experiment using EGFP–Rab27A instead of EGFP–Rab3A to label the secretory granules, similar findings were obtained (Figures 6c and 6d), indicating the reproducibility of the effect on granule distribution.
Localization of EGFP–Rab3A and EGFP—Rab27A in control and cells expressing high levels of wild-type Munc18-1 or Munc18-1 mutants
In neuroendocrine cells, the cortical actin network is able to act as a barrier to prevent granules reaching the plasma membrane . It has been suggested that, in addition to direct effects on granule tethering and docking, another mechanism by which Munc18-1 could increase the number of tethered granules is by disrupting cortical actin . This suggestion was based on the observation of gross changes in the cortical actin in chromaffin cells in which Munc18-1 levels had been modified. We examined the effect of expression of Munc18-1 E466K on cortical actin, which was visualized using staining with phalloidin. No difference was seen in the overall organization of peripheral actin, or in the intensity of actin staining in the cell cortex, ruling out this indirect mechanism for the redistribution of granules to the cell periphery (Figure 7).
Expression of Munc18-1 E466K does not affect the organization of the cortical actin cytoskeleton
One regulated step in vesicle exocytosis is the recruitment of vesicles to the plasma membrane. This has been characterized as being the result of vesicle tethering or docking, although the two processes have been difficult to define. The results presented here address two key questions regarding Munc18-1 function. First, what is the basis for Munc18-1 being able to stimulate vesicle docking alongside Rab3? Second, is the mode 1 interaction of Munc18-1 with closed syntaxin 1 functionally relevant? It has been shown in a number of systems that Rab proteins acting through their effectors can mediate vesicle tethering or docking . The additional requirement for Munc18-1 for vesicle docking and tethering suggests a possible link between Munc18-1 and Rabs. Significantly, recent work has established that increased secretory granule recruitment as a result of Rab3 overexpression requires Munc18-1 . However, the molecular basis for this was not explored. One possibility is that this could involve the interaction of Rab effectors with Munc18-1, such as that described between granuphilin and Munc18-1 . We now provide an alternative explanation by showing that Munc18-1 can interact directly with Rab3A. Whereas we detected novel interactions of Munc18-1 with Rab3A and Sec8, only the binding of Rab3A was increased by the Munc18-1 E466K gain-of-function mutant. Expression of Munc18-1 E466K increased exocytosis in adrenal chromaffin and PC12 cells. In squid Sec1  and rat Munc18-1 , Glu466 is in domain III of the protein and is present in an exposed residue on a surface loop, suggesting that this residue could potentially influence protein interactions. A peptide corresponding to the loop region inhibits neurotransmitter release in the squid giant synapse  and Munc18c-dependent translocation of GLUT4 vesicles to the plasma membrane in adipocytes , supporting the functional significance of the loop and suggesting that the E466K mutation could influence crucial protein interactions involving this region. Therefore we set out to search for any protein interactions that were affected by the E466K mutation, and found a specific increase in Rab3A binding from brain extracts to Munc18-1 E466K. This interaction was the result of direct binding of Rab3A to Munc18-1. Expression of the Munc18-1 E466K gain-of-function mutant was also found to increase the proportion of secretory granules at the cell periphery following expression in PC12 cells, consistent with an increase in granule tethering or docking. Confocal microscopy was used to quantify peripheral fluorescence intensity in cells expressing EGFP-tagged Rab3A or Rab27A, which associate preferentially with newly synthesized granules . The same approach has also been used by others and shown to detect changes in granule recruitment (interpreted as increased docking) following manipulation of granuphilin , rabphilin , Rab3A or Rab27A , with expression at granules detected by expression of tagged granule proteins. This approach is valid, as the newly synthesized granules which are labelled are those that form the most readily releasable pool . We have observed the same phenomenon of increased granule redistribution to the cell periphery following granuphilin overexpression using our assay in PC12 cells (M. T. W. Handley and R. D. Burgoyne, unpublished work) as described previously for insulin secreting cells .
Knockout studies have shown that Munc18-1 is normally required for vesicle tethering and docking [21,22]. Wild-type Munc18-1 can bind Rab3A as well as the Munc18-1 E466K mutant in direct-binding assays. Since Munc18-1 is required for Rab-stimulated granule recruitment , this suggests that one of the physiological functions of Munc18-1 may be to link Rab-dependent tethering with SNARE-dependent docking and fusion. The E466K mutation increased Rab3A binding in brain extracts compared with wild-type Munc18-1, to which the binding of Rab3A was low, but using higher concentrations of Rab3A in a direct-binding assay, it was possible to show that wild-type Munc18-1 also binds Rab3A directly. One possibility is that the E466K mutation removes a mechanism that normally limits the Munc18-1/Rab3 interaction within cells or extracts, or alternatively it increases the overall affinity of the interaction with Rab3A, and this leads to the gain of function. The Sly1-20 mutant, on which the Munc18-1 E466K mutant was modelled, was discovered as a suppressor of a loss-of-function mutation in the yeast Rab protein Ypt1, suggesting that it bypassed the requirement for Ypt1 . It may seem counterintuitive for the Munc18-1 E466K mutant to enhance granule recruitment and exocytosis via an increased interaction with Rab3A. These two situations can be reconciled, however, as more detailed analysis of the Sly1-20 mutant suggested that it allowed another closely related Rab protein, Ypt6p, to substitute for Ypt1p , consistent with an alteration in Rab-binding properties as a result of the mutation. It is possible that the loop in domain III of Munc18-1 and other SM proteins could form a Rab-binding site.
Munc18-1, Rab3A and syntaxin 1 were not all found together in the same complex, but the functional effects of the E466K mutation required the ability of Munc18-1 to interact with syntaxin, as shown by the effect of the R39C mutation. We have used the R39C mutation to probe the importance of the mode 1 Munc18-1 interaction with the closed form of syntaxin 1. This mutation disrupts binding to syntaxin 1 constructs in vitro [23,42], but has no effect on the ability of Munc18-1 to stimulate fusion in a manner which is dependent on mode 2 and mode 3 binding . We have observed no differences between Munc18-1 E466K and Munc18-1 R39C/E466K in their binding properties, apart from the reduced binding of closed syntaxin 1 to the double mutant. For example, both proteins bind Rab3A and Sec8. In addition, low levels of binding of the Hsc70 chaperone protein from brain extract to the Munc18-1 R39C/E466K double mutant showed that it was unlikely to be misfolded compared with wild-type Munc18-1. Finally, we confirmed that both Munc18-1 mutants were expressed to the same extent as wild-type protein after transfection. It is notable that the effects of the Munc18-1 E466K mutant on exocytosis and vesicle distribution were lost when the R39C mutation was introduced, suggesting these effects were mediated by the mode 1 interaction with closed syntaxin. Despite the finding that the Munc18-1 E466K mutant required syntaxin binding for its functional effect, it appeared that Munc18-1, Rab3A and syntaxin 1 may not stably assemble into the same complex. One speculative possibility is the existence of a sequential series of interactions in which Munc18-1 initially interacts with Rab3A and is then handed on during the process of tethering, as the tethered state is stabilized through the closed syntaxin interaction. Munc18-1 would thus form a transient bridge between plasma membrane syntaxin and vesicle Rab3 to allow tethering before the alternative modes of interaction of Munc18-1 with syntaxin and the SNARE complex come into play in later stages, leading to membrane fusion.
This work was supported by grants from The Wellcome Trust and the BBSRC (Biotechnology and Biological Sciences Research Council). M. T. W. H. was supported by a Medical Research Council Studentship.
enhanced green fluorescent protein
heat-shock cognate 70 stress protein
PBS containing 0.3% (w/v) BSA and 0.1% (v/v) Triton X-100
- PC12 cells
soluble N-ethylmaleimide-sensitive factor-attachment protein
vesicle-associated membrane protein
vacuolar protein sorting
These authors contributed equally to this work.