Control of Fusion Pore Dynamics During Exocytosis by Munc18

See allHide authors and affiliations

Science  02 Feb 2001:
Vol. 291, Issue 5505, pp. 875-878
DOI: 10.1126/science.291.5505.875


Intracellular membrane fusion is mediated by the SNARE (soluble N-ethylmaleimide–sensitive factor attachment protein receptor) proteins. All vesicle transport steps also have an essential requirement for a member of the Sec1 protein family, including the neuronal Munc18-1 (also known as nSec1) in regulated exocytosis. Here, in adrenal chromaffin cells, we expressed a Munc18 mutant with reduced affinity for syntaxin, which specifically modified the kinetics of single-granule exocytotic release events, consistent with an acceleration of fusion pore expansion. Thus, Munc18 functions in a late stage in the fusion process, where its dissociation from syntaxin determines the kinetics of postfusion events.

Fusion of vesicles with target membranes is a key aspect of vesicular traffic and neurotransmitter release and is mediated by a core machinery of SNARE proteins (1–3). The three synaptic SNAREs (syntaxin 1, SNAP-25, and VAMP) are sufficient for bilayer fusion in vitro (4). This is, however, orders of magnitude slower than synaptic vesicle fusion, suggesting a role for additional factors (5). Members of the Sec1 family of proteins are required for all intracellular vesicular traffic steps (6–10), and the neuronal Munc18-1 protein (11, 12) is essential for synaptic vesicle exocytosis but not for constitutive exocytosis (6,8). It binds to a closed conformation of syntaxin 1, which is unable to participate in the SNARE complex (13–15) and thus may control the addition of syntaxin into the complex. In current models, syntaxin assembles into a “loose” SNARE complex with Munc18 dissociation and the subsequent zippering of SNAREs into a “tight” complex immediately preceding membrane fusion (2,16). Although SNARE complex assembly is crucial for membrane fusion, it is not known whether the complete assembly of the SNARE complex occurs before, during, or after fusion. Fusion proceeds via the formation of a reversible fusion pore (17), but the relation of the function of the SNARE proteins or Sec1 proteins to fusion pore opening and expansion is unknown.

We examined the effect of overexpression of wild-type Munc18 or a mutant form, Arg39 → Cys39 (R39C), on exocytosis using single-cell amperometry to resolve the frequency and kinetics of individual secretory granule release events. The R39C mutation was investigated for two reasons. First, the equivalent mutation in the Drosophila ortholog Rop, Arg50→ Cys50 in the F3 mutant (9), produces flies showing an increase in evoked neurotransmission (8). Second, the crystal structure of the Munc18–syntaxin 1 complex has revealed that Arg39 makes direct contact with Glu234 of syntaxin, leading to the prediction that the R39C mutation should weaken the binding interaction between the two proteins (18). A reduction of high-affinity binding of Munc18 containing the R39C mutation to syntaxin 1 was confirmed with an in vitro binding assay (11, 19) (Fig. 1A). When binding was assayed over a range of Munc18 protein concentrations (2 to 109 nM), the affinity of R39C for syntaxin was reduced by about fivefold in comparison to the wild type, from 6.5 to 35 nM, respectively. The R39C mutant (Fig. 1B) still bound the other Munc18 binding proteins Doc2 (20) and Mint1 (21) from a rat brain extract (22).

Figure 1

Reduced binding of Munc18 R39C to syntaxin and the effect of expression on evoked GH release. (A) Wild-type Munc18 and the R39C mutant were labeled by in vitro transcription and translation in the presence of [35S]methionine, and binding to GST–syntaxin 1A or GST control was assayed. (i), (iii), and (v) show autoradiograms and (ii) and (iv) show Coomassie blue–stained proteins. No binding of Munc18 was seen in GST control incubations. From a quantitative analysis, the binding of R39C to GST–syntaxin 1A was reduced to 1% of the wild-type binding in three experiments. (B) Pull-down assay showing the binding of Doc2 and Mint1 from rat brain membrane extracts to GST-Munc18 (wild type), GST-Munc18 (R39C), and control beads. (C) Demonstration of overexpression of wild-type or R39C Munc18 in adrenal chromaffin cells that were cotransfected with GFP and stained with anti-Munc18 [anti-rbSec1A (45)] at a concentration (1:200) that gave only background staining of control cells (left) transfected with a vector. The scale bar represents 10 μm. (D) PC12 cells were cotransfected with a control vector or plasmids encoding wild-type Munc18 or the R39C mutant along with a plasmid encoding human GH. Three days after transfection, the cells were permeabilized with digitonin and challenged with or without Ca2+ for 10 min. Released human GH was assayed and expressed as a percentage of total cell content (n = 6). Error bars indicate SEM.

The effect of wild-type or R39C Munc18 on overall dense-core granule exocytosis was first assayed in PC12 cells by using transfection and coexpression of growth hormone (GH) (23–25). Transfection resulted in an ∼10-fold overexpression of wild-type Munc18 (26). Overexpression of wild-type protein had no statistically significant effect on the extent of evoked exocytosis due to 10 μM free Ca2+ in permeabilized cells, but the R39C construct produced a significant (55%) inhibition (Fig. 1D). The difference in effect between the two proteins was not due to differences in expression, as both were expressed in virtually all transfected cells (Fig. 1C). We analyzed the effect of overexpression of the proteins in adrenal chromaffin cells using carbon-fiber amperometry to allow direct analysis of single-granule release events (27–30). Transfected chromaffin cells were detected by coexpression of green fluorescent protein (GFP), allowing untransfected cells in the same dishes to be used as controls. This ensured that all cells had been through the transfection protocol, and the same carbon fibers were used to record from transfected and control cells in each series of experiments. The cells were stimulated by local application of digitonin and Ca2+ to permeabilize the cells and allow Ca2+ to directly activate exocytosis (24, 31). Because the granules in these cells have a half-life of >15 days (32), this assay measured release from preformed granules. Overexpression of wild-type Munc18 or R39C had no effect on the overall evoked responses of the cells (Fig. 2) or on the average number of exocytotic events per cell. For wild-type protein, this was 7.43 ± 0.27 (n = 14 cells) for transfected cells versus 10.4 ± 3.1 (n = 10) for control cells, and for R39C, this was 6.32 ± 1.45 (n = 19) for transfected cells versus 6.88 ± 2.93 (n = 8) for control cells in the first 2 min after perfusion. In contrast, treatment with reserpine, an inhibitor of the vesicular monoamine transporter, to partially deplete granule catecholamine levels (32) had an obvious effect in reducing the peak amplitude of the spikes (Fig. 2, D and E).

Figure 2

Amperometric records of evoked exocytosis in adrenal chromaffin cells. Chromaffin cells were transfected with plasmid encoding wild-type Munc18 or the R39C mutant along with a plasmid encoding GFP. Recordings were made from (A) control untransfected cells and from (B and C) transfected cells expressing GFP in the same dishes by using a carbon-fiber electrode in contact with the cell membrane. The shaded bar shows the period of local perfusion with 20 μM digitonin and 10 μM Ca2+ from a pressure ejection pipette to permeabilize the cells and activate exocytosis. The carbon-fiber amperometry showed discrete events (spikes) due to catecholamine release from individual secretory granules, and the responses appeared to be similar for each condition. Overexpression of wild-type Munc18 or R39C did not affect spike amplitude. In contrast, reserpine treatment to deplete vesicle catecholamine reduced spike amplitude in comparison with controls (D and E). The latter data are typical of 11 control and 14 reserpine-treated cells.

Examination of the characteristics of the individual amperometric spikes, each reflecting catecholamine release from single granules (27), showed few effects of overexpression of wild-type Munc18 (33). In contrast, expression of R39C resulted in marked changes in spike characteristics, as compared to spikes from the corresponding control cells (Fig. 3). The mean height was unaffected, but the total charge, a reflection of the total catecholamine released per granule, was reduced by 45%, consistent with the inhibitory effect of the R39C protein on GH release from PC12 cells. This reduction in charge per spike is probably a consequence of a reduction in the half width of the spikes (Fig. 3C), i.e., the time over which release could occur. It is not due to a reduction in catecholamine content, as we also saw changes in the kinetics of release with decreases in both the rise and fall times of the spikes (Fig. 3, D and E). The reduction in charge per spike in R39C-expressing cells is not a consequence of the more rapid rate of release, as we found no correlation for individual spikes between the total charge and rise time under any conditions (33). A statistically significant correlation was found, however, between total charge and the half width of spikes (33), suggesting that the reduction in catecholamine release is a consequence of a shortening of the release time course. The change in the shape of spikes, showing the average parameters (Fig. 3F), was different in R39C-expressing cells as compared to the effect of depletion of vesicle catecholamine, where only the spike amplitude was changed (Fig. 3G).

Figure 3

Analysis of single amperometric spikes from cells expressing Munc18 R39C. No change in spike height was seen in R39C transfected cells as compared to (A) the respective controls, but statistically significant reductions were observed in (B) total charge, (C) half width, (D) rise time, and (E) fall time of the spikes. The data are derived from experiments on 8 (53 spikes) control and 19 (117 spikes) transfected cells. Error bars indicate SEM. (F) Spikes with average parameters from control and R39C-expressing cells. (G) Spikes from control and reserpine-treated (0.5 μM, for 16 hours) cells.

The change in the rise and fall times seen above suggests an increase in the rate of opening of the initial fusion pore followed by its rapid closure in a kiss-and-run–type fusion event (34). This effect was specific, as amperometric spikes were not modified by reduction in SNARE availability by clostridial neurotoxin treatment (24) or by expression of a dominant-negative α-SNAP mutant that can interact irreversibly with multiple SNARE proteins (35). The kinetics of single release events can be modified, however, under specific conditions (24, 30), including treatment with phorbol esters that increases fusion pore opening rates in other cell types (36). The effect of expression of R39C is identical to the effect of activation of protein kinase C (PKC) (24). PKC phosphorylation of Munc18 reduces its binding to syntaxin 1 (37), so that the phosphorylated form behaves like R39C, suggesting that phosphorylation of Munc18 could regulate kiss-and-run fusion in regulated exocytosis.

The reduction in release from dense-core granules following R39C expression can be reconciled with the findings inDrosophila, where the mutation leads to increased neurotransmission (8). The consequence of the mutation could be to increase the rate of neurotransmitter release and the rate of rise in neurotransmitter concentration in the synaptic cleft. This could explain the increase in postsynaptic current and the reduction in neurotransmission failures in the Drosophila F3 mutant. The reduced affinity of R39C for syntaxin suggests that R39C would be able to dissociate more easily from syntaxin than would wild-type protein and R39C would allow more efficient assembly of the “tight” SNARE complex (18). If this is the case, then our data suggest that dissociation of Munc18 and the full SNARE complex assembly occur very late, either during membrane fusion or during fusion pore expansion. Alternatively, dissociation of Munc18 may be important to allow it to directly exert an effect on the fusion process independent of syntaxin 1 either alone or via a downstream effector. The yeast Sec1 has been suggested to have a late role in constitutive exocytosis after SNARE complex assembly (38, 39), consistent with this interpretation. Accumulating evidence suggests that fusion pore dynamics can be regulated (40–44). Our data show that Munc18 plays a late role in membrane fusion and has a key function in fusion pore dynamics, and the data suggest that it may be a target for the control of kiss-and-run fusion in exocytosis.

  • * To whom correspondence should be addressed. E-mail: Burgoyne{at}


View Abstract

Navigate This Article