Evidence of a role for Munc18-2 and microtubules in mast cell granule exocytosis

Mast cells are secretory cells that undergo an extensive release of their granular content of inflammatory mediators within a few minutes following stimulation. This process, termed degranulation, involves compound exocytosis that requires both fusion of docked secretory granules (SGs) at the plasma membrane (PM) and fusion of SGs from within the cell (Alvarez de Toledo and Fernandez, 1990). Degranulation, like other trafficking steps, depends on the interaction of vesicular or v-SNAREs (soluble N-ethylmaleimide-sensitive fusion factor attachment protein receptor) and target or t-SNAREs to form a core complex that catalyses membrane fusion (Weber et al., 1998; Schoch et al., 2001; Washbourne et al., 2002). In mast cells, these include syntaxin 4 and SNAP-23 as t-SNAREs (Guo et al., 1998; Paumet et al., 2000; Vaidyanathan et al., 2001), while VAMP-2 and VAMP-8 represent candidates for v-SNAREs (Demo et al., 1999; Miesenbock et al., 1998; Paumet et al., 2000).

Vesicle trafficking requires several additional proteins (Chen and Scheller, 2001; Jahn and Südhof, 1999; Zerial and McBride, 2001). Some of these serve to connect membrane fusion with cell signalling. For example, synaptotagmin II and the GTPase Rab3D could play a role in coupling the core fusion machinery to Ca²⁺-activated signals in mast cells (Baram et al., 1999; Pombo et al., 2001). Accumulating evidence also points to the coupling of membrane traffic with cytoskeletal and motor proteins in the spatial and temporal control of vesicle movement in conjunction with Rab GTPases (Hammer and Wu, 2002). However, identification of the molecular constituents in SG exocytosis of mast cells and our understanding of the role of the cytoskeleton in facilitating SG exocytosis is still at its infancy.

Recently, we have begun to appreciate the role of sec1/Munc18 family members in various membrane fusion and trafficking steps (Jahn, 2000). Munc18 proteins, more specifically, have been implicated in exocytosis. They bind to specific sets of syntaxins thereby inhibiting binding to cognate SNARE partners (Jahn, 2000). They respond to cell signalling as they are targets of protein kinases and phosphatases (de Vries et al., 2000; Fletcher et al., 1999; Fujita et al., 1996). In mice, genetic deletion of Munc18-1 abolishes exocytosis at the synapse (Verhage et al., 2000) and dramatically affects large dense core vesicle (LDCV) exocytosis from chromaffin cells (Voets et al., 2001). However, the mechanism of action is still unclear. Morphological studies in chromaffin cells derived from Munc18-1 null mice point to a docking defect as LDCV are dispersed instead of being aligned below the PM (Voets et al., 2001). Munc18 or homologs may also act as chaperones that prevent degradation of syntaxin partners and thus affect the number of available t-SNAREs (Bryant and James, 2001; Voets et al., 2001). Based on studies with Munc18-1 mutants a late postdocking function has also been proposed that could involve regulation of fusion pore expansion (Fisher et al., 2001). Besides neuronal Munc18-1, two more ubiquitously expressed mammalian isoforms, Munc18-2 and Munc18-3, binding to distinct sets of syntaxins have been described (Hata and Südhof, 1995; Tellam et al., 1995). Munc18-2 has been proposed to control apical membrane traffic in epithelial cells (Riento et al., 2000), while Munc18-3 has been implicated in regulated exocytosis of the glucose transporter GLUT4 and in the redirection of secretion from the apical to the basal surface in pancreatic acinar cells (Gaisano et al., 2001; Tamori et al., 1998; Thurmond et al., 1998).

In the present study we investigate the Munc18 isoforms expressed in mast cells and characterise their intracellular localisation pre- and post-stimulation as well as their function in degranulation. We report that Munc18-2 and Munc18-3 are expressed in these cells, are differentially compartmentalised, and respond differently to a stimulus. We further demonstrate that the disruption of the microtubular network affects granule exocytosis and the cellular distribution of Munc18-2. Our results therefore identify Munc18-2 and microtubules as important components of the mast cell secretory process.

Antibodies directed to syntaxin 2, 3 and 4 have been described (Paumet et al., 2000). Munc18 isoform-specific antibodies were prepared by immunising rabbits with peptides (Munc18-2_161-229, Munc18-3_163-234) within the predicted second domain outside the syntaxin binding site fused to glutathion-S-transferase (GST) (GenBank accession nos: Munc18-2, NM_031126; Munc18-3, U19521). Antibodies were affinity-purified and depleted of GST antibodies. Mouse IgE anti-DNP and a mouse monoclonal anti-FcϵRIβ chain have been described (Roa et al., 1997). Monoclonal antibodies directed to RMCP II, serotonin and an anti-Munc18-1 antibody were purchased from Moredun Scientific, UK, Dako A/S, Denmark, and Synaptic Systems, Germany, respectively. F-actin was labelled using phalloidin-FITC (Molecular Probes) and microtubules (MTs) were labelled with an antibody to α-tubulin (Sigma). All secondary antibodies were purchased from Jackson Laboratories and include goat anti-rabbit-FITC/Rhodamine, goat anti-mouse-FITC/Rhodamine or goat anti-rabbit IgG-HRP.

The cDNA encoding rat Munc18-1 and 18-2 were provided by T. C. Südhof (UT Southwestern Medical Center, Dallas, TX). To generate an N-terminal enhanced green fluorescent protein (EGFP) fusion construct, the Munc18-2 cDNA was PCR-amplified and introduced in frame into EGFP-C1 (Clontech). The murine Munc18-3 cDNA was provided as an EGFP fusion construct by J. E. Pessin (University of Iowa, Iowa City, IA) in pEGFP-C2. Amino acids 433-488 of Munc18-2 corresponding to a predicted loop structure were cloned into EGFP-C1. A mutant form fused to EGFP comprising the third domain (domain 3a/b) predicted from the Munc18-1 structure (Misura et al., 2000) was obtained by amplifying amino acids 246-488 of Munc18-2 and cloned into EGFP-C1. All constructs were sequenced. Munc18-1, Munc18-2 and Munc18-3 cDNAs were also cloned into the SRαpuro vector (Roa et al., 1997) for expression in COS-7 cells.

Rat RBL and murine C57.1 mast cell lines as well as monkey COS-7 cells were maintained in DMEM-Glutamax supplemented with 10% fetal calf serum, 100 IU/ml penicillin G and 100 μg/ml streptomycin at 37°C in a humidified 5% CO₂ incubator (Pombo et al., 2001). The bone-marrow-derived mouse MCP5/L mast cell line (Dastych and Metcalfe, 1994) was maintained in RPMI-1640 medium supplemented with L-glutamine, penicillin, streptomycin, 10% FCS and 3 Units of IL-3 (Immugenex, Los Angeles, CA). Bone-marrow-derived mast cells (BMMCs) were obtained from femurs of BALB/c mice and cultured at 37°C in a humidified 5% CO₂ incubator at a starting density of 2×10⁵ cells/ml in complete RPMI-1640 containing 3U IL-3 (Martin et al., 2000).

About 5×10⁴ RBL cells were adhered on glass coverslips in 24-well plates for 1 hour and were incubated overnight in a 37°C humidified 5% CO₂ incubator. In stimulation experiments, cells were sensitised with mouse anti-DNP IgE (1/200) overnight. After washing, cells were stimulated with 100 ng/ml of DNP-HSA for the times indicated in the figure legends. In all experiments cells were fixed with 3% paraformaldehyde for 10 minutes. Cells were saturated and permeabilised in PBS containing 5% goat serum (Gibco-BRL), 0.3% BSA, 0.05% Saponin for 1 hour prior to incubation with primary and secondary antibodies in the same solution. Between each incubation cells were thoroughly washed with PBS containing 0.1% Triton X-100. BMMCs were treated as above except that 1×10⁵ cells were allowed to adhere to glass coverslips pre-coated with L-polylysine for 1 hour prior to staining. Coverslips were mounted in Mowiol containing the anti-fading agent DABCO (Sigma). Confocal laser scanning microscopy was carried out with a Zeiss confocal microscope interfaced with an argon/krypton laser. Simultaneous double-fluorescence acquisitions were made with 488 nm and 543 nm laser lines to excite FITC/GFP and Rhodamine fluorescence, respectively, using a 63× oil-immersion lens. The fluorescence was selected using appropriate double-fluorescence dichroic mirror and band-pass filters (BP525 and LP650). Mathematical analysis of captured images was accomplished by retrieving background of confocal acquisitions with Huygens software (Bitplane AG). After Point Spread Function calculation, the iterative likelihood estimation algorithm was applied to the raw acquisitions series.

RBL cells were transfected with 30 μg of GFP constructs (Paumet et al., 2000). After 48 hours cells were harvested, IgE-sensitised at 1×10⁶ cells/ml for 1 hour in complete medium/25 mM Hepes, pH 7.4 and stimulated with DNP-HSA (100 ng/ml) for 30 minutes. Exocytosis was quantitated by cytofluorometry with a FACSscan and Cell Quest software (Beckton-Dickinson) by measuring surface-exposed PS using biotinylated Annexin V and Streptavidin-Phycoerythrine (Southern Biotechnology) (Martin et al., 2000). Transfected cells were gated based on GFP-fluorescence and viability was checked in parallel with propidium iodide.

Total lysates of the various mast cells used were prepared by lysing cells directly in SDS-Sample buffer. Total brain lysates were obtained from Signal Transduction Laboratories. For immunoprecipitation experiments, RBL cells were either harvested by trypsinisation or left adherent. Cells were solubilised in lysis buffer (25 mM Pipes pH 7.3, 150 mM NaCl, 5 mM KCl, 5 mM MgCl₂, Triton X-100 1%, 1 mM sodium orthovanadate (Sigma), 1000 U/ml aprotinin (Sigma), 10 μg/ml pepstatin, 20 μg/ml leupeptin, 2 μM AEBSF (Alexis) at 5×10⁷ cells/ml (non-adherent cells) or by directly adding 1 ml of lysis buffer to 1-2×10⁷ adherent cells before harvesting them by scraping. Postnuclear supernatants were prepared by centrifugation at 15,000 g for 30 minutes and the protein of interest was immunoprecipitated for 2 hours by adding 2-5 μg of specific antibodies or normal rabbit IgG as a control prebound to protein A-sepharose. Isolated proteins were resolved by SDS-PAGE and transferred onto nitrocellulose membranes (Schleicher & Schuell, Dassel, Germany). The membranes were blocked with 5% nonfat milk for 1 hour and incubated with primary antibodies for 1 hour at room temperature. After several washes, blots were incubated with anti-rabbit IgG HRP (1/30,000) for 45 minutes and were developed by ECL (Amersham Pharmacia Biotech). NIH Image 1.61/fat software was used to quantify western blots experiments.

Changes in the actin microfilament arrays were elicited by cytochalasin D (Calbiochem), added from a 4 mM stock solution in ethanol. Changes in the MT array were elicited by nocodazole (Calbiochem), added from a 33 mM stock solution made in DMSO. Incubation times were as indicated in the figure legends. The release of mediators stored in SGs was monitored using the β-hexosaminidase release assay as described (Roa et al., 1997).

Immunoblot analysis of lysates from COS-7 cells transfected with rat Munc18-1, rat Munc18-2 or murine Munc18-3 established that antibodies generated to Munc18-2 and Munc18-3 reacted with their corresponding isoform. The antibody to Munc18-2 crossreacted with Munc18-1, but not Munc18-3, most likely owing to the significant amino acid identity of Munc18-1 and Munc18-2 (55%) within the immunising peptide. The anti-Munc18-1 and the anti-Munc18-3 reacted only with extracts from cells transfected with their corresponding cDNA (not shown). To assess the expression of Munc18 isoforms in mast cells, total lysates of RBL cells were immunoblotted with Munc18-specific antibodies and compared with rat brain. Fig. 1A demonstrates that rat basophilic leukemia cells (RBLs) do not express neuronal Munc18-1 (≈67 kDa), while species corresponding to Munc18-2 (≈65 kDa) and Munc18-3 (≈69 kDa) are detectable. As Munc18-2 is not expressed in brain (Tellam et al., 1995), the detection by anti-Munc18-2 of a species with slightly lower mobility in brain lysates can be explained by the crossreactivity of anti-Munc18-2 with Munc18-1. The expression of Munc18-2 and Munc18-3 was confirmed in two murine mast cell lines, C57.1 and MCP5/L, as well as in primary bone-marrow-derived mast cells (BMMCs) (Fig. 1B).

As Munc18 proteins interact with distinct syntaxins we studied their association in resting RBL cells. Syntaxin 2, 3 or 4 were immunoprecipitated from cell lysates resolved by SDS-PAGE and probed with antibodies to Munc18-2 or 18-3. Fig. 2A shows that Munc18-2 primarily interacts with syntaxin 3, less with syntaxin 2 and not with syntaxin 4, while Munc18-3 interacted exclusively with syntaxin 4. Conversely, as the anti-Munc18-2 allowed its immunoprecipitation, we confirmed the preferential association with syntaxin 3 (Fig. 2B). A quantitative estimation taking into account immunoprecipitation efficiencies of the syntaxin antibodies, suggested that about 10% and 20% of cellular Munc18-2 was associated with syntaxin 2, and 3, respectively, and about 25% of Munc18-3 was associated with syntaxin 4. We further examined whether FcϵRI stimulation could promote quantitative changes in these interactions. However, no differences were seen between unstimulated and stimulated RBL cells (data not shown).

Membrane targeting of Munc18 proteins is thought to result from their interaction with syntaxins, although other components may also participate (Biederer and Südhof, 2000). We studied the subcellular localisation of Munc18-2 and 18-3 together with the most abundant partners, syntaxin 3 and 4 by confocal microscopy. Fig. 3A illustrates the observed pattern in comparison with the PM stained with an antibody directed to the intracellular N-terminal tail of the β chain of the cell-surface IgE receptor (FcϵRI). Fields of a single cell as well as the merge with several cells are shown. The anti-β mAb clearly delineates the PM, although a diffuse cytoplasmic staining also becomes apparent most likely owing to neosynthesis (Quarto et al., 1985) and presence of the receptor β chain in endosomal compartments (Xu et al., 1998). Munc18-2 does not codistribute with the PM marker but is found in close apposition appearing as granular structures that often seem to extend into lamellipodia-like protrusions beyond the main cell body, which is delineated by the anti-β PM marker (compare with the merge of the DIC image and the PM marker in the inset). Its distribution is not uniform and is restricted to specific zones at the medium-exposed cell surface as revealed by examination of sections running from the ventral to the dorsal surface (not shown). Syntaxin 4 and Munc18-3 show punctuate foci at the cell periphery and overlap to a significant extent with the PM. Likewise, syntaxin 3 reveals PM staining (Fig. 3A) but internal vesicular staining is also detected (Fig. 3B). Owing to its granule-like appearance, we looked at whether Munc18-2 co-localised with mediator-containing SGs as determined with an antibody to rat mast cell protease II (RMCP II), a granule-localised chymase (Schwartz, 1994). Fig. 3B reveals a substantial co-localisation between the Munc18-2-enriched zones and RMCP II in RBL cells, an observation more apparent when looking at several different confocal planes. In BMMCs, Munc18-2 staining also coincided largely with SGs, which were defined in BMMCs with an antibody to serotonin. Syntaxin 3 was present at the PM; however, its distribution also partially overlapped with SGs in RBL cells (Fig. 3B). This overlap was even more evident in BMMCs, where almost all syntaxin 3 staining appeared on SGs (Fig. 3B). No SG staining was observed for syntaxin 4 and Munc18-3 (not shown).

In pancreatic acinar cells PM-localised Munc18-3 redistributes into the cytosol upon stimulation (Gaisano et al., 2001). As Munc18-3 also localises to the PM in mast cells we looked at whether FcϵRI-triggering would similarly promote its redistribution to the cytosol. We therefore looked at the distribution of Munc18-3 and FcϵRI β in stimulated cells using confocal imaging. However, no significant redistribution to the cytosol became apparent and Munc18-3 stays at the PM while the FcϵRI β chain is increasingly endocytosed indicating that cells had effectively been stimulated (data not shown).

Initial experimental data established that Munc18-2 remains associated with the granular compartment after activation through FcϵRI. This indicated that secretory structures that can be quite large persist (Fig. 4A) in stimulated cells similar to what has been seen in scanning force microscopy images (Spudich and Braunstein, 1995). At the same time intragranular RMCP II staining is lost (not shown). Measurement of the secretory structures revealed variable sizes ranging from a few hundred nm to μm (0.2 to 2.1 μm for SGs shown in Fig. 4A), which suggested intragranular fusion events (compound exocytosis). Indeed, mathematical analysis of images using Huygens software revealed that a secretory structure of 2 μm is probably comprised of about 6-7 fused granules (Fig. 4A, inset). The persistence of Munc18-2 on SGs allowed us to examine more closely the relationship between the secretory compartment and the cytoskeleton known to undergo dramatic reorganisation changes during regulated secretion. IgE-sensitised RBL cells were challenged with antigen for 10 minutes to achieve maximal degranulation, and Munc18-2 staining was compared with F-actin and microtubules (MTs) by using phalloidin-FITC and anti-α-tubulin, respectively. As depicted in Fig. 4B (top panel) F-actin, which is essentially subcortical in resting cells (Fig. 4B inset), undergoes an extensive rearrangement characterised by the appearance of membrane ruffles found in forming lamellipodia as cells flatten and spread out on the surface (compare cell size in Fig. 3 and Fig. 4B). A large part of persisting Munc18-2-containing SGs are now redistributed into these lamellipodia. Interestingly, these Munc18-2-containing SGs were excluded from actin ruffles as shown by reconstitution of slices along the z-axis (XZ), which revealed SGs filling the space in-between these ruffles. Stimulation also led to the formation of new MT tracks that extended into the lamellipodia (Fig. 4B, bottom panel). Although no clear co-localisation was apparent, Munc18-2 secretory structures seemed to be aligned along these newly formed MT tracks. This is also corroborated by our XZ analysis that revealed no particular exclusion between MTs and the Munc18-2-stained granule compartment in contrast to the data with F-actin ruffles.

The cytoskeleton plays a central role in determining cell polarity (Drubin and Nelson, 1996). We investigated whether the restricted localisation of Munc18-2- and RMCP II-positive SGs in resting cells was dependent on cytoskeletal elements. Fig. 5 shows co-staining experiments of Munc18-2 with MTs and F-actin in resting cells. As shown before, Munc18-2-positive SGs are polarised. There was no discernible relationship in the localisation of Munc18-2 with subcortical F-actin (Fig. 5A, left panel). Treatment with the actin-depolymerising fungal metabolite cytochalasin D did not affect the polarised staining pattern (Fig. 5A, right panel). When the relationship to the MT network was examined, SGs were found opposite the microtubule organising center (MTOC) at the plus end of MT tracks indicating that their localisation may depend on MTs (Fig. 5B). In agreement with a role for MTs, treatment of RBL cells with the MT-depolymerising drug nocodazole induced the redistribution of Munc18-2 to a diffuse staining pattern (Fig. 5B). Granular staining with RMCP II was still intact indicating that nocodazole-treatment does not affect granular integrity. Yet they seem to be more dispersed throughout the cell body (Fig. 5B).

To further explore the role of cytoskeletal elements in degranulation the effect of cytoskeletal-depolymerising drugs was studied in stimulated cells. In agreement with previous observations (Frigeri and Apgar, 1999), treatment with cytochalasin D led to enhanced degranulation in FcϵRI-stimulated RBL cells (Fig. 6A), although ruffling was abolished (not shown). Yet, as seen in Fig. 6B, Munc18-2- containing SGs still moved to areas outside the main cell body. In contrast, cells treated with EGTA, which is non-permissive for degranulation (Martin et al., 2000), did not show a redistribution of Munc18-2-positive SGs and they were found inside the cell body although ruffling was evident (Fig. 6B). Treatment of cells with nocodazole showed inhibition of secretion at concentrations as low as 0.25 μM (Fig. 6A). After stimulation, nocozadole-treated cells still adhered and spread out on the substratum surface (Fig. 6B). However, as in resting cells, Munc18-2 staining is characterised by a diffuse pattern inside the cell body with no evident relocation into the lamellipodia.

To study the function of Munc18-2 and Munc18-3 isoforms in mast cells we tested the effect of overexpressing wild-type proteins on FcϵRI-stimulated degranulation. N-terminal GFP-tagged Munc18-2, Munc18-3 or a GFP control were transfected, and gated GFP-positive RBL cells were analysed for externalisation of phosphatidylserine (PS) to assess FcϵRI-triggered exocytosis. As shown in Table 1, only the increased expression of Munc18-2 inhibited exocytosis. Transfection of Munc18-3 had no significant effect. These findings implicate Munc18-2 in FcϵRI-stimulated exocytosis. Crystal structure analysis of Munc18-1 has revealed the existence of a potential effector loop containing a residue homologous to the Sly1-20 mutant in the yeast Sly1 protein, which uncouples the sec1/Munc18 family member from Rab effector function (Dascher et al., 1991). A peptide containing part of this loop has been demonstrated to interfere with the Munc18-3-dependent secretion of GLUT4 vesicles after insulin stimulation (Thurmond et al., 2000). Thus, as an alternative approach, we tested whether a homologous peptide loop in Munc18-2 or the predicted domain 3a/b of Munc18-2 (also containing the loop) could interfere with exocytosis. Analysis of gated, transfected cells significantly inhibited FcϵRI-triggered degranulation, as measured by PS externalisation, for both Munc18-2-derived peptides. These findings suggest that Munc18-2-mediated inhibition of degranulation in mast cells, following overexpression, is probably due to a competing and thus sequestering interaction with a protein that binds to the Munc18-2 effector domain loop.

Mast cell degranulation depends on a complex biochemical chain of intracellular events. Little is known about the molecules that relay early FcϵRI-activated signals to the SNARE fusion machinery. Similarly, the connection of this machinery to the extensive cytoskeletal changes that occur during secretion are still poorly understood. Nevertheless, it has been established by various laboratories that the cytoskeleton is intimately linked to regulated exocytosis (Pendleton and Koffer, 2001; Pfeiffer et al., 1985; Valentijn et al., 1999). Here, we identify the syntaxin-binding proteins Munc18-2 and Munc18-3 in mast cells. We show that Munc18-2 participates in FcϵRI-triggered degranulation and reveal the importance of the MT network to Munc18-2 localisation and redistribution as well as to degranulation.

In searching for regulators of SNARE complex formation we demonstrated that mast cells express Munc18-2 and Munc18-3 that bind, respectively, to syntaxin 2 or 3 and syntaxin 4. Their subcellular localisation revealed a strikingly different pattern. Both Munc18-3 and syntaxin 4 located to the PM. Although a polarised staining at the basolateral PM has been seen for Munc18-3 in pancreatic acinar cells (Gaisano et al., 2001) no evidence for this type of distribution was found in RBL cells, similar to the situation in adipocytes (Khan et al., 2001; Thurmond et al., 1998). Interestingly, Munc18-2, which has been demonstrated to localise to the apical PM in intestinal epithelial cells (Riento et al., 1998), localised to SGs in mast cells. Syntaxin 3 was found both on the PM and SGs, implying that the latter is not the sole determinant of Munc18-2 location. Yet, like in epithelial cells, we noted a polarisation of Munc18-2 to restricted zones at the dorsal surface. This raises the possibility that specific sites favourable for membrane fusion exist in mast cells similar to other cells (Zenisek et al., 2000) and that Munc18-2 could participate in the targeting of SGs to these fusion-competent sites. This recruitment could depend on the interaction with other effectors that form a secretion-inducing molecular scaffold (Biederer and Südhof, 2000; Butz et al., 1998). In favour of this, preliminary data using sucrose gradient fractionation showed that a sizeable fraction of Munc18-2, but not Munc18-3, distributes to specialised lipid rafts (I.P., unpublished) that could provide a framework for such interactions in addition to the MT network (see below).

In resting cells actin microfilaments are found at the cell periphery forming an actin-myosin cortex thought to represent a barrier for secretion (Aunis and Bader, 1988; Koffer et al., 1990; Oheim and Stuhmer, 2000). Consistent with this view, fusion of SGs with the PM after stimulation is accompanied by disassembly of the cortex and formation of F-actin ruffles that reach out into lamellipodia (Koffer et al., 1990; Oliver et al., 1992; Pfeiffer et al., 1985). Yet, treatment with a variety of actin-specific drugs, has variable effects on degranulation ranging from enhancement to inhibition, which suggests a complex relationship and that all observed F-actin changes are not absolutely required for degranulation (Frigeri and Apgar, 1999; Pendleton and Koffer, 2001). Other studies point to a role for MTs in the mobilisation of SGs of various cell types (Burkhardt et al., 1993; da Costa et al., 1998; Olson et al., 2001; Radoja et al., 2001). In RBL cells, activation-induced rearrangement of MTs has been observed (Oliver et al., 1992) and secretion is inhibited with nocodazole in rat peritoneal mast cells (Nielsen and Johansen, 1986). New evidence also indicates the combined action of actin-myosin and microtubule networks in vesicle movement (Cordonnier et al., 2001). As the connection of SG localisation and mobilisation with the cytoskeleton has not been investigated in mast cells, we exploited our finding of the persistence of Munc18-2 on empty secretory structures after stimulation to investigate this relationship more closely.

In resting RBL cells Munc18-2-stained SGs appeared adjacent to the actin cortex at focal restricted zones. Treatment with cytochalasin D, which provoked cell rounding and decrease of F-actin staining, did not affect the polarised appearance of Munc18-2. The results were quite different when we looked at MTs. The latter appeared to originate from the MTOC near the nucleus and to radiate towards the PM. Although some MTs extended around the cell periphery, most were enriched at one pole. Nocodazole treatment led to the disappearance of these tracks concomitantly with the dispersal of RMCP II-stained SGs and a change from the polarised distribution of Munc18-2 to a diffuse distribution, suggesting that SG location and their association with Munc18-2 is linked to MTs.

Stimulation of RBL cells led to a pronounced remodelling of F-actin with the formation of ruffles and actin plaques, which can be inhibited by cytochalasin D. As observed before, treatment with this drug did not inhibit secretion but rather led to its enhancement. Although this has been attributed to the actin-dependent downmodulation of cell signalling (Frigeri and Apgar, 1999) a facilitating role of the removal of the actin barrier cannot be completely excluded. Nevertheless, our data further suggest that formation of actin ruffles is not directly required for SG mobilisation as they are particularly excluded from these structures. This clearly differs from what has been observed for the insulin-stimulated and cytochalasin D-sensitive exocytosis of GLUT4 vesicles, which become highly concentrated in these ruffles (Kanzaki and Pessin, 2001; Tong et al., 2001). Conversely, dramatic effects were seen when the relation to MTs was examined. Indeed, IgE-dependent stimulation also leads to the formation of new tubular tracks, and co-staining experiments revealed that Munc18-2 granular structures were aligned along these tracks. Nocodazole-treatment inhibits secretion. Together with the observed dispersal of Munc18-2 staining and effect on granule polarisation in resting cells our results indicate that Munc18-2 localisation as well as the secretory process depends on an intact MT system.

Given the importance of Munc18 proteins to exocytosis in other cells, it is not surprising that they may be essential to mast cell degranulation. Consistent with this view, overexpression of Munc18-2 inhibited the FcϵRI-stimulated degranulation response. The inhibition with peptides containing a homologous effector loop structure further demonstrates the necessity for the interaction of Munc18-2 with particular, albeit unknown, effectors to promote membrane fusion similar to the role of Munc18-3 in GLUT4 vesicle exocytosis (Thurmond et al., 2000). The absence of inhibition by Munc18-3 in mast cells was somewhat surprising given that it binds to syntaxin 4, which is known to play a role in mast cell exocytosis (Paumet et al., 2000). However, we know from previous studies that overexpression of Munc18 isoforms does not always result in inhibition of exocytosis and could even be stimulatory in some cases (Dresbach et al., 1998; Graham et al., 1997; Schulze et al., 1994; Voets et al., 2001). Further studies are necessary to establish the role of Munc18-3.

In conclusion, our findings demonstrate the expression, compartmentalisation and redistribution during FcϵRI-dependent stimulation of Munc18-2 in mast cells. Our findings also implicate Munc18-2 in the FcϵRI-stimulated degranulation response. The polarised and MT-dependent localisation of Munc18-2 to SGs in RBL cells further supports the notion that these proteins participate in a network of scaffolding proteins that are necessary for exocytosis. The demonstration of the connection to the MT network points to similarities with other secretory cells of hematopoietic origin such as cytotoxic T cells (Burkhardt et al., 1993; Radoja et al., 2001). However, in contrast to the latter, which have been demonstrated to depend on coupling of actin-myosin motors via Rab27A for exocytosis (Haddad et al., 2001), mast cells do not show such a coupling (R. Siraganian, personal communication). While further studies are required to obtain a precise picture of the molecular mechanisms in the final steps of mast cell degranulation, we identify both Munc18-2 and the MT cytoskeleton as important components in this process.

We thank T. C. Südhof and J. E. Pessin for cDNA constructs and C. Guérin-Marchand and J. LeMao for helpful discussion. We are grateful to P. Roux for excellent confocal microscopy assistance, and to R. Peronet for help in preparing polyclonal antibodies and BMMCs. The confocal microscope was purchased with a donation from Marcel and Liliane Pollack. This work was supported by Institut Pasteur, CNRS and a grant from the Association de la Recherche sur le Cancer (ARC No 4389). S.M.-V. was supported by a fellowship from the Ministère de la Recherche in France and by ARC. I.P. was a recipient of a PRAXIS XXI Fellowship (Ministério da Ciência e Tecnologia, Portugal).