Munc18-2 and Syntaxin 3 Control Distinct Essential Steps in Mast Cell Degranulation

Mast cells (MC) are key effectors in immunity, but are also notoriously known to cause allergies (1). Activation through IgE receptors (FcεRI) triggers degranulation with release of inflammatory mediators stored in abundant cytoplasmic secretory granules (SGs). This is followed by de novo synthesis and secretion of cytokines/chemokines (2). In contrast to neuronal/neuroendocrine cell exocytosis, in which individual SGs fuse at the plasma membrane (PM), MC degranulation involves granule–granule and SG-PM fusion by compound/multigranular exocytosis (3). Despite these differences, MC and other regulated-secretion competent cells share a common molecular machinery of membrane fusion (2, 4, 5). It includes SNARE fusion (6, 7) and fusion regulatory proteins such as Munc18 (8–10), Munc13 (11, 12), Rab (13–15), complexin (16), SCAMP (17), and Synaptotagmin-family members (18, 19). SNAREs promote fusion by forming a tetrameric complex through their SNARE motif (20). A functional complex in MC comprises the vesicular SNARE VAMP8 and the target SNAREs syntaxin (STX)4 and SNAP-23 (6, 7, 21, 22). In agreement with the heterogeneity of the SG compartment and the possibility of granule–granule fusion, other SNAREs may also play a role (9, 23–25). During degranulation, SNAP-23 redistributes from the PM into nascent degranulation channels, which may be a particularity of compound exocytosis (6).

Mammalian homologs of Caenorhabditis elegans uncoordinated-18 (Munc18) proteins are essential regulators of exocytosis, as evidenced after genetic deletion of the neuronal isoform Munc18-1 (20) and the more ubiquitously expressed isoforms Munc18-2 and Munc18-3 (10, 26–28). Their action involves different SNARE-binding modes (20, 29, 30). Whereas binding to the closed conformation of STX was thought to block SNARE assembly (31), new evidences rather support that this complex serves as an organizer of productive membrane fusion, allowing privileged interactions with other fusion effectors and the promotion of a conformational switch to an open form (20) able to bind tetrameric SNARE complexes at its C terminus, thereby activating fusion (30, 32, 33). Although this defines a minimal fusion entity (32), additional evidence indicates that in living cells Munc18 proteins participate in steps promoting vesicle translocation, tethering, and docking that are required to support fusion (29).

MC express the ubiquitous isoforms Munc18-2 and Munc18-3 (8). Multiple studies support a role for Munc18-2 in degranulation (8–10, 34, 35). Furthermore, familial hemophagocytic lymphohistiocytosis type 5 (FHL-5) patients carrying Munc18-2 mutations causing misfolding show impaired lytic granule exocytosis in NK and cytotoxic T cells (36, 37). In contrast to other immune effector cells (34, 38, 39), a role for PM-localized Munc18-3 in MC has not been demonstrated, and the implication of the neuronal-specific Munc18-1 is controversial (35, 40).

The cytoskeleton also plays an important role in vesicular trafficking. MC activation is accompanied by cortical actin disassembly and membrane ruffling that favors access of SGs to the PM (41–43). Similarly, the actin regulatory proteins coronin 1a/b have recently been shown to differentially effect on degranulation and cytokine production (44). FcεRI stimulation also triggers microtubule formation through a Fyn/Gab2/RhoA signaling pathway necessary for SG translocation (43, 45).

In this study, we investigated the role of Munc18-2 and STX3 in degranulation. We show that they have essential but complementary roles and define a new stimulatory Munc18-2 microtubule-dependent axis enabling SG translocation and fusion.

Rabbit Abs to STX3, Munc18-2, rabbit isotype control anti-GST, mouse isotype control IgG (15.1), and mouse monoclonal anti–DNP-IgE have been described (8, 22, 46). Anti–α-tubulin, DNP human serum albumin (DNP_30–40-HSA, Ag), ionomycin, and PMA were purchased from Sigma-Aldrich. Alexa Fluor 555 wheat germ agglutinin (WGA) was from Invitrogen. Murine IL-3 and stem cell factor (SCF) were from PeproTech (Paris, France). Nocodazole and Paclitaxel were purchased from Merck-Calbiochem (Nottingham, U.K.). The Myc–Munc18-2 plasmid was a gift from V.M. Olkkonen (BIOMEDICUM Helsinki 2U).

The rat RBL-2H3 MC line and COS7 cells were maintained, as described (8, 22, 46). Bone marrow was isolated from femurs of C57BL/6 mice (8–12 wk old), and bone marrow–derived MC (BMMC) were derived, as described (22). Cell cultures were used after 4–8 wk after routine checking for purity by evaluating expression of c-kit and FcεRI (6, 22). Wistar male or female rats (200–400 g) were purchased from the University of Trieste. Rats were euthanized by CO₂ inhalation. Rat peritoneal MC (RPMC) were obtained and stimulated using compound 48/80, as described (22). All animal studies have been approved by the local institutional animal care and use committees.

The following small interfering RNAs (siRNAs) (sense) were used (Eurogentec): Munc18-2-siRNA1 (5′-GGCUCAUCGUGUACAUUG-3′); Munc18-2-siRNA2 (5′-CGCUCACAGUUGCUCAUAA-3′); STX3-siRNA1 (5′-GGCUCAACAUCGACAAGAU-3′), sSTX3-siRNA2 (5′-GGAGCUCCAUGACAUGUUU-3′); and universal scramble siRNA (5′-GCCCCGUCUACAUACAUGU-3′). RBL cells were transfected with annealed siRNAs (0.05 μM each) in two successive electroporations 24 h apart, adjusting cell density each time to 1 × 10⁷/ml in electropermeabilization buffer (120 mmol/L KCl, 10 mmol/L NaCl, 1 mmol/L KH₂PO₄, 10 mmol/L glucose, 20 mmol/L HEPES [pH 7.0]) at 250 V and 2100 μF. Cells were used for functional analysis the next day. For gene silencing of BMMC, short hairpin RNA (shRNA) targeting of Munc18-2 or STX3 was performed using lentiviral vectors. Viruses were produced by transfecting vectors with Munc18-2 or STX3 shRNAs into the packaging cells 293LTV using Lipofectamine 2000 (Invitrogen), as described (47). For infection, 10-d-old bone marrow cells were resuspended in the concentrated viral supernatants for 2 d at 37°C with 5% CO₂. Postinfection, cells were grown in IL-3– and SCF-containing medium for 2 d before initiating selection of transduced cells with 8 μg/ml blasticidinS (Invitrogen) for 2 wk. For transient expression of CD63-GFP, Nucleofection (Amaxa) was performed, according to the manufacturer's instructions, using mouse macrophage nucleofection program. BMMC were cultured in IL-3– and SCF-containing medium for 2 d until experiments.

Release of granule content was determined by measuring β-hexosaminidase release (7). CCL2 release was quantified using a Duoset cytokine ELISA kit (R&D Systems). Degranulation of BMMC was evaluated by a CD63 expression assay. Briefly, BMMC were deprived of SCF overnight, followed by sensitization with DNP-specific IgE in Tyrode’s buffer (7) for 3 h at 37°C. IgE-sensitized BMMC (2 × 10⁶ cells) or unsensitized BMMC were stimulated with DNP-HSA (Ag) or PMA/ionomycin (20 ng/ml 1 μM) for 30 min at 37°C. Cells were incubated with anti-CD63 Abs (2 μg/ml; MBL International) or isotype control (2 μg/ml) at 4°C, followed by incubation with Alexa Fluor 488–conjugated anti-rat IgG before analysis by flow cytometry using a FACSCalibur (BD Biosciences, San Jose, CA) and FlowJo software (Tree Star, Ashland, OR).

IgE-sensitized or unsensitized RBL cells (1 × 10⁷) were resuspended in medium and challenged with Ag (DNP-HSA, 100 ng/ml) or PMA/ionomycin (20 nM/1 μM) for the indicated times. Stimulation was arrested using ice-cold PBS. Lysates were prepared in 50 mM HEPES (pH 7.3) containing 10 mM CHAPS, 0.1% SDS, 150 mM NaCl, 5 mM KCl, 1 mM MgCl₂, 1 mM sodium orthovanadate, and protease inhibitors, as described (6, 22), and used for Munc18-2–tubulin coimmunoprecipitation studies. Munc18-2–STX3 interactions were explored using BMMC (48 h). After washing, IgE-sensitized cells were resuspended in medium and challenged with DNP-HSA (10 ng/ml). Stimulation was arrested using ice-cold PBS containing 0.25 mM N-ethylmaleimide for 15 min, followed by a wash in PBS. Lysates were prepared in 50 mM HEPES (pH 7.2) containing 1% Triton X-100, 150 mM NaCl, 5 mM KCl, 1 mM MgCl₂, 1 mM sodium orthovanadate, and protease inhibitors before immunoprecipitation. Immunoblotting was carried out as previously described (22, 47).

siRNA-transfected RBL cells were seeded in 24-well plates on coverslips for 16 h at 37°C in a humidified atmosphere with 5% CO₂. After PMA/ionomycin stimulation, cells were washed and fixed for 20 min on ice in 10 mM PIPES (pH 6.8), 150 mM NaCl, 5 mM EGTA, 5 mM MgCl₂, and 5 mM glucose containing 4% paraformaldehyde (immunofluorescence [IF] buffer), followed by two washes in IF buffer. Fixed cells were permeabilized in IF buffer containing 0.025% saponin for 20 min at room temperature, followed by blocking in IF buffer containing 0.012% saponin and 7% horse serum (Invitrogen) for 30 min at room temperature (RT). Staining with primary Abs was performed in IF buffer containing 0.012% saponin and 5% horse serum for 2 h at RT, followed by incubation with secondary Abs for 60 min at RT. Cells were mounted in Prolong-Gold antifading reagent (Molecular Probes) and were analyzed using confocal laser-scanning microscope LSM 510 (Zeiss, Oberkochen, Germany).

For real-time imaging of secretory granules, CD63-GFP–expressing granules were tracked. CD63-GFP–transduced BMMC were incubated with Alexa Fluor 555 WGA (5 μg/ml) and placed on poly-d-lysine–coated glass-bottom dishes for 10 min at 37°C. Cells were washed and subsequently stimulated with PMA/ionomycin (20 nM/1 μM). A Zeiss LSM-510 META confocal microscope with 488-nm and 543-nm wavelength light from an argon laser and HeNe was used for SG-PM colocalization measurements. The fluorescence of GFP and Alexa Fluor 555 was detected through a band pass filter (505–530 nm) and a long pass filter (>560 nm), respectively. Fluorescence images were collected every 2 s, and the acquired data were processed by Adobe Photoshop. Colocalization coefficient analysis was made using Medical Image Processing Analysis and Visualization (National Institutes of Health) software. Data of colocalization coefficient were normalized to the starting point (t = 0). Increase of value indicates SG-PM membrane fusion. Imaris software (Bitplane AG, Zurich, Switzerland) was used for tracking of CD63-GFP–expressed SG and for analyzing granule velocities by the spot-detection function using the following parameters: autoregressive motion, gapclose 2, 2 μm maximum distance. Average velocities of each granule were analyzed before (1–20 frames) and after (21–50 frames) stimulation.

Double immunogold labeling of RPMC ultrathin sections was as described before (22, 34). Briefly, RMPC were fixed in 1.5% glutaraldehyde (Serva) diluted in 0.1 M cacodylate buffer (pH 7.4) stored for 20 min at room temperature, postfixed in 1% OsO₄ for 60 min at 4°C, dehydrated in ethanol, and finally embedded in Dow Epoxy Resin (DER 332; Unione Chimica Europea, Milano, Italy). For double immunogold labeling, ultrathin sections, cut by an ultramicrotome (Ultracut UCT; Leica, Wien, Austria), were mounted on nickel grids etched for 1 min with 1% periodic acid and rinsed in distilled water three times (2 min each time). The grids with the section sides facing downward were incubated in 20 mM Tris-HCl (pH 8.2), containing 2% BSA, 1% goat serum, 0.05% Tween 20, and 0.1% Triton X-100 containing rabbit anti–Munc18-2 (50 μg/ml) at 4°C overnight. After washing in 20 mM Tris-HCl (pH 8.2) containing 225 mM NaCl, 20 mM NaN3, 0.05% Tween 20, 0.5% BSA, 0.1% Triton X-100, and goat serum 0.5% grids were incubated for 1 h at room temperature with 10 nm gold-conjugated goat anti-rabbit (British Biocell International, Cardiff, U.K.) diluted 1:50 in Tris-HCl-BSA-Triton. After rinsing thoroughly, grids were turned over with the section sides facing upward, according to Bendayan (48), and exposed to rabbit anti-STX3 (50 μg/ml). The procedure followed thereafter was as described above, with the exception that 20 nm gold-conjugated protein A-G (British Biocell International) diluted 1:50 in Tris-HCl-BSA-Triton was used as revealing system. Sections were analyzed by a Philips EM208 transmission electron microscope mounted to a 4008 × 2672 pixel Morada camera (Olympus). Control experiments using irrelevant IgG performed in parallel showed no significant immunogold labeling.

Results were analyzed using GraphPad Prism (GraphPad Software). Statistical analysis of differences between groups was determined by unpaired, two-tailed Student’s t tests or by one-way ANOVA, as appropriate. Two-way ANOVA was used to examine the overall effects of phenotype and time on the changes in granule-plasma membrane colocalization. Differences were considered significant if the p value was ≤0.05 with the confidence intervals of 95%.

We explored the role of Munc18-2 in MC secretory responses by siRNA-mediated silencing in RBL-2H3 MC. Munc18-2 expression was downregulated in RBL MC using siRNA. Consistent with the extent of specific knockdown (Fig. 1A), IgE-induced release of the granule-stored enzyme β-hexosaminidase was diminished (Fig. 1B). RNA silencing of the Munc18-2–interacting partner and SG-localized SNARE protein STX3 (Fig. 1C) also inhibited IgE-induced degranulation with two independent siRNAs (Fig. 1D and data not shown). Interestingly, combined silencing of STX3 and Munc18-2 resulted in an additive inhibitory effect (Fig. 1D), demonstrating that they can contribute to degranulation independently of each other. Degranulation was also inhibited after stimulation with PMA/ionomycin (Fig. 1D), which bypasses early FcεRI-induced signaling, supporting the view that both effectors function in the late steps of stimulus-secretion coupling. This is also supported by data showing that more upstream responses such as calcium mobilization were not affected (data not shown). No effect of Munc18-2 silencing on STX3 and Munc18-3 expression was observed (Fig. 1C and data not shown). This excludes a chaperon function of Munc18-2 for STX3, as observed for other Munc18-binding proteins (36, 49). We also analyzed the effect on de novo CCL2 chemokine secretion. Neither silencing of Munc18-2 or of STX3, nor a combined knockdown of both, affected IgE-induced CCL2 release when compared with cells transfected with scrambled siRNA (Fig. 1E). As STX3 has not been previously implicated in MC degranulation (7), we evaluated the respective contribution of STX3 and STX4 in the degranulation response. Surprisingly, siRNA experiments (Supplemental Fig. 1) showed that the effect of STX3 silencing was more pronounced than that of STX4. No additive effect was observed when both STXs were downregulated.

To explore the mechanism of action of Munc18-2, in MC we investigated the consequences of silencing its expression on the location of SGs in resting and stimulated RBL MC. SGs were labeled with an Ab to STX3 (a SG resident protein), and their distribution was compared with microtubules, based on findings that SGs are mobilized along microtubules (8, 43, 45). Fig. 2 shows that in resting, siRNA controls, SGs (in green) are randomly distributed along microtubules (in red) and often concentrate around the microtubule-organizing center or in proximity to the PM (Fig. 2, upper left panel). This distribution was not altered in Munc18-2–silenced cells (Fig. 2, lower left panel). PMA/ionomycin stimulation of control siRNA-treated cells induced the formation of microtubular tracks reaching out to the plasma membrane into filopodia-like extensions (Fig. 2, upper right panels). At the same time, SGs aligned along these tracks and substantial STX3 immunolabeling was observed at the periphery, indicating that SGs had translocated and fused with the PM. Similar data were also obtained when cells were stimulated with IgE/Ag, albeit STX membrane translocation was somewhat less prominent as compared with PMA/ionomycin-stimulated cells (Supplemental Fig. 2). Munc18-2 silencing did not affect microtubule formation (Fig. 2, lower right panel); however, SG movement to the periphery was markedly altered with most SG remaining in the cytosol, although they remain attached to microtubules. This indicated that the inhibitory effect on degranulation in the absence of Munc18-2 was due to the inhibition of SG translocation and subsequent incorporation into the PM.

To further confirm and extend these findings, we used dynamic image analysis of primary bone marrow–derived MC (BMMC) after silencing the expression of Munc18-2 or its binding partner STX3. Cells were transduced with lentiviruses encoding shRNAs corresponding to Munc18-2 (shM18-2), STX3 (shSTX3), or scrambled controls (shScr). Fig. 3A shows that partial, but specific silencing of Munc18-2 or STX3 was achieved when compared with wild-type (WT) or scrambled control shRNA. Like in RBL cells, silencing of STX3 had no effect on Munc18-2 expression or vice versa. Viral transduction also did not affect expression levels of FcεRI or the general phenotype or growth properties of these cells (Fig. 3B and data not shown). As observed before in RBL cells, silencing of Munc18-2 or STX3 markedly diminished cell surface expression of CD63 (a surrogate marker of MC degranulation at the PM) after IgE/Ag or PMA/ionomycin-induced degranulation when compared with WT or control shRNA-transduced cells (Fig. 3C–E). Release of granule-stored β-hexosaminidase was similarly inhibited (Supplemental Fig. 3). To follow SG movement and SG fusion events with the PM in real time, we additionally transfected cells with CD63-GFP, which incorporates into MC SGs (43). The PM was labeled by loading cells with Alexa Fluor 555–conjugated WGA, and cells were then stimulated using PMA/ionomycin. The dynamics of SG movement and their tracking using real-time CD63-GFP imaging analysis are shown in Supplemental Movies 1A–D. Representative images of the monitored fusion events during time (yt and xt) as compared with the xy image at the beginning and end of the stimulation period are depicted in Fig. 4A. In WT (Fig. 4A [WT], Supplemental Video 1A) or control shRNA-treated BMMC (Fig. 4A [shScr], Supplemental Video 1B), SGs became highly mobile and moved toward the PM (open arrowhead in Fig. 4A [WT, shScr]). Velocities of these SGs were increased after stimulation (Fig. 4B [WT, shScr]). Furthermore, SGs and PM fusion were observed in WT or control shRNA-treated BMMC (arrows in Fig. 4A [WT, shScr]). Following specific silencing of either Munc18-2 or STX3, SGs showed distinct features and differential behavior when compared with control-treated or WT BMMC. When Munc18-2 was silenced (Fig. 4A [shM18-2], Supplemental Video 1C), SGs lost their direction and moved randomly in the cytosol (open arrowhead). In addition, SGs in Munc18-2–silenced BMMC showed strongly reduced velocities compared with WT or shScr controls. Conversely, silencing of STX3 (Fig. 4A [shSTX3], Supplemental Video 1D) showed that SGs remained highly mobile, yet upon reaching the PM they rarely fused (indicated by the arrowhead in Fig. 4A [shSTX3]). To quantify the occurring fusion events, we measured colocalization of SG and PM markers as a function of time by correlation coefficient analysis (Fig. 4C). An increase in value indicates improved fusion of CD63-GFP–positive granules with the PM. The number of fusion events in WT and control shRNA-transduced cells rapidly rose to a maximum before starting to decline at the end of the monitoring period. In Munc18-2–silenced cells, a small rise in fusion events was observed; however, this rise was delayed and was minor relative to control cells. Importantly, due to the decreased and random mobility of CD63-GFP–positive granules, fewer SGs reached the PM (Fig. 4A, 4B). In contrast, STX3 silencing showed that, although CD63-GFP–positive SGs were observed in close proximity to the PM, fusion events were diminished.

During analysis of SG dynamics, we noticed that in some instances SGs remained immobile in areas that appeared to be restricted or relatively sparse in microtubular networks, whereas they readily moved and fused in other areas (Supplemental Video 2). This argued for the need of cytoskeletal reorganization to coordinate SG mobility and fusion. In light of the deficiency in translocation, we explored whether Munc18-2 could be part of a complex that couples the SG fusion machinery with microtubules. As shown in Fig. 5A (upper panel), myc-tagged Munc18-2 transfected into COS7 cells was specifically coimmunoprecipitated with α-tubulin. Similarly, endogenous Munc18-2 was found to associate with α-tubulin in RBL MC (Fig. 5A [lower panel], 5B). This association was decreased in cells treated with the microtubule-dissolving drug nocodazole (Fig. 5B). We next examined whether the Munc18-2–binding partner STX3 and the noninteracting protein STX4 cognate SNARE fusion proteins also coimmunoprecipitated with tubulin. However, although we confirmed previous data that Munc18-2 interacts with STX3, but not with STX4 or SNAP23, neither of these Abs was able to coimmunoprecipitate tubulin supporting a Munc18-2–dependent but STX3-independent interaction (Fig. 5C). We further analyzed whether the Munc18-2–tubulin interaction was modulated upon stimulation. Our kinetic analysis showed a time-dependent decrease of the interaction in both IgE/Ag and PMA/ionomycin-stimulated cells (Fig. 5D).

We next used immunogold labeling and electron microscopy to visualize at the ultrastructural level the distribution of Munc18-2 (10 nm gold particles [GP]) and STX3 (20 nm GP) in RPMC. The latter contain a highly differentiated granular compartment and upon stimulation release granular content by a process involving both SG-SG fusion and SG-PM fusion. Representative images of the GP distribution in resting and stimulated cells are depicted in Fig. 6. Quantitative analysis of GP distribution in resting cells shows that Munc18-2 is majorly (80%) associated with SGs (Table I) both on the surface (Fig. 6A, 6C, 6E, black arrows), but to a large part also inside the matrix (Fig. 6A, 6E, black arrowheads). The latter finding could relate to the fact that MC granules and those of other secretory cells have been found to contain exosomes (25, 50). A sizable proportion of Munc18-2 (20%) was also found in zones connecting SGs each to the other. These connections appeared to be constituted of filamentous tangles that appear attached to membranous extensions of the SGs (Fig. 6A, 6C, thin black arrows). PM localization, although occasionally seen, was rare. Similar to Munc18-2, the majority of STX3 was found on the SG surface (22.5%) (Fig. 6A, 6C, white arrows) and within SGs (68.1%) (Fig. 6A, 6E, white arrowheads). STX3 was also clustered at sites with filamentous connections (9.4%) often together with Munc18-2 (Fig. 6A, thin white arrows). Treatment with microtubule-depolymerizing or -stabilizing drugs induced the disappearance of Munc18-2 and STX3 labeling at the level of the filamentous connection in nocodazol- and taxol-treated RPMC, respectively (Supplemental Fig. 4, Table I). Quantitation of immunogold labeling also showed that the amount of the granule matrix GP in these cells did not change, although on the average an increase of granule surface labeling was noted (Supplemental Fig. 4, Table I). After stimulation, numerous fusion events with the expansion of the granular matrix can be observed (Fig. 6B). Interestingly, large clusters of STX3 and Munc18-2 can frequently be seen often coinciding with possible sites of fusion between granules (Fig. 6B [white arrowheads], 6D [white arrowheads at higher magnification]), but also between granules and the PM (Fig. 6F).

In this study, we report that the SNARE regulator Munc18-2 and its STX3-binding partner have distinct, but complementary roles in MC degranulation. Munc18-2 supports exocytosis by facilitating translocation of SGs, whereas STX3 plays a direct role in fusion. Nonetheless, STX3 function is apparently intimately linked to Munc18-2, as indicated by the observed coclustering at fusion sites in degranulating cells. Our findings also demonstrate that Munc18-2 is coupled to cell signaling, which modulates its interaction with the microtubule cytoskeleton.

Munc18 proteins are essential fusion effectors in regulated exocytosis; however, their precise function remains incompletely understood (20). In addition to their direct fusion-promoting role (20, 30), Munc18 proteins were proposed to regulate additional steps such as vesicle translocation, docking, postdocking, and modulation of cytoskeletal functions (29, 51, 52). In this line, Munc18-1–deficient chromaffin cells revealed a largely altered distribution of dense core vesicles, attributed to a docking defect (53). The defect was overcome by overexpressing STX1, which prevents formation of nonproductive SNARE (STX1-STX1-SNAP-25) complexes (53), suggesting that Munc18-1 could serve as an organizer of productive SNARE assembly. This view is also validated by reconstitution experiments (33).

Early studies on the ubiquitously expressed Munc18-2 showed that it controls exocytosis in various epithelial cells (54, 55), amylase release from pancreatic acinar cells (56), or insulin release from pancreatic β cells (57). Munc18-2 was also implicated in SG release by immune cells, including MC and neutrophils (8–10, 34). The discovery of Munc18-2 mutations in immune deficiency FHL-5 patients leading to misfolding was shown to affect STX11 expression and to lead to a defect in lytic granule exocytosis in NK and cytotoxic T cells (CTL) (36, 37). Although these results indicate that Munc18-2, similarly to Munc18-1, plays an important role in regulated secretion, the functions of these two SNARE accessory proteins may not be completely identical. Thus, in MC and other cell types (34, 57), the granular localization of Munc18-2 and STX3 is strikingly different. In epithelial cells, Munc18-2 was associated with polarized secretory events, as it was found to regulate, together with STX3, apical secretion in both intestinal and kidney epithelial cells (54, 55, 58). Interestingly, a recent study performed with pancreatic acinar cells suggested a preponderant role of Munc18-2 in sequential fusion (59). Other differences are the specific binding of calcium EF-hand protein Cab45 to Munc18-2 in the absence of calcium, whereas Munc18-1 binds only in its presence (60). It has been proposed that this interaction, which occurs in the absence of STX3, might regulate the association of Munc18-2 with STX partners during exocytosis (57).

Concerning MC, our data show that downregulation of either Munc18-2 or STX3 expression markedly inhibited MC degranulation, whereas CCL2 chemokine secretion was not affected, arguing that both effectors function only in release from storage organelles. These experiments implicate STX3, in addition to STX4 (7, 61), in regulating degranulation. STX11 has also been a potential candidate for regulating MC degranulation. However, although STX11 decreases in Munc18-2–deficient MC (35) in regulating degranulation, STX11 has also been a potential candidate for regulating MC degranulation. However, although STX11 decreases in Munc18-2–deficient MC, no effect on degranulation was seen in MC from STX11-deficient mice (35, 40, 62). Interestingly, STX3 downregulation had a more pronounced effect on degranulation than downregulation of STX4. Given the SG localization of the former, this finding supports the hypothesis that sequential granule–granule fusion is an important contributor in MC degranulation. Nonetheless, STX3 was also found in clusters at the membrane, suggesting that during this process it is able to relocate to this compartment in a manner similar to SNAP23, which in a reverse manner relocates from the plasma membrane to SG during degranulation (6). No additive effect on degranulation was observed, when both STX3 and STX4 were simultaneously downregulated, suggesting that they function independently. By contrast, inhibitory effects on degranulation were enhanced when Munc18-2 and STX3 expression were jointly reduced, suggesting unique, but complementary functions for each protein in this process.

Importantly, confocal imaging and live cell imaging studies of the degranulation process support that Munc18-2 controls granule translocation contrasting with STX3, which, as expected, directly impacts on fusion. As degranulation involves microtubule-dependent SG transport (43, 45, 63), this suggested that Munc18-2 might play a role in connecting the fusion apparatus with the microtubule cytoskeleton. In support, previous studies have shown that Munc18-2–labeled SG align along microtubules (8). Similarly, a sizable proportion of Munc18-2 can be found at cytoskeletal granular connections. Furthermore, neuronal Munc18-1 was shown to bind to the Kinesin-1 adaptor protein, fasciculation and elongation protein zeta-1 (FEZ1) (64), providing a link to axonal-dependent vesicular transport. We therefore asked whether Munc18-2 could interact with microtubules. Using coimmunoprecipitation experiments showed that Munc18-2 associated in a specific manner with α-tubulin. This interaction decreased in the presence of nocodazole. Importantly, the interaction was downmodulated after stimulation, suggesting a dynamic coupling during cell signaling. Likely, Munc18-2 is not directly responsible for this interaction, as SGs remain attached to microtubules after Munc18-2 silencing (Fig. 2). Likewise, attachment seems to take place independently of Munc18-2 binding to STX3, as tubulin was not found to coimmunoprecipitate with anti-STX3 and specific knockdown of STX3 did not affect the interaction (data not shown). However, to date, we have been unable to detect interactions of Munc18-2 with other microtubule-associated proteins like kinesin or the FEZ1 homolog FEZ2 (data not shown). Based on our results that SG translocation is clearly affected, it is possible that Munc18-2 could promote dynamic exchanges between the fusion and the cytoskeletal transport machinery, perhaps by enabling multiple cycles of docking. This could probably take place independent of its binding to STX3, as also suggested by the additive effect of downregulation of both STX3 and Munc18-2. These data are also in agreement with the described docking defects in Munc18-1 null chromaffin cells (53), but suggest that docking could be a dynamic process involved in granule translocation. In agreement, FHL-5 patients carrying mutations in Munc18-2 retain persistent membrane-trafficking defects in epithelial cells even after hematopoietic stem cell transplantation with inappropriate accumulation of granules in the cytoplasm similar to what is seen in MC (65).

The importance of microtubules in organizing the cellular distribution of the core fusion machinery is also supported by ultrastructural studies. Munc18-2 GP are found mainly in association with secretory granules, confirming our previous data (8), but also labeled granule-connecting filamentous structures. This distribution, in particular those present in the filamentous connections, is decreased in the presence of microtubule-targeting drugs, supporting the above hypothesis that it could depend on dynamic interactions of Munc18-2 with the microtubule network. This is in agreement with early transmission electron microscopy observations in RPMC, which revealed an important cytoskeletal meshwork around SGs and microtubules connecting the SG surface to the subplasmalemmal network or to the adjacent granule surface (66, 67). On some occasions, however, large clusters of both molecules accumulated at sites of granule–granule or granule–PM fusion, suggesting that at the fusion site productive interactions are maintained to support fusion. Taken together, these data support the important role of the dynamic interactions of the secretory machinery with the cytoskeleton, as already shown in the case of Rab3D, that could play a role in actin coating of secretory granules during exocytosis (68, 69) or Rab27 isoforms that could regulate the transition of microtubule to actin-based motility of granules (70).

Collectively, our findings establish Munc18-2 as an important regulator of MC-prestored mediator release that functions by enhancing SG translocation. Concerning STX3, our findings support a major role in SG-SG and SG-PM fusion. We also define a new Munc18-2–microtubule–dependent regulatory axis that coordinately organizes productive interactions between promoters of SG translocation and fusion with the cytoskeleton.

We thank Samira Benadda, IFR-02, for help with confocal imaging. We thank V.M. Olkkonen for providing the Myc–Munc18-2 plasmid construct.

The authors have no financial conflicts of interest.