Munc18-1 redistributes in nerve terminals in an activity- and PKC-dependent manner

Synaptic vesicle fusion is executed by Sec1/Munc18 (SM) proteins and the multisubunit SNARE protein complex of synaptobrevin-2/VAMP2, syntaxin-1, and SNAP25 (Jahn and Scheller, 2006; Jahn and Fasshauer, 2012; Rizo and Südhof, 2012). The SM protein Munc18-1 is a soluble protein initially found as an interacting partner of syntaxin-1 (Hata et al., 1993; Garcia et al., 1994; Pevsner et al., 1994). Deletion of munc18-1 completely arrests synaptic transmission (Verhage et al., 2000), whereas increased Munc18-1 levels result in a larger readily releasable vesicle pool (RRP) and increased synaptic efficacy (Toonen et al., 2006b). Mutations in munc18-1 are found in patients with epilepsy and intellectual disability, and Munc18-1 dysregulation is implicated in Alzheimer’s disease and schizophrenia (Jacobs et al., 2006; Saitsu et al., 2008; Hamdan et al., 2009; Milh et al., 2011; Vatta et al., 2012; Mastrangelo et al., 2013; Urigüen et al., 2013). Hence, Munc18-1 is required for normal brain function and its global expression levels scale with synaptic strength. However, the molecular mechanisms that control its synaptic levels are largely unknown.

Munc18-1 binds syntaxin-1 with high affinity, clamping syntaxin-1 in a conformation unable to bind other SNARE proteins (Dulubova et al., 1999; Misura et al., 2000). This interaction may stabilize both proteins and support their trafficking, as abrogated Munc18-1 expression results in aberrant targeting of syntaxin-1 in heterologous cells (Rowe et al., 2001; Medine et al., 2007; McEwen and Kaplan, 2008) and reduced expression levels of syntaxin-1 in neurons (Verhage et al., 2000; Voets et al., 2001; Zhou et al., 2013). In contrast, Munc18-1 levels are reduced in syntaxin-1 knockdown neurons (Zhou et al., 2013).

In developing neurons, syntaxin-1 is transported on vesicles via pathways dependent on the kinesin adaptor proteins syntabulin (Su et al., 2004) and FEZ1 (Chua et al., 2012). Munc18-1 is present in syntaxin–FEZ1–kinesin transport complexes (Chua et al., 2012). At this stage, the syntaxin-1/Munc18-1 dimer may function as a co-chaperone complex aiding transport and preventing premature breakdown of its constituents. However, in mature neurons syntaxin-1 is mainly transported via lateral diffusion along the plasma membrane (Mitchell and Ryan, 2004; Ribrault et al., 2011), and it is unknown whether at this stage Munc18-1 depends on syntaxin-1 for targeting to fusion sites.

To investigate the dynamics of endogenous Munc18-1 in neurons, we created mice expressing fluorescently tagged Munc18-1 from the endogenous munc18-1 locus. We characterized Munc18-1 dynamics using FRAP in cultured hippocampal neurons at rest, during stimulation, and upon application of several active compounds. Munc18-1 trafficked through axons and to synapses with membrane-bound syntaxin-1. During stimulation, synaptic Munc18-1 rapidly dispersed from synapses and reclustered within minutes. This was independent of syntaxin-1 and synaptic vesicle fusion but required calcium influx and PKC activity. Hence, the presence of Munc18-1 in synapses is tightly regulated; during periods of activity Munc18-1 becomes more dynamic and reclusters at synapses in a phosphorylation-dependent way. Finally, we show that increased Munc18-1 recruitment correlates with increased strength of individual synapses.

Munc18-1-Venus knock-in mice were generated with Venus cDNA, replacing the stop codon of exon 20 of munc18-1 (Fig. 1 A) via homologous recombination in embryonic stem cells (Fig. 1 B). Wild-type (WT; +/+), Munc18-1-Venus heterozygous (+/m), and homozygous (m/m) genotypes were detected by PCR (Fig. 1 C). Total brain protein levels of Munc18-1-Venus in m/m mice were indistinguishable from Munc18-1 in +/+ littermates (Fig. 1, D and F), although Munc18-1-Venus levels were slightly reduced in +/m littermates (Figs. 1 D and S1), as was found in Munc13-1-EYFP mice (Kalla et al., 2006). Munc18-1-Venus localization was comparable with Munc18-1 in WT brain with high expression in axonal fibers and mossy fiber terminals of the stratum lucidum in the hippocampus (Fig. 1 G).

The morphology and synapse number of hippocampal neurons from Munc18-1-Venus mice at 8 and 14 d in vitro (DIV) was similar to WT littermates (Fig. S1 and Fig. 2, A–D). Excitatory postsynaptic current (EPSC) amplitudes (Fig. 2 E), short-term plasticity (Fig. 2, H and K), spontaneous miniature EPSC frequency and amplitude (Fig. 2 H), and RRP size and recovery (Fig. 2, I–K) of Munc18-1-Venus autaptic hippocampal cultures were similar to WT. Hence, Munc18-1-Venus neurons can be used as functional reporters of Munc18-1 dynamics in living cells.

In cultured neurons, membrane-bound syntaxin-1 diffuses laterally through the axon (Ribrault et al., 2011). Munc18-1 colocalizes with syntaxin-1 throughout the axon, and both proteins may be transported as a binary complex (Garcia et al., 1995; Ribrault et al., 2011). To test this, we imaged hippocampal neurons of Munc18-1-Venus mice at 14 DIV, when synaptic contacts are functional, and compared Munc18-1-Venus dynamics in axons with soluble GFP and membrane-bound syntaxin-1–enhanced YFP (Stx-YFP) using FRAP analysis. We bleached extended regions of single axons (Fig. 3, A–C) to discriminate fast recovery of freely diffusing GFP (Fig. 3 A) from relatively slow recovery typical for membrane-bound molecules like Stx-YFP (Fig. 3 B). Fluorescence recovery of Stx-YFP was gradual and bidirectional, but markedly different in speed and extent from soluble GFP (Fig. 3, A, B, and E). Bleaching of axonal Munc18-1-Venus resulted in similar bidirectional and gradual recovery (Fig. 3, C and D). Mean Munc18-1-Venus recovery was similar to Stx-YFP and significantly slower than GFP (Fig. 3 E). In developing neurons Munc18-1 is bound to transport vesicles (Chua et al., 2012). We observed Munc18-1-Venus puncta entering the bleached area in only 16% of Munc18-1-Venus axons (Fig. S2), with a minimal effect on total fluorescence recovery. Hence, the majority of Munc18-1 in axons of 14 DIV neurons, in contrast to developing neurons, does not bind transport vesicles. Munc18-1 also does not diffuse freely like GFP, but may move through axons by interacting with slowly diffusing proteins, like syntaxin-1.

To test this, we expressed botulinum neurotoxin C (BoNT/C) light chain in Munc18-1-Venus neurons 6–12 h before FRAP experiments. BoNT/C cleaves syntaxin-1 at the C terminus just above the transmembrane region (Blasi et al., 1993; Schiavo et al., 1995), detaching the remaining protein from the membrane and arresting synaptic vesicle fusion (de Wit et al., 2006). Fluorescence recovery of axonal Munc18-1-Venus in cells expressing BoNT/C was significantly faster than control cells (Fig. 3 F). BoNT/C expression does not affect electrical properties of neurons or neuronal morphology at the ultrastructural level (de Wit et al., 2006), making it unlikely that indirect effects of syntaxin-1 cleavage contributed to the observed Munc18-1 transport phenotype. Together, these results show that the majority of Munc18-1 does not diffuse freely but binds to syntaxin-1 for transport through the axon via lateral diffusion in the membrane.

Munc18-1 is expressed in synapses (Toonen and Verhage, 2007). Munc18-1-Venus fluorescence was high in presynaptic terminals identified by synapsin-mCherry (Fig. 4 A). The synapse-to-axon ratio of Munc18-1-Venus was significantly higher than a membrane-targeted Venus protein (mVenus, Fig. 4 B), which suggests that Munc18-1-Venus is retained at synapses. To test this, we used FRAP of Munc18-1-Venus, Stx-YFP, and GFP on single synapses (Fig. 4 C). A focused laser spot aimed at a single synapsin-mCherry punctum (Fig. 4 D, open arrowhead) bleached the synaptic region, and fluorescence recovery was followed over time (Fig. 4 D). The initial (at 10 and 30 s) recovery of synaptic Munc18-1-Venus was markedly slower than GFP and not significantly different from Stx-YFP, although the mobile fraction of Munc18-1-Venus was slightly smaller than Stx-YFP (Fig. 4 E). We then compared FRAP of Munc18-1-Venus in the presence or absence of BoNT/C. Mean fluorescence recovery of synaptic Munc18-1-Venus in BoNT/C-expressing cells was significantly larger than in control cells (Fig. 4 F) and resulted in an increased mobile fraction of Munc18-1-Venus (Fig. 4 G). Hence, deletion of syntaxin-1 increases the dynamics of synaptic Munc18-1-Venus, which indicates that synaptic retention of Munc18-1 is largely dependent on syntaxin-1. A small immobile Munc18-1-Venus pool remained upon BoNT/C treatment (Fig. 4 G), which suggests that this pool is retained by interactions with other synaptic proteins.

Next, we tested whether Munc18-1 dynamics are modulated by activity by stimulating Munc18-1-Venus neurons while bleaching synaptic fluorescence during stimulation. Initial Munc18-1-Venus fluorescence recovery (after 10 and 30 s) in synapses stimulated with 600 action potentials (AP) at 20 Hz was indistinguishable from control synapses (Fig. 5 A). However, the mobile fraction (measured at 160 s) was significantly larger in stimulated neurons (Fig. 5, A and B). We also probed Stx-YFP mobility using FRAP and found, in line with Ribrault et al. (2011), that initial recovery of Stx-YFP fluorescence was not affected by stimulation (Fig. 5 C). However, the mobile fraction (measured at 160 s) was increased (Fig. 5, C and D), similar to Munc18-1-Venus (Fig. 5, A and B). Thus, strong stimulation increases synaptic Munc18-1 and syntaxin-1 dynamics in hippocampal neurons.

Stimulation causes dispersion of several soluble synaptic proteins (Chi et al., 2001; Sankaranarayanan et al., 2003; Star et al., 2005; Denker et al., 2011), vesicle SNAREs (Li and Murthy, 2001; Fernández-Alfonso et al., 2006), and target SNAREs (Degtyar et al., 2013). To test whether Munc18-1 disperses from the synapse during stimulation, we measured intensity changes upon stimulation in Munc18-1-Venus neurons. Munc18-1-Venus was enriched in synapses identified by synapsin-mCherry before stimulation (Fig. 6 A). During stimulation, fluorescence of Munc18-1-Venus and synapsin-mCherry at synapses decreased (Fig. 6 A): Munc18-1-Venus fluorescence dispersed from the start of stimulation (Fig. 6 B), similar to synapsin-mCherry (Chi et al., 2001; Fig. 6 B, inset). The intensity changes represented Munc18-1-Venus movement and not changes in Venus fluorescence due to proton-induced quenching (Nagai et al., 2002), as the fluorescence decrease of an mVenus was ninefold (P < 0.001) lower than the decrease in Munc18-1-Venus fluorescence (Fig. 6 C). Stx-YFP also dispersed from synapses (Fig. 6 B), although significantly less than Munc18-1-Venus (Fig. 6 C). Dispersion of Munc18-1-Venus and synapsin-mCherry required calcium influx and was impaired by superfusion of P/Q-, L-, and N-type calcium channel blockers (Fig. S5). To test whether Munc18-1 dispersion required active vesicle release and syntaxin-1, we expressed BoNT/C. This did not affect Munc18-1-Venus dispersion (Fig. 6, D and E). Thus, stimulation-induced calcium influx triggers the dispersion of a fraction of synaptic Munc18-1-Venus into the axon. This dispersion is largely independent of syntaxin-1 and does not require synaptic vesicle release.

Synapsin-1a fully reclusters to synapses within 10 min after synaptic activity (Chi et al., 2001). To test whether Munc18-1-Venus also reclusters at synapses, we followed synaptic Munc18-1-Venus fluorescence after stimulation. We observed a large variation in synaptic Munc18-1-Venus intensity changes (Fig. 7, A and B) that did not correlate with initial Munc18-1-Venus levels (Fig. S4). However, the mean response of all synapses showed dispersion during the stimulus, leading to a net increase in Munc18-1-Venus levels at t = 160 s (Fig. 7 C). This reclustering was much faster than synapsin-mCherry and was not observed for mVenus (Fig. S3). We defined two sets of synapses, one with initial dispersion (ΔF/F0 < 0 at t = 10 s) followed by a net increase of fluorescence (ΔF/F0 > 0 at t = 160 s, called Δ+) and one with initial dispersion (ΔF/F0 < 0 at t = 10 s) followed by a net decrease of fluorescence (ΔF/F0 < 0 at t = 160 s, called Δ−). Nearly half of all synapses fell in the Δ+ subset and over one-third in the Δ− subset (Fig. 7 C). Unlike the increase observed for Munc18-1-Venus, Stx-YFP mean fluorescence remained below initial levels after dispersion with <25% Δ+ synapses (Fig. 7 D). Reclustering was also slower than Munc18-1-Venus (Fig. 7 D). This suggests that Munc18-1-Venus reclusters faster than syntaxin-1 (and synapsin-mCherry), and independently of syntaxin-1. Indeed, synaptic Munc18-1-Venus fluorescence increased faster upon BoNT/C expression (Fig. 7, E and F) and resulted in higher levels at t = 160 s than in control neurons. Also, the percentage of Δ+ synapses was higher upon BoNT/C (Fig. 7 E). Hence, reclustering of synaptic Munc18-1 does not require syntaxin-1, and is even enhanced in the absence of syntaxin-1. Munc18-1 reclustering was absent in neurons incubated with calcium channel blockers (Fig. 7, G and H; and Fig. S5, A–C). Hence, calcium influx as a result of synaptic activity increases Munc18-1 mobility and redistributes synaptic Munc18-1 levels such that on average synapses contain more Munc18-1 after a strong stimulus.

PKC-dependent phosphorylation of Munc18-1 is essential for DAG-induced potentiation of synaptic transmission (Wierda et al., 2007) and vesicle pool replenishment in chromaffin cells (Nili et al., 2006). To test if PKC activity modulates synaptic Munc18-1 dynamics, we applied the DAG analogue PMA to Munc18-1-Venus neurons. This did not significantly affect synaptic Munc18-1 levels (Fig. 8, A and B). PMA application during electrical stimulation did not further increase Munc18-1 recruitment after stimulation (Fig. 8, C and D) or its dynamics (Fig. S5). Hence, PMA application does not have an additive effect when PKC activity is triggered by calcium influx. However, application of the specific PKC inhibitor Ro 31-8220 (Davis et al., 1989) resulted in significantly less dispersion during stimulation and an almost complete block of Munc18-1-Venus reclustering after stimulation (Fig. 8, E and F). Ro 31-8220 application significantly reduced the number of synapses and mean recruitment in the Δ+ subset (Fig. S5). Hence, PKC activity is required for reclustering of Munc18-1 after strong stimulation. Finally, to test whether PKC-dependent phosphorylation of Munc18-1 itself is required for its reclustering, we expressed a fully functional but PKC-insensitive mutant (PKC_i) of Munc18-1-EGFP (Wierda et al., 2007) in munc18-1 null mutant neurons and compared its dynamics with WT Munc18-1-EGFP on the same null mutant background. In contrast to WT Munc18-1-EGFP, which behaved like Munc18-1-Venus, the PKC-insensitive mutant did not show reclustering after electrical stimulation (Fig. 8, G and H; and Fig. S5). This shows that activity-dependent PKC phosphorylation of Munc18-1 is essential to recruit Munc18-1 to synapses after neuronal stimulation.

Munc18-1 expression levels in neurons control synaptic strength by modulating the number of release-ready vesicles (Toonen et al., 2006b). To determine whether activity-dependent modulation of synaptic Munc18-1 levels affects presynaptic strength on a synapse level we probed vesicle release using FM4-64 uptake during 30 AP at 1 Hz, before and after high-frequency stimulation (Fig. 9 A). Individual synapses in naive neurons showed a large variation in Munc18-1-Venus levels and FM4-64 levels (Fig. 9 B). Interestingly, and in line with our previous results in Munc18-1–overexpressing neurons (Toonen et al., 2006b), synaptic Munc18-1-Venus levels showed a significant correlation with FM4-64 loading (Spearman’s correlation r_s = 0.27, P < 0.001; Fig. 9 C), but not with synapse size (Fig. S4). Next, we tested whether a change in synaptic Munc18-1-Venus levels after high-frequency stimulation correlated with concomitant modulation in synaptic strength. We compared the change in FM4-64 uptake before and after 600 AP high-frequency stimulation between synapses with a net positive reclustering of Munc18-1-Venus and synapses with a net negative reclustering. Synapses with positive reclustering had significantly more FM4-64 uptake compared with synapses having a net negative reclustering after high-frequency stimulation (20 Hz; Fig. 9 D). Synapses that were stimulated at low frequency (2 Hz; Fig. 9 D) did not show such a correlation. Together, these results show that synaptic Munc18-1-Venus levels scale with release efficiency and that changes in levels after high-frequency stimulation correlate with increased synaptic strength.

Here, we investigated the distribution and dynamics of endogenous Munc18-1 in axons and synapses using Munc18-1-Venus knockin mice. In line with its essential role in exocytosis, we found that Munc18-1 is highly expressed in synapses. The majority of Munc18-1 traffics through axons via lateral diffusion together with syntaxin-1. Exchange rates of Munc18-1 at individual synapses are high compared with active zone components Bassoon and Munc13-1 (Kalla et al., 2006; Tsuriel et al., 2009). During periods of synaptic activity, Munc18-1 becomes even more dynamic, rapidly disperses, and subsequently reclusters at synapses. This requires calcium influx and PKC-dependent phosphorylation of Munc18-1, but is independent of syntaxin-1 interaction. Importantly, synaptic Munc18-1 levels scale with release efficiency, and higher Munc18-1 levels after stimulation increase synaptic strength.

During neuronal development, Munc18-1/syntaxin-1 and syntaxin-1 are transported by FEZ1–KIF5C (Chua et al., 2012) and syntabulin–KIF5B (Su et al., 2004; Cai et al., 2007) transport complexes. Our FRAP experiments in axons (Fig. 3) showed identical recovery of Munc18-1-Venus and Stx-YFP and increased Munc18-1-Venus recovery after cleavage of syntaxin-1, which indicates that lateral diffusion via membrane-bound syntaxin-1 rather than active transport of organelles is the main mode of delivering Munc18-1 to release sites in mature neurons. This is in line with a study on syntaxin-1 dynamics in mature rat neurons (Ribrault et al., 2011), underscoring the notion that after initial membrane delivery via fusion of transport vesicles, Munc18-1–syntaxin-1 complexes diffuse through the axonal membrane. Only rarely were moving Munc18-1-Venus puncta observed, which suggests that vesicular transport of Munc18-1–syntaxin-1 complexes does continue in mature neurons but that diffusion dominates. Hence, during neuronal maturation, the mode of syntaxin/Munc18 transport switches from predominantly vesicular transport (Cai et al., 2007; Chua et al., 2012) to lateral diffusion.

Munc18-1-Venus showed a strong synaptic preference (Fig. 4). Our FRAP measurements revealed that synaptic Munc18-1 is very mobile. Munc18-1-Venus exchange is best described with a fast (τ = 4.6 s) and a slow component (τ = 60 s), comparable to syntaxin-1 (Ribrault et al., 2011), whereas Bassoon (Tsuriel et al., 2009; Schröder et al., 2013), Munc13-1 (Kalla et al., 2006), liprin-α2 (Spangler et al., 2013), and synapsin-1 (Tsuriel et al., 2006) turnover at synapses is much slower, with exchange rates in the order of minutes to hours. Exchange rates of the total synaptic vesicle pool are even slower (Orenbuch et al., 2012b), although individual synaptic vesicle exchange between boutons is relatively fast (Darcy et al., 2006; Staras et al., 2010; Herzog et al., 2011). Synaptic activity increased the mobility of Munc18-1-Venus and Stx-YFP (Fig. 5), in line with increased mobility of synapsin (Chi et al., 2001; Tsuriel et al., 2006) and Rab3 (Star et al., 2005). In contrast, Munc13-1 and Bassoon mobility are not affected by acute stimulation (Kalla et al., 2006; Tsuriel et al., 2009). Hence, synapses exchange their components at different speeds. Proteins of the secretory machinery are more mobile than active zone scaffolding proteins and become even more dynamic during increased activity. The scaffolding proteins may therefore provide synapses with stability to maintain synapse integrity (Tsuriel et al., 2009), whereas intersynaptic exchange of presynaptic molecules and vesicles provides plasticity during increased network activity.

At synapses, ∼40% of Munc18-1-Venus resided in an immobile fraction. Cleavage of syntaxin-1 resulted in a larger mobile Munc18-1-Venus fraction, but did not fully abolish the immobile fraction. This suggests that Munc18-1 also interacts with other binding partners at the synapse. Here, Munc18-1 may be released when syntaxin-1 interacts with other SNAREs, which lowers the binding affinity of Munc18-1 for syntaxin-1 (Xu et al., 2010). Strong interaction between Munc18-1 and syntaxin-1 in the axon may ensure proper delivery of Munc18-1 to synapses and protect syntaxin-1 from nonspecific interactions. A more dynamic interaction at the synapse provides plasticity to modulate SNARE complex formation and interactions of Munc18-1 with other proteins.

Synaptic activity triggers synapsin dispersion from synapses (Chi et al., 2001; Fig. 6). In the same synapses, we observed significant Munc18-1-Venus dispersion during stimulation. Stx-YFP also dispersed, although to a smaller extent, which is in line with recent results (Degtyar et al., 2013). Dispersion of Munc18-1-Venus was not affected by BoNT/C expression (Fig. 6). Thus, calcium influx triggers dispersion irrespective of vesicle fusion, thereby increasing the concentration of soluble and dynamic Munc18-1.

Munc18-1 repopulated the synapse within 2 min after stimulation, much faster than synapsin (Chi et al., 2001; Orenbuch et al., 2012a) and Rab3a (Star et al., 2005), and reclustering already started during the stimulation train. On average, synapses contained more Munc18-1 after stimulation. Interestingly, poststimulation changes in Munc18-1-Venus levels were highly variable: some synapses showed an ∼20% increase in fluorescence, whereas others showed an ∼10% net decrease (Fig. 7). Such distinct and acute changes in synaptic levels have not been reported before. As Munc18-1 expression levels correlate with synaptic strength and stamina (Toonen et al., 2006b), this provides a mechanism that helps individual synapses to quickly tune their output and maintain vesicle release during periods of high demand. The wide range in initial synaptic Munc18-1-Venus levels is in line with the notion that individual synapses can have very different release probabilities (Rosenmund et al., 1993; Murthy et al., 1997; Matz et al., 2010; Ariel et al., 2012). This presynaptic mechanism adds to other well-known factors that tune synaptic output like postsynaptic target identity (Reyes et al., 1998; Sylwestrak and Ghosh, 2012) and synaptic position on the dendrite (de Jong et al., 2012), and may partly underlie the effects of neuronal activity (Branco et al., 2008) and calcium influx (Ermolyuk et al., 2012) on synaptic output.

Activation of the DAG–PKC pathway potentiates synaptic transmission (Malenka et al., 1986; Shapira et al., 1987) and is essential for post-tetanic potentiation (Brager et al., 2003; Fioravante et al., 2011). Furthermore, we have shown that DAG–PKC–dependent synaptic potentiation requires Munc18-1 phosphorylation, and that Munc18-1 phosphorylation is essential for maintaining synaptic release during repetitive stimulation and for synaptic recovery after stimulation (de Vries et al., 2000; Wierda et al., 2007). Here we show that Munc18-1-Venus reclustering after stimulation is calcium-dependent and almost absent after PKC inhibition. Moreover, a PKC-insensitive mutant Munc18-1 showed strongly impaired reclustering after intense stimulation. Thus, PKC-dependent phosphorylation of Munc18-1 is required for its reclustering and consequent changes in synaptic levels. Together, this provides a plausible mechanism for the fact that efficacy of synaptic vesicle release depends on Munc18-1 expression levels (Toonen et al., 2006b) and PKC activity (Majewski and Iannazzo, 1998; Stevens and Sullivan, 1998; Francis et al., 2002). We have previously shown that the PKC–Munc18-1 pathway translocates synaptic vesicles to the active zone (Wierda et al., 2007) and that Munc18-1 overexpression promotes membrane delivery of secretory vesicles in chromaffin cells (Toonen et al., 2006a). Hence, PKC-dependent recruitment of Munc18-1 temporarily increases its synaptic levels, thereby facilitating synaptic vesicle recruitment. Importantly, FM4-64 loading correlated with Munc18-1-Venus expression levels in individual synapses (Fig. 9). Moreover, synapses that recruited more Munc18-1 after high-frequency stimulation had a larger releasable vesicle pool. Hence, at the single synapse, Munc18-1 expression levels correlate with synaptic strength.

Together, our data support a model in which soluble Munc18-1 is dependent on plasma membrane–bound syntaxin-1 for delivery to the synapse at rest (Fig. 10, 1). At the synapse, Munc18-1 resides with syntaxin-1 and SNAP25 in microdomains at the plasma membrane (Lang et al., 2002; Sieber et al., 2006; Pertsinidis et al., 2013). Munc18-1 also interacts with other binding partners, and a fraction of Munc18-1 may transfer from syntaxin-1 (Fig. 10, 2) to these partners at some point after arrival in the synapse. Calcium influx temporarily increases Munc18-1 dynamics, which are independent of vesicle release or syntaxin-1 binding (Fig. 10, 3). Soluble Munc18-1 can disperse or may bind syntaxin-1 (Fig. 10, 4). After stimulation, PKC controls reclustering by phosphorylation of Munc18-1 (Fig. 10, 5). PKC-dependent reclustering of Munc18-1 temporarily increases the release of the synaptic vesicle pool.

A Venus cDNA fragment from pVenus-N1 (Nagai et al., 2002) was subcloned using restriction enzymes (SmaI–NotI and SmaI, respectively) into pUC21 (Vieira and Messing, 1991), resulting in pUC21VENUS. A loxNEOlox cassette (containing neomycin resistance gene and thymidine kinase promoter flanked by two LoxP sites) was subcloned using SalI–XbaI into pUC21VENUS in an antisense orientation 3′ of the Venus stop codon. The left cloning arm, a genomic sequence 5′ of, and including, exon 20 of the munc18-1 gene from 129/Sv embryonic stem (ES) cell DNA, was subcloned using PCR into pGEM-T Easy (Promega), resulting in pGEM-T Easy-Left-Arm. The right cloning arm, a genomic sequence 3′ of exon 20 of the munc18-1 gene from 129/Sv ES cell DNA, was subcloned using PCR into pGEM-T Easy, resulting in pGEM-T Easy-Right-Arm. The right cloning arm was subcloned into pGEM-T Easy-Left-Arm using KpNI–AvrII and KpNI–SpeI. Venus and the loxNEOlox cassette from pUC21VENUS were subcloned between the left and right arm, with Venus replacing the stop codon of exon 20 in frame, creating the targeting vector (Fig. 1 A). Mice carrying the munc18-1-venus-neo gene were generated by homologous recombination in ES cells and identified by Southern blotting (Fig. 1 B) or PCR (Fig. 1 C). Heterozygous mice (+/m) were crossed with EIIa-cre mice, which express Cre recombinase in early embryonic stages (Lakso et al., 1996), to remove the neomycin resistance cassette. All imaging experiments were conducted on hippocampal m/m neurons of +/m × +/m crosses. Munc18-1 null mutant mice (deleting exon 2–6) have been described previously (Verhage et al., 2000). All animals were housed and bred according to the institutional and Dutch governmental guidelines for animal welfare.

Dissociated hippocampal cultures were obtained from embryonic day 18 (E18) m/m mice as described previously (Meijer et al., 2012). In brief, hippocampi were dissected in Hepes-buffered HBBS (Invitrogen) and digested with 0.25% trypsin (Invitrogen) at 37°C for 20 min. After washing and trituration, cells were plated at a density of 25,000 cells/well for low-density cultures on top of a pregrown rat glia feeder layer on 18-mm coverslips. For autaptic cell cultures, cells were plated at a density of 2,000 cells/well on micro islands of rat glia as described previously (Meijer et al., 2012). Micro-islands were made by stamping 200-µm-diameter poly-d-lysine (0.15 mg/ml; Sigma-Aldrich)/rat tail collagen (3.66 mg/ml; BD) dots on 18-mm coverslips pretreated with 0.15% agarose (Type II-a). Cultures were kept in neurobasal medium (containing 2% B27, 18 mM Hepes, 0.5 mM GlutaMAX, and penicillin/streptomycin; all obtained from Invitrogen), and half the medium was replaced once every week in low-density cultures. Where applicable, Munc18-1-Venus neurons were cultured together with WT neurons in a 1:50 ratio to image single fluorescent neurons. Neurons were used for experiments at 14–21 DIV.

For imaging experiments on WT neurons, syntaxin-1-EYFP, GFP, mVenus, and synapsin-mCherry were transfected using calcium phosphate precipitate at 6–10 DIV as described previously (Köhrmann et al., 1999). For FM4-64 loading experiments in Munc18-1-Venus neurons, synapsin-ECFP was transfected using calcium phosphate precipitate at 6–10 DIV. For all other imaging experiments on Munc18-1-Venus neurons, synapsin-mCherry was expressed by adding lentiviral particles before plating. After trituration, lentiviral particles were added and cells were incubated at 37°C/5% CO₂ for 2 h. After three washes with neurobasal medium supplemented with 10% fetal bovine serum to inactivate the lentiviral particles, cells were plated on glia feeder layers. BoNT/C-ECFP was expressed using Semliki viral particles 6–12 h before imaging in autaptic cultures. ECFP signal in the soma of a single neuron on a micro island ensured BoNT/C expression in the imaged synapses.

For imaging of Munc18-1-EGFP and PKC-insensitive Munc18-1-EGFP on munc18-1 null mutant neurons (Verhage et al., 2000), E18 embryos were obtained from timed munc18-1 heterozygous null mutant matings and processed for WT neurons. Dissociated null mutant neurons were infected in solution before plating on glia micro islands with lentiviral particles expressing Munc18-1-EGFP WT or PKC-insensitive mutant together with synapsin-mCherry particles and placed in a cell culture incubator for 1.5 h. After two washing steps in MEM with 10% FCS, neurons were counted and plated at a density of 2,000 cells/18-mm coverslip.

Syntaxin-1-EYFP is full-length rat Syntaxin1a in pEYFP-N1 (CMV promoter; Takara Bio Inc.), and has been described previously (Toonen et al., 2005). Expression plasmid pCMV BoNT/C light chain was a gift from T. Galli (Institut National de la Santé et de la Recherche Médicale, Paris, France). Synapsin-mCherry (pcDNA3.1) was a gift from A. Jeromin (Allen Brain Institute, Seattle, WA). Synapsin-ECFP was generated by replacing mCherry in Synapsin-mCherry with ECFP. mVenus was generated by the addition of the last 20 amino acids (KMSKDGKKKKKKSKTKCVIM) of K-Ras2B, representing a canonical palmitoylation signal (Welman et al., 2000). Synapsin-mCherry was subcloned into pLenti vectors, and viral particles were produced as described previously (Naldini et al., 1996). Munc18-1-EGFP WT and PKC-insensitive mutant Munc18-1-GFP (Munc18_PKCi: S306A, S312A, S313A; Wierda et al., 2007) were cloned into pLenti vectors and used to rescue munc18-1 null mutant neurons.

Electrophysiological recordings were performed on autaptic glutamatergic hippocampal neurons at room temperature (21–24°C) as described previously (Weber et al., 2010). In brief, isolated hippocampal neurons were plated on astrocyte micro islands (Bekkers and Stevens, 1991) in neurobasal medium (Invitrogen) supplemented with B-27 (Invitrogen), 17.3 mM Hepes, 1% GlutaMax-I (Invitrogen), and 1% penicillin/streptomycin (Invitrogen). Autaptic cells between 10 and 14 DIV were used for experiments. The patch-pipette solution included 135 mM potassium gluconate, 10 mM Hepes, 1 mM EGTA, 4.6 mM MgCl₂, 4 mM Na-ATP, 15 mM creatine phosphate, 50 U/ml phosphocreatine kinase, and 300 mOsm, pH 7.3. The standard extracellular medium consisted of 140 mM NaCl, 2.4 mM KCl, 10 mM Hepes, 10 mM glucose, 4 mM CaCl₂, 4 mM MgCl₂, and 300 mOsm, pH 7.3. Cells were whole-cell voltage clamped at −70 mV with an EPC-9 amplifier (HEKA) under control of Pulse 8.80 software (HEKA). Currents were low-pass filtered at 1 or 5 kHz and stored at either 10 or 20 kHz. Pipette resistance ranged from 4 to 6 MΩ. The series resistance was compensated for 75%. Only cells with series resistances <20 MΩ were analyzed. EPSCs were evoked by depolarizing the cell from −70 to 0 mV for 2 ms.

Cultured neurons from 1 to 14 DIV were fixed in 3.7% formaldehyde in PBS, pH 7.4, for 20 min at RT. Neurons were washed three times in PBS, permeabilized in 0.5% Triton X-100 (Sigma-Aldrich) in PBS for 5 min, and incubated with blocking solution (PBS containing 2% normal goat serum, 2% bovine serum albumin, and 0.1% Triton X-100) for 30 min at RT. Neurons were then incubated with primary antibodies in blocking solution for 1 h at RT, washed three times with PBS, and incubated with secondary antibodies conjugated to Alexa Fluor in blocking solution for 1 h at RT (1:1,000; Invitrogen). Primary antibodies used were chicken polyclonal MAP2 (1:20,000; Abcam), mouse monoclonal AnkyrinG (1:1,000; SC-12719; Santa Cruz Biotechnology, Inc.), Smi-312 (1:500; Covance), and polyclonal synapsin (1:2,000; Synaptic Systems). Finally, cells were washed three times in PBS and coverslips were mounted in Mowiol on glass slides. Images were acquired on a confocal microscope (LSM 510 Meta; Carl Zeiss) with a 63× oil objective lens, NA 1.4, and analyzed in MATLAB (MathWorks, Inc.) using SynD (Schmitz et al., 2011).

For labeling of brain slices, P14-P21 WT and Munc18-1-Venus mice were perfused transcardially with 4% PFA in PBS, then brains were removed and post-fixed in 4% PFA in PBS overnight. Subsequently, brains were submerged in 30% sucrose in PBS for 3 d before cryosectioning. Before incubation in blocking solution (5% normal goat serum, 2.5% BSA, and 0.2% Triton X-100 in PBS) for 1 h, 35-µm cryosections were incubated in 1% H₂O₂ for 30 min and rinsed with PBS. Sections were incubated with primary antibody in blocking solution overnight at 4°C on a shaker. Primary antibodies used were rabbit polyclonal Munc18-1b (1:200; Synaptic Systems), chicken polyclonal MAP2 (1:200; Abcam), and mouse monoclonal VAMP-2 (1:500; Synaptic Systems). Cryosections were then washed four times in PBS and incubated with secondary antibody (Alexa Fluor; Invitrogen) diluted in blocking solution for 2 h on a shaker. Finally, cryosections were mounted in 2.5% DABCO (Invitrogen) in Mowiol on glass slides. All steps were performed at RT, unless otherwise stated. Images were acquired on a confocal microscope (LSM 510 Meta; Carl Zeiss) with either a 10× air objective lens and 0.7× mechanical zoom (Fig. 1 G, top) or a 40× objective (1.3 NA) and 0.7× mechanical zoom (Fig. 1 G, bottom).

All live imaging experiments, except FM4-64 synapse labeling (Fig. 9), were conducted on a custom-built tandem illumination microscope (TIM; Olympus) consisting of an inverted imaging microscope (IX81; Olympus) and an upright laser scanning microscope. The inverted microscope part was used for imaging fluorescence using an MT20 light source (Olympus), appropriate filter sets (Semrock), and a 60× oil objective lens (NA 1.49) with or without a 1.6× additional magnification on an EM charge-coupled device (CCD) camera (C9100-02; Hamamatsu Photonics). Xcellence RT imaging software (Olympus) was used for controlling the microscope and recording the images.

The upright microscope part (TIM; Olympus) guided and focused a 473-nm laser (full width at half maximum = ∼1 µm) used for fluorescence bleaching through a 60× water immersion objective lens (NA 0.90) on the plane of focus of the inverted microscope. For bleaching, one or multiple positions were selected and prepared in the Xcellence RT imaging software, and a transistor–transistor logic (TTL) pulse in between two imaging frames triggered the solid-state laser and laser scanner (UGA40; both from Rapp OptoElectronic). Laser intensity and pulse duration for bleaching were optimized to reach >60% fluorescence decrease. Fluorescence recovery was monitored by time-lapse acquisitions at 1–2 Hz for the first 30 s and 0.25–0.5 Hz for the remaining time.

Experiments were conducted on an inverted microscope (Axio Observer Z; Carl Zeiss) for imaging fluorescence, using an illumination unit (Polychrome V; Till Photonics), appropriate filter sets (Chroma and Semrock), and a 40× oil objective lens (NA 1.30) equipped with automated focus adjustment (Definite Focus; Carl Zeiss) on an EM-CCD camera (C9100-02; Hamamatsu Photonics). Synaptic vesicle pools were labeled by field stimulation for 30 s at 1 Hz in the presence of 10 µM FM4-64 (Invitrogen). FM4-64, dissolved in imaging solution, was applied locally using a barrel pipette 30 s before, during, and 1 min after stimulation to ensure complete labeling of all recycling vesicles. The cells were subsequently washed by applying dye-free imaging solution for 15 min via the barrel pipette. The FM4-64 labeling experiments consisted of seven sequential steps: a baseline of 10 frames each (Synapsin-ECFP, Munc18-1-Venus, FM4-64); FM4-64 synapse labeling and wash with 1 frame each every 2 min; baseline of 10 frames each; 30 s of 600 AP at 20 Hz, or 600 AP at 2 Hz with 30 s of 0.2 Hz imaging + 5 min 1 frame every 30 s after Munc18-1-Venus reclustering; baseline of 10 frames each; FM4-64 synapse labeling and wash with 1 frame each every 2 min; baseline of 10 frames each; 30 s of 600 AP at 20 Hz; and baseline of 10 frames each.

Coverslips were placed in an imaging chamber and perfused with imaging solution (Tyrode’s: 2 mM CaCl₂, 2.5 mM KCl, 119 mM NaCl, 2 mM MgCl₂, 20 mM glucose, and 25 mM Hepes, pH 7.4). Electrical field stimulation by parallel platinum electrodes was applied by a Master-8 system (A.M.P.I.), a stimulus isolator (A385RC; World Precision Instruments) delivering 30-mA, 1-ms pulses, and in the presence of 50 µM (2R)-amino-5-phosphonovaleric acid (AP5; Tocris Bioscience) and 10 µM 6,7-dinitroquinoxaline-2,3-dione (DNQX; Tocris Bioscience) to block network activity by inhibiting glutamatergic transmission. PMA (Sigma-Aldrich) was bath applied for long incubations (≥2 min) or barrel applied for short incubations (30 s). Ro 31-8220 (EMD Millipore) was bath applied. All imaging experiments were performed at RT (21–24°C).

Image stacks from time-lapse recordings were loaded into ImageJ (National Institutes of Health), and multiple regions of interest were placed: 6-pixel-diameter circles in the middle of bleached areas and/or unbleached synapses, background regions devoid of any cells, and a control region for acquisition bleaching. Raw traces were extracted and loaded into Excel (Microsoft) for further analysis. All raw traces were corrected for background with F(t) = F_raw(t) − F_b(t), where F_raw(t) and F_b(t) are the intensity of a raw trace and the background trace, respectively, at time t, and subsequently corrected for ongoing photobleaching by multiplying with F_cntrl(0)/F_cntrl(t), where F_cntrl(0) is the background-corrected intensity of the control trace at time 0. For synaptic levels analysis, the corrected traces were normalized to the fluorescence at the start of the experiment with the following: ΔF/F0 = (F(t) − F(0))/F(0). For fluorescence recovery (F_rec(t)), the corrected traces of bleached regions were normalized to the fluorescence before (F(t_prebleach)) and immediately after bleaching (F(t_bleach); Siggia et al., 2000) with:

A fit of the mean Munc18-1-Venus recovery trace was made according to a biexponential function:

where M_f and M_s are the mobile fraction of the fast and slow component, respectively, and τ_f and τ_s are the recovery time constants for the fast and slow component, respectively. The NonlinearLeastSquares function implemented in MATLAB was used to calculate the best fit. For synapse-to-axon ratio analysis, a linescan (width of 3 pixels) from the middle of the synapse (identified by Synapsin-mCherry) up to 4 µm into the axon (without passing other synapses) was extracted from an image loaded into ImageJ, and the trace was background corrected and normalized to intensity at the indicated lengths with F_norm(x) = F(x)/F(x_a), where F(x) is the background-corrected fluorescence along the linescan and F(x_a) is the background-corrected fluorescence at a specified point in the axon (x_a: 3–4 µm from synapse center). For kymographs, a line (width of 3 pixels) was placed over an axon in an image loaded into ImageJ, and the kymograph function was used accordingly.

Nimopedine, omega-agatoxin, and omega-conotoxin were obtained from Tocris Bioscience. Ro-31-8220 from Sigma-Aldrich. AP-V and DNQX were from Ascent.

Differences between two groups were tested for significance using a Student’s t test for unpaired data when data passed a normality test (Kolmogorov-Smirnov) and a Mann-Whitney (M-W) test when it did not. Differences between more than two groups were tested for significance using a Kruskal-Wallis (K-W) test, and pairwise comparisons using M-W tests were used where applicable. P-values were adjusted using a false discovery rate (FDR) correction (Benjamini and Hochberg, 1995) for the number of multiple comparisons in short FDR[number] corrections. For testing significant differences in the amount of synapses per group under different conditions, the Pearson’s χ² test was used. Correlation between two variables was tested for significance with Spearman rank correlation. Differences were regarded as significant when P < 0.05. Statistics were performed in SPSS (IBM). All data are plotted as mean ± SEM (error bars or shaded area) in arbitrary units.

Fig. S1 shows protein expression levels in brain lysate of +/m and −/m mice together with example images of Munc18-1-Venus neuron morphology. Fig. S2 shows kymographs and quantification of bleached axons with and without transported vesicles. Fig. S3 shows the intensity change of mVenus during and after stimulation. Fig. S4 shows the relation between initial synaptic Munc18-1-Venus fluorescence and fluorescence after stimulation together with the relation between synaptic Munc18-1-Venus fluorescence and synapse size. Fig. S5 shows the intensity change of Munc18-1-Venus subsets + P/Q-, L-, and N-type calcium channel blockers; Munc18-1-Venus FRAP + PMA; Munc18-1-Venus subsets + Ro 31-8220; and subsets of PKC-insensitive Munc18-1 mutant. Additional data are available in the JCB DataViewer at http://dx.doi.org/10.1083/jcb.201308026.dv.

We thank Robbert Zalm for expert help with virus production and cloning and Emmeke Aarts for excellent advice on statistics.

This work is supported by the Netherlands Organization for Scientific Research (ZonMw-VENI 916-66-101 and ZonMW-TOP 91208017 to R.F. Toonen, Pionier/VICI 900-01-001 and ZonMW 903-42-095 to M. Verhage, and ZonMw-TOP 91208017 to J.P. Weber). This work is also supported by the EU (EUSynapse project 019055, EUROSPIN project HEALTH-F2-2009-241498, and HEALTH-F2-2009-242167 SynSys project to M. Verhage).

The authors declare no competing financial interests.

T. Cijsouw, M. and R.F. Toonen designed the research; T. Cijsouw and J.H. Broeke performed and analyzed live cell imaging experiments; I. Saarloos, D. Schut, T. Kroon, and J.A.C. Broek performed biochemistry, and immunocytochemistry; J.P. Weber performed and analyzed electrophysiological recordings; and T. Cijsouw, M. Verhage, and R.F. Toonen wrote the manuscript.