The Munc18-1 domain 3a loop is essential for neuroexocytosis but not for syntaxin-1A transport to the plasma membrane

Synaptic transmission relies on the rapid and regulated fusion of synaptic vesicles with the plasma membrane. Essential to this process are the SNAREs (soluble N-ethylmaleimide-sensitive factor-attachment protein receptors), syntaxin-1A (Sx1a) and SNAP25, which reside on the plasma membrane, and VAMP2/synaptobrevin found on the synaptic vesicles (Chen and Scheller, 2001; Jahn et al., 2003; Weber et al., 1998). The formation of a complex between these proteins, involving the zippering of their SNARE domains into a four-helix parallel coiled-coil, provides the energy to drive the fusion of the two opposing membranes required for secretion of vesicular neurotransmitters (Poirier et al., 1998; Sutton et al., 1998). A number of regulatory proteins are also necessary for SNARE-mediated membrane fusion. One of the most highly studied but still enigmatic regulatory proteins is Munc18 (Han et al., 2010; Toonen and Verhage, 2007; Verhage et al., 2000). Munc18-1, the primary neuronal isoform (Pevsner et al., 1994), can bind directly to Sx1a (‘closed’ conformation) in a binary complex, proposed to regulate the anterograde transport of Sx1a to the cell surface (Arunachalam et al., 2008; Han et al., 2011; Han et al., 2009; Malintan et al., 2009; Rickman et al., 2007), and can also associate directly with the SNARE complex, where it is required for vesicle fusion (Deák et al., 2009; Shen et al., 2007; Tareste et al., 2008). Depletion of Munc18 from neurosecretory cells results in a transport defect associated with reduced Sx1a levels (Arunachalam et al., 2008; Han et al., 2009; Toonen et al., 2005; Verhage et al., 2000; Voets et al., 2001) and Sx1a mislocalisation to intracellular compartments (Arunachalam et al., 2008; Han et al., 2009; McEwen and Kaplan, 2008), leading to the abrogation of stimulated secretion (Arunachalam et al., 2008; Han et al., 2009; Toonen et al., 2005; Voets et al., 2001).

A number of distinct modes of binding between Munc18-1 and Sx1a have been proposed, based upon both structural studies and predicted interactions (see Fig. 1A). The first complex of Munc18-1 with the soluble region of Sx1a was solved by X-ray crystallography (Burkhardt et al., 2008; Misura et al., 2000), and demonstrated Munc18-1 binding to closed Sx1a. In this complex the C-terminal SNARE helix and N-terminal Habc domain of Sx1a are held in a compact auto-inhibitory conformation that precludes Sx1a interaction with other SNAREs. This structure constitutes the predominant form required for transport to and/or stabilisation of Sx1a at the plasma membrane. More recently, a second ‘open’ Sx1a conformation was observed by small-angle X-ray scattering studies (Christie et al., 2012). A short N-terminal peptide in Sx1a is critical for the formation of this open Sx1a conformation, which is required for SNARE complex formation, but the mechanism driving this conformational change is still not understood. It has also been shown that a loop in Munc18-1 domain 3a can undergo a significant conformational change to an extended helical structure, which is incompatible with binding of closed Sx1a and may therefore be important for subsequent SNARE complex formation (Hu et al., 2011). This loop has also been suggested to act as a platform for interaction with SNAREs or other proteins during downstream SNARE complex assembly (Boyd et al., 2008; Hu et al., 2011). Although structural studies are beginning to shed light on the molecular details of the different binding modes, their exact functional role, the mechanisms that regulate their inter-conversion, and the spatiotemporal regulation of their assembly remain unclear (Rizo and Südhof, 2012).

In the present study we analysed the role of the Munc18-1 domain 3a loop in both Sx1a transport to the cell surface and secretory granule (SG) fusion using PC12 cells deficient in Munc18-1/-2 (Han et al., 2009; Malintan et al., 2009). Intriguingly, we found that although a partial deletion of the domain 3a loop resulted in a Munc18-1 mutant that could bind normally to Sx1a or the SNARE complex, the loop mutant was unable to rescue membrane fusion in vivo, despite fully rescuing the transport of Sx1a to the plasma membrane. A companion study, published in this issue of Journal of Cell Science, reveals that serendipitous insertion of several residues within this loop led to similar phenotypes, including a block of neuroexocytosis without disruption of Sx1a transport to the plasma membrane (Han et al., 2013). Together these results provide direct evidence for a defined role for Munc18-1 in promoting neuroexocytosis that is clearly distinct from its function in Sx1a transport, and further suggest that additional regulatory elements are required to promote the function of Munc18-1 in supporting membrane fusion.

From recent structural analyses of Munc18-1 binding to Sx1a, we hypothesised that a conformational change in domain 3a amino acids 317–333, from a compact loop to an extended helical structure, may help to promote the conversion of Sx1a from a closed to an open conformation (Fig. 1A) (Hu et al., 2011). Furthermore, the presence of a proline residue adjacent to the domain 3a loop (P335) suggested that it might act as a hinge residue important for this extension (Hu et al., 2011). In order to test these hypotheses we constructed two mutants of Munc18-1, one in which the predicted extended region of domain 3a was deleted (Munc18-1^Δ317–333), and a second in which P335 was replaced with alanine (Munc18-1^P335A; Fig. 1B).

To analyse the effect of these Munc18-1 mutations on SG fusion we used a PC12 cell line stably depleted of endogenous Munc18-1 and Munc18-2 (DKD-PC12), which is deficient in stimulated exocytosis (Han et al., 2009). Expression constructs of wild-type Munc18-1 (Munc18-1^WT), Munc18-1^P335A and Munc18-1^Δ317–333 carrying silent nucleotide mutations within the RNAi target sequence (Arunachalam et al., 2008) were generated, both as untagged proteins and with a C-terminal emGFP tag, and expressed in DKD-PC12 cells. Consistent with the small deletion, Munc18-1^Δ317–333 (tagged or untagged) was detected on western blots at a slightly lower molecular mass than Munc18-1^WT (Fig. 1C). For both mutations, the protein expression level was similar to that of the wild-type protein.

To examine the ability of the Munc18-1 mutants to restore stimulated SG fusion, emGFP-tagged Munc18-1^WT, Munc18-1^P335A or Munc18-1^Δ317–333 was re-expressed in DKD-PC12 cells together with neuropeptide Y (NPY) fused to the catalytic domain of human placental alkaline phosphatase [NPY–hPLAP (Arunachalam et al., 2008)]. Munc18-1^WT and Munc18-1^P335A were able to rescue stimulated NPY release to a similar extent. However, there was no significant rescue of NPY-stimulated release following expression of Munc18-1^Δ317–333 (Fig. 2A). To directly examine the effect of the Munc18-1^Δ317–333 deletion on SG fusion, DKD-PC12 cells were subsequently co-transfected with control vectors, Munc18-1^WT–emGFP or Munc18-1^Δ317–333–emGFP and NPY–mCherry. The fusion of individual SGs containing NPY–mCherry with the cell surface was analysed by total internal reflection fluorescence (TIRF) microscopy before and during stimulation (Fig. 2B–D). Consistent with previous results, Munc18-1^WT but not Munc18-1^Δ317–333 restored the number of SG fusion events detected to a level statistically indistinguishable from that in control PC12 cells. The lack of stimulated secretion and vesicle fusion did not result from a reduction in SGs, as analysis of both the number of SGs within the TIRF plane (Fig. 2F) and immunolabelling for the SG protein synaptotagmin-1 (Fig. 2E) demonstrated a similar number and distribution of SGs in both Munc18-1^WT- and Munc18-1^Δ317–333-expressing DKD-PC12 cells. These data confirm that, although the putative hinge residue P335 is not required, the loop region of domain 3a is necessary for Munc18-1 function in SG fusion during stimulated secretion and exocytosis.

We have previously shown that Sx1a transport to the cell surface is defective in DKD-PC12 cells (Han et al., 2009; Malintan et al., 2009). Most importantly, it was shown that the inability of various Munc18-1 mutants to mediate fusion and exocytosis is highly correlated with their inability to rescue Sx1a plasma membrane localisation (Han et al., 2011). To determine if the Munc18-1^Δ317–333 mutant behaves similarly, we first analysed the capacity of Munc18-1^WT or Munc18-1^Δ317–333 to restore Sx1a plasma membrane localisation. Confocal microscopy of transfected cells revealed that the majority of the expressed Munc18-1 resided in the cytosol, although a small plasma membrane-associated pool could be detected in a subset of cells (Fig. 3A). In control transfected or untransfected DKD-PC12 cells a low level of endogenous Sx1a was detected in intracellular compartments, with none detectable at the cell surface. In contrast, following re-expression of either Munc18-1^WT–emGFP or Munc18-1^Δ317–333–emGFP, Sx1a was predominantly present at the cell surface and little labelling was detected in intracellular compartments. Quantification of Sx1a labelling at the cell surface revealed that Munc18-1 and Munc18-1^Δ317–333 were capable of rescuing Sx1a plasma membrane localisation to the same degree (Fig. 3B).

To further analyse Sx1a transport, we then expressed Sx1a–GFP in DKD-PC12 cells. Unlike in heterologous cell types (Medine et al., 2007; Rickman et al., 2007; Rowe et al., 1999), we detected a low level of expressed Sx1a–GFP at the surface of DKD-PC12 cells lacking Munc18 (Fig. 3C–E). An additional intracellular pool was detected in the perinuclear region of these cells. Upon co-expression of either Munc18-1^WT or Munc18-1^Δ317–333 there was a significant shift in Sx1a–GFP localisation, with increased cell surface fluorescence and a loss of intracellular fluorescence (Fig. 3D,E). Together these results show that although the loop region in domain 3a of Munc18-1 is required for the functional activity of Sx1a in membrane fusion, its deletion does not affect the ability of Munc18-1 to facilitate the transport and/or localisation of Sx1a at the cell surface.

We next asked whether the inability of Munc18-1^Δ317–333 to facilitate vesicle fusion was due to differences in binding to Sx1a, by comparing the interaction between Munc18-1^WT and Munc18-1^Δ317–333 with the cytoplasmic domain of Sx1a (Sx1a^1–261). Using immobilised C-terminally His-tagged Sx1a^1–261 we found that both Munc18-1^WT and Munc18-1^Δ317–333 showed equivalent binding in pulldown assays (Fig. 4A). Analysis of the thermodynamic properties of Munc18-1 binding to soluble Sx1a^1–261 by isothermal titration calorimetry (ITC) revealed only minor differences between Munc18-1^WT and Munc18-1^Δ317–333 (Fig. 4B). Munc18-1^WT bound with typical high affinity (K_d = 6.4±2.8 nM) and an enthalpy of −26.3±0.9 kcal/mol, similar to previous reports (Burkhardt et al., 2008; Deák et al., 2009; Malintan et al., 2009). Munc18-1^Δ317–333 bound with a statistically indistinguishable K_d of 13.5±9.4 nM but with a slightly reduced binding enthalpy of −20.5±1.7 kcal/mol. We therefore conclude that the Munc18-1 loop deletion does not significantly alter binding to the isolated Sx1a protein, in line with its ability to mediate Sx1a plasma membrane transport.

We next examined the binding between Munc18-1^Δ317–333 and the pre-formed SNARE complex (Sx1a–SNAP25–VAMP2). Immobilised His–Sx1a was incubated with purified SNAP25 and VAMP2 overnight prior to the addition of either Munc18-1^WT or Munc18-1^Δ317–333 for up to 2 hours. Intriguingly, both Munc18-1^WT and Munc18-1^Δ317–333 interacted with the pre-formed SNARE complex to a similar degree (Fig. 4C,D), despite the inability of Munc18-1^Δ317–333 to mediate SG fusion in cells.

Although our data show that Munc18-1^Δ317–333 binds to both monomeric Sx1a and the SNARE complex, and rescues Sx1a localisation at the plasma membrane, it is unable to rescue SG fusion in vivo. We therefore examined the binding of Munc18-1 to Sx1a in DKD-PC12 cells. Munc18-1^WT–emGFP or Munc18-1^Δ317–333–emGFP was expressed in DKD-PC12 cells and Munc18-1 isolated by immunoprecipitation using the GFP tag. As expected, a robust interaction was detected between Munc18-1^WT–emGFP and Sx1a (Fig. 5A; ∼50% of Sx1a bound to Munc18-1). Surprisingly, however, despite the normal transport and localisation of Sx1a to the plasma membrane, the amount of Sx1a bound to Munc18-1^Δ317–333–emGFP was greatly reduced in a cellular context. Conversely, we also co-expressed Sx1a–GFP and either Munc18-1^WT or Munc18-1^Δ317–333 in DKD-PC12 cells and immunoprecipitated Sx1a using the GFP tag (Fig. 5B). Consistent with the Munc18-1 immunoprecipitation, very little interaction was observed between Sx1a–GFP and Munc18-1^Δ317–333, whereas there was significant binding to Munc18-1^WT. These data clearly indicate that, although deletion of the domain 3a loop of Munc18-1 does not affect the transport and stabilisation of Sx1a at the cell surface, it leads to a dramatically perturbed binding equilibrium with Sx1a, either in its apo state or within the SNARE complex itself.

The dual nature of the functional interaction between Munc18-1 and Sx1a has hindered direct analysis of the role of Munc18-1 in SG fusion. For all previously identified Munc18-1 mutations that impact neuroexocytosis, the effect can be directly correlated with both reduced binding affinity and cell surface recruitment of Sx1a (Han et al., 2011). In the present study, we have identified a loop region in domain 3a of Munc18-1 that plays a clear role in exocytosis independent of Sx1a binding affinity or anterograde transport. Furthermore, we have shown that the consequences of deleting the Munc18-1 domain 3a loop are exclusively detected in vivo, suggesting the involvement of additional regulatory factors acting through or on Munc18-1 that are not present in the in vitro assays.

The observation that expression of Sx1a in heterologous cell types results in its localisation to intracellular compartments unless Munc18-1 is co-expressed was the first indication that Munc18-1 may be required to chaperone Sx1a through the biosynthetic pathway (Medine et al., 2007; Rickman et al., 2007; Rowe et al., 1999). Subsequent analysis of the expression levels and localisation of Sx1a in Munc18-1-deficient neurons and neuroendocrine cells confirmed that Sx1a levels are significantly reduced in the absence of Munc18-1, with the remaining Sx1a being restricted to intracellular compartments (Arunachalam et al., 2008; Han et al., 2009; McEwen and Kaplan, 2008; Rickman et al., 2007; Toonen et al., 2005; Verhage et al., 2000; Voets et al., 2001). Interestingly, in comparison to heterologous cell types, we found that expressed Sx1a–GFP was able to traffic to the cell surface in Munc18-deficient DKD-PC12 cells, albeit at a much reduced level and with less efficiency than upon re-expression of Munc18-1. The reason for this is currently unknown, but could reflect the involvement of specific factors present in neurosecretory cells that can partially compensate for the loss of Munc18 in regulating the fidelity of Sx1a targeting. Munc18-1^WT and Munc18-1^Δ317–333 were able to restore Sx1a transport to the cell surface to an equivalent extent, and to reduce intracellular Sx1a–GFP accumulation, consistent with the indistinguishable binding affinities identified in vitro.

Our data, and that of the companion paper (Han et al., 2013), clearly establish a positive role for Munc18-1 in facilitating membrane fusion that is separate from its role in transporting Sx1a at the cell surface. Our results are consistent with the view that deletion of the extendable loop of domain 3a diminishes the equilibrium binding of Munc18-1 and Sx1a at the cell surface and fails to restore membrane fusion. In addition, they strongly suggest that the closed conformation binding required for transport to the cell surface contributes relatively little to the overall interaction between Munc18-1 and Sx1a in cells. Following the transport of the Sx1a–Munc18-1 binary complex to the cell surface, Munc18-1 plays a role in both docking and fusion of SGs (Han et al., 2010; Toonen and Verhage, 2007). Although one of the primary defects present in Munc18-1-deficient neuroendocrine cells is a reduction in the number of docked SGs, this may be partly, or even primarily, the result of the dramatic loss of Sx1a at the cell surface. In Munc18-1-null chromaffin cells, overexpression of the plasma membrane SNARE protein, SNAP25, which can form a binary complex with Sx1a and stabilise its cell surface recruitment, is able to restore the docking of vesicles but not subsequent fusion. This suggests an essential function of Munc18-1 in membrane fusion downstream of SG docking (de Wit et al., 2009), consistent with the results of electrophysiological analyses (Gulyás-Kovács et al., 2007; Jorgacevski et al., 2011; Schütz et al., 2005). Here we show a similar effect from specific mutation of Munc18-1 itself, providing direct evidence for a specific role of Munc18-1 in the fusion reaction in vivo. Previous studies of domain 3a have identified a number of point mutations located close to the 317–333 loop that exert effects on Munc18-1 function. Overexpression of one of these mutations (Y337L) has been reported to affect the characteristics of individual fusion events in bovine chromaffin cells without influencing the overall extent of fusion, and to result in reduced binding to SNARE complexes isolated from brain lysates by GST–complexin II affinity chromatography (Boyd et al., 2008). Clearly, further studies on domain 3a are needed to better understand this region in the context of neuroexocytosis. As Munc18-1^Δ317–333 exhibited binding to the SNARE complex that was indistinguishable from that of wild-type Munc18-1, but was unable to support SG fusion, our data support mounting evidence suggesting that Munc18-1 binding to the SNARE complex itself is not inherently required for exocytosis (Han et al., 2011; Malintan et al., 2009; Meijer et al., 2012).

Although Munc18-1^Δ317–333 was indistinguishable from wild-type Munc18-1 in binding to Sx1a both in a binary complex and in the context of the SNARE complex in vitro, significant differences were identified in the interaction with Sx1a and in SG fusion in cells. The reasons for these differences are currently unknown but are likely to stem from regulatory factors present in cells that are absent from the in vitro assays. Possibilities include post-translational modifications of either Sx1a or Munc18-1, or disrupted binding to other interacting molecules. Sx1a and Munc18-1 are known to interact directly or indirectly with a wide range of proteins, including other key components of the secretory machinery such as Munc13 and complexin, in addition to the cognate t-SNARE, SNAP25, and v-SNARE VAMP2/synaptobrevin (reviewed by Latham and Meunier, 2007; Malsam et al., 2008; Rizo and Südhof, 2012), and lipid regulatory proteins (Lee et al., 2004). The effect of mutating the domain 3a loop of Munc18-1 on these other interactions warrants further investigation. Furthermore, Sx1a is known to form 40–70 nm nanoclusters at the cell surface (Bar-On et al., 2012; Rickman et al., 2010), which correlate with sites of SG exocytosis (Barg et al., 2010). While our study has shown that Sx1a is delivered to the cell surface by Munc18-1^Δ317–333, possible effects on the distribution of Sx1a into microdomains, and the underlying nanoarchitecture of these domains, remains unknown. Finally, the domain 3a loop lies close to previously identified PKC phosphorylation sites. Phosphorylation of Munc18-1 is known to prevent Sx1a binding in vitro (Fujita et al., 1996) and alter vesicle release kinetics in chromaffin cells (Barclay et al., 2003). Although outside the scope of this study, it would be interesting to determine whether PKC-mediated phosphorylation of Munc18-1 affects the function of the loop in facilitating membrane fusion.

In summary we report here on a mutation in Munc18-1 domain 3a that has a defined exocytic role distinct from its chaperoning function in Sx1a transport to the plasma membrane. Furthermore, our study suggests that the function of Munc18-1 in membrane fusion in vivo is facilitated by additional regulatory elements, such as post-translational modifications or accessory proteins and lipids, which are absent from our in vitro binding assays. Identification of these regulatory components will prove an important endeavour for future studies.

Mouse anti-Sx1a (S0644, clone HPC-1) and mouse anti-β-actin (A5316, clone AC-74) were obtained from Sigma-Aldrich (Sydney, NSW, Australia), mouse anti-Synaptotagmin-1 (#105 011) was obtained from Synaptic Systems (Goettingen, Germany), and mouse anti-Munc18-1 (#610336) was obtained from BD Biosciences (Sydney, NSW, Australia). GFP-trap beads were provided by the University of Queensland Protein Expression Facility. pCMV-Munc18-1, pCMV-Munc18-1-emGFP, NPY–hPLAP and NPY–mCherry have been described previously (Arunachalam et al., 2008, Tomatis et al., 2013). Munc18-1^P335A and Munc18-1^Δ317–333 were generated using the QuikChange Lightning site-directed mutagenesis kit (Stratagene, Santa Clara, CA, USA) and mutational primers 5′-GATGCTGAAGAAAATGGCCCAGTACCAGAAGG-3′ and 5′-GACTTTTCCTCTAGCAAGAGGATGATGCCCCAGTACCAGAAGGAGC-3′, respectively. All constructs were sequenced in the Australian Genome Research Facility, University of Queensland. Control PC12 cells and Munc18-1/Munc18-2 double knockdown PC12 cells (DKD-PC12) were maintained as described previously (Han et al., 2009) supplemented with 2.5 µg/ml plasmocin (ant-mpt, InvivoGen, San Diego, CA, USA).

Recombinant proteins Sx1a (residues 1–261), Munc18-1^WT or Munc18-1^Δ317–333 used for the ITC experiments were prepared as previously described (Christie et al., 2012; Deák et al., 2009). SNAP25 and VAMP2 were expressed and purified as described previously (Hu et al., 2011).

Pulldown assays were carried out using GST-cleaved Munc18-1^WT and Munc18-1^Δ317–333. His–Sx1a was incubated with Munc18-1^WT or Munc18-1^Δ317–333, at 4°C with 20 µl of TALON™ for 2 hours. Beads were then washed in buffer containing 25 mM Tris-HCl pH 8, 150 mM NaCl, 10% glycerol, 5 mM imidazole, 0.1% Triton X-100 and 2 mM β-mercaptoethanol. Bound protein was analysed by reducing SDS-PAGE stained with Coomassie Blue.

For experiments performed with pre-formed ternary complex, His–Sx1a was incubated with an excess of SNAP25 and VAMP2 overnight at 4°C. Unbound proteins were washed with binding buffer (25 mM Tris-HCl pH 8, 150 mM NaCl, 10% glycerol, 15 mM imidazole, 0.1% Triton X-100 and 2 mM β-mercaptoethanol) and incubated with Munc18-1^WT or Munc18-1^Δ317–333 for up to 2 hours. Unbound protein was removed by washing with 25 mM Tris-HCl pH 8, 150 mM NaCl, 10% glycerol, 5 mM imidazole, 0.1% Triton X-100 and 2 mM β-mercaptoethanol buffer before analysis using SDS-PAGE.

Proteins were buffer exchanged and further purified by gel filtration chromatography using 25 mM HEPES (pH 8.0), 150 mM NaCl, 1 mM dithiothreitol, 10% glycerol (ITC buffer). Isothermal titration calorimetry (ITC) was carried out at 298°K using a MicroCal iTC200 (GE Healthcare, Sydney, NSW, Australia), with 16×2.5 µl injections of 50 µM Sx1a into 5 µM Munc18-1. Integration of the titration curves was performed using ORIGIN software (OriginLab, Northampton, MA, USA) to extract thermodynamic parameters, stoichiometry N, equilibrium association constant K_a ( = K_d⁻¹) and the binding enthalpy ΔH. The Gibbs free energy of binding ΔG was calculated from the relation ΔG = −RTln(K_a) and the binding entropy DS was deduced from the equation (ΔG = ΔH−TΔS). Experiments were performed with protein concentrations well within the recommended range for the c-value (concentration of protein in cell/K_d∼200). Binding parameters were calculated as the average of at least three independent experiments ± standard error of the mean (s.e.m.).

For immunoprecipitation, transfected cells were homogenised in 50 mM Hepes, pH 7.2, 100 mM NaCl, 1 mM EDTA and protease inhibitors (Roche, Sydney, NSW, Australia), and 500 µg aliquots solubilised in 1% Triton X-100 at 4°C for 1 hour. Tagged proteins were immunoprecipitated using GFP-Trap beads and bound proteins eluted and analysed by SDS-PAGE and western blotting. Blots were visualised and quantified using the Li-Cor Odyssey system (Li-Cor Biosciences, Lincoln, NE, USA).

DKD-PC12 cells were transfected with Munc18-1 constructs and immunolabelled as described previously (Martin et al., 2009) after permeabilisation using 0.1% Triton X-100. Cells were imaged using a Zeiss LSM510 confocal microscope. Sx1a plasma membrane labelling was quantified using ImageJ. Images were compiled using Adobe Photoshop CS3.

Transfected PC12-DKD cells were visualised with a TIRF microscope (Marianas, SDC Everest™, Intelligent Imaging Innovations Inc., Denver, CO, USA) fitted with a 100×oil immersion objective (NA = 1.46, Carl Zeiss Pty Ltd, Sydney, NSW, Australia) using an EMCCD camera (QuantEM 512sc, Photometrics, Tucson, AZ, USA) and Slidebook software (version 4.2.0.27-33) for image acquisition. Cells were bathed in Buffer A (145 mM NaCl, 5 mM KCl, 1.2 mM Na₂HPO₄, 10 mM D-glucose, 20 mM Hepes, pH 7.4) before the addition of 2 mM BaCl, and a time-lapse movie using simultaneous acquisition was captured at 2 frames/second for the indicated period. The density of NPY–mCherry-labelled SGs was measured using Imaris (version 7.31, Bitplane AG, Zurich, Switzerland).

PC12-DKD cells were co-transfected with NPY–hPLAP plasmid and the indicated Munc18-1 plasmid for 72 h. Cells were washed and incubated with PSS buffer as a control (145 mM NaCl, 5.6 mM KCl, 2.2 mM CaCl₂, 0.5 mM MgCl₂, 5.6 mM glucose, 15 mM Hepes-NaOH, pH 7.4) or stimulated with high K⁺-PSS buffer (81 mM NaCl, 70 mM KCl, 2.2 mM CaCl₂, 0.5 mM MgCl₂, 5.6 mM glucose, 15 mM Hepes–NaOH, pH 7.4) for 15 minutes at 37°C. After collecting the supernatants, cells were lysed with 0.2% Triton X-100 in order to measure total NPY–hPLAP. The released and total NPY–hPLAP were measured using the Phospha-Light™ chemiluminescent reporter gene assay system (Applied Biosystems, Melbourne, VIC, Australia), according to the manufacturer's instructions, and the results expressed as a percentage of total NPY–hPLAP.

Unless otherwise stated, statistical significance was determined using an unpaired, two-tailed, Student's t-test assuming equal variance. Results shown are means ± s.e.m., unless otherwise stated.

The authors thank Rowan Tweedale for critical reading of the manuscript.