Introduction

Hormones, neurotransmitters and other signalling molecules are packaged and stored in secretory vesicles within cells. Exocytosis is the process by which pre-packaged intracellular vesicles fuse with the plasma membrane to release all or only some of their contents to the extracellular space. Upon cell stimulation, they move towards the plasma membrane, fuse with it and release their contents through a narrow and transient fusion pore.1, 2, 3 The fusion pore allows for the ability to finely regulate both the rate and amount of released cargo. Vesicles are then retrieved via endocytosis to be re-used for further secretion, either by refilling or by direct reuse. This highly regulated process occurs in response to acute physiological stimuli, resulting in the tight coupling of cellular stimulation to graded hormone/neurotransmitter secretion.

The simplest model on the nature of the fusion pore is that it is a channel composed of protein and lipid, involving two pore states. The early fusion pore forms a narrow neck and often remains in an unstable state where the pore is thought to be able to ‘flicker’ open and closed for a variable period of time.4 The pore is then thought to expand and be maintained in a stable open form or it closes (collapses). If the pore closes the vesicle returns to the cytosol, then only partly depleted of transmitter in a process termed kiss-and-run exocytosis. If instead fusion pore expansion (FPE) occurs, then it typically leads to full collapse fusion of the vesicle into the plasma membrane, accompanied by a larger increase in transmitter release of its entire contents.

A consequence of two fusion pore states is two modes of endocytosis. Endocytosis after full collapse fusion is well characterised, but vesicle endocytosis after kiss-and-run occurs directly after fusion pore closure, and cannot readily be separated from exocytosis.3 Exocytosis and endocytosis are tightly coupled through sharing of the same proteins. For example, the exocytic proteins synaptotagmin-1, SNAP-25, syntaxin1A, SCAMP-1 and voltage-activated calcium channels are also required for endocytosis.3, 5, 6, 7, 8, 9 Conversely, some classical endocytic proteins such as AP180 or intersectin are required for exocytosis.5, 10, 11

Dynamin and its major binding partner syndapin also appear to couple exo- and endocytosis through their ability to control FPE.12, 13, 14 Dynamin is a mechanochemical enzyme that can self-assemble to single rings (26–32 copies of the protein) in the absence of templates like lipids, nanotubes or microtubules, or it can be stimulated by these templates to assemble as long helices (hundreds to thousands of proteins) around the templates. Dynamin oligomerisation is responsible for stimulating its GTPase activity.15, 16 Dynamin oligomerises in vitro and in cells into helices around negatively charged lipids present at the neck of budding endocytic vesicles. Dynamin alone is sufficient to mediate membrane fission at this site, but in some systems it appears to be aided by tension applied to the budding vesicle via coupled actin polymerisation.17 The actin cytoskeleton is also involved in exo- and endocytosis coupling.18, 19, 20 Stabilising F-actin structures increases kiss-and-run events21 and dynamin directly binds short F-actin filaments and promotes their elongation by displacing the actin capping protein gelsolin.22 In turn, F-actin promotes dynamin oligomerisation into rings.22 The acto-myosin II cortical network regulates both the translocation of secretory vesicles to their release sites23 and the fusion pore dynamics21 thereby controlling both catecholamine and neuropeptide release.

Dynamin has a well-characterised role as a regulator of FPE, through which it controls transmitter release, independently of its endocytic role12, 24, 25 or vesicle translocation.23 Dynamin I is enriched in excitable cells such as neurons and both dynamins I and II have been found in chromaffin cells, and have been found to regulate the amount of release12, 24 by acting on FPE.26, 27 Dynamin GTPase activity is required for the bi-directional control of fusion pore development, with mutations that decrease or increase GTPase activity associated with longer or shorter pre-spike foot (PSF) signals, respectively, in amperometry recordings.26 However, the dynamin-1 mutant T141A which was used in this study for its ability to elevate dynamin GTPase activity26 only has moderate, if any, stimulatory effect on dynamin GTPase activity.28 Additionally, the effects of those mutations only on the amperometric PSF (not the amperometric spike itself)26 are at odds with the well-characterised role of dynamin in regulating the activity-dependent increase in quantal release (amperometric spike size).21, 27, 29, 30, 31

The actions of dynamin on the fusion pore are mostly attributed to the dynamin I isoform,24, 32 but there is the potential for dynamin II to also be involved.12 Proteins known to directly bind dynamin also alter FPE; these include F-actin, syndapin and calcineurin.22, 33, 34 Therefore, dynamin may represent one component of a larger protein complex or signalling system that controls FPE.

Previous studies in this area have used genetic approaches to modify dynamin or other proteins likely to modulate FPE. An important issue with these studies is the days or weeks required for transfections, knockdowns or knockouts. During this time, major rearrangements of cellular trafficking are known to occur and off-target proteins are frequently downregulated35 or mislocalised in the cell.36 An alternative approach that avoids these issues is the use of small molecules that rapidly and reversibly affect dynamin function in cells. Like genetic manipulations, small molecules have off-target actions of their own;36 however, the ability to use them in an acute (minutes) and reversible context minimises these issues and they have far better utility in genetically intractable systems. Ideally, genetic and small molecule studies should be combined to strengthen mechanistic conclusions. There are now compounds available that either inhibit37, 38, 39, 40, 41, 42 or activate43 dynamin through effects on specific dynamin domains.38, 42 Two earlier studies have used dynasore to unravel a role of dynamin in FPE.12, 44 Novel compounds with 10–40 times greater potency than dynasore are now widely available38, 42 and in addition to these dynamin inhibitors, the first dynamin activator small molecules have been developed.43 Given that previous work reporting a bi-directional effect of dynamin GTPase on the fusion pore26 used an 'activating' mutant that has an assembly-stimulated GTPase activity 27-fold lower than wild-type dynamin,28 these small molecule dynamin effectors now allow for true bi-directional dynamin modulation to be investigated in FPE with much improved accuracy. We have now used these small molecules to demonstrate that dynamin is a bi-directional modulator of FPE during exocytosis in chromaffin cells, and that this mechanism involves both actin and myosin II.

Materials and Methods

Primary culture and isolation of mouse chromaffin cells

Mice (male, 10-week-old C57Bl/6) were euthanized humanely by gassing with isoflurane and performing a cervical dislocation, as approved by the Flinders University Animal Welfare Committee. Both adrenal glands were removed and stored in a 15-ml falcon tube containing ice-cold Locke’s solution (154 mM NaCl, 5.6 mM KCl, 3.6 mM NaHCO3, 5.6 mM D-Glucose, 5.0 mM HEPES, pH 7.4). The adrenal medulla was dissected away from the adipose tissue and the adrenal cortex and digested in 5 ml of 3 mg ml−1 collagenase type A for three intervals of 15, 10 and 5 min at 37 °C in the shaking water bath. During this time, the suspension was triturated with a pipette. The collagenase was then diluted in cold Locke’s solution and the cells pelleted in a refrigerated centrifuge for 10 min at 4 °C and 400 g. The Locke’s solution was removed and the pellet of tissue was resuspended in 5 ml Dulbecco’s modified Eagle’s medium (DMEM supplemented with 10% (v/v) heat inactivated fetal calf serum, 100 units per ml penicillin and 100 mg ml−1 streptomycin; Invitrogen, Carlsbad, CA, USA) before being pipetted through a 40-μm nylon mesh filter (Sigma-Aldrich, St Louis, MO, USA) to remove connective tissue and centrifuged for 10 min at 4 °C and 400 g. DMEM was removed and the pellet of cells resuspended in 1 ml of fresh DMEM. Cells were plated on plastic tissue culture dishes (Corning Incorporated, Corning, NY, USA) and incubated at 37 °C with 5% CO2 for 3–5 days before experimentation.

Amperometric measurement of catecholamine release from chromaffin cells

Catecholamine release from chromaffin cells was measured using carbon fibre amperometry.45, 46 Krebs solution (140 mM NaCl, 5 mM KCl, 2 mM CaCl2, 1 mM MgCl2, 10 mM glucose, 10 mM Hepes; pH 7.4) was the standard bath solution and the temperature was controlled using an automatic temperature controller at 35–37 °C (TC-344B; Warner Instrument Corporation, Hamden, CT, USA). All solutions were applied to cells using a gravity-facilitated perfusion system. A carbon fibre electrode (5 μm diameter, ProCFE, Dagan Corporation, Minneapolis, MN, USA) was lowered to a few microns above a single chromaffin cell and the peak oxidation current (+800 mV) for catecholamine was applied to the bath under voltage clamp conditions. The cell was stimulated for 60 s with 70 mM K+ Krebs solution. Current signals attributable to catecholamine oxidation were relayed to a patch clamp amplifier (EPC‐7; List Medical GmbH, Darmstadt, Germany) controlled with Pulse software (V8.78; HEKA Electronik GmbH, Lambrecht/Pfalz, Germany), sampled at 10 kHz and low‐pass filtered at 1 kHz.

Amperometry data analysis

The quantitative Pulse data files were converted to Axon Binary Files (ABF Utility, version 2.1, Synaptosoft, Decatur, GA, USA) and secretory spikes were analysed for the period of 60 s stimulation. Cells were selected for analysis only if the spike number was greater than 10 in the stimulation period. Each spike selected in the stimulation period for analysis was required to meet threshold criteria. Only spikes with an amplitude of 10 pA were selected and overlapping events were excluded to avoid distorting the data. PSF and stand-alone foot (SAF) signals were selected for analysis only if they were equal to or greater than 1 ms in duration and equal to or above 4 times the root-mean-squared noise of the baseline. The rise time spike parameter was calculated from 50 to 90% rising phase to eliminate any PSF signals that may skew the data. Spikes that met threshold criteria were used to analyse spike kinetics. The spike values were pooled into their relevant groups and an average taken of each parameter, including amplitude, area, rise time, decay time and half-width. The spike frequency from each cell was calculated in Hz. PSF and SAF signal kinetics were also analysed from pooled recordings and average amplitude, area and duration calculated. The duration of the PSF signal was calculated from the time interval between the onset (when the signal exceeds the peak-to-peak baseline noise) and the end of the PSF. The end of the PSF signal is defined as the inflection point between the PSF signal and the spike. SAF duration was calculated in the same way but the end point was derived when the spike current met the baseline. PSF and SAF amplitude were calculated by subtracting the baseline from the peak of these signals. As the spike data have a non-parametric distribution, spike kinetic values in each data set were evaluated using the Kruskal–Wallis ANOVA test with a Dunn’s post hoc test to evaluate differences across multiple comparisons. Samples sizes were based on a priori data, n values are displayed in text, data were obtained from at least three replicate cell culture preparations. Data sets were considered as statistically significant when P<0.05.

Total internal reflection fluorescence microscopy

Chromaffin cells were isolated from bovine adrenal glands as described previously47, 48 and maintained in DMEM supplemented with 10% serum supreme, 2.5 μg ml−1 fungizone, 50 μg ml−1 gentamycin and 10 mM HEPES on 0.1 mg ml−1 poly-D-lysine-coated culture dishes (MatTek Corporation, Ashland, MA, USA). Cells were transfected using electroporation (Amaxa Basic Nucleofector Kit for primary neurons (Lonza, Basel, Switzerland), Program X-001) and cultured at 37 °C/5% CO2 for at least 24 h before experimentation. Transfected cells were plated onto poly-D-lysine-coated glass-bottomed culture dishes (MatTek Corporation) and processed for imaging 48–72 h post transfection. pCMV-NPY-emerald GFP was provided by S Sugita. For immunofluorescence microscopy, chromaffin cells transfected with NPY-emGFP and seeded on poly-D-lysine-coated coverslips were washed in PBS, fixed with 4% paraformaldehyde/PBS for 15 min, quenched with 50 mM NHCl4/PBS 10 min, washed in PBS, permeabilised in 0.1% TX100/PBS for 10 min and then blocked in 0.2% BSA/0.2% fish skin gelatin/PBS 10 min. Coverslips were incubated at RT with primary antibodies (anti-dynamin II, Abcam Cat# AB30457, Cambridge, MA, USA) for 60 min, washed with PBS, and revealed using an anti-mouse-Alexa647 (Life Technologies Cat# 21236, Grand Island, NY, USA), followed by PBS washing and mounting with Prolong Gold (Invitrogen, Melbourne, Victoria, Australia). Cells were imaged using a Roper Scientific TIRF microscope (Roper Scientific, Martinsried, Germany).23

For live-cell total internal reflection fluorescence (TIRF) analysis, transfected cells on glass-bottomed culture dishes (MatTek) were visualised on a Roper Scientific TIRF microscope with an ILas2 double laser illuminator (Roper Scientific) and fitted with a Nikon CFI Apo TIRF 100 × objective (Nikon Instruments, Melville, NY, USA) using an Evolve512 delta EMCCD camera (Photometrics, Tucson, AZ, USA) and Metamorph software (version 7.7.8, Molecular Devices, Sunnyvale, CA, USA). Cells were bathed in Buffer A and pre-incubated with DMSO or 1 μM Ryngo 20 min before imaging. Cells were stimulated with 100 μM nicotine in the presence of 2 mM CaCl2. Time-lapse TIRF movies were captured at 25 frames per second for 3 min at 37 °C. Fusions events were analysed from the acquired time-lapse images using ImageJ. To classify fusion events, we used a 1-μM2 circular region of interest (ROI) centred on the observed vesicle that encompasses both the vesicle and part of its immediate surrounding and measured changes in fluorescent intensity inside this ROI. A release event leads to a characteristic burst in fluorescence intensity within the ROI. Full vesicle fusions lead to the complete disappearance of a vesicle whereas during kiss-and-run events the NPY-mEmerald-positive vesicle was still observable after a release event. The duration of release events was measured as rise times, that is, the time during which fluorescence intensity increased within the selected ROI during a release event.

Vesicle size was determined from TIRF images of NPY-emGFP transfected cells. All images were background subtracted using ImageJ (rolling ball) and vesicles were detected using the Region Competition Tool of the Mosaic plugin for ImageJ/FIJI, which allows unbiased detection of object within an image.49 From those detected objects, we then obtained the vesicle radii.

Results

We used Ryngo 1-23 (2-cyano-N-{3-[2-cyano-3-(3,4,5-trihydroxyphenyl) acryloylamino] propyl}-3-(3,4,5-trihydroxyphenyl)acrylamide)39, 43 to test whether dynamin activation affects FPE and vesicle release. Ryngo 1-23 stimulates dynamin GTPase activity by promoting its oligomerisation in vitro and in cells.43 Chromaffin cells were pre-incubated with or without Ryngo 1-23 for 30 min. The release of catecholamines from chromaffin cells in response to 70 mM K+ was monitored as amperometric spikes (Figure 1a). Cells incubated in Ryngo 1-23 maintain stimulus-induced exocytosis (Figure 1b) with the area of each spike appearing to increase (Figures 1a and b, inset). The average number of release events was unaffected (Figure 1c) but a significant increase in the average spike charge (this reflects spike area) for cells treated with either 0.5 μM (806±47.7 fC) or 1 μM Ryngo 1-23 (1200±86.9 fC) compared with control cells (494.3±19.5 fC, P<0.001, Figure 1d) was observed, demonstrating an increase in the amount released per fusion event. Spike amplitude was unaffected by Ryngo 1-23 (Figure 1e), while rise time, decay time and half width were all increased at both concentrations tested (P<0.01). The duration of these events is significantly longer in the presence of Ryngo 1-23 (Figure 1f). The frequency distribution of spike charge in cells treated with either 0.5 μM (P<0.001) or 1 μM (P<0.001) Ryngo 1-23 shifted to the right, indicating dynamin activation caused a greater proportion of release events in which more catecholamine is released (that is, increased spike charge, Figure 1g).

Figure 1
figure 1

Dynamin activation increases vesicle release during exocytosis. (a) Stimulation of chromaffin cells with 70 mM KCl triggers catecholamine release, with typical spike size observed (inset) by amperometry. (b) Stimulated secretion also occurs in the presence of the dynamin activator, Ryngo 1-23 (1 μM), with the size of single spikes typically being larger (inset). Spike frequency (c), charge (d), amplitude (e) and duration (f) are also shown. (g) Relative charge distribution is shifted rightward, indicating more large release events with Ryngo 1-23 (***P<0.001 for both concentrations). The number of cells from which recordings were taken is shown in each bar in (c). Number of spikes in (d–g)=866 (control), 439 (0.5 μM) and 308 (1 μM). Data in bar graphs represent mean±s.e.m. **P<0.01, ***P<0.001 vs control. ##P<0.05 vs Ryngo 1-23 0.5 μM. Scale bar in (a and b) represents 200 pA and 5 s and 50pA and 0.5 s in inset.

PowerPoint slide

Full size image

We then tested two, chemically distinct, dynamin inhibiting small molecules, Dynole 34-2 and Dyngo 4a, on FPE. Dynole 34-2 inhibits dynamin39 and promotes disassembly of dynamin helices, while Dyngo 4a inhibits dynamin without disrupting its oligomerisation into helices around lipid templates.40 Both compounds are up to 40-fold more potent than dynasore.39, 40 Chromaffin cells were pre-incubated with or without Dyngo 4a or Dynole 34-2 for 30 min. Stimulus-evoked exocytosis was still observed in the presence of either inhibitor (Figures 2a and b), but individual release events appeared smaller (Figures 2a and b, inset). Both Dyngo 4a (Figure 2c) and Dynole 34-2 (Figure 2d) reduced the number of release events. Catecholamine release per fusion event was also reduced (Figures 2e and f). Charge distribution shifted to the left with either dynamin inhibitor (Figure 2g), illustrating a greater proportion of small release events. Dyngo 4a and Dynole 34-2 also reduced spike amplitude (Figure 2h), duration (Figure 2i), half-width, rise time and decay time (data not shown). The incidence of PSF signals was reduced (Figure 2j) to such an extent that so few events occurred preventing their analysis.

Figure 2
figure 2

Dynamin inhibition decreases vesicle release during exocytosis. (a) Catecholamine release occurs in the presence of either Dyngo 4a or (b) Dynole 34-2; however, the typical spike size observed by amperometry is reduced (inset). Spike frequency is reduced by both Dyngo 4a (c) and Dynole 34-2 (d), as is spike charge (e and f). (g) Relative charge distribution is shifted leftward (1 μM shown for both indicated compounds), indicating more small release events with both dynamin inhibitors (***P<0.001 for both drugs). Spike amplitude (h), duration (i) and PSF frequency (j) are all reduced by Dyngo 4a and Dynole 34-2 (0.5 μM shown for both indicated compounds, total of 546 pre-spike foot (PSF) in control, 13 in Dyngo 4a and 5 in Dynole 34-2). Number of cells from which recordings were taken is shown in each bar in (c and d). Number of spikes for Dynole 34-2=623 (control), 470 (0.1 μM) and 171 (0.5 μM) and for Dyngo 4a=778 (control), 559 (0.1 μM) and 198 (0.5 μM). **P<0.01, ***P<0.001 vs control. Scale bar in (a and b) represents 50 pA and 5 s and 50 pA and 0.5 s in inset.

PowerPoint slide

Full size image

We utilised TIRF microscopy to visualise single vesicle release events from chromaffin cells transfected with fluorescence-labelled NPY (as -NPY-emerald GFP or NPY-emGFP), which is packaged into secretory vesicles and released upon secretagogue stimulation.50 We tested whether we could detect dynamin associated with NPY-emGFP labelled vesicles in bovine chromaffin cells. To this end, chromaffin cells expressing NPY-emGFP were immunolabelled for dynamin II (Figure 3a). Indeed, we observed clear colocalization between NPY-emGFP and dynamin II structures, suggesting that the NPY-emGFP-positive vesicles were capable of recruiting dynamin II in good agreement with previous reports.12, 51

Figure 3
figure 3

Total internal reflection fluorescence (TIRF) analysis reveals dynamin activation slows fusion pore opening. (a) TIRF images of a fixed chromaffin cell labelled for dynamin II and neuropeptide Y (NPY). The merged image is shown, with the box representing the far right image (scale=5 μM). Arrowheads indicate vesicles containing dynamin. (b) Schematic representation of our analysis approach of live-cell TIRF data, enabling us to measure the rate of vesicle release. (c) Time lapse TIRF images of the fast full fusion release of a single NPY vesicle. Right: Measuring the NPY fluorescence using the method in (b) provides kinetics information on the rate of NPY release through the fusion pore. We also observe (d) slower full fusion events as well as (e) partial release of NPY representing kiss-and-run exocytosis. (f) Analysis of NPY vesicle radii demonstrates no change in vesicle size (n=171 control, 182 Ryngo 1-23). (g) In the presence of Ryngo 1-23, there is a reduced number of kiss-and-run events. (h) The kinetics of full fusion events becomes slower. Number of cells=8 in control and 8 in Ryngo 1-23. *P<0.05, **P<0.01, Student’s t-test.

PowerPoint slide

Full size image

To measure single exocytosis events, we placed a 1-μM2 sized circular ROI over each vesicle and measured changes in NPY-emGFP fluorescence intensity within those ROIs. Stimulus-evoked (nicotine, 100 μM) single exocytosis events were identified as a characteristic short burst of NPY fluorescence due to NPY being released from the vesicle followed by a complete or partial disappearance of the fluorescent vesicle (Figure 3b). The kinetics of these events was classified as three modes: fast full fusion events (Figure 3c, Supplementary Movie 1), slower full fusion events (Figure 3d, Supplementary Movie 2) or partial release of NPY via kiss-and-run exocytosis (Figure 3e, Supplementary Movie 3).

To analyse fusion kinetics, we opted to measure the rise time of NPY-emGFP-fluorescence in an ROI larger than the actual vesicle (1 μm2) during release events. We first attempted to measure release decay times, that is, measure the loss in fluorescence intensity during the time in which fluorescent material is released from the vesicle using an ROI that was exactly the size of a vesicle. However, due to pH-dependent NPY-emGFP deprotonation and NPY-emGFP molecules coming into closer proximity to the TIR-plane during fusion, fluorescence intensity measured within such an ROI can increase or plateau transiently during release and thus delay the fluorescent decay associated with the loss of released material. We therefore looked for a more accurate method to determine release kinetics. We found that the time during which fluorescence intensity increases due to the aforementioned effects can be utilised as a useful measurement when we included a portion outside the vesicle into our measurements. This allowed us to record a transient increase in fluorescent intensity during vesicular release (Figure 3b) resulting from deacidification of the vesicular lumen and the diffusion of NPY towards the TIR plane. The rise time in the selected ROI (Figure 3b) provided very reliable data for both full fusion and kiss-and-run events and was not influenced by the diffusion time of NPY-emGFP molecules (D=200 μm2 s−1)1 away from the vesicle (Supplementary Figure 1B). We also tested ROIs that surrounded the vesicle as an annulus but found no further improvement to our previous measurement (Supplementary Figure 1A).

To determine whether the different release times could potentially stem from altered vesicle sizes resulting from Ryngo 1-23 treatment, we analysed the radii of vesicles using TIRF microscopy. Ryngo 1-23 (1 μM) had no effect on the distribution of vesicle size compared with control cells (control 122±4 nm vs Ryngo 1-23 119±3 nm, Figure 3f). An equal number of full fusion and kiss-and-run events were observed in control cells while Ryngo 1-23 reduced the number of kiss-and-run events (Figure 3g) but not full fusion events. The time taken for the fluorescence surrounding each NPY vesicle to peak was analysed to identify whether release rate is altered by Ryngo 1-23. This was unchanged in kiss-and-run events (data not shown) but prolonged for full fusion events (Figure 3h). Thus, dynamin activation with Ryngo 1-23 shifts exocytosis from kiss-and-run to full fusion and slows vesicle release during full fusion exocytosis.

Changes in kiss-and-run kinetics may be difficult to assess in our TIRF experiments due to the limitations associated with the temporal resolution of image capture (40 ms) and the faster speed of these transient fusion events compared with full fusion events. Therefore to further examine whether Ryngo 1-23 alters kiss-and-run kinetics, we examined SAF signals from our amperometry experiments. SAF signals constitute a foot signal that returns to baseline before an amperometric spike is observed due to re-closure of the fusion pore (Figure 4a).52 In amperometric experiments, SAF signals strictly represent slow kiss-and-run events.53 We analysed these events and find the average SAF number per recording is reduced in the presence of Ryngo 1-23 (Figure 4b) as is the relative number of these events compared with events containing a spike in each recording (Figure 4c). The SAF kinetics including amplitude (Figure 4d), duration (Figure 4e) and area (Figure 4f) are all increased by Ryngo 1-23. Thus, activation of dynamin with Ryngo 1-23 reduces the frequency of slow kiss-and-run events, and those remaining events have a prolonged duration. We observed no SAF signals in cells treated with either Dyngo 4a or Dynole 34-2 (P<0.01 for either drug vs control).

Figure 4
figure 4

Kiss-and-run exocytosis is regulated by dynamin. (a) An example stand-alone foot (SAF) signal. Scale=10 ms and 10 pA. (b) Ryngo 1-23 reduces the number of SAF signals and (c) the ratio of SAF events to full spikes. Ryngo 1-23 also increased the (d) amplitude, (e) duration and (f) area of SAF signals. *P<0.05, **P<0.01, ***P<0.001 vs control. Number of SAF=171 (control), 57 (0.5 μM) and 34 (1 μM). Data represent mean±s.e.m.

PowerPoint slide

Full size image

Ryngo 1-23-stimulated dynamin activity and oligomerisation also promotes dynamin binding to F-actin and actin elongation.43 To test whether the action of Ryngo 1-23 on vesicle release involves actin, cells were pre-incubated for 30 min with two actin polymerisation inhibitors, cytochalasin B (0.5 μM, Figure 5a) or latrunculin B (0.1 μM, Figure 5b). Both abolished the ability of Ryngo 1-23 (0.5 μM) to increase vesicle release. Since actin interacts with myosin II to affect the fusion pore in these cells,54 the effect of inhibiting myosin II activity was also tested. Pre-incubation for 30 min in the myosin II inhibitor, blebbistatin (0.5 μM) increased the amount released per fusion event by 32% (P<0.001). Blebbistatin also blocked any effect of Ryngo 1-23 on vesicle release (Figure 5c). This was further substantiated by the lack of shift in charge distribution by Ryngo 1-23 in the presence of cytochalasin B (Figure 5d), latrunculin B (Figure 5e) or blebbistatin (Figure 5f).

Figure 5
figure 5

Dynamin regulates the fusion pore via the actin/myosin network. Inhibition of actin polymerisation with either (a) cytochalasin B (control n=8 cells, 635 spikes; cytochalasin B n=9 cells, 355 spikes; Cyt+Ryngo 1-23 n=10 cells, 591 spikes) or (b) latrunculin B (control n=8 cells, 498 spikes; Latrunculin B n=9 cells, 516 spikes; LatB+Ryngo 1-23 n=13 cells, 710 spikes) blocks the increase in spike charge caused by Ryngo 1-23. (c) The same result is seen with the myosin II inhibitor blebbistatin (control n=5 cells, 443 spikes; Blebbistatin n =6 cells, 216 spikes; Blebbi+Ryngo 1-23 n=10 cells, 423 spikes). (d) Cytochalasin B, (e) latrunculin B and (f) blebbistatin block the rightward sift in spike charge distribution caused by Ryngo 1-23. ***P<0.001 blebbistatin vs control.

PowerPoint slide

Full size image

The PSF signal (Figure 6a), which represents the release of catecholamine through an early, small, unstable and flickering fusion pore,52 was measured to examine the effect of dynamin activation on the early fusion pore. Both PSF amplitude (Figure 6b) and duration (Figure 6c) were increased by Ryngo 1-23 and the distribution of PSF charge shifted to the right (Figure 6d). This indicates the early fusion pore is wider and remains in an unstable state longer, when dynamin is activated. The rate of early FPE is estimated from the PSF duration, reflecting neurotransmitter efflux from the vesicle through a narrow flickering fusion pore.55 Ryngo 1-23 caused more events with a longer fusion pore dilation (Figure 6e) and this effect was blocked when either actin (Figures 6f and g) or myosin II (Figure 6h) was inhibited. Myosin II blockade alone caused a significant rightward shift in this distribution (P<0.01). Thus, dynamin regulation of the early fusion pore is also dependent on actin and myosin.

Figure 6
figure 6

Dynamin regulates the pre-spike foot (PSF) in an actin- and myosin-dependent manner. (a) An example amperometric spike with the red section indicating the start and finish of a PSF. Scale=20 ms and 100 pA. PSF amplitude (b) and duration (c) are also increased by Ryngo 1-23, causing a rightward sift in the relative PSF charge distribution (d). More exocytosis events display a longer PSF duration in the presence of 0.5 μM (red) or 1 μM (blue) Ryngo 1-23 (e). This shift is blocked by cytochalasin B (f), latrunculin B (g) or blebbistatin (h). In (f–h) dashed line is drug alone and solid line is drug and Ryngo 1-23 (1 μM). *P<0.05, ***P<0.001 vs control and in (d), (e) and (h) represents both drug groups vs control.

PowerPoint slide

Full size image

Discussion

The results of this study show that dynamin bi-directionally regulates transmitter release from vesicles by acting on FPE through a mechanism involving actin and myosin II. We have utilised small molecule effectors of dynamin applied in an acute manner to demonstrate that dynamin has a critical role in expanding and stabilising the fusion pore thereby controlling the amount of transmitter released per vesicle during exocytosis.

Our study provides the first demonstration that a suite of small molecules acting via distinct mechanisms on a single target protein can be used to unravel the bi-directionally control the exocytosis fusion pore by dynamin. The amperometry results support a role of dynamin GTPase activity and actin binding ability in regulating the fusion pore and transmitter release. In chromaffin cells treated with the chemically unrelated dynamin inhibitors Dynole 34-2 or Dyngo 4a, significant reductions in vesicle release were observed. In contrast, dynamin activation with a small molecule, Ryngo 1-23, known to activate GTPase activity, actin binding and dynamin oligomerisation,43 resulted in a significant increase in catecholamine released per fusion event. TIRF data further demonstrated that Ryngo 1-23 reduced the frequency of kiss-and-run fusion and prolonging the time taken for full fusion to be completed. The former result is further consolidated by our finding that the frequency of SAF signals is also reduced. SAF signals unequivocally represent kiss-and-run exocytosis, as the fusion pore closes before an amperometric spike is observed.52 We cannot claim that any event displaying an amperometric spike is definitely a full fusion event, because if all events containing a spike were full fusion events, Ryngo 1-23 would not be able to augment quantal release as it does. However, SAF signals do represent a subset of kiss-and-run events, likely to be due to prolonged pore flickering and re-closure.52 Additionally, our analysis of vesicle size using TIRF demonstrates that dynamin is associated with these vesicles near the plasma membrane and that the effects we observed are not due to alterations in vesicle size by these dynamin effectors. Given the intravesicular concentration of catecholamines within chromaffin vesicle is constant,56 this lack of change in vesicle size indicates that fusion of different sized vesicles does not explain our data.

Our PSF duration, SAF duration and spike duration data indicate that FPE slows as dynamin GTPase activity, ring stabilisation and actin polymerisation increase, to enable more catecholamine to be released per exocytosis event. The fact that two chemically distinct inhibitors of dynamin GTPase activity have the opposite effect on release kinetics to Ryngo 1-23 strengthens the notion that these chemical effects are specific to dynamin function. This is further strengthened by the ability of both actin and myosin II antagonists to block the Ryngo 1-23 influence on vesicle release. These data agree with previously described roles of both dynamin27, 29, 30, 31 and actin/myosin II21, 23 in the regulation of activity-dependent increases in catecholamine release. Additionally, inhibition of myosin II in our experiments caused a moderate increase in spike charge. Thus, not only does actin and myosin II underlie the mechanism by which Ryngo 1-23 affects vesicle release, myosin II may additionally affect the rate of catecholamine extrusion independently of this pathway. The mechanisms by which myosin II can affect the fusion pore independently of actin and dynamin remain unknown.

The TIRF data demonstrate that Ryngo 1-23 reduces the number of kiss-and-run events and slows the kinetics of full fusion, but not kiss-and-run, exocytosis. As Ryngo 1-23 augments the amount released per event in our amperometric recordings, the reduction in kiss-and-run frequency may be due to a shift towards full vesicle fusion, resulting in less filled vesicles being available for repeated kiss-and-run. The slower rate of NPY release during full fusion events could potentially be caused by either a more narrow fusion pore or a slowing of the FPE rate. The increased PSF duration caused by Ryngo 1-23 supports the latter explanation. Additionally, increased PSF and SAF amplitude, a marker of increased fusion pore width, was observed in the presence of Ryngo 1-23. A previous report that PSF duration increases proportionally with increasing dynamin GTPase activity26 should be interpreted with some caution. The 'activating' mutant used (T141A) has only a moderate (from 105±47 to 185±55 per minute) increases in GTPase activity and this increase was not statistically confirmed.28 Furthermore, a T141Q mutant has similar Km and Vmax values to wild-type dynamin and inhibits GTPase activity in COS-7 fibroblasts.57 Thus, that approach seemed based on the use of a poorly characterised and unconfirmed dynamin construct.

Dynamin oligomerization is known to be regulated by direct dynamin–actin interactions. Ryngo 1-23 promotes the intrinsic propensity of dynamin to oligomerize into rings in an actin-dependent manner43 and is thus an ideal reagent to study the novel role of actin-dependent dynamin oligomerization within physiological pathways. Actin-dependent dynamin oligomerization is also dependent on dynamin's ability to bind and hydrolyse GTP,43 an effect blocked by Dynole 34-2.39 We propose that the promotion of dynamin GTPase activity, dynamin oligomerisation and ring formation and actin polymerisation by Ryngo 1-23, all act together to slow FPE and stabilise the fusion pore. This then results in increased pore open time, which would explain the increased amperometric spike charge and the reduced incidence of kiss-and-run events observed in the TIRF experiments. Dynamin activation by Ryngo 1-23 also slows FPE, as suggested by the longer PSF duration and the longer rise times of full fusion events in both amperometry and TIRF experiments. The much faster spikes observed with both Dyngo 4a and Dynole 34-2 indicate that dynamin GTPase inhibition may cause the fusion pore to be unstable and collapse soon after opening. Conversely, the increased GTPase activity and dynamin ring stabilisation in the presence of Ryngo 1-23 appears to stabilise and enlarge the fusion pore to allow increased vesicle release. As the rise time of kiss-and-run events is not affected by Ryngo 1-23 in the TIRF experiments, it does not appear that dynamin is involved in these transient events. This further indicates that different mechanisms may underlie full fusion and transient kiss-and-run as previously suggested.46

The dynamin-binding proteins actin, the SNARE proteins, syndapin, calcineurin and synaptotagmin all regulate the fusion pore.22, 33, 34 In chromaffin cells, the dephosphorylation of dynamin I by the Ca2+ and calmodulin-dependent protein phosphatase, calcineurin, recruits the BAR domain protein syndapin 3 to promote FPE and increases catecholamine release upon high intensity stimulation.31 Our data utilising small molecule effectors are in agreement with such a role, and dynamin may represent one component of a larger protein complex that bi-directionally controls FPE and stabilisation. Dynamin-dependent increases in vesicle release occur dynamically in response to increased stimulation intensity.30, 31 Such changes are far too rapid to be due to increased vesicle loading, require dynamin phosphorylation31 and are due to altered fusion pore opening.30 One alternative explanation for this, and the effect we see with Ryngo 1-23, is that more large vesicles undergo fusion to produce increased vesicle release. However, the mechanism underlying such an alternative theory is difficult to envisage, and such an alternative seems unlikely given that two completely different dynamin-activating pathways produce the same outcome on FPE and vesicle release. Thus, we believe the effect on vesicle release observed by these dynamin effector molecules to be due to the known role of dynamin as a regulator of the fusion pore.

Thus, by using a pharmacological approach to acutely modulate dynamin function, we have demonstrated multiple roles of the dynamin GTPase domain in FPE and that dynamin is unique in that it bi-directionally regulates FPE and vesicle release during exocytosis. Increasing GTPase activity and dynamin oligomerisation stabilises the fusion pore and increases the incidence of full fusion events. This GTPase activity and dynamin oligomerisation is also essential to avoid fusion pore collapse soon after it has developed. Actin polymerisation and myosin II activity are clearly required to enable dynamin to stabilise the fusion pore and increase vesicle release. These compounds represent the first suite of chemical tools targeted at a single protein that will enable researchers to examine the bi-directional control of the fusion pore and quantal release in a multitude of cell types and species. Such a straightforward chemical biology strategy can be readily transferred to multiple cell types to modulate cell signalling and to develop a better understanding of the role of dynamin within the brain and other physiological systems. Furthermore, because the fusion pore can limit the release of vesicle contents, such as neurotransmitters and peptide hormones, these compounds hold some promise in the development of therapeutic drugs across an array of human health disorders linked to both the nervous system and beyond.