fnins-13-01047 October 1, 2019 Time: 16:31 # 1

ORIGINAL RESEARCH published: 02 October 2019 doi: 10.3389/fnins.2019.01047

Differential Release of Exocytosis Marker Dyes Indicates Stimulation-Dependent Regulation of Synaptic Activity

Andreas W. Henkel1*†, Abdeslam Mouihate1 and Oliver Welzel2

1 Department of Physiology, Faculty of Medicine, Kuwait University, Kuwait City, Kuwait, 2 Department of Psychiatry and Psychotherapy, University Hospital Erlangen, Erlangen, Germany

There is a general consensus that synaptic vesicular release by a full collapse process is the primary machinery of synaptic transmission. However, competing view suggests that synaptic vesicular release operates via a kiss-and-run mechanism. By monitoring the release dynamics of a synaptic vesicular marker, FM1-43 from individual synapses Edited by: Dominique Massotte, in hippocampal , we found evidence that the release of was CNRS UPR 3212 Institut des delayed by several seconds after the start of field stimulation. This phenomenon was Neurosciences Cellulaires et associated with modified opening kinetics of fusion pores. Detailed analysis revealed Intégratives (INCI), France that some synapses were completely inactive for a few seconds after stimulation, Reviewed by: John J. Wagner, despite immediate calcium influx. This delay in vesicular release was modulated by University of Georgia, United States various stimulation protocols and different frequencies, indicating an activity-dependent Min-Yu Sun, Washington University in St. Louis, regulation mechanism for exocytosis. Staurosporine, a drug known United States to induce “kiss-and-run” exocytosis, increased the proportion of delayed synapses as *Correspondence: well as the delay duration, while fluoxetine acted contrarily. Besides being a serotonin Andreas W. Henkel reuptake inhibitor, it directly enhanced vesicle mobilization and reduced synaptic fatigue. [email protected] Exocytosis was never delayed, when it was monitored with pH-sensitive probes, †ORCID: Andreas W. Henkel synaptopHlourin and αSyt-CypHerE5 antibody, indicating an instantaneous formation orcid.org/0000-0002-2627-0215 of a fusion pore that allowed rapid equilibration of vesicular lumenal pH but prevented FM1-43 release because of its slow dissociation from the inner vesicular membrane. Specialty section: This article was submitted to Our observations suggest that synapses operate via a sequential “kiss-and-run” and Neuropharmacology, “full-collapse” exocytosis mechanism. The initially narrow vesicular pore allows the a section of the journal Frontiers in Neuroscience equilibration of intravesicular pH which then progresses toward full fusion, causing Received: 10 July 2019 FM1-43 release. Accepted: 18 September 2019 Keywords: hippocampal neurons, exocytosis, fusion pore, FM1-43, kiss-and-run Published: 02 October 2019 Citation: Henkel AW, Mouihate A and INTRODUCTION Welzel O (2019) Differential Release of Exocytosis Marker Dyes Indicates Stimulation-Dependent Regulation The predominant neurotransmitter exocytosis mechanism “kiss-and-run” or “full-collapse fusion” of Synaptic Activity. (Heuser and Reese, 1973; Ceccarelli and Hurlbut, 1980) in the brain is still under debate, despite Front. Neurosci. 13:1047. a wealth of evidence collected for both models in many different cell- and vesicle types (Albillos doi: 10.3389/fnins.2019.01047 et al., 1997; Schikorski and Stevens, 2001; Richards et al., 2005). While it is widely accepted that

Frontiers in Neuroscience| www.frontiersin.org 1 October 2019| Volume 13| Article 1047 fnins-13-01047 October 1, 2019 Time: 16:31 # 2

Henkel et al. Delayed Release of FM1-43

both mechanisms occur in mast- and neuro-endocrine cells as exo-endocytosis mechanism (“classical exocytosis” or “kiss-and well as large dense core vesicles (Elhamdani et al., 2006; Xia et al., run”) was operating at synapses under different conditions 2009; Balseiro-Gomez et al., 2016), either as separate mechanisms (Henkel and Betz, 1995; Klingauf et al., 1998; Henkel et al., or in combinations of both, their functional significance for CNS- 2000). We had previously shown that fluoxetine could overcome synapses with small synaptic vesicles is much less conclusive synaptic fatigue after excessive stimulation (Henkel et al., 2010). (Harata et al., 2006; Granseth et al., 2009; Zhang et al., 2009; But since the mechanism remained unknown, we used a Rudling et al., 2018). selection of activity-dependent staining techniques and drugs to It is not clear yet, whether the variation in the process investigate variability, timing, and kinetic mechanisms of exo- of vesicular release is due to the difference in types of and endocytosis in individually visualized synapses in order to synapses (GABAergic, glutamatergic, etc.), difference in the experimentally document a consistent model of vesicle fusion. regulation of release of two and more transmitters from a single synapse (Jonas et al., 1998), or the change in the exocytosis kinetics of a specific synapse class (Gan and Watanabe, 2018). MATERIALS AND METHODS Many research groups have shown that information transfer between neurons can be modulated by regulating the release of Experiments and Animals presynaptic neurotransmitter and or by affecting postsynaptic Experiments were conducted on primary cell cultures, obtained receptor densities, leading to such phenomena as LTP and LTD from 2- to 4-day-old Wistar rats (supplied by the Animal (Castillo, 2012). These features enable a dynamic processing of Resource Center in Kuwait and the Franz-Penzoldt-Zentrum, information, reaching far beyond classical linear information Erlangen, Germany) of both sexes, at the laboratories of transfer and rather may form the basis of complex cognitive the Psychiatric University Clinic in Erlangen, Germany, the functions (Fernandes and Carvalho, 2016). We have observed Research Core Facility at the Kuwait University and the that individual synapses in hippocampal cultures exhibit a laboratories of the Department of Physiology, Kuwait University. large heterogeneity of release kinetics that may add a temporal We decided to use neonatal preparation with fully component to the synaptic information processing mechanism functional synapses because it was not possible to obtain healthy (Henkel, 2015). functional synapses from neurons of adult rats. All animal Individual synapses, even of the same class, are able to adjust work conducted in Erlangen was approved by the Kollegiales neurotransmission not only quantitatively but also temporally. Leitungsgremium of the Franz-Penzoldt Zentrum, Erlangen, The mechanisms underlying highly dynamic variation of synaptic Germany, in accordance with the Animal Protection Law of the vesicle release have been shown to be linked to different vesicular Federal Republic of Germany and Animal care while all handling pools, variable modes of mobilization, docking access, calcium procedures conducted in Kuwait complied with standards of the levels, protein phosphorylation (Petrov et al., 2015), and a variety International Council of Laboratory Animals Sciences and were of endocytosis mechanisms. approved by the Kuwait University Research Administration A detailed view on the exocytosis of individual vesicles Ethics Committee. originates mostly from electrophysiological experiments on large granules, since their fusion process could directly be monitored Cell Culture with “cell-attached” capacitance measurements (Albillos et al., Neonatal hippocampal neurons and associated glial cells were 1997; Henkel et al., 2000). Smaller vesicles (<50 nm in diameter) cultured as described (Klingauf et al., 1998; Welzel et al., 2011) are much harder to analyze in detail by electrophysiological with minor modifications. In brief, newborn rats were sacrificed techniques due to technical limitations. One exception is the large by decapitation, whole hippocampi were removed from the brain, synapse “Calyx of Held,” where single vesicle fusions have been transferred into ice-cold Hank’s salt solution, and cleaned from resolved and analyzed (He et al., 2006). adhesive tissue and blood vessels. After digestion with trypsin Much of our knowledge was compiled from indirect exo- and (5 mg/mL), cells were triturated in a glass pipette, and plated endocytosis analyses with fluorescent activity-dependent dyes in minimal essential medium (MEM), supplemented with 10% (Betz et al., 1992; Harata et al., 2006). These optical techniques fetal calf serum and 2% B27 Supplement (all from Invitrogen, allowed the simultaneous monitoring of many synapses (200– Karlsruhe, Germany) on 18 mm glass coverslips, coated 1500 synaptic buttons per image frame) in hippocampal cultures with Matrigel (BD Biosciences, San Jose, CA, United States). (Klingauf et al., 1998). The synapses do not necessarily belong to The coverslips were placed in 24-well plates and the plates a homogenous neurotransmitter class but were rather composed were wrapped with cling wrap to maintain iso-osmolarity of different synapse types, most of them being glutamatergic (320 mOsmol). The wrapping allowed diffusion of gases like CO2 (Megias et al., 2001). We addressed the question of whether the but retained evaporated water. Antibiotic/antimycotic (Gibco, release kinetics of FM1-43 in a selected individual synapse was Grant Island, United States) solution was added. Cells were used constant or dynamic under changing stimulation conditions. In between 20 and 30 days in culture. particular, we focused on the mechanistic kinetic process that affected the delay of dye release in individual synapses under Transfection With SynaptopHlourin increasingly stronger stimulation and concurrent application To monitor synaptic vesicle exocytosis by measuring the collapse of fluoxetine or staurosporine (Henkel, 2015). The non- of the vesicular proton gradient, neurons were transfected with specific kinase inhibitor staurosporine was used to test which synaptopHlourin, a pH-sensitive GFP variant that was fused to

Frontiers in Neuroscience| www.frontiersin.org 2 October 2019| Volume 13| Article 1047 fnins-13-01047 October 1, 2019 Time: 16:31 # 3

Henkel et al. Delayed Release of FM1-43

the lumenal domain of synaptobrevin (VAMP), controlled with a (Nikon, Melville, United States), using a Nikon Cy5 filter set synapsin promoter (Sankaranarayanan et al., 2000). Transfection (excitation: 604–644 nm, emission: 672–712 nm). was done on the 3rd day in vitro with a modified calcium phosphate technique, described elsewhere (Threadgill et al., 1997; Calcium Measurement Welzel et al., 2010). The experiments were conducted on the 25th day in vitro. Cells, grown on 18 mm cover slips, were loaded with 5 µM Ca-GreenAM (Thermo Fisher Scientific, Waltham, MA, FM1-43 Loading, Destaining, and United States) or 1 µM Fura-2AM (Calbiochem- now Merck, Imaging Darmstadt, Germany), dissolved in Modified Eagle’s Medium Cells, attached to 18 mm cover slips, were transferred to a (MEM) for 90 min at 37◦C. Cells were then transferred to an stimulation chamber, containing rat Ringer’s solution (in mM): electrical stimulation chamber, mounted on the microscope stage, NaCl 145, KCl 4, MgCl2 2.5, CaCl2 2.5, glucose 10, and HEPES and stimulated with trains of bidirectional 1 ms pulses at 30 Hz 10, pH 7.2, 300 mOsmol, mounted on the microscope stage for 30 s. The Ca-GreenAM signal was imaged by a conventional and experiments were conducted for up to 45 min at room FITC filter set (490 nm excitation/520 nm emission) while Fura- temperature 22–24◦C. The experimental protocol was employed 2 fluorescence was monitored using filter set with 380 nm for activity-dependent vesicle staining and subsequent measuring excitation and 520 nm emission. No quantitative 340/380 nm of exocytosis. Synapses were loaded by electrical stimuli trains ratio-metric measurement was used for Fura-2 excitation. at 30 Hz for 20–30 s with 2.5 µM FM1-43 (Thermo Fisher Scientific, Waltham, MA, United States), dissolved in the Ringer’s Image Quantification and Data Analysis solution as previously described (Henkel et al., 2010). Stimulation Images were processed and analyzed using a package of self- was administered through two parallel platinum wires, spaced written programs for image processing and data analysis: “Image 1 cm apart. Supplementary Figure S1 depicts the extended Processing and Data Analyzing Software” (IPDAS, SynoSoft, experimental “sequential loading procedure,” used in some Kuwait) (Henkel et al., 2014; Mouihate et al., 2019). experiments. The preparation was stimulated with 1 ms bipolar At first, fluorescent peaks corresponding to synapses and non- current pulses (50 mA) at 30 Hz for 30 s, in the presence of FM1- synaptic artifact spots were detected and their intensity course 43 for loading synaptic vesicles. The solution containing FM1-43 was quantified in all frames by ImageStar V. 2.18 software, was immediately removed, briefly rinsed, and washed six times SynoSoft. The peak detection algorithm excluded hazy areas for 1 min. The preparation was illuminated for 30 s with 30 flashes and overlapping or odd-shaped spots, but included inactive (not for 250 ms each, to attenuate strong initial bleaching. Destaining destaining) spots. Spot detection parameters included a general was triggered by stimulation at 30 Hz for 20 s. Images were taken intensity threshold, a shape parameter fit, a local background from 10 s before the start of the stimulation to the end at 1 detection threshold, and a peak density-minimum-distance image/s, 250 ms exposition time. In the case of sequential staining threshold, all build in ImageStar. procedures, the staining and destaining sequence was repeated as The individual fluorescence intensity traces from all these exemplified in Supplementary Figure S1. Drugs [staurosporine, spots were stored and further analyzed by the self-written 2 µM (Calbiochem, United States); fluoxetine, 10 µM (Sigma- program “Exocytosis Analysis,” SynoExA, Ver. 1.36. SynoExA Aldrich, United States)] were either applied 30–60 min before the selected only fluorescence traces that met minimum data quality experiments or in between the “sequential loading procedure” for criteria, which included noise level (rms noise < 1.5), trace course 10 min as indicated and were present in all subsequent washing (exclusion of brighter growing and wave-like traces), and bleach and stimulation media. In control experiments, drugs were correction, derived from pre- and post-stimulation intensity omitted and experiments were conducted as described above. fluorescence sections. A trace was defined as synaptic, when the difference between pre- and post-stimulation regression lines at the mid-stimulation period was >0.1 dF/F of the initial Labeling Synapses With αSyt1-CypHerE5 max fluorescence (Fmax). These regression lines were also used to Antibody calculate the fluorescence loss during stimulation. Fluorescence αSyt1-CypHer5E (Synaptic Systems, Göttingen, Germay) is an change of each individual spot was quantified from the start to the antibody against the vesicle-specific protein synaptotagmin, end of the stimulation period and were expressed as normalized coupled to a modified, pH−sensitive derivative of Cy5 (Adie change (dF/Fmax)(Henkel et al., 2010). Synapses, whose dF/Fmax et al., 2002) that permanently labels the lumen of recycled was larger than 0.35 were defined as very active synapses (VAS). synaptic vesicles. In contrast to synaptopHlourin, its fluorescence Delay release time (Dt) was defined as the time difference is quenched at pH ∼7.2 and becomes brighter once transferred between start of stimulation (Stim0) and the onset of dye inside synaptic vesicles at pH < 5.5. Synapses were loaded with release on low-pass filtered synaptic traces. The threshold for αSyt1-CypHer5E (50 µg/ml) by incubation in high K+-Ringer’s release onset was arbitrarily defined as the time, when the initial ◦ (40 mM) for 10 min at 37 C, washed two times with normal fluorescence (dF/Fmax) dropped below 4% of its pre-stimulation Ringer’s, and were quickly imaged thereafter. The preparation value. A synapse was designated as “delayed,” if its fluorescence could be stimulated several times at 30 Hz, since the fluorescence decrease was <4% within 5 s after stimulation onset. decay recovered after endocytosis and re-acidification of vesicles. The software was also used to select traces according to Fluorescence was monitored on a TI-Eclipse inverted microscope defined kinetic properties and assigned them to specific synaptic

Frontiers in Neuroscience| www.frontiersin.org 3 October 2019| Volume 13| Article 1047 fnins-13-01047 October 1, 2019 Time: 16:31 # 4

Henkel et al. Delayed Release of FM1-43

classes. Threshold settings for selection parameter were defined acceptance level was set to p < 0.05. To assess the effect size of an by the user and indicated in the section “Results.” Only traces experimental intervention, Cohen’s D was calculated to compare derived from spots without increasing intensities, which were experimental groups. The effects size was determined to: below a minimum noise level of 1.5 rms and showed sufficient 0 < 0.2 “no effect,” 0.2–0.3 “small effect," around 0.5 “medium trace steadiness (no wave-like traces), were included in the effect,” >0.8 “large effect” (Cohen, 1988). data analysis. Table 1 lists the available class selection criteria, Categorical data for two population proportions, like implemented into SynoExA. differences between groups, expressed as percentages, were tested Several kinetic parameters were determined in each of the for significance with the Z-test, where the significance acceptance synapse classes to characterize class-specific release properties. level was set to p < 0.05. Table 2 lists all available parameters.

Statistics RESULTS Statistical analysis was performed with SynoStat 1.23 (SynoSoft) and Microsoft Excel 2016 (Microsoft). Kolmogorov–Smirnov test Kinetically Different Synapse Classes in and Shapiro–Wilk test were used to check if the data followed a Hippocampal Preparations normal distribution. One-way ANOVA analyses were performed The kinetics of FM1-43 release from individual synaptic boutons if more than two conditions were compared. appeared to be highly variable although the average of several Results from single data series were expressed as hundreds of fluorescence traces generally followed a single or mean ± standard deviation. Results, derived from data series, second-order exponential decay function, which was the result of collected from repeated parallel experiments were expressed mutual compensatory effects of fast and slow destaining synapses. as weighted mean (x¯ ) ± weighted standard deviation (sd ), w w Destaining patterns of individual synapses were analyzed in to account for different synapse numbers in independent high detail and three main kinetic classes were defined, namely: experiments. Weights (w ) were set according to their relative i (1) “Classical synapses” showed a dye release kinetic that contribution (number of synapses) in individual experiments; followed a single exponential decay function, (2) “Two-speed n = number of experiments; m = number of non-zero weights. synapses,” characterized by a double exponential decay function, Pn and (3) “Linear synapses” exhibiting a constant release rate. i=1 wixi x¯w = n A special type of synapses was discovered and defined as P w i=1 i “Delayed synapse” (DelSyn), whose release kinetics belonged sPn 2 to either of the classes’ (1–3), but whose onset of release wi(xi −x ¯i) sd = i=1 was considerably delayed after the start of the stimulation. w Pn (m−1) i=1 wi m Representative examples of the three defined synaptic kinetic classes are shown in Figure 1A, while Figure 1B presented an Normally distributed data were tested for significant differences example of a DelSyn. The depicted traces were averaged from between groups with Student’s t-test, where the significance individual synapses, extracted from the original fluorescent spot database by SynoExA. The initial challenge for this study was the reliability for detection of synapses from all fluorescently FM1- TABLE 1 | Synaptic class definitions. 43 stained objects, a task that was conducted by “ImageStar”

1. Active synapses, mono-exponential decay, [dF/Fmax], >0.1 = active Ver. 2.18. Figure 1C shows a typical hippocampal preparation synapses, [dF/Fmax], >0.35 = VAS. after an activity-dependent FM1-43 staining, marking all detected 2. Delayed synapses, onset of fluorescence release by stimulation [Dt]. round or oval shaped spots that laid outside high background 3. Fast destaining synapses. areas. The intensity traces derived from all detected spots were 4. Linear synapses, defined by constant linear release rate. separately analyzed, as depicted in the left panel of Figure 1D, 5. Two-speed-component synapses with double exponential decay function. and traces were defined as “VAS,” if they followed a large fluorescence decrease (dF/Fmax > 0.35) during the stimulation TABLE 2 | Kinetic parameter for synapse characterization. period (right panel).

Initial fluorescence intensity (Fmax) (range: 0–255) A Subpopulation of VAS Shows Delayed Absolute fluorescence release (dF) (range: 0–255) Release of FM1-43 Relative fluorescence release (dF/Fmax) (range: 0–1) Figure 2A Half-maximal release time (tau50) (s) shows that fluorescent spots, after background

Fast release, defined by maximal release velocity (Vmax) (dF) subtraction by a deconvolution filter, exhibited a wide variability Fast release, defined by initial release slope in morphological shapes and fluorescence intensities. The spot

Time of Vmax after stimulation start (t−Vmax) (s) traces were also broadly different with regard to release velocity, Fluorescence release delay after stimulation start (s) release quantity, kinetic course, and onset of destaining. Traces Linearity of release (linear fit regression coefficient, linear regression of regular and DelSyns could be clearly distinguished, when correlation > 0.997) their traces were directly overlaid (inset in Figure 2A). DelSyns Percentage of active synapses (%) were extracted from a dataset of VAS (n = 242, from 10

Frontiers in Neuroscience| www.frontiersin.org 4 October 2019| Volume 13| Article 1047 fnins-13-01047 October 1, 2019 Time: 16:31 # 5

Henkel et al. Delayed Release of FM1-43

FIGURE 1 | (A) Three examples of synaptic classes, selected by SynoExa according to specified kinetic courses and fitting functions were overlaid onto mean traces during the stimulation period. Left panel: “Classical synapses” derived from n = 453 individual synapses, r = 0.997; middle panel: “Two-speed synapses,” n = 342, tau1: r = 0.998, tau2: r = 0.999; and right panel: “Linear synapses,” n = 126, r = 0.999. (B) Pseudo-colored image series (imaging rate: 1 Hz) shows a single delayed synapse that started destaining 3 s after start of stimulation (red bar). (C) Hippocampal synapses, stained with FM1-43. Individual spot intensities were detected and extracted with ImageStar (red box markers), and stored in a database together with their x–y positions. (D) Fluorescent traces, derived from individual spots, normalized to start of stimulation, overlaid onto the mean trace (black–red–black). Left panel, Traces from all identified spots; right panel, selected traces from synapses (VAS) with >0.35 dF/Fmax FM1-43 release.

experiments, dF/Fmax > 0.35) and analyzed in detail. It was exponential decay function to a 10 s section after the start of found that the distribution of Dt could be fitted by a Gaussian fluorescence decay. bell curve with a weighted mean of 4.75 ± 1.52 s, ranging Spatial analysis revealed that fluorescent spots were not evenly from 1 to 10 s). When Dt was measured on all synaptic traces, distributed but formed chain-like structures and large clusters on Dt distribution was shifted to about half this value, while the cell aggregates (Supplementary Figure S2). Although DelSyns VAS subset showed a reduction of long Dts, as demonstrated were mostly concentrated in clusters, they were sometimes seen in Figure 2B. Figure 2C shows typical examples of individual close to fast destaining synapses. High resolution images of two synaptic fluorescence traces with increasing Dt and how Dt was such closely adjacent synapses are shown in Figure 2D. The measured and visualized at the intersection of a linear regression delayed synapse (#2) showed almost no destaining after 5 s, when line that was fitted to the pre-stimulation section and a single its faster neighbor (#1) had already released >50% of its initial

Frontiers in Neuroscience| www.frontiersin.org 5 October 2019| Volume 13| Article 1047 fnins-13-01047 October 1, 2019 Time: 16:31 # 6

Henkel et al. Delayed Release of FM1-43

FIGURE 2 | (A) Fluorescent spots from Figure 1C after background subtraction. The traces from two labeled synapses are shown in the inset. Inset: classical synapse (1) and a delayed onset synapse (2). (B) Histogram of first-order derivatives of mean traces indicate Dt distribution of presented kinetic subpopulations as indicated by color. (C) Four panels showing individual synaptic fluorescence traces with increasing Dt and corresponding single exponential fits to trace sections after the start of FM1-43 release. Pre-stimulation sections were fitted with linear regression lines. (D) Comparison of two neighboring synapses with different Dt. The graph shows overlaid traces from regions, indicated in the images.

intensity. The release kinetics were very similar when destaining S1 and described in detail previously (Henkel et al., 2010). finally set off, 6 s after stimulation start. During the second cycle, control preparations showed that the proportion of DelSyns was significantly increased, while Repetitive FM1-43 Loading and the proportion of VAS was significantly reduced (p < 0.01) (Figure 3A). The same reduction was also observed in Unloading Modifies the Relative double – exponential – and fast synapses. In order to test Proportion Between Kinetic Classes whether these changes were related to pathological intra-synaptic We investigated whether specific synaptic classes remained mechanisms, fluoxetine was applied between the first and the kinetically stable after repeated staining and destaining cycles, second stimulation cycles. Fluoxetine, a classical SSRI-class anti- a procedure that subjected them to considerable stress. The depression drug, was found to reduce excitotoxicity and to experimental protocol was presented in Supplementary Figure increase the mobilization of synaptic vesicles (Kim et al., 2013;

Frontiers in Neuroscience| www.frontiersin.org 6 October 2019| Volume 13| Article 1047 fnins-13-01047 October 1, 2019 Time: 16:31 # 7

Henkel et al. Delayed Release of FM1-43

FIGURE 3 | (A) Quantification of specified kinetic synapse classes in two subsequent staining/destaining cycles and effect of fluoxetine on class change in between both stimulation cycles (9 control experiments, 10,569 synapses; 11 fluoxetine experiments, and 11,299 synapses). ANOVA analysis indicated significant (p < 0.001) differences between the three groups in all five experimental setups. (B) Green color indicates staining of synapses with FM1-43, while red color shows direct subsequent stimulation-dependent destaining. Two single synapses (arrows) shows no destaining during the 1st stimulation cycle but recovered, and destained in the 2nd cycle. (C) Correlation plot of FM1-43 release between the 1st and the 2nd staining/destaining cycles (n = 424 matched synapses). (D) Direct comparison of individual synapses in subsequent staining/destaining cycles. Upper traces show the course of fluorescence and lower traces their corresponding first-order derivatives. First cycle (red), 2nd cycle (cyan), and stimulation period (red bar). The image pairs and the traces were directly obtained with SynCom Ver. 1.61, software. (E,F) The pie-charts display the differences between two successive staining/destaining cycles when no drug (control) or 10 µM fluoxetine was administered between the cycles, control n = 424, fluoxetine n = 387 synapses. (E) Differences in FM1-43 dye release from synapses between 1st and 2nd stimulation cycle. (F) Variances in the percentage of synapses with delayed release onset (DelSys) between 1st and 2nd stimulation cycle.

Jung et al., 2014). Application of fluoxetine reversed and even synapses (DelSyns and “linear synapses”) was significantly over-compensated these kinetic changes. Figure 3A shows that reduced by >50% (p < 0.01). Figure 3B shows an example of two fluoxetine enhanced the fraction of both fast releasing- and synapses that did not destain during the first stimulation cycle, double exponential synapses, despite the repetitive staining– but became active during the second cycle after the application of destaining cycle. The relative proportion of slow releasing fluoxetine. These observations suggest that individual synapses

Frontiers in Neuroscience| www.frontiersin.org 7 October 2019| Volume 13| Article 1047 fnins-13-01047 October 1, 2019 Time: 16:31 # 8

Henkel et al. Delayed Release of FM1-43

may change their release kinetics in response to external stimuli Delayed Onset of Exocytosis Is Not or pharmacological interventions. When the release quantity Caused by Delayed Calcium Entry was compared between both cycles, it was found that d / F Fmax Calcium has be shown to regulate the speed and magnitude of was a little (14.7%), but significantly ( < 0.01) reduced p exocytosis (Neher and Sakaba, 2008); therefore, we investigated and poorly correlated ( = 0.66). The low correlation points r whether the delay in FM1-43 release was associated with to a high variability of release activity within an individual a delayed calcium entry. We conducted experiments on synapse, rather than toward an activation of previously silent hippocampal neurons that were preloaded with either Ca-Green synapses (Figure 3C). AM or Fura-2 A, two well-known calcium sensors. Figure 4A Next, we checked whether this variability was mostly due shows a typical preparation that was stained with Ca-green AM to a change in the kinetic release mode of the same synapse. and subjected to two successive stimulation cycles. None of the Therefore, individual synapses were localized and matched active synapses (n = 465) showed a delay after the start of between the first and the second stimulation cycles and their the stimulation within our time resolution of 1 s (Figure 4B). release kinetics were directly compared. Figure 3D shows typical Similarly, synapses loaded with Fura-2 AM showed an invariably traces from three individual synapses that were identified and sudden onset of synaptic calcium influx that started to decay matched between two successive staining/destaining cycles (first even during the course of the stimulation (Figure 4C). These cycle in red, second cycle in cyan). The two lower traces observations excluded the possibility that a delayed calcium represented the filtered first-order derivatives of the upper ones. entry could account for the observed delayed release of FM1- The left panel shows a fast responding synapse that largely 43 in a sub-population of synapses. It is noteworthy that our retained its release kinetics during the second cycle. The middle experimental setup did not allow the measurement of calcium in panel shows a synapse that showed a delayed onset of release micro-domains inside synapses which could have accounted for during the second cycle, while the synapse on the right panel high localized calcium concentrations. showed a fast-responding one during the second cycle. The proportion of dye release was very variable between two successive stimulation cycles. Less than a third of all synapses Staurosporine Increases the Proportion showed identical release proportions (dF/Fmax) in both cycles of Delayed Synapses in High-Releasing while almost two-thirds showed reduced exocytosis (Figure 3E). Synapses A few synapses (∼10%) released more dye in the second cycle, We then addressed the question of whether a modulated opening indicating a facilitation effect. Treatment with fluoxetine before kinetics of the fusion pore could cause the observed delay the 2nd cycle caused synaptic facilitation in about a third of of FM1-43 release. There is evidence in several exocytosis all synapses, while far fewer synapses reduced release compared model systems that the non-specific protein kinase inhibitor to untreated synapses during the 2nd cycle. About 40% of the staurosporine inhibits fusion pore expansion and prevents vesicle synapses performed the same in both cycles. collapse into the plasma membrane (Klingauf et al., 1998; Figure 3F shows that the onset of release (Dt) was delayed Henkel et al., 2001; Rudling et al., 2018). FM1-43 staining of in some synapses to a variable degree. Although the majority staurosporine-treated preparations was indistinguishable from (65%) of active individual synapses didn’t change their release controls (Cohen’s D = 0.14). Staurosporine led to a significant kinetics between consecutive staining/destaining cycles, we found reduction of dye release from active synapses (16.1%, Cohen’s deviations when synapses were classified according to Dt. In D = 4.8) and resulted in a significantly slower half time of order to clearly separate groups, synapses were considered as release (tau50 = 10.0%) (Cohen’s D = 1.07). Furthermore, “Changed,” if their individual Dt – change between the 1st staurosporine treatment significantly increased the proportion of and the 2nd cycle was longer than 2 s. Fluoxetine significantly delayed synapses by 58.6%, when active synapses with release increased the proportion of unchanged synapses when it was rates >0.2 dF/Fmax were analyzed (Figure 4D). This result compared to controls (Zscore = −6.55, p = 0.01). Thus we tested was even more pronounced, when all synapses (including low if individual DelSyns retained their prolonged Dts during the quantity releasing synapses) were included (data not shown). The second stimulation cycle. When we compared active synapses weighted mean delay time (Dt) was also 61.7% slower (Cohen’s with a delay of 3 or more seconds, we found that the vast D = 2.22) in staurosporine-treated synapses, compared to majority (>90%) remained unchanged or decreased the delay controls (Figure 4E). These observations support the hypothesis times, despite the fact that the total number (Figure 3A) of that inhibition of fusion pore expansion might be the underlying DelSyns was increased in the second cycle. This indicated that mechanism of delayed FM1-43 release. previously fast responding synapses showed a delayed exocytosis onset. Fluoxetine retained the delay time between both cycles Synaptic Vesicle Exocytosis Is Not to the largest extent and reduced the proportion of delay- Delayed in SynaptopHlourin-Transfected changing synapses distinctively. Dts, measured in DelSyns during the second stimulation cycle, were slightly reduced in response Neurons to fluoxetine. However, this reduction was not statistically The use of pH-sensitive dyes provides direct visual evidence significant (control: Dt = 4.48 ± 2.05, n = 172; fluoxetine: on the existence of a hydrophilic connection (fusion pore) Dt = 4.1 ± 1.23, n = 56). The size effect was also small (Cohen’s between the extracellular space and the vesicular lumen D = 0.226). (Sankaranarayanan and Ryan, 2000). Thus, we transiently

Frontiers in Neuroscience| www.frontiersin.org 8 October 2019| Volume 13| Article 1047 fnins-13-01047 October 1, 2019 Time: 16:31 # 9

Henkel et al. Delayed Release of FM1-43

FIGURE 4 | (A) Synapses loaded with calcium sensor Ca-green AM. (B) Traces from five regions in panel A, green bars: electrical field stimulation. (C) Synapses loaded with calcium sensor Fura-2 AM, traces are from single synapses in the corresponding images, red bar indicates electrical field stimulation. (D) Quantification of staurosporine (SSP)-treated preparations, percentage of active synapses (dF/Fmax > 0.2), labeled with FM1-43 labeled, control: n = 3758, from five independent experiments, SSP: n = 4865, from six independent experiments. (E) Corresponding weighted mean delay times. Asterisks indicate a significant difference between SSP and control (p < 0.01).

transfected hippocampal neurons with the pH-sensor the escape of FM1-43 from the vesicle. Transfected neurons synaptopHlourin, to test whether the delayed release and FM1-43-stained preparations were stimulated in seven phenomenon was due to a transient arrest of exocytosis or independent experiments. All traces were normalized, since if a narrow fusion pore was formed, which could prevent the synaptopHlourin signal increase was much weaker than

Frontiers in Neuroscience| www.frontiersin.org 9 October 2019| Volume 13| Article 1047 fnins-13-01047 October 1, 2019 Time: 16:31 # 10

Henkel et al. Delayed Release of FM1-43

FIGURE 5 | (A) Normalized averaged synaptic exocytosis traces from active synapses. Normalization was performed by stretching each individual trace between the start image and the first image after the end of stimulation to neutralize brightness variability and enhance the kinetic course. The red bar indicates electrical field stimulation at 30 Hz and the black vertical line the 5th second after the end of stimulation. Left panel: DelSyns, (FM1-43), subsequent panels as indicated: synaptopHlourin transfected synapses (Syn-pH), synaptopHlourin transfected synapses incubated with staurosporine (Syn-pH + SSP), and αSyt1-CypHerE5 labeled synapses (CypHer5E). (B) Quantification of weighted mean delay times [FM1-43: n = 1994 synapses (nine experiments), Syn-pH: n = 354 (six experiments), Syn-pH + SSP: n = 398 (five experiments)]. Syn-pH values were below the noise level in the automated SynoExa measurements, ANOVA analysis p < 0.001, the asterisk indicates a significant difference (p < 0.01) between FM1-43 and all other experimental conditions. (C) Second-order derivative traces (long-pass filtered) of VAS fluorescence, color-coded as indicated in the graph. The arrows point to the time positions of maximal speed change that correspond to the onset of exocytosis (Dt). (D) Percentage of synapse classes at stimulation with three different frequencies: independent experiments at: 10 Hz n = 5, 20 Hz n = 5, 30 Hz n = 4; number of synapses at 10 Hz n = 2066, 20 Hz n = 2092, 30 Hz n = 1534 synapses, ANOVA analysis p < 0.001; asterisks indicate statistically significant difference (Student’s t-test, ∗∗p < 0.01, ∗p < 0.05). (E) Weighted mean delay times from active synapses (five independent experiments for each frequency) at indicated stimulation frequencies showed a significant increase of Dt at 10 Hz (p < 0.01) and at 20 Hz (p < 0.05) compared to 30 Hz.

Frontiers in Neuroscience| www.frontiersin.org 10 October 2019| Volume 13| Article 1047 fnins-13-01047 October 1, 2019 Time: 16:31 # 11

Henkel et al. Delayed Release of FM1-43

FIGURE 6 | Mechanistic models of exocytosis monitoring. (A) FM1-43 labels exclusively recycled vesicles and its fluorescence destains only after full fusion but not upon fusion pore opening. (B) CypHerE5 permanently labels exclusively recycled vesicles and its fluorescence is transiently quenched upon fusion pore opening and restored after endocytosis and re-acidification of the vesicle. Recycled vesicle may become part of other functional pools upon repetitive stimulation. (C) SynaptopHlourin (Syn-pHluorin) labels all vesicles and reports fusion pore openings by increased fluorescence. Endocytosis quenches fluorescence reversible after re-acidification of the vesicle. (D) Proposed models of exocytosis from an individual vesicle: neurotransmitter is slowly released through a narrow fusion pore under the premises that it is partially sequestered onto a intravesicular matrix, FM1-43 is not released through a narrow pore, CypHerE5 (darkening) and SynaptopHlourin (brightening) indicate instant opening of the pore.

the magnitude of FM1-43 destaining. Figure 5A shows fluid occurred instantaneously upon stimulation, most likely normalized synaptic exocytosis traces and their corresponding through a membrane spanning fusion pore. quantification (Figure 5B). Active synapses stained with FM1-43 had a weighted mean Dt of 4.64 ± 1.9, seconds, n = 1994, while synaptopHlourin-transfected synapses showed no apparent delay Recycled Vesicles, Labeled With (Figures 5A,B). Application of staurosporine did not change Syt1-CypHerE5 Show No Delay the instantaneous exocytosis onset in transfected synapses. In order to test whether recycled vesicles were directed into a Second-order derivatives of exocytosis traces, representing transient reserve pool that could potentially prolong a direct re- exocytotic velocity change during the stimulation period, showed use, we permanently labeled them with a pH-sensitive antibody that synaptopHlourin transfected VAS showed an immediate αSyt1-CypHerE5 (CypHer) against the inner vesicular domain response to the stimulus and were distinctively faster than VAS of synaptotagmin. FM1-43 and CypHer shared the same labeling stained with FM1-43 (Figure 5C). These observations suggest mechanism, namely the uptake of the probe during a complete that an intravesicular pH equilibration with the extracellular cycle of exo/endocytosis but recycled vesicle could mix with other

Frontiers in Neuroscience| www.frontiersin.org 11 October 2019| Volume 13| Article 1047 fnins-13-01047 October 1, 2019 Time: 16:31 # 12

Henkel et al. Delayed Release of FM1-43

functional pools upon repetitive stimulation. This experiment the system of signal processing (Sakaba, 2008; Gu et al., 2012; could ascertain whether a transient fusion pore existed that Somogyi et al., 2014). allowed fast equilibration of lumenal vesicular pH. In clear To our knowledge, this is the first report that documented a contrast to observations on FM1-43-stained synapses, none of delayed onset of FM1-43 release of a few seconds in a sizable the active CypHer-labeled synapses showed a delay after the fraction of synapses. Therefore, we had to exclude experimental start of stimulation. This immediate response to stimulation artifacts that may have been imposed by field stimulation was not affected by staurosporine, demonstrating that recycled inhomogeneity or imprecise timing. We did not find evidence vesicles were directly available for reuse (Figures 5A,B). The that the local distribution of delayed synapses was correlated contrast between the dynamic of FM1-43 and CypHer is to their to the strength of the depolarizing field gradient. The increased fundamentally different reporter mechanisms. Indeed, FM1-43 prevalence of delayed synapses on cell clusters and their absence dye can depart from the vesicular membrane only after complete on elongated structures argued strongly against this possibility fusion, while the CypHer dye needs only a transient fusion (Supplementary Figure S2). If a strength gradient would have pore to destain. been present, it would have shown an inhomogeneous synapse distribution depending on the distance to the stimulation Lower Frequency Stimulation Increases electrodes. The presence of DelSyns clusters suggested rather the Percentage and Delay Time in that an unknown local factor was the cause for the delay. Synapses A timing artifact could also be excluded by the observation that occasionally two directly neighboring synapses showed different The effect of stimulation frequency on delayed release of FM1-43 onset of FM1-43 release (Figure 2D). was determined to analyze whether the level of delay (kiss-and- First, we asked whether delayed FM1-43-stained synapses run) and of full collapse are modulated by neuronal activity. belonged to a neurotransmitter-specific type of synapse. We Therefore, synapses were bleach-corrected and normalized matched individual synapses in two consecutive stimulation before analysis. Figure 5D showed a significant increase in the series, measured their release kinetics, and formed two kinetic percentage of delayed synapses and a corresponding decrease groups: (1) “Delayed onset,” D = 3 s and (2) “Fast onset,” D = 1 s. of fast-onset synapses at 10 Hz frequency compared to our t t Even though this procedure underestimated the real number of standard stimulation protocol (30 Hz). Twenty and 30 Hz did changed synapses, the results suggested it was unlikely that a not change the percentage of fast-onset synapses significantly. neurotransmitter-specific synapse was associated with a unique The percentage of active synapses was decreased to about 50% release mode, when about 35% could change their individual at 10 Hz, indicating that the release probability was slowed down D by >2 s, either toward slower or faster release onset. It is compared to higher frequency stimulations. Figure 5E showed t noteworthy that the absolute percentage of synapses depended that the mean delay time was significantly (p < 0.01) longer at on the definition of “delayed onset.” Slightly different timing 10 Hz, compared to both 30 and 20 Hz, while D at 30 Hz was t definitions may result in different absolute values. However, the significantly (p < 0.05) shorter than at 20 Hz. original data were always compared to their respective controls by unbiased computer-based techniques, which provide stronger DISCUSSION validation to our observation. A lower stimulation frequency (10 Hz) increased the Synapses have been classified according to a variety of percentage delayed synapses while increasing their delay neurotransmitter types, release kinetics, and plasticity (Castillo, time, pointing to an activity regulated general mechanism 2012; Emes and Grant, 2012; Nusser, 2018). However, it of neurotransmitter release control. We did not observe a was still unclear whether all synapses of a defined type are correspondingly large effect at 20 Hz compared to 30 Hz functionally homogeneous or whether they exhibit individually stimulation, an indication that stimulation frequency may not be distinct release kinetics. In the present work, we provide directly correlated to the exocytosis mode and could be rather experimental evidence that synapses within a given neuron governed by non-linear calcium concentration (Blank et al., show heterogeneous behaviors. Thus, synapses can fine-tune 2001). Taken together, these observations strongly suggest that at their exo-endocytosis kinetics independent from their immediate lower neuronal activity, synapses routinely work in a slow fusion neighboring synapses. Indeed, individual synaptic activity pore expansion mode and accelerate toward the full fusion mode seemed to be governed by intra-synaptic factors. Since the through a non-linear kinetic function (Wu et al., 2005). experiments were conducted on neurons from newborn rats, it Application of fluoxetine significantly stabilized the release could not be excluded that neurons from adult animals would mode, reduced the proportion of delayed synapses, and increased express a different activity. fast and bi-phasic synapses. This effect could be due to the We observed that prevalent synaptic exocytosis kinetics, like increased size of the synaptic vesicle recycling pool and increased single exponential release functions, were the result of averaging facilitation of exocytosis by fluoxetine (Jung et al., 2014). If our of many synapses, while some individual synapses employed bi- assumption was correct that delayed FM1-43 release onset was phasic or linear release kinetics. It appears that complex synaptic caused by the transient opening of a fusion pore that preceded networks might include a temporal component for information the full fusion event, we could speculate that increased availability transfer and processing which involves dynamic change from fast of vesicles for release shortened this pore transition toward full to slow synapses and vice versa, adding a further dimension to collapse. This, however, was not supported by our results. When

Frontiers in Neuroscience| www.frontiersin.org 12 October 2019| Volume 13| Article 1047 fnins-13-01047 October 1, 2019 Time: 16:31 # 13

Henkel et al. Delayed Release of FM1-43

Dts of delayed synapses were compared between control and synapses never showed a delayed onset of exocytosis, excluding fluoxetine-treated preparation, we found no significant difference a temporary hold of vesicle translocation to the fusion site. It of individual Dts or mean Dts between both groups. Therefore, also supported the hypothesis that a fusion pore was immediately the effect of fluoxetine seems to be not directed onto the fusion formed after stimulation had started. This observation was in pore kinetics itself, but rather to a reduced number of pores, line with the results of experiments where single vesicles were capable of delaying full fusion. Previous studies have shown that labeled with quantum dots, suggesting a prolonged fusion pore fusion pore expansion was mediated by phosphorylation (Scepek connection that preceded full-fusion events of vesicles (Zhang et al., 1998; Fisher et al., 2001). Recent studies provided evidence et al., 2009). This observation was also made on hippocampal that fluoxetine operates by enhancing the kinase activities of large dense core vesicles that released neuropeptides from sites MZeta and Akt/GSK-3beta (Yan et al., 2017; Yi et al., 2018). It is on neuronal somata (Xia et al., 2009). plausible that fluoxetine effect on the release dynamics of synaptic The next test for the temporary fusion pore hypothesis was vesicles seen in the present study is mediated through its kinase- the application of staurosporine that inhibited exocytosis by enhancing activity. This could be investigated in detail in a set of preventing the full collapse vesicle fusion (Henkel and Betz, future experiments where one would use more specific inhibitors 1995; Klingauf et al., 1998; Henkel et al., 2000, 2001). Although for these kinases prior to fluoxetine application. this mechanism has been challenged later in neuromuscular In the next sets of experiments, we tested whether the synapses (Becherer et al., 2001). In our experiments with FM1-43, delayed FM1-43 release could be explained by either activation staurosporine significantly increased the proportion of delayed of different release modes or by shifts in vesicle pool dynamics. synapses and extended the onset of dye release, supporting the FM1-43 was the first widely successful optical probe to monitor idea that a transient fusion pore was the cause of the observed the fusion of synaptic vesicles (Betz et al., 1992) in two delay. This conclusion was supported by the observation that complete consecutive cycles of exo-endocytosis (1st staining – synaptopHlourin (proton-sensitive probe) never showed a delay 2nd destaining cycle). This mechanism labeled exclusively either in the presence or the absence of staurosporine. In this recycled vesicles but excluded vesicles from other pools and model, the vesicles in delayed synapses immediately opened, as newly synthesized ones. Using FM1-43, stimulation-dependent demonstrated in experiments with synaptopHlourin and Cypher. exocytosis was only measurable, when the stained vesicle The pore expanded to full collapse fusion after few seconds, and fused completely with the plasma membrane, but remained released FM1-43. undetectable, if the vesicular lumen was only transiently Many studies have shown consistently that the staurosporine connected to the extracellular space through a fusion pore inhibited intrasynaptic vesicle movement (Kraszewski et al., (Richards et al., 2005). Although the mean FM1-43 release from 1996; Becherer et al., 2001; Jordan et al., 2005). Assuming a stimulated hippocampal neuronal preparation could be fitted that vesicle translocation toward the active zone was restricted with a single exponential decay function (Ratnayaka et al., 2012), by staurosporine, one would not expect a fast response to we observed that this was due to an averaging effect of fast and stimulation as it was always observed with synaptopHlourin and slow destaining synapses. Some individual synapses showed a CypHer, independent of drug application. Movement-restriction complete block of any destaining for up to several seconds after by staurosporine would have prevented the translocation of the start of electrical field stimulation. Calcium measurements, CypHer-labeled recycled vesicles from a distant recycling pool using two different sensor probes, showed consistently an to the active zone. Therefore, it appears likely that the majority instantaneous rise of Ca2+ in all imaged synapses without of recycling took place very closely to the active zones and any delay, excluding the hypothesis that quiescent calcium required a full collapse, followed by endocytosis. This mechanism channels were responsible for the effect. We then investigated is plausible because the uptake of large CypHer antibodies two other possible hypotheses, namely the transient parking of into vesicles through a small fusion pore was very unlikely newly recycled vesicles in a recycling pool that delayed their because of their small diameter (Wu et al., 2016; Chang et al., re-use and the formation of a transient fusion pore, since a 2017). Limitations with regard to relatively non-specific effects full-collapse fusion was a prerequisite for FM1-43 release. To of staurosporine could be overcome by future analysis in investigate the fusion pore hypothesis, cells were transfected synaptotagmin-7 knock-out neurons. Indeed, it has been shown with synaptopHlourin, a pH-sensitive reporter molecule that that this protein may be responsible for repetitive open–close labeled pool-independent all vesicles. These experiments showed cycles of a fusion pore in chromaffin cells. This could prevent full no delayed exocytosis onset in any of the monitored synapses, fusion of synaptic vesicles (Rao et al., 2014). supporting the instantaneous formation of an aqueous channel In this study we used several different fluorescent probes between the vesicular lumen and the extracellular space. We then to propose a mechanistic model that was in line with tested whether lingering in the recycling pool did transiently our observations (Figures 6A–D). While FM1-43 and prevent the utilization of recycled vesicles for subsequent CypHer antibody did exclusively stain recycled vesicles, exocytosis. The recycling vesicle pool was specifically stained synaptopHlourin labeled the whole pool (Betz and Bewick, with Cypher, an antibody against the lumenal domain of the 1992; Sankaranarayanan and Ryan, 2000; Adie et al., 2002). vesicle marker synaptotagmin. In contrast to FM1-43, this CypHer and synaptopHlourin reported “classical exocytosis” and probe reported exocytosis by increased fluorescence upon pH “kiss-and-run” by visualizing a breakdown of an intravesicular increase that occurred when a narrow fusion pore on the plasma pH gradient, while FM1-43 indicated only full collapse fusion membrane was formed. Like synaptopHlourin, CypHer labeled by its dissociation from the vesicle membrane. Since CypHer

Frontiers in Neuroscience| www.frontiersin.org 13 October 2019| Volume 13| Article 1047 fnins-13-01047 October 1, 2019 Time: 16:31 # 14

Henkel et al. Delayed Release of FM1-43

became permanently attached to the inner vesicular membrane Erlangen, Germany, in accordance with the Animal Protection after initial contact, it acted as a permanent reporter for Law of the Federal Republic of Germany and by the Kuwait repetitive vesicle-fusion pore-plasma membrane interactions. University Research Administration Ethics Committee. Finally, inner-synaptic transport pathways between a spatially separated recycling pool and release sites could be monitored by FM1-43 and CypHer. AUTHOR CONTRIBUTIONS Delayed release of FM1-43 in combination with immediate fluorescence loss of CypHer and instantaneous increase of AH designed and performed the experiments. AH and AM synaptopHlourin fluorescence point to the instantaneous coordinated the experiments, analyzed the data, and co-wrote the calcium-triggered formation of a transient fusion pore, followed manuscript. OW designed and conducted the experiments. by full exocytosis and subsequent endocytosis. The staurosporine data support this interpretation if one assumes a general vesicle mobility inhibition by this drug required a local FUNDING recycling mechanism for CypHer. The other proposed action of staurosporine, namely the promotion of “kiss-and-run” release, This work was generously supported by the General Facility explains the appearance of more delayed synapses and extended Grant, Kuwait University, No. SRUL02/13. delay times (Dt). Taken together, our data support the idea that an instantaneous opening of a fusion pore allowed neurotransmitter release and proceeded later toward “full collapse vesicle fusion,” ACKNOWLEDGMENTS followed by endocytosis. The time interval between the first opening and the subsequent full fusion can be regulated Experiments were done in part in the Research Unit for by neuronal activity and kinase-modulating chemicals. Our Genomics, Proteomics and Cellomics Sciences of the HSC suggestion that the differential opening characteristics of a Kuwait University, Kuwait, and in the Laboratories of Molecular fusion pore was the main cause for the observed effects is only Neurobiology, Department of Psychiatry and Psychotherapy, indirect, since we could not study individual vesicle fusion. University Hospital Erlangen, Erlangen, Germany. Our special However, very recently it was observed that individual synaptic thank goes to Prof. Dr. J. Kornhuber and Dr. T. Groemer, who vesicles undergo several kiss-and-run events within 10 s after provided the imaging facilities to parts of our project. We further the start of stimulation and finally fuse completely with the thank Mrs. Anju Devassy and Mrs. Katrin Ebert for their excellent plasma membrane (Qin et al., 2019). These recent findings technical help in all cell culture experiments. are in complete agreement with our mechanistic model and provide further support to our conclusions. In summary, in addition to the classical quantitative regulation of synaptic SUPPLEMENTARY MATERIAL transmission, we provide experimental evidence to support a novel process by which inter-synaptic information flow can be The Supplementary Material for this article can be found temporally modulated. online at: https://www.frontiersin.org/articles/10.3389/fnins. 2019.01047/full#supplementary-material

DATA AVAILABILITY STATEMENT FIGURE S1 | Sequential stain exocytosis assay for hippocampal neurons, modified from Henkel et al.(2010). Hippocampal neurons were stained with The datasets generated for this study are available on request to FM1-43 in the absence of fluoxetine, washed, and pictures were taken during the the corresponding author. period, indicated by dots. Fluoxetine was added (red line) after the first staining/destaining series and remained in the bath during the further rest of the experiment. The staining/destaining procedure was repeated in a second series in ETHICS STATEMENT presence of the drug and pictures were taken. These two resulting image series were subsequently compared and analyzed.

The animal studies were reviewed and approved by the FIGURE S2 | (Top) FM1-43-stained hippocampal synapses. (Bottom) Kollegiales Leitungsgremium of the Franz-Penzoldt Zentrum, Distribution of active synapses (green), delayed active synapses (red).

REFERENCES Balseiro-Gomez, S., Flores, J. A., Acosta, J., Ramirez-Ponce, M. P., and Ales, E. (2016). Transient fusion ensures granule replenishment to enable repeated Adie, E. J., Kalinka, S., Smith, L., Francis, M. J., Marenghi, A., Cooper, M. E., et al. release after IgE-mediated mast cell degranulation. J. Cell Sci. 129, 3989–4000. (2002). A pH-sensitive fluor, CypHer 5, used to monitor agonist-induced G Becherer, U., Guatimosim, C., and Betz, W. (2001). Effects of staurosporine on protein-coupled receptor internalization in live cells. Biotechniques 33, 1156– exocytosis and endocytosis at frog motor nerve terminals. J. Neurosci. 21, 1157. 782–787. doi: 10.1523/jneurosci.21-03-00782.2001 Albillos, A., Dernick, G., Horstmann, H., Almers, W., Alvarez de Toledo, G., and Betz, W. J., and Bewick, G. S. (1992). Optical analysis of synaptic vesicle recycling Lindau, M. (1997). The exocytotic event in chromaffin cells revealed by patch at the frog neuromuscular junction. Science 255, 200–203. doi: 10.1126/science. amperometry. Nature 389, 509–512. doi: 10.1038/39081 1553547

Frontiers in Neuroscience| www.frontiersin.org 14 October 2019| Volume 13| Article 1047 fnins-13-01047 October 1, 2019 Time: 16:31 # 15

Henkel et al. Delayed Release of FM1-43

Betz, W. J., Mao, F., and Bewick, G. S. (1992). Activity-dependent fluorescent Jonas, P., Bischofberger, J., and Sandkuhler, J. (1998). Corelease of two fast staining and destaining of living vertebrate motor nerve terminals. J. Neurosci. at a central synapse. Science 281, 419–424. doi: 10.1126/ 12, 363–375. doi: 10.1523/jneurosci.12-02-00363.1992 science.281.5375.419 Blank, P. S., Vogel, S. S., Malley, J. D., and Zimmerberg, J. (2001). A kinetic Jordan, R., Lemke, E. A., and Klingauf, J. (2005). Visualization of synaptic analysis of calcium-triggered exocytosis. J. Gen. Physiol. 118, 145–156. doi: vesicle movement in intact synaptic boutons using fluorescence fluctuation 10.1085/jgp.118.2.145 spectroscopy. Biophys. J. 89, 2091–2102. doi: 10.1529/biophysj.105.061663 Castillo, P. E. (2012). Presynaptic LTP and LTD of excitatory and inhibitory Jung, J., Loy, K., Schilling, E. M., Röther, M., Brauner, J. M., Huth, T., et al. synapses. Cold Spring Harb Persp. Biol. 4:a005728. doi: 10.1101/cshperspect. (2014). The fluoxetine mobilizes vesicles to the recycling pool a005728 of rat hippocampal synapses during high activity. Mol. Neuro. 49, 916–930. Ceccarelli, B., and Hurlbut, W. P. (1980). Vesicle hypothesis of the release of quanta doi: 10.1007/s12035-013-8569-5 of acetylcholine. Physiol. Rev. 60, 396–441. doi: 10.1152/physrev.1980.60.2.396 Kim, H. J., Kim, T. H., Choi, S. J., Hong, Y. J., Yang, J. S., Sung, K. W., et al. (2013). Chang, C. W., Chiang, C. W., and Jackson, M. B. (2017). Fusion pores and their Fluoxetine suppresses synaptically induced [Ca(2)(+)]i spikes and excitotoxicity control of neurotransmitter and hormone release. J. Gen. Physiol. 149, 301–322. in cultured rat hippocampal neurons. Brain Res. 1490, 23–34. doi: 10.1016/j. doi: 10.1085/jgp.201611724 brainres.2012.10.062 Cohen, J. E. (1988). Estimating the effects of successful malarial control Klingauf, J., Kavalali, E. T., and Tsien, R. W. (1998). Kinetics and regulation of fast programmes on mortality. Popul. Bull. UN 6–26. endocytosis at hippocampal synapses. Nature 394, 581–585. doi: 10.1038/29079 Elhamdani, A., Azizi, F., and Artalejo, C. R. (2006). Double patch clamp reveals Kraszewski, K., Daniell, L., Mundigl, O., and De Camilli, P. (1996). Mobility of that transient fusion (kiss-and-run) is a major mechanism of secretion in calf synaptic vesicles in nerve endings monitored by recovery from photobleaching adrenal chromaffin cells: high calcium shifts the mechanism from kiss-and-run of synaptic vesicle-associated fluorescence. J. Neurosci. 16, 5905–5913. doi: to complete fusion. J. Neurosci. 26, 3030–3036. doi: 10.1523/jneurosci.5275-05. 10.1523/jneurosci.16-19-05905.1996 2006 Megias, M., Emri, Z., Freund, T. F., and Gulyas, A. I. (2001). Total number Emes, R. D., and Grant, S. G. (2012). Evolution of synapse complexity and and distribution of inhibitory and excitatory synapses on hippocampal CA1 diversity. Annu. Rev. Neurosci. 35, 111–131. doi: 10.1146/annurev-neuro- pyramidal cells. Neuroscience 102, 527–540. doi: 10.1016/s0306-4522(00) 062111-150433 00496-6 Fernandes, D., and Carvalho, A. L. (2016). Mechanisms of homeostatic plasticity Mouihate, A., Kalakh, S., AlMutairi, R., and Alashqar, A. (2019). Prenatal in the excitatory synapse. J. Neurochem. 139, 973–996. doi: 10.1111/jnc.13687 inflammation dampens neurogenesis and enhances serotonin transporter Fisher, R. J., Pevsner, J., and Burgoyne, R. D. (2001). Control of fusion pore expression in the of adult female rats. Med. Princ. Pract. 28, dynamics during exocytosis by Munc18. Science 291, 875–878. doi: 10.1126/ 352–360. doi: 10.1159/000499658 science.291.5505.875 Neher, E., and Sakaba, T. (2008). Multiple roles of calcium ions in the regulation of Gan, Q., and Watanabe, S. (2018). Synaptic vesicle endocytosis in different model neurotransmitter release. Neuron 59, 861–872. doi: 10.1016/j.neuron.2008.08. systems. Front. Cell Neurosci. 12:171. doi: 10.3389/fncel.2018.00171 019 Granseth, B., Odermatt, B., Royle, S. J., and Lagnado, L. (2009). Comment on Nusser, Z. (2018). Creating diverse synapses from the same molecules. Curr. Opin. "The dynamic control of kiss-and-run and vesicular reuse probed with single Neurobiol. 51, 8–15. doi: 10.1016/j.conb.2018.01.001 nanoparticles. Science 325:1499. doi: 10.1126/science.1175790 Petrov, A. M., Zakyrjanova, G. F., Yakovleva, A. A., and Zefirov, A. L. (2015). Gu, Z., Lamb, P. W., and Yakel, J. L. (2012). Cholinergic coordination of Inhibition of protein kinase C affects on mode of synaptic vesicle exocytosis presynaptic and postsynaptic activity induces timing-dependent hippocampal due to cholesterol depletion. Biochem. Biophys. Res. Commun. 456, 145–150. . J. Neurosci. 32, 12337–12348. doi: 10.1523/JNEUROSCI. doi: 10.1016/j.bbrc.2014.11.049 2129-12.2012 Qin, X., Tsien, R. W., and Park, H. (2019). Real-time three-dimensional tracking of Harata, N. C., Choi, S., Pyle, J. L., Aravanis, A. M., and Tsien, R. W. (2006). single synaptic vesicles reveals that synaptic vesicles undergoing kiss-and-run Frequency-dependent kinetics and prevalence of kiss-and-run and reuse at fusion remain close to their original fusion site before reuse. Biochem. Biophys. hippocampal synapses studied with novel quenching methods. Neuron 49, Res. Commun. 514, 1004–1008. doi: 10.1016/j.bbrc.2019.05.043 243–256. doi: 10.1016/j.neuron.2005.12.018 Rao, T. C., Passmore, D. R., Peleman, A. R., Das, M., Chapman, E. R., and He, L., Wu, X. S., Mohan, R., and Wu, L. G. (2006). Two modes of fusion pore Anantharam, A. (2014). Distinct fusion properties of synaptotagmin-1 and opening revealed by cell-attached recordings at a synapse. Nature 444, 102–105. synaptotagmin-7 bearing dense core granules. Mol. Biol. Cell. 25, 2416–2427. doi: 10.1038/nature05250 doi: 10.1091/mbc.E14-02-0702 Henkel, A. W. (2015). Analysis of kinetically distinguished synaptic subtypes in Ratnayaka, A., Marra, V., Bush, D., Burden, J. J., Branco, T., and Staras, K. (2012). hippocampal neurons. Biophys. J. 108, 101a–101a. Recruitment of resting vesicles into recycling pools supports NMDA receptor- Henkel, A. W., Alali, H., Devassy, A., Alawadi, M. M., and Redzic, Z. B. (2014). dependent synaptic potentiation in cultured hippocampal neurons. J. Physiol. Antagonistic interactions between dexamethasone and fluoxetine modulate 590, 1585–1597. doi: 10.1113/jphysiol.2011.226688 morphodynamics and expression of cytokines in astrocytes. Neuroscience 280, Richards, D. A., Bai, J., and Chapman, E. R. (2005). Two modes of exocytosis 318–327. doi: 10.1016/j.neuroscience.2014.09.012 at hippocampal synapses revealed by rate of FM1-43 efflux from individual Henkel, A. W., and Betz, W. J. (1995). Staurosporine blocks evoked release of vesicles. J. Cell. Biol. 168, 929–939. doi: 10.1083/jcb.200407148 FM1-43 but not acetylcholine from frog motor nerve terminals. J. Neurosci. 15, Rudling, J. E., Drever, B. D., Reid, B., and Bewick, G. S. (2018). Importance of 8246–8258. doi: 10.1523/jneurosci.15-12-08246.1995 full-collapse vesicle exocytosis for synaptic fatigue-resistance at rat fast and Henkel, A. W., Kang, G., and Kornhuber, J. (2001). A common molecular slow muscle neuromuscular junctions. Int. J. Mol. Sci. 19:E1936. doi: 10.3390/ machinery for exocytosis and the ‘kiss-and-run’ mechanism in chromaffin cells ijms19071936 is controlled by phosphorylation. J. Cell Sci. 114, 4613–4620. Sakaba, T. (2008). Two Ca(2+)-dependent steps controlling synaptic vesicle fusion Henkel, A. W., Meiri, H., Horstmann, H., Lindau, M., and Almers, W. (2000). and replenishment at the cerebellar basket cell terminal. Neuron 57, 406–419. Rhythmic opening and closing of vesicles during constitutive exo- and doi: 10.1016/j.neuron.2007.11.029 endocytosis in chromaffin cells. EMBO J. 19, 84–93. doi: 10.1093/emboj/ Sankaranarayanan, S., De Angelis, D., Rothman, J. E., and Ryan, T. A. (2000). The 19.1.84 use of pHluorins for optical measurements of presynaptic activity. Biophys. J. Henkel, A. W., Welzel, O., Groemer, T. W., Tripal, P., Rotter, A., and Kornhuber, J. 79, 2199–2208. doi: 10.1016/s0006-3495(00)76468-x (2010). Fluoxetine prevents stimulation-dependent fatigue of synaptic vesicle Sankaranarayanan, S., and Ryan, T. A. (2000). Real-time measurements of vesicle- exocytosis in hippocampal neurons. J. Neuro. 114, 697–705. doi: 10.1111/j. SNARE recycling in synapses of the central . Nat. Cell. Biol. 2, 1471-4159.2010.06795.x 197–204. doi: 10.1038/35008615 Heuser, J. E., and Reese, T. S. (1973). Evidence for recycling of synaptic vesicle Scepek, S., Coorssen, J. R., and Lindau, M. (1998). Fusion pore expansion in membrane during transmitter release at the frog neuromuscular junction. J. Cell horse eosinophils is modulated by Ca2+ and protein kinase C via distinct Biol. 57, 315–344. doi: 10.1083/jcb.57.2.315 mechanisms. EMBO J. 17, 4340–4345. doi: 10.1093/emboj/17.15.4340

Frontiers in Neuroscience| www.frontiersin.org 15 October 2019| Volume 13| Article 1047 fnins-13-01047 October 1, 2019 Time: 16:31 # 16

Henkel et al. Delayed Release of FM1-43

Schikorski, T., and Stevens, C. F. (2001). Morphological correlates of functionally Xia, X., Lessmann, V., and Martin, T. F. (2009). Imaging of evoked dense- defined synaptic vesicle populations. Nat. Neurosci. 4, 391–395. doi: 10.1038/ core-vesicle exocytosis in hippocampal neurons reveals long latencies and 86042 kiss-and-run fusion events. J. Cell Sci. 122, 75–82. doi: 10.1242/jcs.03 Somogyi, P., Katona, L., Klausberger, T., Lasztoczi, B., and Viney, T. J. (2014). 4603 Temporal redistribution of inhibition over neuronal subcellular domains Yan, W., Liu, J. F., Han, Y., Zhang, W., Luo, Y. X., Xue, Y. X., et al. (2017). Protein underlies state-dependent rhythmic change of excitability in the hippocampus. kinase Mzeta in medial prefrontal cortex mediates depressive-like behavior Philos Trans. R. Soc. Lond. B Biol. Sci. 369:20120518. doi: 10.1098/rstb.2012. and antidepressant response. Mol. Psychiatry 23, 1878–1891. doi: 10.1038/mp. 0518 2017.219 Threadgill, R., Bobb, K., and Ghosh, A. (1997). Regulation of dendritic growth and Yi, J. H., Zhang, J., Ko, S. Y., Kwon, H., Jeon, S. J., Park, S. J., et al. (2018). Fluoxetine remodeling by Rho. Rac, and Cdc42. Neuron 19, 625–634. doi: 10.1016/s0896- inhibits natural decay of long-term memory via Akt/GSK-3beta signaling. Mol. 6273(00)80376-1 Neurobiol. 55, 7453–7462. doi: 10.1007/s12035-018-0919-x Welzel, O., Henkel, A. W., Stroebel, A. M., Jung, J., Tischbirek, C. H., Ebert, K., Zhang, Q., Li, Y., and Tsien, R. W. (2009). The dynamic control of kiss-and-run et al. (2011). Systematic heterogeneity of fractional vesicle pool sizes and release and vesicular reuse probed with single nanoparticles. Science 323, 1448–1453. rates of hippocampal synapses. Biophys. J. 100, 593–601. doi: 10.1016/j.bpj.2010. doi: 10.1126/science.1167373 12.3706 Welzel, O., Tischbirek, C. H., Jung, J., Kohler, E. M., Svetlitchny, A., Henkel, A. W., Conflict of Interest: The authors declare that the research was conducted in the et al. (2010). Synapse clusters are preferentially formed by synapses with large absence of any commercial or financial relationships that could be construed as a recycling pool sizes. PLoS One 5:e13514. doi: 10.1371/journal.pone.0013514 potential conflict of interest. Wu, W., Xu, J., Wu, X. S., and Wu, L. G. (2005). Activity-dependent acceleration of endocytosis at a central synapse. J. Neurosci. 25, 11676–11683. doi: 10.1523/ Copyright © 2019 Henkel, Mouihate and Welzel. This is an open-access article jneurosci.2972-05.2005 distributed under the terms of the Creative Commons Attribution License (CC BY). Wu, Z., Auclair, S. M., Bello, O., Vennekate, W., Dudzinski, N. R., Krishnakumar, The use, distribution or reproduction in other forums is permitted, provided the S. S., et al. (2016). Nanodisc-cell fusion: control of fusion pore nucleation original author(s) and the copyright owner(s) are credited and that the original and lifetimes by SNARE protein transmembrane domains. Sci. Rep. 6, 27287. publication in this journal is cited, in accordance with accepted academic practice. No doi: 10.1038/srep27287 use, distribution or reproduction is permitted which does not comply with these terms.

Frontiers in Neuroscience| www.frontiersin.org 16 October 2019| Volume 13| Article 1047