Retrograde regulation of synaptic vesicle endocytosis and recycling

Sustained release of neurotransmitter depends upon the recycling of synaptic vesicles. Until now, it has been assumed that vesicle recycling is regulated by signals from the presynaptic bouton alone, but results from rat hippocampal neurons reported here indicate that this need not be the case. Fluorescence imaging and pharmacological analysis show that a nitric oxide (NO) signal generated postsynaptically can regulate endocytosis and at least one later step in synaptic vesicle recycling. The proposed retrograde pathway involves an NMDA receptor (NMDAR)-dependent postsynaptic production of NO, diffusion of NO to a presynaptic site, and a cGMP-dependent increase in presynaptic phosphatidylinositol 4,5-biphosphate (PIP2). These results indicate that the regulation of synaptic vesicle recycling may integrate a much broader range of neural activity signals than previously recognized, including postsynaptic depolarization and the activation of NMDARs at both immediate and nearby postsynaptic active zones.     

Regulation of synaptic vesicle recycling by calcineurin in different vesicle pools

The synaptic vesicles keep recycling by the processes of endocytosis and exocytosis to maintain the normal synaptic transmission. The synaptic vesicles are classified as the readily releasable pool (RRP) and the reserve pool (RP). In the endocytosis process, calcineurin (CaN), a Ca2+/calmodulin-dependent protein phosphatase, has been shown to play important roles. However, it is unclear about its roles in different vesicle pools. Here, we investigated the role of CaN in the regulation of vesicle recycling in the RRP and RP. Vesicle recycling was monitored by using fluorescent dyes FM1-43 and FM4-64 in the primary cultures of hippocampal neurons. Inhibition of CaN by FK506 and cyclosporin A suppressed the endocytosis in the RP, but not in the RRP. Inhibition of CaN also restrained the exocytic process triggered by 10 Hz stimulation, but had no effect on 3-5 Hz stimulation-induced exocytosis. FK506 also reduced the total vesicle pool size in the synaptic terminals. A synthesized CaN inhibitory peptide showed the similar effects as FK506 and cyclosporin A. These results revealed a novel mechanism that CaN plays critical roles in the distinct vesicle recycling processes.     

Cholesterol-dependent balance between evoked and spontaneous synaptic vesicle recycling

Cholesterol is a prominent component of nerve terminals. To examine cholesterol's role in central neurotransmission, we treated hippocampal cultures with methyl-β-cyclodextrin, which reversibly binds cholesterol, or mevastatin, an inhibitor of cholesterol biosynthesis, to deplete cholesterol. We also used hippocampal cultures from Niemann-Pick type C1-deficient mice defective in intracellular cholesterol trafficking. These conditions revealed an augmentation in spontaneous neurotransmission detected electrically and an increase in spontaneous vesicle endocytosis judged by horseradish peroxidase uptake after cholesterol depletion by methyl-β-cyclodextrin. In contrast, responses evoked by action potentials and hypertonicity were severely impaired after the same treatments. The increase in spontaneous vesicle recycling and the decrease in evoked neurotransmission were reversible upon cholesterol addition. Cholesterol removal did not impact on the low level of evoked neurotransmission seen in the absence of synaptic vesicle SNARE protein synaptobrevin-2 whereas the increase in spontaneous fusion remained. These results suggest that synaptic cholesterol balances evoked and spontaneous neurotransmission by hindering spontaneous synaptic vesicle turnover and sustaining evoked exo-endocytosis.     

Genetic and molecular analysis of synaptic vesicle recycling in Drosophila

Following exocytosis, one of the major presynaptic events is replenishing synaptic vesicles (SVs) to ensure the possibility of continuous synaptic transmission. The nerve terminal is thought to recycle SVs through clathrin-mediated endocytosis and by a clathrin-independent pathway called "kiss and run". This review highlights the use of the genetic model organism, the fruit fly (Drosophila melanogaster), in dissecting the molecular mechanisms of clathrin-mediated endocytosis in recycling SVs at neuromuscular junctions (NMJs). Analyses of endocytotic mutants in Drosophila indicate that clathrin-mediated endocytosis may be essential for SV recycling, including a putative fast recycling mechanism uncovered recently. Further, a rather complex picture begins to emerge suggesting that clathrin-mediated endocytosis involves several sequential steps mediated by a large number of proteins. Finally, these studies also reveal that SV proteins may be selectively retrieved into nascent SVs by clathrin accessory proteins and defects in protein retrieval have significant impacts on synaptic transmission. Following the completion of the Drosophila Genome Project and the development of gene targeting and RNAi approaches, genetic studies in Drosophila have become increasingly efficient. Hence, Drosophila is expected to continue to serve as an important model organism for studies of SV recycling.     

Matrix metalloproteinase-7 modulates synaptic vesicle recycling and induces atrophy of neuronal synapses

Matrix metalloproteinase-7 (MMP-7) belongs to a family of zinc dependent endopeptidases that are expressed in a variety of tissues including the brain. MMPs are known to be potent mediators of pericellular proteolysis and likely mediators of dynamic remodelling of neuronal connections. While an association between proteases and the neuronal synapse is emerging, a full understanding of this relationship is lacking. Here, we show that MMP-7 alters the structure and function of presynaptic terminals without affecting neuronal survival. Bath application of recombinant MMP-7 to cultured rat neurons induced long-lasting inhibition of vesicular recycling as measured by synaptotagmin 1 antibody uptake assays and FM4-64 optical imaging. MMP-7 application resulted in reduced abundance of vesicular and active zone proteins locally within synaptic terminals although their general levels remained unaltered. Finally, chronic application of the protease resulted in synaptic atrophy, including smaller terminals and fewer synaptic vesicles, as determined by electron microscopy. Together these results suggest that MMP-7 is a potent modulator of synaptic vesicle recycling and synaptic ultrastructure and that elevated levels of the enzyme, as may occur with brain inflammation, may adversely influence neurotransmission.     

Monitoring synaptic vesicle recycling in frog motor nerve terminals with FM dyes

Ultrastructural observations made in the study of the frog neuromuscular junction (NMJ) almost three decades ago showed that synaptic vesicle cycling functions through a slow pathway, requiring the use of clathrin-coated vesicles and an endosomal compartment. Simultaneously, a conceptually simpler model emerged, postulating rapid retrieval of vesicle membrane through a mechanism similar to a reversal of vesicle fusion. With the advent of fluorescence imaging which allows the investigator to monitor recycling in living nerve-muscle preparations, new data appeared which reconcile at least in part the two models, indicating that both may be important at this synapse. Two different synaptic vesicle pools can be defined, a readily releasable pool (RRP), consisting of quanta that are immediately available for release, and a reserve pool (RP) that is exocytosed only after prolonged stimulation. Vesicles in the RRP recycle through a fast endocytic pathway, which does not rely on an endosomal compartment, while vesicles in the RP cycle more slowly through formation of infoldings and endosomes and their subsequent severance into vesicles. The two pools mix slowly, and their recycling may be regulated by different mechanisms.     

Impairment of synaptic vesicle exocytosis and recycling during neuromuscular weakness produced in mice by 2,4-dithiobiuret

Chronic treatment of rodents with 2,4-dithiobiuret (DTB) induces a neuromuscular syndrome of flaccid muscle weakness that mimics signs seen in several human neuromuscular disorders such as congenital myasthenic syndromes, botulism, and neuroaxonal dystrophy. DTB-induced muscle weakness results from a reduction of acetylcholine (ACh) release by mechanisms that are not yet clear. The objective of this study was to determine if altered release of ACh during DTB-induced muscle weakness was due to impairments of synaptic vesicle exocytosis, endocytosis, or internal vesicular processing. We examined motor nerve terminals in the triangularis sterni muscles of DTB-treated mice at the onset of muscle weakness. Uptake of FM1-43, a fluorescent marker for endocytosis, was reduced to approximately 60% of normal after either high-frequency nerve stimulation or K(+) depolarization. Terminals ranged from those with nearly normal fluorescence ("bright terminals") to terminals that were poorly labeled ("dim terminals"). Ultrastructurally, the number of synaptic vesicles that were labeled with horseradish peroxidase (HRP) was also reduced by DTB to approximately 60%; labeling among terminals was similarly variable. A subset of DTB-treated terminals having abnormal tubulovesicular profiles in their centers did not respond to stimulation with increased uptake of HRP and may correspond to dim terminals. Two findings suggest that posttetanic "slow endocytosis" remained qualitatively normal: the rate of this type of endocytosis as measured with FM1-43 did not differ from normal, and HRP was observed in organelles associated with this pathway- coated vesicles, cisternae, as well as synaptic vesicles but not in the tubulovesicular profiles. In DTB-treated bright terminals, end-plate potential (EPP) amplitudes were decreased, and synaptic depression in response to 15-Hz stimulation was increased compared with those of untreated mice; in dim terminals, EPPs were not observed during block with D-tubocurarine. Nerve-stimulation-induced unloading of FM1-43 was slower and less complete than normal in bright terminals, did not occur in dim terminals, and was not enhanced by alpha-latrotoxin. Collectively, these results indicate that the size of the recycling vesicle pool is reduced in nerve terminals during DTB-induced muscle weakness. The mechanisms by which this reduction occurs are not certain, but accumulated evidence suggests that they may include defects in either or both exocytosis and internal vesicular processing.     

STXBP1: the second synaptic vesicle release gene involved in epilepsy

De novo mutations in the gene encoding STXBP1 (MUNC18‐1) cause early infantile epileptic encephalopathy
Saitsu et al. (2008)
Nature Genetics 40: 782–788     

The synaptic vesicle cluster: a source of endocytic proteins during neurotransmitter release

Over the past few years significant progress has been achieved in understanding the molecular steps underlying the fusion and recycling of vesicles at central synapses. It still remains unclear, however, how the fusion event is linked with vesicle membrane retrieval. Several factors promoting the transition from exo- to endocytosis have been extensively studied, including levels of intracellular Ca2+, the synaptic proteins involved at both sides of the vesicle cycle, posttranslational modification of endocytic proteins, and the lipid composition of recycled membranes. Recent studies in glutamate synapses indicate that vesicle clusters accumulated at the sites of synaptic contacts have a more complex organization than has previously been thought. Many endocytic proteins reside in the vesicle pool at rest and undergo cycles of migration between the active and periactive zones during synaptic activity. We propose that the local migration of endocytic proteins triggered by Ca2+ influx into the nerve terminal functions as one of the molecular mechanisms coupling exo- and endocytosis in synapses.     

Functionally heterogeneous synaptic vesicle pools support diverse synaptic signalling

Synaptic communication between neurons is a highly dynamic process involving specialized structures. At the level of the presynaptic terminal, neurotransmission is ensured by fusion of vesicles to the membrane, which releases neurotransmitter in the synaptic cleft. Depending on the level of activity experienced by the terminal, the spatiotemporal properties of calcium invasion will dictate the timing and the number of vesicles that need to be released. Diverse presynaptic firing patterns are translated to neurotransmitter release with a distinct temporal feature. Complex patterns of neurotransmitter release can be achieved when different vesicles respond to distinct calcium dynamics in the presynaptic terminal. Specific vesicles from different pools are recruited during various modes of release as the particular molecular composition of their membrane proteins define their functional properties. Such diversity endows the presynaptic terminal with the ability to respond to distinct physiological signals via the mobilization of specific subpopulation of vesicles. There are several mechanisms by which a diverse vesicle population could be generated in single presynaptic terminals, including distinct recycling pathways that utilize various adaptor proteins. Several additional factors could potentially contribute to the development of a heterogeneous vesicle pool such as specialized release sites, spatial segregation within the terminal and specialized delivery pathways. Among these factors molecular heterogeneity plays a central role in defining the functional properties of different subpopulations of vesicles. IntroductionHighly specialized functions of various subcellular elements endow neurons with a wide range of physiological information‐processing capabilities. Dendrites detect incoming information and axons transmit the processed information. A single axon has several hundreds of synaptic terminals each containing large numbers of synaptic vesicles. Morphologically these vesicles are quite similar: their shape, size and spatial distribution is rather uniform. However, in spite of the uniform morphological features of synaptic vesicles, their functional heterogeneity was suggested by several recent studies (Sakaba & Neher, 2001 a; Hua et al. 2011). This functional heterogeneity can be envisioned via the integration of different sets of proteins into the vesicular membrane (Takamori et al. 2006). This complexity within the presynaptic terminal could play a key role in various forms of synaptic communication. There are several factors that can define distinct vesicle pools. Vesicles are released during different presynaptic activities, they contribute to various forms of release, they have different sets of membrane proteins, and they can be generated via diverse endocytotic recycling pathways. How do these factors contribute to the functionally diverse forms of neurotransmitter release exhibited by presynaptic terminals and what is the relationship between these categories?

Functional heterogeneity among vesicles based on their availability for release

Traditionally vesicles are sorted into categories based on their availability for release and the degree of difficulty to trigger their fusion with the membrane (Fig. 1 A). According to these criteria, three pools of synaptic vesicles were identified as the readily releasable pool (RRP), the recycling pool and the reserve or resting pool (Sudhof, 2000; Zucker & Regehr, 2002; Rizzoli & Betz, 2005; Alabi & Tsien, 2012). While the number and distribution of synaptic vesicles in the three pools differs substantially between organisms and synapses, several common characteristics are observed and will be emphasized below.Open in a separate windowFigure 1 Heterogeneity among vesicles Synaptic vesicles in the presynaptic terminal can be sorted into distinct pools based on how strongly the terminal needs to be stimulated for their release (A), via which endocytotic pathway they recycle (B), the form of neurotransmitter release they contribute to (C), and the particular sets of membrane protein vesicles expressed (D). RRP, readily releasable pool.Vesicles belonging to the RRP are the first to be recruited during neuronal activity. Most of these vesicles are docked at the active zone or located in its vicinity, where activation of release machinery can trigger their release with short latency (Schikorski & Stevens, 2001; Dittman & Ryan, 2009). Although this pool accounts for only a small fraction of the total number of vesicles available (1–2%) in several synapses, including CA1 hippocampal excitatory synapses (Schikorski & Stevens, 1997) and the frog neuromuscular junction (Rizzoli & Betz, 2005), their availability for release and their fast recycling allow them to support the majority of physiological neurotransmission. Vesicles located in the recycling pool are not exocytosed right away during weak neuronal activity. Vesicles in this pool are largely scattered within the synaptic terminal but display high mobility. Synaptic vesicles in the recycling pool encompass a larger fraction of the total number of vesicles than the RRP and represent approximately 10–20% of the total vesicles available in several synapses, including hippocampal presynaptic terminals (Harata et al. 2001). Together, these characteristics suggest that the recycling pool acts as an effective backup mechanism to secure synaptic activity on longer time scales. Although it largely dominates the other pools in terms of numbers (80–90% of total synaptic vesicles), mobilization of the resting pool vesicles to the membrane requires strong activity, demanding tens of seconds of high‐frequency electrical stimulation (Richards et al. 2000; Denker et al. 2009). The conditions required for its complete depletion remain mysterious at certain synapses, including the calyx of Held, as approximately 50% of the pool could never be released (de Lange et al. 2003; Fernandez‐Alfonso & Ryan, 2008; Wyatt & Balice‐Gordon, 2008).Why should a larger portion of vesicles inside synaptic terminals participate minimally in neurotransmission? Intriguingly, experiments performed for longer periods of time in vivo or using physiological firing frequencies suggest that all vesicles from the reserve pool could be released. Experiments performed at the frog neuromuscular junction for periods up to 8 h using low‐frequency stimulation (2 Hz) were able to induce fusion of the full vesicle pool (Ceccarelli et al. 1972; Betz & Henkel, 1994) and challenge the view that the resting pool contributes little to physiological neurotransmission (Denker et al. 2011). As such, the frontiers between the recycling pool and the resting pool may be thinner than previously thought. Vesicles could move through vesicle pools in a state‐dependent manner. This evidence supports the notion that vesicles evolve as they move through different pools following endocytosis (Denker & Rizzoli, 2010). The level of activity exhibited by the presynaptic terminal will also influence the form of endocytosis released vesicles will go through. Vesicles can recycle via three distinct mechanisms: during the kiss‐and‐run type of endocytosis vesicles do not fuse completely with the plasma membrane (Fesce et al. 1994); clathrin‐dependent recycling involves full vesicle fusion and endocytosis via clathrin‐coated vesicles; and bulk‐endocytosis occurs when vesicles recycle via endosome‐like intermediate organelles (for review see: Rizzoli, 2014; Jähne et al. 2015; Kononenko & Haucke, 2015). Kiss‐and‐run and clathrin‐dependent endocytosis occur at lower activity levels; however, bulk endocytosis is usually activated after prolonged, extensive activity. In this way, the composition of the vesicle pool based on the route of recycling is a result of the previous history of the synapse (Fig. 1 B).

Spatial localization of vesicles in the terminal

Why are certain pools easier to release than others? It is tempting to speculate that physical distance from the active zone determines how likely it is to release vesicles from specific pools. Vesicles belonging to the RRP need to be docked at the active zone to be released with millisecond precision. In accordance with this function, a portion of vesicles belonging to the RRP are found close to the active zone (Schikorski & Stevens, 2001). Considering the increased level of activity required for mobilization of the recycling and resting pool, the most obvious hypothesis would be that the recycling and resting pools are located gradually further away from the active zone (Fig. 2 A). While this is the case at the drosophila neuromuscular junction (NMJ) (Kuromi & Kidokoro, 1998), no evidence supports the precise spatial segregation of the recycling and the resting pools in CNS synapses (Rizzoli & Betz, 2005). Indeed, calcium uncaging experiments showed that the position of vesicles within the terminal could not explain the release properties of distinct pools (Schneggenburger et al. 1999; Bollmann et al. 2000; Schneggenburger & Neher, 2000; Denker & Rizzoli, 2010).Open in a separate windowFigure 2 Potential mechanisms of functional heterogeneity within the presynaptic terminal Heterogeneity on the presynaptic side could be achieved if vesicles situated at various distances from the active zone responded to different levels of calcium (A), if certain active zones or synapses are specialized to support distinct forms of release (B), if pools are composed of vesicles exhibiting particular sets of membrane proteins endowing them with unique functional properties (C), or if filaments within the terminal ensure the delivery of distinct vesicles to the active zone (D).

Functional heterogeneity among vesicles based on their contribution to particular release type

Three modes of vesicle release govern chemical synaptic transmission: spontaneous, synchronous and asynchronous release (Fig. 1 C). Here again, the activity pattern of the presynaptic terminal plays a key role in determining through which release type neurotransmission occurs. Indeed, activity‐dependent release of neurotransmitters is tightly regulated by the spatiotemporal dynamics of calcium elevation in the presynaptic terminal. Spontaneous release, fast synchronous release and asynchronous release are sequentially observed with increasing intraterminal calcium concentration.The three different modes of vesicle release can be associated with specific neuronal network states and serve specialized physiological roles for intercellular communication. For instance, spontaneously occurring release of vesicles was demonstrated to be a key element regulating homeostatic plasticity (Sutton et al. 2006). Synchronous neurotransmitter release is well‐known to pace neuronal network activity on a very precise time scale. A striking example is the generation of hippocampal oscillations by interconnected interneuron networks. Indeed, synchronous release of GABA at interneuron‐to‐interneuron synapses is essential in generating hippocampal gamma oscillations (Bartos et al. 2002). On the other hand, many studies support the notion that the longer‐lasting and delayed asynchronous release is ideal to smooth the synaptic transmission over a longer period of time. For example, asynchronous glutamate release contributes to up states in striatal medium spiny neurons (Plenz & Kitai, 1998). Consistent with a role of elevating the general level of activity in neuronal networks, asynchronous glutamate release was found to sustain network reverberations in cultured hippocampal neurons (Lau & Bi, 2005). Given that different modes of release support specific physiological functions, vesicles liberated during a given mode of release could originate from designated pools; this provides an interesting model combining structure and physiological function. It would be straightforward to propose that the RRP mediates spontaneous and fast synchronous release while the recycling and resting pools are progressively recruited during asynchronous release. As whether various forms of release utilize the same vesicle pool or not is currently extensively debated, both possibilities will be explored.First, vesicles contributing to different types of release could originate from the same pools (Prange & Murphy, 1999; Groemer & Klingauf, 2007). Several lines of evidence suggest that spontaneously and synchronously released vesicles share a common origin (Groemer & Klingauf, 2007). Initial studies by Prange & Murphy (1999) positively correlated the probability of spontaneous synaptic activity with the synchronous release probability, indicating that the two modes of release shared a common vesicle pool. By imaging labelled vesicles in hippocampal neurons, Groemer & Klingauf (2007) showed that vesicles released spontaneously and following stimulation emerged from the same pool. Furthermore, studies indicate that fast synchronous and asynchronous release recruit vesicles from the same pool (Hagler & Goda, 2001; Otsu et al. 2004). Using electrophysiological approaches in cultured hippocampal neurons, Hagler & Goda (2001) found that asynchronous release could deplete the vesicle pool mediating fast synchronous release. In accordance with this view, Otsu et al. (2004) showed that both synchronous and asynchronous release compete for the same pool of vesicles. Therefore, both release modes could access the same pool of vesicles in discrete time windows.On the other hand, reports show that distinct pools contribute to different types of physiological activity (Sara et al. 2005; Fredj & Burrone, 2009). Indeed, a large body of evidence demonstrates that spontaneous and synchronous releases are mediated by specific pools (Sara et al. 2005; Fredj & Burrone, 2009). Using tracking of fluorescently labelled vesicles and immunohistochemistry, Sara et al. (2005) showed that vesicles were recycled at rest and that, following fusion, spontaneously released vesicles were directed to a small pool of vesicles. Furthermore, this specific pool would then be more likely to provide vesicles that would be released during the subsequent event of spontaneous release. In accordance with these results, fast synchronous and asynchronous release do not compete for the recruitment of vesicles at glutamatergic synapses in the nucleus accumbens (Hjelmstad, 2006). Furthermore, exocytosed vesicles during evoked synchronous or asynchronous release are segregated from spontaneously released vesicles (Chung et al. 2010). According to these latter studies, vesicles released during specific types of activity would originate from distinct pools. The studies that lead to these opposing views on vesicle heterogeneity are using similar technical approaches. The reason behind this discrepancy is not immediately apparent: however, an interesting idea proposing that developmental factors could be behind these disparate results has recently emerged (Truckenbrodt & Rizzoli, 2014). While at the moment this debate is not resolved, recent findings identifying distinct regulatory processes associated with spontaneous release such as postsynaptic signalling and vesicle fusion are increasingly difficult to reconcile with the idea of a homogeneous vesicle population (Sutton et al. 2006; Sutton & Schuman, 2006; Leitz & Kavalali, 2014).On the postsynaptic side, can different modes of release requiring different pools of vesicles recruit specific signalling cascades? Postsynaptic cells are modulated differently by spontaneously released synaptic signals compared with action potential‐triggered events; spontaneous release can influence homeostatic plasticity via the regulation of dendritic protein synthesis through the deactivation of eukaryotic elongation factor 2 (eEf2) kinase (Sutton et al. 2006; Sutton & Schuman, 2006). Endocytosis of spontaneously released vesicles shows kinetic properties that are different from those after action potential invasion of the terminal, while their fusion is distinctly regulated (Leitz & Kavalali, 2014).

Heterogeneity among vesicles based on their molecular composition

Which feature of synaptic vesicles determines under which conditions they are released? Differences in morphological features such as diameter or shape could be envisioned as a significant factor in determining when a vesicle is released. However, ultrastructural studies failed to identify a key morphological parameter that would define different vesicle pools (Rizzoli & Betz, 2005). Similar to neuronal membranes (Stoeckenius, 1962), synaptic vesicles are composed of proteins embedded in a lipid bilayer and exhibit a 60% protein to lipid ratio (Takamori et al. 2006). Although neither the protein/lipid ratio nor the basic structure composition can distinguish between functionally different synaptic vesicles (Takamori et al. 2000), the complex protein structures assembled on the surface of vesicles coupled with a vast repertoire of proteins could support synaptic vesicle specialization (Poudel & Bai, 2014). Indeed, a potential source of vesicle diversity lies within the combination of membrane proteins that is incorporated into vesicle membranes. It was revealed that over 400 different proteins can be found on synaptic vesicles (Takamori et al. 2006; Gronborg et al. 2010). This wide array of possible proteins can provide the molecular cues to distinguish vesicles and associate them to their respective pools of belonging. Incorporation of specific proteins in addition to the required basic elements (Takamori et al. 2006; Dittman & Ryan, 2009) at the vesicle surface could be envisaged to explain how vesicles within the same presynaptic terminal possess different functional properties (Fig. 1 D).Synapsin is a molecular marker for vesicles in the reserve pool and these vesicles did not seem to participate in physiological neurotransmission (Denker et al. 2011). As such, fewer than 5% of the total vesicles were recycled over hours of neuronal activity (Denker et al. 2011). However, the number of vesicles being recycled was increased to 30% for the same amount of activity in drosophilas that did not express synapsin. Therefore, synapsin appears to be specifically incorporated in vesicles belonging to the reserve pool. Additionally, it is considered to control the entry and the exit of synaptic vesicles from the reserve pool (Verstegen et al. 2014).As additional support for vesicle sorting based on their molecular composition, zinc was found to be only present in a subpopulation of vesicles in hippocampal mossy fibre presynaptic terminals. Evidence also shows that the presence of zinc‐containing vesicles was crucial in maintaining normal levels of spontaneous glutamatergic activity (Lavoie et al. 2011), but that these vesicles were preferentially released during high‐frequency stimulation. Thus zinc‐containing vesicles contribute to specific modes of neurotransmission. This is consistent with the concept that these vesicles belong to a specific pool of vesicles. How is this molecular heterogeneity of molecule incorporation in synaptic vesicles created? In addition to vesicles, zinc staining was also found in synaptic endosomes at mossy fibre to CA3 pyramidal cell synapses, suggesting that zinc‐containing vesicles were derived from bulk endocytosis (Lavoie et al. 2011). Therefore, these zinc‐containing synaptic vesicles could be generated through a specific recycling pathway. In return, this suggests that the recycling pathway could be responsible for tagging vesicles with the appropriate proteins or molecules to direct them to specific pools. In accordance with this idea, vesicles derived from distinct recycling pathways populate specific pools in terminals of cerebellar granule neurons (Cheung et al. 2010).

Potential mechanisms of distinct modes of neurotransmitter release

How is a certain mode of release selected over another in presynaptic terminals? Synapses can be specialized to a specific mode of release or support multiple modes of release. First, the demonstration that spontaneous and evoked release can co‐exist in the same synaptic terminal implies that highly specialized molecular machinery can discriminate between different modes of release (Ramirez & Kavalali, 2011). One such example is the calcium affinity of sensors and their position relative to the calcium source, which dictates the recruitment of a specific pool of vesicles during a given type of activity (Wen et al. 2010) (Fig. 2 B ). On the other hand, recent findings suggest that specialized active zones in the same synaptic terminal independently support spontaneous and evoked neurotransmitter release at the fly NMJ (Melom et al. 2013). Additionally, an interesting avenue for the separation of function of release modes could be the specialization of synapses in mediating a specific mode of release. Indeed, recent evidence indicates that evoked and spontaneous neurotransmitter release are mediated at different sets of synapses expressing specific postsynaptic receptors in drosophila NMJ (Peled et al. 2014). In addition, neurotransmitter release from these highly specialized terminals activates different sets of postsynaptic receptors, consistent with data showing that spontaneous release and fast synchronous release activate different sets of postsynaptic receptors in cultured neurons (Atasoy et al. 2008). Taken together, these results support the fact that certain synapses are specialized to mediate either spontaneous or evoked release. While these two modes of release can originate from the same synaptic terminal, distinct active zones can also mediate spontaneous and evoked release to activate specific postsynaptic receptors.Synchronous and asynchronous release are differentially affected by slow and fast calcium chelators (EGTA and BAPTA) (Adler et al. 1991; Cummings et al. 1996; Atluri & Regehr, 1998; Lu & Trussell, 2000). This differential effect is generally interpreted as asynchronous release resulting from the slow expansion of the localized calcium entry near the active zones (Borst & Sakmann, 1996; Meinrenken et al. 2002). However, according to data obtained from zebrafish NMJ, the two types of release could also be triggered by calcium originating from different sources. While synchronous release was triggered by calcium entry via P/Q‐type channels localized in synaptic boutons, asynchronous release was boosted by calcium sources from off‐synaptic locations (Wen et al. 2013). While distinct release sites and various calcium sources could explain some aspects of various forms of neurotransmitter release occurring in response to different stimuli, none of these scenarios explain why and how these vesicles respond to different calcium dynamics. In addition, it is also unclear at the moment whether these contributors play a key role in small CNS synapses. Spatial segregation of distinct modes of release would need to occur in the same small terminal or various boutons would need to specialize to certain types of release; currently neither of these scenarios are supported by experimental evidence.

Molecular determinants of functional heterogeneity

The molecular machinery involved in spontaneous, synchronous and asynchronous release also appears to be distinct (Fig. 2 C), as shown by the selective expression of proteins on vesicles preferentially recruited during specific modes of release. For example, Vps10p‐tail‐interactor‐1a is selectively found on a population of vesicles from the resting pool that supports spontaneous neurotransmitter release (Ramirez et al. 2012). While VAMP2‐expressing vesicles were released during synchronous release (Schoch et al. 2001), VAMP4 was selectively observed on synaptic vesicles associated with asynchronous release (Raingo et al. 2012).Molecular heterogeneity of vesicles also makes it possible to distinguish between vesicles originating from the same pool. Both synchronous and asynchronous release can be mediated by vesicles from the RRP (Sun et al. 2007), but the calcium sensors involved were different. In this report, the authors demonstrated that deletion of synaptotagmin 2 abolished synchronous release, but asynchronous release was kept intact. An additional example of the selectivity of sensors for a given type of release is found at GABAergic synapses. Deletion of the calcium sensor synaptotagmin 1 blocked synchronous release of glutamate and GABA, while asynchronous release was unaffected (Maximov & Sudhof, 2005). Confirming these findings regarding specific sensors being associated with particular release types, at the zebrafish NMJ, knockdown of synaptotagmin 7 decreased the asynchronous release while synaptotagmin 2 was key in mediating fast synchronous release (Wen et al. 2010). A similar conclusion was reached using cultured hippocampal neurons. In this study, synaptotagmin 7 ablation in synaptotagmin 1‐deficient synapses suppressed asynchronous release (Bacaj et al. 2013). In contrast to its function in asynchronous release, synaptotagmin 7 was rather suggested to regulate Ca²⁺‐dependent synaptic vesicle replenishment in mammalian synaptic terminals (Liu et al. 2014). Similarly to synaptotagmin 7, the role of Doc2, a calcium sensor, in synaptic functions is intensely debated. Recent findings show that Doc2 is required for spontaneous activity (Groffen et al. 2010; Pang et al. 2011; Walter et al. 2011). Indeed, Doc2 deletion results in significant decrease in the frequency of spontaneous EPSCs, while its deletion has no effect on asynchronous release. In sharp contrast with the aforementioned findings, Doc2 has been identified as a calcium sensor that is involved in asynchronous release in hippocampal neurons (Yao et al. 2011). Based on the kinetics of Doc2 interaction with calcium, the authors linked its activity to asynchronous release. Furthermore, the level of asynchronous release was intimately linked with the expression of Doc2.These studies adhering specific calcium sensors to distinct modes of release support the idea that inclusion of distinct sets of sensors to the vesicle membrane could link different presynaptic calcium dynamics with distinct release modes. Specific sensors and their position relative to the source of calcium could dictate through which mode of release the vesicle will fuse to the presynaptic terminal, even when vesicles belong to the same pool.Theoretical data suggest that heterogeneity among vesicles is necessary for neurotransmission over a wide range of physiological activities (Sakaba & Neher, 2001 b). In order to maintain an appropriate level of neurotransmitter release, modelling studies predict that at least two distinct pools of vesicles need to exist in presynaptic terminals. At the calyx of Held, two pools of vesicles can be recruited in an activity‐dependent manner (Sakaba & Neher, 2001 a). First, fast‐releasing vesicles were emptied quickly following activity but were recycled more slowly and their recovery was strongly associated with the level of Ca²⁺ in the terminal (Sakaba & Neher, 2001 a). On the other hand, slow‐releasing vesicles were recruited later during activity, but were endocytosed faster from the membrane following activity (Sakaba & Neher, 2001 a). Consistent with their timing of fusion with the membrane, fast‐releasing vesicles are involved in synchronous release while slowly releasing vesicles mediate asynchronous release (Sakaba, 2006). Molecular markers expressed on vesicles could dictate such precise recruitment of vesicle pools during neurotransmission. In support of this idea is the fact that both pools respond differently to pharmacological treatments (Sakaba & Neher, 2001 a). Additionally, expression of various V‐SNARE proteins could predict the fate of vesicles (Scheuber et al. 2006; Fig. 2 A). For example, VAMP7 is highly expressed in resting pool vesicles, while VGLUT1 is found in vesicles from the recycling pool. Not only do these vesicles demonstrate molecular heterogeneity, but VAMP7‐positive vesicles are also more likely to be released during spontaneous activity than VGLUT1‐containing vesicles (Hua et al. 2011). Therefore, the release of specific vesicles during different modes of neurotransmission appears guided by the differential expression of vesicular proteins. The heterogeneous populations of synaptic vesicles are preferentially released in response to different intensities of synaptic activity. How subsets of vesicles contribute to neurotransmission and their role in neuronal coding is currently being actively investigated. Different calcium sensors could promote the fusion of vesicles in function of the activity sensed in the presynaptic terminal (Fig. 2 C).

How are different pools of vesicles generated following membrane fusion and vesicle recycling?

Recent experiments tracking synaptic vesicles in real‐time indicate that vesicles will be reassigned to their pool of origin following neurotransmitter release and endocytosis (Park et al. 2012). Since vesicles could be sorted into pools without spatial consideration, certain mechanisms should exist to allow for the allocation of vesicles to the right pool. Although vesicles are recycled and sorted in their respective pool of origin, our current understanding of how this process occurs is rather limited. Expression of specific molecular markers could be ideally suited to sort vesicles in their respective pools (Lavoie et al. 2011). In agreement with this view, data demonstrate that an amino acid sequence present in the VGAT transporter is required for proper shipping of the vesicle to its pool of residence (Santos et al. 2013). Further evidence for molecule‐guided vesicle sorting is the effect of the presence of adaptor protein 3 (AP‐3) on vesicles, which routes vesicles to the correct pool (Voglmaier et al. 2006; Voglmaier & Edwards, 2007). The functionally distinct role of these vesicles was demonstrated in a subsequent study. Vesicles generated via AP‐3‐dependent recycling contribute to asynchronous release demonstrating how various recycling pathways can contribute to the generation of a functionally diverse vesicle pool in single presynaptic terminals (Evstratova et al. 2014). Alternatively, one could envision special delivery routes for vesicles contributing to various forms of release working similarly to conveyor belts. Filaments connecting vesicles within the presynaptic terminal were visualized using high‐pressure freezing (Siksou et al. 2007) (Fig. 2 D). While it is intriguing to imagine that these filaments could support various forms of release, experimental evidence regarding the mechanism of action by which the cytomatrix regulates transmitter release is currently not available.

Conclusion

Multiple studies support the existence of various functionally distinct vesicle pools in presynaptic terminals. This heterogeneity has been linked to spatial segregation in large NMJ terminals; however, no definite spatial position can be attributed to these pools within small CNS synapses. Heterogeneity of vesicles cannot be explained by morphological characteristics, but rather by specific expression of vesicular proteins. Molecular heterogeneity among vesicles can explain how small CNS synapses utilizing limited vesicle pools can exhibit different forms of neurotransmitter release in response to distinct physiological stimuli (Fig. 3). When vesicles are specifically tagged with membrane proteins they can respond to a particular physiological stimulus. With the utilization of a heterogeneous vesicle population even a small presynaptic terminal can show a wide range of responses, depending on the level of activity, by releasing specific populations of vesicles. This vast heterogeneity of possible responses widens the range over which the synapse can transmit physiological information. Various forms of neurotransmitter release and vesicle endocytosis are tightly linked to the level of activity experienced by the presynaptic terminal. Therefore, the composition of a heterogeneous vesicle pool at any given moment is the direct result of the previous history of the neuron; this could potentially allow the presynaptic terminal to adapt to changes in neuronal activity levels by altering the ratio between different vesicle pools.Open in a separate windowFigure 3 Distinct physiological signals trigger the release of different subpopulations of vesicles Molecular heterogeneity (green and orange) of synaptic vesicles could endow a single presynaptic terminal with the ability to respond to different (A and B) physiological stimuli with the release of distinct sets of vesicles.

Future directions

In recent years we have seen a steady increase in the number of studies that identified various processes and molecules that are specific for certain subgroups of vesicles. Identification of the molecular signatures of functionally distinct vesicle groups will be necessary steps towards answering several questions regarding the functional importance of vesicle heterogeneity. First, the proper understanding of downstream effects of various modes of neurotransmitter release relies on the demonstration of selective stimulation of distinct signalling pathways. Second, identification of the unique sets of membrane proteins will also be crucial for determining the mechanisms by which distinct pools are regulated by calcium dynamics. And third, molecular ‘fingerprints’ of various vesicle groups will be vital for obtaining a complete picture of the biogenesis of various vesicle groups.It will also be important to understand how rigid these groups are, how previous activity can alter the composition of vesicle pools, and via which mechanism this activity‐dependent response can be achieved. Vesicle heterogeneity is only one, albeit crucial, element of the complex output patterns single terminals can generate; other factors such as the composition of ion channels and signalling cascades in a given terminal endow different types of presynaptic terminals with functional properties that are uniquely suited to their function.     

Neurotransmitter depletion by bafilomycin is promoted by vesicle turnover

Accumulation of neurotransmitter into synaptic vesicles is powered by the vacuolar proton ATPase. We show here that, in brain slices, application of the H(+)-ATPase inhibitors bafilomycin or concanamycin does not efficiently deplete glutamatergic vesicles of transmitter unless vesicle turnover is increased. Simulations of vesicle energetics suggest either that bafilomycin and concanamycin act on the H(+)-ATPase from inside the vesicle, or that the vesicle membrane potential is maintained after the H(+)-ATPase is inhibited.     

Immunochemical comparison of synaptic plasma membrane and synaptic vesicle membrane antigens

Summary A synaptic vesicle fraction and a synaptic plasma membrane fraction obtained after subfractionation of synaptosomes from chick forebrain have been used to produce antisera in rabbits.Immunofluorescence histology with the two antisera revealed that they reacted strongly with synaptic terminal regions present in the chick forebrain, cerebellum and spinal cord. In addition, the synaptic plasma membrane antiserum (but not the synaptic vesicle antiserum) reacted with preterminal axons in the cerebellum and spinal cord.Comparison of the two antisera by two-dimensional immunoelectrophoresis, revealed the presence of common antigens in the synaptosomal vesicle and plasma membrane fractions.Incubation of synaptosomesin vitro with the synaptosomal vesicle antiserum and complement produced a dose-dependent inhibition of synaptosome swelling up to a maximum of 55% of that obtained with the synaptosomal plasma membrane antiserum. The results of this test are consistent with the hypothesis that some synaptosomal vesicle antigens may be present also in the synaptosomal plasma membrane and imply that they face the external surface of the synaptosomes.The fate of vesicle membrane components in synaptosomal plasma membranes is not known. The possibility is discussed that they may be recycled locally by a mechanism similar to that proposed by Heuser and Reese (1973) for re-use of synaptic vesicle membranes at the neuromuscular junction.     

Neurotransmitter release and its facilitation in crayfish

Quantal synaptic currents were recorded at nerve terminations on the opener muscle of crayfish using a macro-patch-clamp electrode, and the release was elicited by depolarizing current pulses applied to the terminal through the same electrode. After 2 ms depolarization pulses at low temperature, release started with about 2 ms delay after the onset of depolarization, and the maximum rate of release occurred at about 4 ms delay. Large variations in Ca inflow during the pulses were concluded from the facilitation of test EPSCs. The time course of release proved to be remarkably invariant in spite of large changes in release. If a conditioning train of depolarization pulses preceded the test pulse, release due to the test pulse was facilitated up to 60-fold, but the shapes of distributions of quantal delays were practically not affected by this facilitation. Facilitation by the conditioning trains must have raised the [Ca]i level at the onset of the test pulse. The invariance of the time course of release with respect to the level of [Ca]i cannot be explained by theories in which [Ca]i alone controls the time course of release. The time courses of reactions controlling release were explored by mathematical analysis and simulation. A reaction scheme in which the activation of "release sites" directly by depolarization had rate limiting control on the release reactions, in which rise of [Ca]i only was a promoting cofactor, and in which a cooperative reaction involving the complex of release sites and Cai, (SCai) was one of the final steps eliciting release, was able to predict the delayed onset of release and the substantial latency between the end of the depolarization pulse and the maximum of the rate of release. Reaction schemes in which the direct effect of depolarization on release occurred at one or more steps following the entry of Ca could be excluded generally by showing conflict with the experimental findings.     

Roles of calmodulin and calmodulin-binding proteins in synaptic vesicle recycling during regulated exocytosis at submicromolar Ca2+ concentrations

Calcium ion is required at various concentrations for vesicular recycling in the presynaptic terminal. Although calmodulin (CaM) is the most abundant Ca2+-binding protein and has a submicromolar affinity for Ca2+, it is not the Ca2+ sensor for vesicular fusion because this process requires Ca2+ concentrations above 1 microM. Several lines of evidence, however, suggest that CaM mediates the regulation of vesicular recycling by submicromolar Ca2+ via novel protein-protein interactions. In this review, we discuss recent findings on how CaM regulates synaptic vesicle recycling by controlling the SNARE mechanism, which is the molecular machinery that mediates exocytosis.     

Neurotransmitter release and its facilitation in crayfish

Excitatory postsynaptic currents (EPSCs) were recorded extracellularly from synaptic spots on crayfish opener muscles. Release of transmitter was determined by counting the average number of quanta which appear after a stimulus. When [Mg]₀ was increased from 2.5 to 12.5 mM, release was inhibited. Quantitatively the effect of [Mg]₀ could be described by a competitive inhibition of the entry (not of the release) of Ca²⁺ after an impulse, with apparent dissociation constants K_Mg between 1.4 and 18 mM [Mg]₀, assuming saturation kinetics for entry of Ca²⁺ and release. At constant [Ca]₀, twin pulse facilitation (F _s ) for short intervals (about 10 ms) increased when [Mg]₀ was raised from low values, reached a maximum at a certain {ie237-1} and unexpectedly decreased again at higher [Mg]₀. At higher [Ca]₀, {ie237-2} shifted to higher values. This maximum of facilitation is predicted qualitatively by our theoretical model. However, the amplitude of facilitation was larger than predicted theoretically, and the {ie237-3} were smaller than predicted. The theoretical possibilities to correct these discrepancies within the framework of residual calcium based facilitation and saturation kinetics of entry and release were analyzed, but all were in conflict with experimental findings. It is concluded that an essential element is missing in the present theory of facilitation.Supported by the Deutsche Forschungsgemeinschaft and by the USA-Israel Binational Foundation     

Neurotransmitter release and its facilitation in crayfish

Excitatory synaptic currents (EPSCs) were recorded extracellulary from synaptic spots on crayfish opener muscles. The decay of facilitation after a first pulse was measured by a second pulse given at increasing intervals; the duration of facilitationT _F was the interval after which the second EPSC had 1.1 times the amplitude of the first one. Increasing [Mg]₀ in the range from 0.5–12.5 mM at low [Ca]₀ (1.7–4.5 mM) led to a monotonic prolongation of facilitation.T _F showed a S-shaped dependence on [Mg]₀, rising very steeply at 2–5 mM [Mg]₀. At higher [Ca]₀, and also at half normal [Na]₀, an increase of [Mg]₀ did not affect the decay of facilitation appreciably. As shown before, the decay of facilitation is due to two Ca_i-removal processes,R ₁ andR ₂. Mg₀ inhibits only theR ₁ process, which is also inhibited by high [Ca]₀, is dependent on normal [Na]₀ and has the characteristics of a Ca_iNa₀ exchange. As one possible mechanism, competition of Mg₀ with Na₀ at the extracellular loading site of the exchange is discussed quantitatively.Supported by the Deutsche Forschungsgemeinschaft and by the US-Israel Binational Foundation