Differential Regulation of Spontaneous and Evoked Neurotransmitter Release at Central Synapses

Abstract

Recent studies have begun to scrutinize the presynaptic machinery and vesicle populations that give rise to action potential evoked and spontaneous forms of neurotransmitter release. In several cases this work produced unexpected results which lend support to the notion that regulation, mechanisms, postsynaptic targets and possibly presynaptic origins of evoked and spontaneous neurotransmitter release differ. Furthermore, the list of regulatory pathways that impact spontaneous and evoked release in a divergent manner is rapidly growing. These findings challenge our classical views on the relationship between evoked and spontaneous neurotransmission. In contrast to the well-characterized neuromodulatory pathways that equally suppress or augment all forms of neurotransmitter release, molecular substrates specifically controlling spontaneous release remain unclear. In this review, we outline possible mechanisms that may underlie the differential regulation of distinct forms of neurotransmission and help demultiplex complex neuronal signals and generate parallel signaling events at their postsynaptic targets.

Introduction

During activity, synaptic nerve terminals release neurotransmitter in response to presynaptic action potential firing and ensuing Ca²⁺ influx. In addition, several substances, often called secretagogues, including α-latrotoxin, lanthanides, and certain neuromodulators can trigger release [1, 2, 3]. However, since the early work of Bernard Katz and colleagues, it has been wellknown that neurotransmitter release can also occur spontaneously in the absence of presynaptic action potentials or secretagogues, albeit with a low probability [4].

Recent studies have suggested several physiological roles for this stimulus-independent form of neurotransmitter release [5, 6]. Spontaneous neurotransmitter release, which typically corresponds to fusion of a single synaptic vesicle, can be sufficient to trigger postsynaptic action potential firing or regulate postsynaptic excitability and spike timing [3, 7, 8]. In addition, several forms of homeostatic plasticity seen after cessation of neuronal activity can be strongly modulated in duration and magnitude by spontaneous release events [9, 10, 11, 12, 13]. Furthermore, these low probability stochastic fusion events have been shown to suppress dendritic local protein translation and help maintain the stability of synaptic responses [10]. In agreement with these observations, recent studies have suggested a spatial segregation for the two forms of release [14, 15], supporting the premise that they activate parallel signal transduction pathways with minimum overlap [6].

Differential regulation of spontaneous and evoked release

In contrast to the emerging consensus on the distinct functional role of spontaneous neurotransmission in neuronal signaling [6, 16, 17], the presynaptic mechanisms that underlie spontaneous release and the degree of its segregation from the machinery giving rise to action potential evoked neurotransmission has been under intense debate [18, 19, 20, 21, 22, 23, 24, 25]. Nevertheless, several studies have provided strong evidence for specific regulation of spontaneous neurotransmitter release by certain signaling pathways and, in some cases, actions of these pathways have been shown to alter spontaneous and evoked release in the opposite direction [26, 27].

Mechanistically, it is straightforward to envision how neuromodulators may selectively target evoked release or modify both evoked and spontaneous release in the same direction. This regulation can be achieved by inhibition of voltage-gated Ca²⁺ channel function or block of presynaptic excitability leading to impaired evoked release [28, 29]. Some neuromodulators such as adenosine [30], serotonin [31] or GABA, acting via GABA_B receptors [32], target the release machinery as well as Ca²⁺ channels, thus suppressing both evoked and spontaneous release. Exogenous application of phorbol esters or diacylglycerol generation leads to augmentation of both evoked and spontaneous release [33, 34, 35, 36, 37]. In contrast to these well-characterized neuromodulatory pathways, mechanisms controlling selective regulation of spontaneous release but not evoked release or opposite regulation of spontaneous and evoked release remain unclear.

Examples of this anomalous regulation include selective inhibition of spontaneous but not Ca²⁺-dependent evoked release by activation of presynaptic group II metabotropic glutamate receptors in rat cerebellar slices [38], specific enhancement of spontaneous miniature excitatory postsynaptic currents (mEPSCs) but not evoked EPSCs in response to BDNF application in immature visual cortical neurons [39], and selective activity-dependent decrease in the frequency of miniature EPSCs triggered by inhibition of DNA methyltransferases, key enzymes that methylate DNA and regulate gene expression [40]. Induction of endoplasmic reticulum stress also causes a dramatic four-fold increase in spontaneous excitatory transmission coupled with only a small increase in evoked neurotransmitter release probability [41]. Finally, loss of presenilins, integral components of γ-secretase, a key enzyme involved in the etiology of Alzheimer's disease, or treatment with a γ-secretase inhibitor augments the rate of spontaneous release without altering evoked synaptic transmission [42]. This regulation appears to be achieved by modulation of tonic calcium influx into presynaptic terminals, consistent with the proposed role of presenilins in regulation of neuronal Ca²⁺ homeostasis [43, 44]. With regard to opposing regulation of spontaneous and evoked release, certain nitric oxide-related species inhibit evoked neurotransmission but enhance spontaneous mEPSCs [45]. Additionally, cholesterol depletion or inhibition of cholesterol synthesis in neurons causes an increase in the rate of spontaneous transmission but impairs evoked neurotransmission [26, 27].

This growing list of examples challenges our classical views on the relationship between evoked and spontaneous neurotransmitter release. Several recent studies of the presynaptic machinery and the vesicle populations have produced unexpected results which support specific regulation of spontaneous neurotransmitter release. In the following sections, we outline four possible mechanisms that emerged from recent work and may underlie the differential regulation of distinct forms of neurotransmission.

Mechanisms underlying differential regulation of spontaneous and evoked release

1. Segregation of evoked and spontaneous release at the level of individual presynaptic terminals

Numerous fluorescent imaging studies have documented substantial co-localization of spontaneous and evoked synaptic vesicle recycling in individual synaptic boutons [18, 19, 20, 22, 46, 47]. These findings come with the caveat that typical fluorescent identification criteria used in imaging experiments select for synapses with robust recycling properties, over ones that may manifest a more inefficient release property [18, 22, 47]. Indeed, immature synaptic boutons typically favor spontaneous release and fail to respond to action potential stimulation [48, 49, 50], which raises the possibility that a population of nascent synapses in an otherwise mature synaptic network may selectively sustain spontaneous release. Interestingly, a recent study from our group also revealed a set of presynaptic terminals that support action potential driven release with negligible concurrent spontaneous vesicle exocytosis [14]. The prevalence of these types of synapses is hard to ascertain due to the inherent bias associated with optical analysis in identification of functional synaptic boutons. In this study, majority of synapses (~80%) were capable of both evoked and spontaneous release, although the kinetics of the two forms of release did not show correlation in a given synapse [14]. These findings suggest that spontaneous and evoked release have substantial overlap in their sites of origin, but they may not be linked with respect to their relative activities. Taken together, although complete segregation of spontaneous and evoked neurotransmitter release into different synapses remains an unlikely possibility, there is evidence supporting existence of synaptic niches dominated by one form of release versus another.

2. Differential Ca²⁺-dependence of spontaneous versus evoked neurotransmitter release

Tonic levels of Ca²⁺ in the presynaptic milieu or pulsatile release of Ca²⁺ from internal stores, called Ca²⁺ sparks, can potently regulate neurotransmitter release under resting conditions [51]. Even in the absence of presynaptic action potentials, nerve terminals manifest brief bursts of high fusion activity, clearly deviating from the low frequency, random nature of spontaneous release [52, 53]. These presynaptic bursts are typically triggered by fluctuations in intracellular Ca²⁺ [54], although some studies suggest that resting neurotransmitter release also strictly relies on intracellular Ca²⁺ without these large Ca²⁺ fluctuations [55]. In contrast to the steep Ca²⁺ dependence of evoked transmission, spontaneous neurotransmission displays close to linear Ca²⁺ dependence [34, 55, 56, 57]. This suggests that at low Ca²⁺ concentrations, Ca²⁺ signaling may selectively impact spontaneous release but not evoked release [Figure 1]. This phenomenon may underlie the role of presenilins in regulation of spontaneous release. A further implication of this idea suggests that limited Ca²⁺ influx may impact spontaneous but not evoked release, depending perhaps on the type of channel being activated or on the distance of the channel from the presynaptic release machinery. Calcium entry through non-voltage-gated channels has been shown to affect spontaneous glutamate burst activity (via nicotinic acetylcholine receptor activation and subsequent calcium-induced calcium release) [3] and both evoked and spontaneous asynchronous release (via capsaicin receptor TRPV1 activation) [58]. Finally, in some preparations, spontaneous release is sensitive to blockade of voltage-gated Ca²⁺ channels. In mouse brainstem, both excitatory and inhibitory spontaneous release are substantially decreased when N-type channel function is perturbed either by deletion of the neurexins, a class of trans-synaptic cell adhesion molecules, or by application of a specific channel blocker [29]. Furthermore, phorbol ester treatment of hippocampal neurons potentiates spontaneous release through L-type channel activation, though evoked release was not similarly potentiated [37]. These examples suggest differential calcium signaling contributes substantially to the regulation of spontaneous versus evoked release.

Figure 1. Calcium-dependence of spontaneous and evoked neurotransmitter release.

Diagram depicts the calcium-dependent increase in the rate of vesicle fusion [see 34, 57]. At low intracellular Ca2+ concentrations indicated by the yellow zone, small changes in Ca2+ signaling may selectively alter spontaneous release without significantly modifying evoked release properties.

Ca²⁺-dependent regulation of the spontaneous release rate may also require divergent molecular players compared to evoked release. At least two calcium sensors operate during neurotransmitter release at central synapses. Spontaneously recycling synaptic vesicles possess synaptotagmin1, the canonical Ca²⁺ sensor for synchronous evoked release [18]. Although evoked asynchronous release may utilize an additional sensor or sensors with decreased calcium cooperativity, the majority of spontaneous release appears to be mediated by synaptotagmin1 [55]. However, specific Ca²⁺ binding residues within synaptotagmin1 that support the Ca²⁺-dependence of spontaneous release differ from the key residues that determine cooperativity of evoked release [55]. In addition, synaptotagmin1 appears to function as a fusion clamp for the second, more sensitive sensor exclusively driving spontaneous release, as loss of this protein produced a paradoxical increase in mIPSC frequency [55]. Double C2 domain 2b (doc2b) has recently been identified as an alternate Ca²⁺ sensor specifically mediating spontaneous release [56]. The Ca²⁺-dependence and fusion propensity of spontaneous release can also be modified by other synaptic vesicle proteins such as synaptotagmin12 [59].

3. Differences between presynaptic fusion machineries giving rise to spontaneous and evoked release

Several recent findings indicate that although both forms of fusion largely utilize the same molecular machinery, they rely on distinct molecular interactions of the same components for normal function. Structure-function analyses of neuronal SNAREs (soluble N-ethylmaleimide-sensitive factor attachment protein receptors) including plasma membrane-associated SNAP-25 and vesicular synaptobrevin2 (also called VAMP2), which together with syntaxin1 comprise the core synaptic vesicle fusion machinery [60, 61], revealed three key differences between molecular interactions that give rise to spontaneous and evoked fusion. First, loss of SNAP-25 or synaptobrevin2 in central neurons largely abolishes Ca²⁺-dependent evoked release but leaves some spontaneous release intact [62, 63, 64] suggesting a role for alternate SNAREs in mediating low levels of spontaneous release [65]. Several additional SNAREs which exhibit a domain structure similar to that of synaptobrevin2 are expressed at low levels on synaptic vesicles, including VAMP4, VAMP7 and Vps10p tail interactor 1 a (Vti1a) [66, 67, 68, 69].

Non-canonical SNAREs represent an attractive possibility to mediate specific forms of neurotransmission; indeed, recent studies implicate VAMP7 in the regulation of asynchronous and spontaneous release at the mossy fiber terminals [69]. Additionally, the secretagogue α-latrotoxin can augment resting levels of release without relying on the canonical SNARE machinery components, implicating that a separate complement of molecules may support spontaneous transmission [1].

Second, in synaptobrevin2-deficient synapses, spontaneous release could be rescued by expression of a synaptobrevin2 construct with an insertion of 12 residues between the SNARE motif and transmembrane region, whereas the same construct was not able to restore action potential evoked release [70], indicating the physical constraints on SNARE complex assembly are less stringent for spontaneous release. Finally, expression of SNAP-25 mutants destabilizing the C-terminal end of the SNARE-bundle in neurons obtained from SNAP-25 null mice abolished spontaneous neurotransmitter release but caused only a small reduction in evoked release probability. In contrast, destabilizing the middle or deleting the N-terminal end of the SNARE-bundle potentiated the propensities of both spontaneous and evoked fusion. Interestingly, both manipulations caused a more dramatic change in spontaneous release compared to evoked neurotransmission [71]. These biophysical differences in SNARE complex formation between spontaneous and evoked vesicle fusion may depend on the interactions of specific auxiliary molecules, such as complexin. Indeed, complexin knockdown produces reciprocal effects on spontaneous and evoked release in cultured cortical neurons, where the frequency of excitatory mEPSCs is increased fourfold whereas evoked EPSC amplitudes are decreased fourfold [72]. This growing body of evidence describes clear distinctions between SNARE complexes driving spontaneous and evoked vesicle fusion, which may be regulated by a specific molecule or molecules at the level of SNARE complex formation.

4. Diversity of synaptic vesicle populations giving rise to spontaneous and evoked neurotransmission

Synaptic vesicles are classically assigned to one of three pools based primarily on their latency to release upon stimulation: the readily releaseable pool (RRP), the recycling pool, and the reserve pool [73]. Although synaptic vesicle pools are indistinguishable at the ultrastructural level (apart from the RRP vesicles docked at the active zone) and no molecular basis for separation of the recycling and reserve pools has been described, several studies have suggested that the two vesicle populations do not overlap and function to maintain evoked and spontaneous transmission, respectively [18, 19, 20], supporting distinctions not only in fusion mechanisms but also in the overall identity of vesicles that sustain the two forms of release [Figure 2]. Key evidence supporting the “two pools” hypothesis comes from the finding that spontaneously recycled vesicles preferentially exocytose spontaneously rather than with activity, and that blocking vesicle refilling at rest by preventing vesicle re-acidification selectively decreased spontaneous transmission without significantly affecting evoked transmission [18]. This idea is controversial, as some studies have described contradictory findings whereby vesicles from the same population are released spontaneously and with activity [22, 23, 24]. While the majority of studies addressing this issue utilize a number of classical and novel fluorescent imaging methodologies to monitor synaptic vesicle release and recycling [18, 19, 20, 21, 22, 23, 24], new evidence from the postsynaptic side points to a model in which distinct receptor populations are activated by spontaneous or evoked glutamate release in hippocampal neurons [14]. This observation is also consistent with the total internal reflection microcopy based observations of spontaneous and evoked vesicle fusion in goldfish retinal bipolar cells [15]. The notion that postsynaptic receptors detecting evoked or spontaneous transmitter release are spatially segregated on the same synapse was convincingly demonstrated with optical imaging and modeling approaches [14, 15], and has precedence in the finding that low (40 action potentials at 2 Hz) frequency stimulation recycles vesicles to a specific pool that is kinetically faster than those vesicles mobilized by high frequency stimulation, and is organized closer to the active zone [75]. Moreover, acute application of dynasore, a reversible inhibitor of the essential endocytic protein dynamin, showed that evoked synchronous and asynchronous release originate from the same vesicle pool that recycles rapidly in a dynamin-dependent manner, while a distinct vesicle pool sustains spontaneous release independent of dynamin activation [2]. These findings imply that the distinct identities of spontaneous and evoked recycling vesicles are not perturbed upon exocytosis-endocytosis. The premise of two separate pools for spontaneous and evoked release is also consistent with the prevalence of synapses that only support spontaneous neurotransmission and spontaneous synaptic vesicle recycling at early stages of synapse maturation [48, 49, 50]. Interestingly, purified synaptic vesicles show an intrinsic tendency for unregulated constitutive fusion [76], suggesting that evoked regulated fusion may constitute a gain-of-function which is attained gradually during synapse maturation. Accordingly, mature synapses may also contain a population of these “immature” vesicles that are unable to respond to brief action potential stimulation but fuse and recycle constitutively [14, 48]. Ultimately, in light of very recent publications questioning the distinction between vesicles used for spontaneous or evoked transmission [23, 24], controversy still remains as to whether vesicles released spontaneously or with stimulation populate the same or different pools. Future studies with better molecular specificity will help resolve this issue.

Figure 2. Segregation of evoked and spontaneous neurotransmission.

(A) A recent study using total internal reflection fluorescence microscopy on retinal bipolar cell presynaptic terminals showed that spontaneous fusion events were largely excluded from synaptic ribbons which comprised the preferential site for evoked fusion [15].

(B) Work in hippocampal synapses suggested that spontaneous and evoked fusion may be carried out via separate pools of vesicles, which may recycle independently [18, 19, 20]. Taken together with the evidence from the postsynaptic side [14], these observations points to a model in which distinct receptor populations are activated by spontaneous or evoked glutamate release in hippocampal neurons [14]. This model is also consistent with the total internal reflection microcopy based observations of spontaneous and evoked vesicle fusion in goldfish retinal bipolar cells [15].

(C) A study by Fredj and Burrone (2009) [19] suggested that spontaneous neurotransmitter release is preferentially maintained by the resting pool of vesicles that do not typically contribute to evoked neurotransmission.

(D) Immature synaptic boutons typically favor spontaneous release and fail to respond to action potential stimulation [48, 49, 50], which raises the possibility that a population of nascent synapses in an otherwise mature synaptic network may selectively sustain spontaneous release. In addition, some presynaptic terminals may support action potential driven release with negligible concurrent spontaneous vesicle exocytosis [14]. Therefore, some synapses may have a strong propensity for spontaneous fusion whereas others may preferentially release neurotransmitter in response to action potentials.

Conclusions

Recent studies suggest a framework where certain neuronal signaling pathways specifically regulate spontaneous neurotransmitter release. This release mode selective modulation of neurotransmission supports the premise that spontaneous and evoked neurotransmission comprise independent neuronal signal transduction pathways that may operate in a spatially segregated manner [6, 77]. Elucidation of molecular substrates that underlie this regulatory selectivity will allow dissection of neurotransmitter signaling with respect to its origins, downstream targets as well as behavioral consequences. This approach can open new avenues in neuronal signaling by enabling manipulation of neurotransmission in a release mode specific manner.