The signalling roles of Ca2+ic (intracellular Ca2+) stores are well established in non-neuronal and neuronal cells. In neurons, although Ca2+ic stores have been assigned a pivotal role in postsynaptic responses to Gq-coupled receptors, or secondarily to extracellular Ca2+ influx, the functions of dynamic Ca2+ic stores in presynaptic terminals remain to be fully elucidated. In the present paper, we review some of the recent evidence supporting an involvement of Ca2+ic in presynaptic function, and discuss loci at which this source of Ca2+ may impinge. Nerve terminal preparations provide good models for functionally examining putative Ca2+ic stores under physiological and pathophysiological stimulation paradigms, using Ca2+-dependent activation of resident protein kinases as sensors for fine changes in intracellular Ca2+ levels. We conclude that intraterminal Ca2+ic stores may, directly or indirectly, enhance neurotransmitter release following nerve terminal depolarization and/or G-protein-coupled receptor activation. During conditions that prevail following neuronal ischaemia, increased glutamate release instigated by Ca2+ic store activation may thereby contribute to excitotoxicity and eventual synaptopathy.
It is well established in presynaptic nerve terminals that depolarization-dependent Ca2+ec (extracellular Ca2+) influx through VDCCs (voltage-dependent Ca2+ channels) leads to the exocytosis of synaptic vesicles as well as the activation of signalling cascades leading to stimulation of protein kinases and phosphatases. Increasingly, a role of store-resident or Ca2+ic (intracellular Ca2+) in presynaptic function is being intimated, certainly in modulating neurotransmitter release and presynaptic plasticity , but also in the direct mediation of exocytosis in some cases .
In early studies with presynaptic terminals, the paucity of morphological evidence substantiating presynaptic Ca2+ic stores, akin to those found in the neuronal cell bodies, was posited as an argument for the exclusive regulation of neurotransmitter release by voltage-dependent Ca2+ec influx. However, more recently, improvements in ultrastructural techniques have provided increasing support for the existence of intraterminal Ca2+ stores . Moreover, underpinned by this physical evidence, functional data for the operation of Ca2+ic stores, in a number of presynaptic models , have resulted in the early question of ‘are Ca2+ic stores relevant presynaptically?’ being superseded with ‘how are Ca2+ic stores involved in nerve terminal function?’
Triggering of exocytosis: are Ca2+ic stores sufficient for supporting synaptic vesicle fusion?
The determination of the triggering of vesicular exocytosis is contingent on three aspects relating to Ca2+, i.e. the localization of the Ca2+ source, the kinetics of the [Ca2+] change and the affinity of Ca2+ sensors expediting exocytosis or the regulation thereof . Synchronous release is postulated to occur through a trigger with a Ca2+ affinity in the tens of micromolar range, and highly co-operative binding of the ion (fourth or fifth power) [5,6]. Synaptotagmins 1, 2 and 9 fit this profile of fast Ca2+-binding proteins and are thereby cited as the Ca2+ sensors detecting localized increases in [Ca2+] in nano- and/or micro-domains established transiently following the activation of P/Q- and/or N-type VDCCs by single action potentials .
Asynchronous release appears to occur as a result of extended elevation of cytosolic [Ca2+], usually following repetitive stimulation of nerve terminals. Although this was initially thought to point to a Ca2+ sensor for asynchronous release with a higher affinity for Ca2+ than the aforementioned synaptotagmins, the functional data in fact evince an, as yet, unidentified trigger with an affinity comparable with that for synchronous release, but with slower binding and markedly lower co-operativity (second power) . Finally, spontaneous release, perhaps surprisingly, depends on synaptotagmin 1 as it is affected by mutations altering the Ca2+-binding affinity of the sensor. However, when synaptotagmin 1 is ablated, spontaneous release evidently adopts the Ca2+ sensor attributed to asynchronous release .
Notwithstanding the identity of Ca2+ sensors, if Ca2+ic store-derived Ca2+ is to directly expedite exocytosis, the spatiotemporal characteristics of the elevation of cytosolic [Ca2+] produced must, to some degree, mimic those occurring following Ca2+ec influx (Figure 1). That is, to contribute to neurotransmitter release directly, the release of Ca2+ic must be of sufficient magnitude and speed in the context of a single action potential. In situations where these strict criteria are not fulfilled, Ca2+ic stores may nevertheless contribute to the facilitation of neurotransmitter release, through modulatory influences ultimately coding presynaptic plasticity.
Presynaptic pathways in the regulation of Ca2+ic and downstream effectors/signalling
Key factors that would be expected to affect the impact of Ca2+ic release include the proximity of the Ca2+ic stores to active zones, the intraterminal buffering capacity, the degree of loading of the Ca2+ic, and the frequency of stimulation. There is ultrastructural evidence to suggest that putative Ca2+ stores are often juxtapositioned to active zones . Coupled with the small size of many presynaptic boutons, it is feasible that Ca2+ic stores could contribute to the Ca2+ microdomains that determine neurotransmitter release triggering. This is, however, obviously dependent on intraterminal Ca2+ buffers, but even with the diffusional constraints imposed by buffers, [Ca2+]ic is predicted to peak in a few milliseconds or less in small terminals of ~1–2 μm radius. The size of the Ca2+ic signal will also rather depend on the extent to which the store is loaded before stimulation; this may not only determine the amount of Ca2+ released when a store is replete, but also could alternatively reflect net Ca2+ uptake/buffering if the compartment is relatively depleted. Finally, the impact of store-derived Ca2+ic may possibly be realized under conditions of repetitive stimulation. Under this type of stimulation paradigm, residual Ca2+ levels following Ca2+ec influx would in any case rise as a result of some mitigation of buffering capacity, but, importantly, this may additionally produce conditions under which the Ca2+ic store-derived signal can actually manifest in directly triggering neurotransmitter release (Figure 1).
Ca2+ic stores and pharmacological regulators
The major Ca2+ic stores derived from smooth endoplasmic reticulum, which are virtually omnipresent in mammalian cells, are those activated by RyRs (ryanodine receptors) and IP3Rs (InsP3 receptors) . Stores with RyRs support direct CICR (Ca2+-induced Ca2+ release), whereas IP3Rs underpin IICR (InsP3-induced Ca2+ release), which occurs following second messenger InsP3 liberation from membrane PtdIns(4,5)P2 metabolism by GPCR [G (Gq)-protein-coupled receptor] -activated PLC (phospholipase C). Indeed, as IICR appears to require the synergistic action of InsP3 and Ca2+, there may well be signal amplification through reciprocal interaction between IICR and CICR from independent or co-resident Ca2+ic stores.
RyR- and IP3R-sensitive Ca2+ic stores are established and maintained by SERCA (sarcoplasmic/endoplasmic reticulum Ca2+-ATPase) pumps, which primarily contribute a major component of the Ca2+ sequestration in cells . This capacity can be abrogated by SERCA pump inhibitors such as thapsigargin and cyclopiazonic acid. Interestingly, together with RyR- and IP3R-activated Ca2+ic stores, there is accumulating evidence for another Ca2+ic store that is insensitive to thapsigargin, but is depleted by proton pump inhibitors such as bafilomycin, implying an acidic storage compartment . This pool has several distinguishing characteristics, but most conspicuous is its mobilization by NAADP (nicotinic acid–adenine dinucleotide phosphate) and functional co-operation with RyR- and IP3R-activated Ca2+ic stores .
Presynaptically, even though Ca2+ic stores have been difficult to delineate morphologically, there is now good evidence for the presynaptic localization of both RyRs and IP3Rs, therefore suggesting the existence of active intraterminal Ca2+ic stores . Indeed, functional studies evince presynaptic activity-dependent mobilization of RyR-dependent Ca2+ic stores by CICR, as well as IICR, by presynaptic GPCR-coupled recruitment of IP3R-dependent Ca2+ic pools . The involvement of Ca2+ic stores in presynaptic function has been pharmacologically studied by the use of SERCA pump inhibitors such as thapsigargin or cyclopiazonic acid, which abrogate the Ca2+ sequestration aspect of RyR- and IP3R-dependent stores. Additionally, putative modulators of RyRs, such as caffeine/theophylline (+), ryanodine (+/−), Ruthenium Red (−), dantrolene (−), and IP3Rs, such as 2-APB (2-aminoethoxydiphenyl borate) (−) and xestospongin C (−), have also been employed to respectively elucidate the roles of CICR and IICR in a number of models, albeit with some caution given the multiplicity of targets for some of these agents. The following account reviews some of these studies in relation to neurotransmitter release at central synapses.
Can Ca2+ic store-derived increases in cytosolic [Ca2+] trigger exocytosis?
A number of studies argue for the sufficiency of Ca2+ic stores in directly supporting exocytosis with experiments that correlate Ca2+ic store mobilization with spontaneous neurotransmitter release. Random increases in terminal [Ca2+], variously described as ‘Ca2+ sparks’ and spontaneous Ca2+ transients [12–14], underlie increases in spontaneous neurotransmitter release indicated by increased frequencies of mIPSCs (miniature inhibitory postsynaptic currents) and mEPSCs (miniature excitatory postsynaptic currents) at inhibitory [12,13,15,16] and excitatory [17,18] synapses respectively.
Perhaps experimentally less tractable has been the delineation of the involvement of Ca2+ic stores in evoked release, not least because of the overarching predominance of extracellular Ca2+ec influx through VDCCs supporting exocytosis during nerve terminal depolarization (Figure 1). Ca2+ responses to depolarization of hippocampal mossy fibre terminals are certainly sensitive to pharmacological manipulation of terminal and axonal Ca2+ic stores [19,20]. However, it would appear that, although these Ca2+ic stores play a modulatory role in synaptic plasticity at excitatory synapses (see below), they do not significantly affect single evoked EPSCs (excitatory postsynaptic currents) [17,21]. In contrast with the aforementioned situation at excitatory synapses, effects of CICR regulation have been noted on evoked IPSCs (inhibitory postsynaptic currents) in cerebellar Purkinje cells produced by GABA (γ-aminobutyric acid) release from basket cell terminals . Arguably, the most convincing evidence for the sufficiency of Ca2+ic stores in supporting exocytosis has come from examples where depolarization-dependent activation of Ca2+ic stores and peptide release can be shown to occur in the absence of external Ca2+, therefore not consequent on CICR as such, but as a result of L-type VDCCs potentially being physically coupled to the ryanodine-sensitive stores [23,24]. Interestingly, this and other studies looking at large dense-core vesicle release may well provide clear examples of exocytosis supported by Ca2+ic stores because of the higher-affinity Ca2+ sensor/trigger thought to be associated with aminergic and peptidergic exocytosis .
Ca2+ic stores in the modulation of neurotransmitter release
Notwithstanding the accumulating evidence for the direct role of Ca2+ic stores, certainly in spontaneous exocytosis, but also perhaps in evoked release in some instances, there are arguably firmer grounds for invoking Ca2+ic stores in the modulation of neurotransmitter release and thereby underpinning aspects of presynaptic plasticity. Apart from repetitive stimulation and influx of Ca2+ec via VDCCs, some ionotropic presynaptic receptors instigate Ca2+ec influx to initiate CICR, and subsequently contribute to presynaptic plasticity. For instance, kainate-type glutamate-receptor-mediated LTP (long-term potentiation) at mossy fibre–CA3 pyramidal synapses in the hippocampus is contingent upon receptor-mediated Ca2+ entry effecting CICR in the excitatory presynaptic terminal . In other studies, activation of single mossy fibre terminals supported kainate-receptor-dependent and Ca2+ic store-sensitive paired-pulse facilitation .
Interestingly, Ca2+-influx following NMDA (N-methyl-D-aspartate) receptor activation also supports CICR and appears to underpin synaptic down-regulation in LTD (long-term depression) at both excitatory CA3–CA3 pyramidal cell neuron synapses  and inhibitory interneuron–CA3 pyramidal cell synapses in the hippocampus . In the cerebellum, retrograde signalling to presynaptic NMDA receptors in response to glutamate released by target Purkinje cells leads to CICR, which increases mIPSC frequencies as well as evoked GABA release from inhibitory interneurons . Other than the aforementioned glutamate-receptor-mediated CICR, Ca2+-conducting nicotinic acetylcholine receptor activation also leads to Ca2+ic store mobilization and modulation of glutamate release from hippocampal mossy fibre terminals [30,31] and isolated nerve terminals from the prefrontal cortex .
Receptor-mediated activation of PLC leading to InsP3 production may also play a regulatory role in presynaptic plasticity through the activation of IICR. Group 1 mGluRs (metabotropic glutamate receptors) are found presynaptically and facilitate the release of glutamate [33,34]. Recent observations indicate that part of the modulation of glutamate release instigated by the PLC-coupled receptors may be underpinned by nerve terminal Ca2+ic stores (Figure 1) . The role of IICR in neurotransmitter release has been posited in several models. One clear example of this has arisen from work characterizing the molecular target of the spider toxin latrotoxin. It transpires that the toxin binds to a GPCR termed latrophilin, which, through the activation of a PLC, mobilizes an IP3R-dependent Ca2+ic store in nerve terminals, and thereby promotes neurotransmitter release [36,37]. In general, it is likely that IICR and CICR are functionally interdependent. Not only can Ca2+-influx and CICR invoke Ca2+-dependent PLC activation to effect mGluR -independent production of InsP3 (Figure 1), but it is evident that IICR may also be activated through high-affinity Ca2+-binding proteins which act as agonists at IP3R, without the need for InsP3 itself . Intriguingly, physical bridging of group 1 mGluRs with IP3R-dependent Ca2+ic stores by Homer proteins  has been suggested to effect agonist-independent mobilization of postsynaptic Ca2+ stores, although it remains to be seen whether a similar mechanism operates presynaptically and indeed affects modulation of neurotransmitter release.
High-affinity presynaptic effector targets for Ca2+ic
The foregoing data indicate that, in some circumstances, even when Ca2+ic levels are insufficient to affect exocytosis directly, Ca2+ic stores may nonetheless be of paramount importance in a regulatory role. Thus, mechanistically, Ca2+ic store-derived Ca2+ may contribute to the modulation of steps upstream of exocytosis itself, e.g. during vesicle recruitment and priming. This is feasible because some of these steps are contingent on the activation of high-affinity Ca2+ sensors, which would be activated by cytosolic [Ca2+] that may be an order of magnitude lower than the activation threshold for the relatively low-affinity Ca2+ triggering of exocytosis, but above the resting cytosolic [Ca2+] (50–100 nM). The major candidate in this regard is the prototypic Ca2+-binding protein CaM (calmodulin), which, with a Ca2+ affinity in the sub-micromolar range (500 nM) , is ideally suited to sense regulatory bulk changes in cytosolic [Ca2+] resultant from Ca2+ec influx or indeed Ca2+ic store activation. Ca2+-binding proteins with even higher affinity may well respond to resting [Ca2+] .
CaM has numerous neuronal effectors, including proteins such as Munc13, myosin V, Rab3A and synaptobrevin, representing effectors that are either essential for exocytosis or are important for the modulation of neurotransmitter release . Interestingly, in this regard, the essential Munc13 family of vesicle priming proteins are modulated by Ca2+ alone at low cytosolic [Ca2+], but employ Ca2+/CaM at the higher [Ca2+] that prevails during nerve terminal activity . One CaM effector that has major relevance in the activity-dependent modulation of presynaptic function is CaMKII (Ca2+/CaM kinase II). The role of this effector in neurotransmitter release and its potential as a high-affinity sensor for Ca2+ic store activation is discussed below.
CaMKII is a multifunctional protein kinase [44,45] with a conspicuous presence on synaptic vesicles [46,47]. As such, it was shown some years ago to facilitate neurotransmitter release  through a mechanism involving synapsin I , one member of a family of neuronal- and nerve-terminal-specific proteins (Figure 1) . Synapsin I cross-links small synaptic vesicles with the actin cytoskeleton  and, in so doing, defines a reserve pool (Figure 1) of neurotransmitter release. Activity-dependent stimulation of CaMKII effects phosphorylation of synapsin I at specific sites and promotes the recruitment of synaptic vesicles  into a pool of vesicles that can be primed for entry into a readily-releasable pool (Figure 1). CaMKII activation can evidently occur with increases of cytosolic [Ca2+], which are subthreshold with respect to triggering of exocytosis, but sufficient to effect synapsin I phosphorylation . Consequently, CaMKII-dependent phosphorylation of synapsin I therefore represents a potential target for Ca2+ released by Ca2+ic store activation, through either CICR or IICR (Figure 1). The former could well form the basis of action potential integration by Ca2+-dependent synapsin phosphorylation underpinned by ‘residual’ and/or Ca2+ic store-derived Ca2+. The predicted facilitation of neurotransmitter release could contribute to asynchronous release in the first instance .
Addressing Ca2+ic store activation in isolated nerve terminals (synaptosomes), an ideal biochemical model for looking at neurotransmitter release, proves difficult because the small size of the terminals precludes the optical delineation of stores using fluorimetric reporters, even if reporters of appropriate affinity for detecting small changes in [Ca2+] are available. Alternatively, the detection of discrete changes in cytosolic [Ca2+] can been expedited by virtue of the ability of the cation to activate CaM-dependent phosphorylation. Notably in this regard, activation of CaMKII  and synaptosomal ERK (extracellular-signal-regulated kinase) 1/2 [55,56] can be used as sensors for Ca2+ic store mobilization (Figure 1).
Ca2+ influx into synaptosomes through VDCCs activated by the K+-channel blocker 4-AP (4-aminopyridine)-induced depolarization, or directly using the Ca2+ ionophore ionomycin , evince CICR sensitive to the SERCA pump inhibitor thapsigargin (Figure 1) . Thapsigargin is a well-established tool which depletes RyR- and IP3R-dependent Ca2+ic stores by way of the inherent leakiness of these compartments, or the inability of stores to refill after previous activation/depletion. Tentative evidence for IICR arises in this model from the sensitivity of Ca2+-dependent kinase phosphorylation/activation to the IP3R inhibitor 2-APB. Although the latter agent is known to have other targets , credence for a GPCR/PLC/IP3R signalling cascade has been provided by experiments using the mGluR1/5 receptor agonist DHPG (3,5-dihydroxyphenylglycine). mGluR1/5 activation couples, through Gq, to stimulate PLC to generate InsP3 from PtdIns(4,5)P2 . DHPG-mediated increases in phosphorylation  were dependent on CaM (i.e. inhibited by the CaM inhibitor, W7) and were sensitive to thapsigargin pre-treatment (Figure 1) , invoking the role of Ca2+ic stores in the mGluR1/5-mediated facilitation of glutamate release reported previously [33,60].
Dysfunction of the regulatory roles of the invoked presynaptic Ca2+ic stores, could contribute to synaptopathies. Following neuronal/nerve terminal ischaemia, one immediate consequence of the resultant reduced levels of ATP would be a compromised Na+/K+-ATPase activity, which is responsible for the electrogenic maintenance of Na+-gradients (3 Na+ out/2 K+ in) across neurons and other excitable cells. To model this pathophysiological paradigm, synaptosomal Na+/K+-ATPase activity can be inhibited pharmacologically using the glycoside ouabain (Figure 1). This treatment has the advantage of leaving ATP levels intact, thus obviating any general effects on kinases which obviously use the nucleotide as a substrate. Ouabain treatment of synaptosomes increases Ca2+-dependent kinase phosphorylation/activation, which was again sensitive to pre-treatment with the calmodulin inhibitor W7 and SERCA pump inhibitor thapsigargin (Figure 1) . These data suggest that activation of CICR following Na+/K+-ATPase inhibition may lead to stimulation of signalling resulting in overactivation of exocytosis . Thus, at central synapses, ‘inappropriate’ mobilization of intraterminal Ca2+ic stores (and downstream effects) may be one deleterious consequence of a compromised Na+/K+-ATPase activity. For instance, following ischaemic insult, initial hyperactivation of general Ca2+-dependent signalling instigated by Ca2+ic store-derived Ca2+ may initiate a sequence of events leading to an excessive stimulation of glutamate release, known to be instrumental in excitotoxicity.
There is increasing evidence to suggest that intracellular Ca2+ic stores resident in nerve terminals can directly support spontaneous neurotransmitter release in some models for neurotransmitter release, and also evoked release in other studies. Most compelling is the increasing evidence that Ca2+ic stores are instrumental in modulatory events upstream of exocytosis which may impinge on neurotransmitter release and presynaptic plasticity. Given the part played by Ca2+ stores in nerve terminal function, dysfunction of this mechanism could underlie the pathophysiology following nerve terminal ischaemia.
Synaptopathies: Dysfunction of Synaptic Function: A Biochemical Society Focused Meeting held at The Hotel Victoria, Newquay, U.K., 2–4 September 2009. Organized and Edited by Nils Brose (Max Planck Institute for Experimental Medicine, Göttingen, Germany), Vincent O'Connor (Southampton, U.K.) and Paul Skehel (Centre For Integrative Physiology, Edinburgh, U.K.)
Ca2+/CaM-dependent kinase II
Ca2+-induced Ca2+ release
InsP3-induced Ca2+ release, IP3R, InsP3 receptor
miniature excitatory postsynaptic current
metabotropic glutamate receptor
miniature inhibitory postsynaptic current
sarcoplasmic/endoplasmic reticulum Ca2+-ATPase
voltage-dependent Ca2+ channel
S.N. is supported by a Biotechnology and Biological Sciences Research Council Ph.D. studentship, and V.W.Y.L. and J.D. were supported by Medical Research Council studentships. P.L. was a recipient of a studentship from the School of Pharmacy (London). J.N.J. and T.S.S. have been supported by the Biotechnology and Biological Sciences Research Council. We thank A.L.K. Sihra-Jovanovic and L.A.K. Sihra-Jovanovic for their helpful discussion.