The Trinity of Ca²⁺ Sources for the Exocytotic Glutamate Release from Astrocytes

Abstract

Astrocytes can exocytotically release the transmitter glutamate. Increased cytosolic Ca²⁺ concentration is necessary and sufficient in this process. The source of Ca²⁺ for the Ca²⁺-dependent exocytotic release of glutamate from astrocytes predominately comes from endoplasmic reticulum (ER) stores with contributions from both inositol 1,4,5 trisphosphate- and ryanodine/caffeine- sensitive stores. An additional source of Ca²⁺ comes from the extracellular space via store-operated Ca²⁺ entry due to the depletion of ER stores. Here transient receptor potential canonical type 1 containing channels permit entry of Ca²⁺ to the cytosol, which can then be transported by the store-specific Ca2+-ATPase to (re)fill ER. Mitochondria can modulate cytosolic Ca²⁺ levels by affecting two aspects of the cytosolic Ca²⁺ kinetics in astrocytes. They play a role in immediate sequestration of Ca²⁺ during the cytosolic Ca²⁺ increase in stimulated astrocytes as a result of Ca²⁺ entry into the cytosol from ER stores and/or extracellular space. As cytosolic Ca²⁺declines due to activity of pumps, such as the smooth ER Ca²⁺-ATPase, free Ca²⁺ is slowly released by mitochondria into cytosol. Taken together, the trinity of Ca²⁺ sources, ER, extracellular space and mitochondria, can vary concentration of cytosolic Ca²⁺ which in turn can modulate Ca²⁺- dependent vesicular glutamate release from astrocytes. An understanding of how these Ca²⁺ sources contribute to glutamate release in (patho)physiology of astrocytes will provide information on astrocytic functions in health and disease and may also open opportunities for medical intervention.

A developing body of evidence implicates astrocytes as active participants in multi-directional signaling in the central nervous system (CNS) [reviewed in (Haydon and Carmignoto, 2006; Theodosis et al., 2008)]. Astrocytes communicate with neurons (Nedergaard, 1994; Parpura et al., 1994), endothelial cells of the cerebrovasculature (Zonta et al., 2003; Mulligan and MacVicar, 2004), and microglia (Davalos et al., 2005). They appear to be an integral part of neural communication throughout the CNS. For example, in the hippocampus, astrocytes integrate neuronal signals (Perea and Araque, 2005), potentiate transmitter release at single synapses (Perea and Araque, 2007), enhance long-term potentiation (Pascual et al., 2005), and synchronize neuronal activity (Fellin et al., 2004). In the thalamus, astrocytes have been shown to drive neuronal activity as a consequence of astrocytic intrinsic Ca²⁺ oscillations (Parri et al., 2001). In the supraoptic nucleus of the hypothalamus, morphological plasticity of astrocytes (Hatton, 2004) leading to differential synaptic coverage by astrocytic processes can modulate the activity of the output neurons (Panatier et al., 2006).

The abundance of astrocytes and their proximity to neurons enables their communication with neurons. Early studies indicated the numerical preponderance of glial cells in the CNS. For example, in layers I and IV of the rat cerebral cortex, glial cells consisting of astrocytes and oligodendrocytes, outnumbered neurons 3 to 1 (Bass et al., 1971). As more sophisticated tools for cell counting emerged, it appears, however, that the glia to neuron ratio in human cortex is ~ 1.65:1, while in rodent this ratio is about 0.3:1 (Nedergaard et al., 2003; Sherwood et al., 2006). Astrocytes occupy distinct domains within the CNS, only at the cells’ peripheries do their processes overlap (Bushong et al., 2002; Ogata and Kosaka, 2002). Within their individual domains astrocytes have extensive morphological interactions with neurons. For example, in the cortex of adult mice, one astrocyte may contact 4 to 8 neurons and surround ~300 to 600 neuronal dendrites (Halassa et al., 2007). In the hippocampus, astrocytes are positioned to contact even more synapses. In adult rats, one astrocyte is estimated to contact ~140,000 synapses of CA1 pyramidal cells (Bushong et al., 2002). Classically, the intimate position of astrocytic processes around the neurons and the synapses has been solely attributed to glutamate uptake and regulation of extracellular K⁺ by astrocytes. To provide these services, astrocytes possess excitatory amino acid transporters (EAAT’s) (Huang and Bergles, 2004; Tzingounis and Wadiche, 2007; Yang and Rothstein, 2009) and K⁺ inward rectifying (Kir) channels (Sontheimer et al., 1994; Kofuji and Newman, 2009), respectively. However, it is also at synapses where astrocytes have been shown to have modulatory effects on the activities of presynaptic and postsynaptic neurons by astrocytic ability to release a variety of transmitters using many different mechanisms [reviewed in (Malarkey and Parpura, 2009)]. This led to the concept of a functional tripartite synapse (Araque et al., 1999), a topic that has been recently reviewed elsewhere (Ni et al., 2007; Halassa and Haydon, 2009).

In this review we focus on the Ca²⁺ sources for the exocytotic release of the excitatory transmitter glutamate from astrocytes, which can readily occur under physiological conditions (Parpura and Haydon, 2000). We start with a historical brief on the Ca²⁺-dependent glutamate release from astrocytes [reviewed in detail in (Montana et al., 2004)]. We then discuss the trinity of Ca²⁺ sources for this release: intracellular stores, extracellular space and mitochondria. An understanding of how these Ca²⁺ sources contribute to the glutamate release will provide information on glial functions in heath and disease and may also introduce opportunities for medical intervention.

Ca²⁺-dependent glutamate release in astrocytes: a historical brief

Astrocytes can couple their inherent intracellular Ca²⁺ excitability to the exocytotic release of transmitters including but not limited to glutamate (Parpura et al., 1994). Throughout this review we interchangeably use Ca²⁺ -dependent and exocytotic attributes of glutamate release from astrocytes. Excitation-secretion coupling in astrocytes involves the molecular machinery for vesicular fusion, including the core members of the soluble N-ethyl maleimide-sensitive fusion protein attachment protein receptor (SNARE) complex, as well as proteins responsible for packaging of glutamate inside vesicles, vacuolar type proton ATPase and vesicular glutamate transporters [reviewed in (Montana et al., 2006; Malarkey and Parpura, 2008)].

Evidence for Ca²⁺-dependent release of glutamate from astrocytes was originally demonstrated in experiments where the application of the Ca²⁺ ionophore ionomycin stimulated the release of glutamate from astrocytes in the presence of external free Ca²⁺ (2.4 mM). However, this ionophore failed to cause glutamate release when applied to astrocytes that were equilibrated in solution with low external free Ca²⁺ (24 nM), for 40–60 minutes, thus causing the depletion of internal Ca²⁺ stores and preventing the Ca²⁺ entry from the extracellular space (Parpura et al., 1994). These data indicate that Ca²⁺ is sufficient and necessary to cause glutamate release from astrocytes. Indeed, depleting internal Ca²⁺ stores by application of thapsigargin, a blocker of the sarco-endoplasmic reticulum Ca²⁺-ATPase (SERCA), or buffering cytoplasmic Ca²⁺ with the membrane permeable Ca²⁺ chelator, 1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid acetoxymethyl ester (BAPTA-AM), resulted in reduction of glutamate release (Araque et al., 1998a; Bezzi et al., 1998; Innocenti et al., 2000; Hua et al., 2004; Montana et al., 2004). Furthermore, alternative stimuli that directly increased astrocytic intracellular Ca²⁺ concentration([Ca²⁺]_i), such as mechanical stimulation (Parpura et al., 1994; Araque et al., 1998a; Araque et al., 1998b; Hua et al., 2004; Montana et al., 2004), photostimulation (Parpura et al., 1994), and photolysis of Ca²⁺ cages (Araque et al., 1998b; Parpura and Haydon, 2000), all caused release of glutamate.

The trinity of Ca²⁺ sources

The source of Ca²⁺ for Ca²⁺-dependent release of glutamate from astrocytes is tripartite, from (i) internal ER stores, (ii) extracellular space and (iii) mitochondria (Figure 1).

Figure 1.

Tripartite Ca²⁺_cyt handling in Ca²⁺- dependent vesicular glutamate release from astrocytes. A drawing of sources of Ca²⁺ for [Ca²⁺]_cyt increase. [Ca²⁺]_cyt can be affected by the endoplasmic reticulum (ER), store-operated Ca²⁺ (SOC) entry and mitochondria. Ca²⁺_cyt accumulation could be caused by the entry of Ca²⁺ from the internal stores of the ER that possess IP₃ and ryanodine receptors. The pre-incubation of astrocytes with ryanodine (Ry) can block Ca²⁺ release from ryanodine/caffeine-sensitve ER stores, while diphenylboric acid 2-aminoethyl ester (2-APB), an IP₃R antagonist, reduces Ca²⁺ release from an IP₃-senistive stores. Thapsigargin (Thaps) and cyclopiazonic acid (CPA), blockers of store specific Ca²⁺-ATPase, prevent filling of these stores by Ca²⁺ and, thus, deplete stores of this ion. Ultimately, the (re)filling of ER Ca²⁺ stores requires Ca²⁺ entry from the extracellular space (ECS) through SOC channels, in particular those containing canonical type 1 transient receptor potential (TRPC1) protein. A specific TRPC1 antibody blocks TRPC1 containing SOC channels and reduces Ca²⁺ entry to the cytosol. A negative membrane potential (about −160 mV) exists across the inner mitochondrial membrane. Mitochondrial Ca²⁺ uptake is mediated by the uniporter as the electropotential gradient drives Ca²⁺ into the matrix. Ru360 blocks Ca²⁺ influx through the uniporter. Free Ca²⁺ exits the mitochondrial matrix through the Na⁺/Ca²⁺ exchanger, and this process can be blocked by CGP37157. Ca²⁺ accumulation in mitochondria is increased by cyclosporin A (CsA) via preventing formation of the mitochondrial permeability transition pore, which releases Ca²⁺ and other components of the mitochondrial matrix at high Ca²⁺ loads. Increase of Ca²⁺_cyt is sufficient and necessary to cause vesicular fusions and release of glutamate, which requires action by proteins of the ternary SNARE complex and their associated proteins (not shown). Additionally, vesicles possess the vacuolar type H⁺-ATPase which creates the proton gradient used by vesicular glutamate transporters to fill these organelles with glutamate (Glu). Drawing is not to scale.

(i) Inositol 1,4,5 trisphosphate (IP₃)- and ryanodine/caffeine-sensitive ER stores

The ER is the major source of intracellular Ca²⁺, and it appears to be the main determinant of Ca²⁺ excitability in astrocytes (Verkhratsky, 2006; Deitmer et al., 2008). The concentration of free Ca²⁺ inside the ER ([Ca²⁺]_ER) ranges from 100 to 800 μM (Burdakov et al., 2005) while the basal concentration of this ion in cytosol ([Ca²⁺]_cyt) is ~100 nM. Ca²⁺ can be released from the ER through IP₃ receptors (IP₃R) and Ca²⁺ -induced Ca²⁺ release (CICR) by activation of caffeine/ryanodine sensitive receptors (RyR) (Simpson et al., 1998). These two ER receptor/channel types may gate distinct Ca²⁺ stores (Golovina and Blaustein, 2000), although this notion has not been widely accepted [e.g., (Hua et al., 2004); see below]. The [Ca²⁺]_ER can affect the sensitivity of the RyRs to cytosolic Ca²⁺ as well as the activity of the SERCA pumps (Burdakov et al., 2005). Thus, at the level of the ER, there is regulation of Ca²⁺ release and entry.

The initial, but indirect indication that ER stores play role in Ca²⁺-dependent release of glutamate from astrocytes came from experiments in which the neuroligand bradykinin which caused cytosolic Ca²⁺ elevation in astrocytes also induced glutamate release from these glial cells (Parpura et al., 1994). Bradykinin binds to G-protein coupled plasma membrane receptors which activate the IP₃ pathway, and cause release of Ca²⁺ from the ER (Cholewinski et al., 1988; Stephens et al., 1993). In addition to bradykinin, other ligands including ATP can induce Ca²⁺ release from the ER (Peuchen et al., 1996) through the IP₃ pathway (Kastritsis et al., 1992). Indeed, ATP stimulation of purinergic 2 receptors caused an increase in astrocytic cytosolic Ca²⁺ which coincided with an increase in glutamate release from cultured astrocytes to the extracellular space (Jeremic et al., 2001). This ATP action was greatly reduced when astrocytes were pre-incubated with thapsigargin, implicating the role of ER stores in providing cytosolic Ca²⁺ for exocytotic glutamate release.

Hua et al. (2004) demonstrated that both IP₃- sensitive and ryanodine/caffeine-sensitive stores play a role in Ca²⁺ -dependent mechanically-induced glutamate release from astrocytes. The Ca²⁺ source for this exocytotic release is predominately from internal stores, as indicated by reduction of mechanically-induced glutamate release in the presence of thapsigargin. To test whether IP₃- sensitive internal stores mediate Ca²⁺-dependent glutamate release from astrocytes, these cells were bathed in diphenylboric acid 2-aminoethyl ester (2-APB) solution, a cell-permeant IP₃R antagonist. This agent greatly reduced exocytotic glutamate release. Next, the role of ryanodine/caffeine-sensitive ER stores was assessed by incubating astrocytes with ryanodine, which at concentrations used (10 μM) blocked the release of Ca²⁺ from the ryanodine/caffeine- sensitive stores. Ryanodine also attenuated mechanically -induced glutamate release. Furthermore, the sustained presence of caffeine that depleted ryanodine/caffeine stores, also reduced mechanically -induced glutamate release. It should be noted, however, that the functionality of ryanodine receptors in astrocytes is still debated, since the lack of their activity in astrocytes in situ had been reported (Beck et al., 2004). Nonetheless, when astrocytes in culture were pre-treated with the combination of 2-APB with ryanodine or caffeine, there was no additive effect on reduction of mechanically-induced glutamate release when compared to treatments with one pharmacological agent only. Since all pharmacological agents used in this study showed cytosolic Ca²⁺ changes that were in good agreement with glutamate release, these data suggested that exocytotic glutamate release requires the co-activation of both IP₃- and ryanodine/caffeine-sensitive internal Ca²⁺ stores and that these stores are operating jointly (Figure 1).

(ii) Extracellular space

Astrocytes utilize extracellular Ca²⁺ as well as ER Ca²⁺ for exocytotic glutamate release (Figure 1). This is evidenced by the reduction of Ca²⁺-dependent glutamate release from astrocytes not only in the presence of thapsigargin, but also in the presence of extracellular Cd²⁺, a blocker of Ca²⁺ entry from the extracellular space (Hua et al., 2004). It should be noted that remaining glutamate release in stimulated astrocytes pre-treated with thapsigargin or Cd²⁺ were ~30% and 55% of control, respectively, indicating that the internal stores are the predominant source of Ca²⁺ for exocytotic release.

Ca²⁺ entry across the astrocytic plasma membrane involves a variety of channels [reviewed in (Verkhratsky et al., 1998); see below]. Ultimately, the (re)filling of ER Ca²⁺ stores requires Ca²⁺ entry from the extracellular space through so called store-operated Ca²⁺ (SOC) channels (Takemura and Putney, 1989; Golovina, 2005). Canonical transient receptor potential (TRPC) channels, which have been implicated in SOC entry, have been found in astrocytes where they play a role in the regulation of Ca²⁺ homeostasis (Pizzo et al., 2001; Grimaldi et al., 2003; Golovina, 2005). TRPC1 functionally contributes to Ca²⁺-dependent glutamate release from astrocytes. Malarkey et al. (2008) detected the presence of TRPC1 in cultured and freshly-isolated astrocytes, along with two other isoforms with which TRPC1 is known to form heteromultimeric channels, TRPC4 and TRPC5 (Strubing et al., 2001; Hofmann et al., 2002; Strubing et al., 2003). It should be noted that TRPC1 does not commonly form homomeric channels (Strubing et al., 2003). When looking at the function of these proteins, Malarkey et al. (2008) measured the extent of SOC entry in cultured astrocytes. Astrocytic internal Ca²⁺ stores were depleted by applying cyclopiazonic acid (CPA), which blocks SERCA and (re)filling the stores. Concomitantly, the entry of external Ca²⁺ was prevented by bathing the astrocytes in a Ca²⁺ free external saline solution. This depletion of Ca²⁺ from internal stores led to opening of SOC channels, which was evident by the increased cytosolic Ca²⁺ when the bathing solution was changed to external solution containing Ca²⁺. This increase was pharmacologically confirmed to be due to Ca²⁺ entry through SOC channels using La³⁺, Gd³⁺ and MRS-1845, compounds known to inhibit SOC channels (Harper et al., 2003; Rychkov and Barritt, 2007).

In experiments using a similar SOC entry assay to the one described above, Golovina (2005) demonstrated significant reduction of SOC entry in cultured astrocytes after an antisense knockdown of the TRPC1 gene. To address whether the acute block of TRPC1 affects SOC entry, Malarkey et al. (2008) used an antibody against TRPC1 that was designed to bind to the pore forming region of TRPC1 and block the functioning of the channel (Wang et al., 1999). This antibody caused a significant decrease in the measured amount of SOC entry in astrocytes. Such immunological interference with TRPC1 channels also showed that the SOC entry through TRPC1 plays a role in mechanically-induced cytosolic Ca²⁺ elevations and the consequent Ca²⁺-dependent release of glutamate from astrocytes (Figure 1). The reduction of glutamate release in astrocytes pretreated with antibody was ~ 39% of control, thus at similar level to that seen in astrocytes treated with thapsigargin (Hua et al., 2004).

It is obvious by now that astrocytes are unlike neurons in which the predominant source of Ca²⁺ for exocytotic release at synaptic terminals is extracellular, mediated via Ca²⁺ entry though voltage-gated Ca²⁺ channels (VGCCs). However, VGCCs were found in astrocytes (Latour et al., 2003; Mulligan and MacVicar, 2004). Since the driving force for the for Ca²⁺ entry into astrocytes is similar to that in neurons with the extracellular Ca²⁺ concentration ([Ca²⁺]_ext) ~2 mM being four orders greater than [Ca²⁺]_cyt, the inability to detect the activity of VGCCs in hippocampal astrocytes in situ may implicate the low level of expression and/or low plasma membrane density of VGCCs (Carmignoto et al., 1998). Although the role of VGCCs in glutamate release from astrocytes is unclear at present, there is some evidence in support of their involvement in this process. Using freshly prepared slices from ventrobasal thalamus, Parri et al. (2001) showed that astrocytes in situ display intrinsic [Ca²⁺]_cyt oscillations. These oscillations were not driven by neuronal activity since they could not be blocked by tetrodotoxin. Astrocytes in situ displaying spontaneous [Ca²⁺]_cyt oscillations could cause N-methy-D-aspartate (NMDA) receptor-mediated neuronal excitability. Thus astrocytic spontaneous [Ca²⁺]_cyt oscillations lead to glutamate release from these glial cells. Although oscillations where driven by Ca²⁺ released from intracellular stores since they were blocked by SERCA blockers thapsigargin and CPA, they were also inhibited to the similar extent by the dihydropyridine (DHP) antagonist nifedipine (Parri et al., 2001). This demonstrates the involvement of DHP-sensitive L-type VGCCs in astrocytic cytosolic Ca²⁺ dynamics. Conversely, the DHP agonist BayK8644 increased the number of astrocytes displaying cytosolic Ca²⁺ oscillations (Parri and Crunelli, 2003). Additionally, increasing the driving force for Ca²⁺ entry by increasing [Ca²⁺]_ext (to 5 mM) increased the number of spontaneously active astrocytes and the number of cytosolic Ca²⁺ transient increases displayed by individual astrocytes (Parri and Crunelli, 2003). Even though the effect of VGCC manipulations on glutamate release from astrocytes of ventrobasal thalamus has not been assessed, it is tempting to speculate that these channels may be a conduit for Ca²⁺ supply for glutamate release from astrocytes (Figure 1).

Ionotropic neurotransmitter receptors represent an additional entry of Ca²⁺ in astrocytes [reviewed in (Verkhratsky, 2009)]. The role of ligand-gated ionotropic receptors in exocytotic glutamate release from astrocytes is provisional at the moment, however. Given that astrocytes can respond to glutamatergic and cholinergic synaptic inputs by increases in [Ca²⁺]_cyt (Dani et al., 1992; Porter and McCarthy, 1996; Araque et al., 2002), we briefly discuss nicotinic acetylcholine receptors, and glutamatergic, α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) and NMDA, receptors on astrocytes. Recent findings indicate that astrocytes express functional nicotinic acetylcholine receptors, particularly those with α7 subunits (Sharma and Vijayaraghavan, 2001). Activation of these receptors on astrocytes can lead to [Ca²⁺]_cyt increases, which require CICR and the activation of ryanodine/caffeine-sensitve internal stores. The dominant glutamatergic ionotropic receptors on astrocytes are AMPA receptors, which are found in astroglial cells throughout the brain (Verkhratsky and Steinhauser, 2000; Seifert and Steinhauser, 2001; Aguirre et al., 2004). Astrocytes in hippocampus express GluR2 subunit and thus display low Ca²⁺ permeability through their AMPA channels (Seifert et al., 2003). Hence, a challenge of hippocampal astrocytes with AMPA does not cause glutamate release (Bezzi et al., 1998). However, some types of astrocytes do not express the GluR2 subunit. For example, Bergmann glia, an astrocytic cell found in the cerebellum, expresses AMPA receptors that are Ca²⁺ permeable. These cells ensheathe parallel fiber-Purkinje cell synapses (Grosche et al., 1999). There, electrical stimulation of parallel fibers to release glutamate caused rapid and transient [Ca²⁺]_cyt elevations within Bergmann glial processes (Grosche et al., 1999). Functional astrocytic NMDA receptors were only recently detected in cortical astrocytes (Lalo et al., 2006). Interestingly, astrocytic NMDA channels grossly differ from neuronal since they do not show a Mg²⁺ block at resting membrane potentials (around −80 mV). Whether activation of these, and other ionotropic receptors expressed by astrocytes leading to increases in [Ca²⁺]_cyt would also lead to Ca²⁺-dependent glutamate release is unknown at the moment.

Another Ca²⁺ entry point that needs further investigation is at plasma membrane Na⁺/Ca²⁺ exchangers. While they are known to extrude Ca²⁺ during high cytosolic loads of this cation in conjuction with the plasma membrane Ca²⁺ -ATPase (PMCA), several reports have shown that the exchanger works in reverse to increase [Ca²⁺]_cyt (Goldman et al., 1994; Benz et al., 2004; Rojas et al., 2007). Hence, release of homeocysteic acid (HCA) from cultured astrocytes can be Ca²⁺-dependent and mediated by activation of glutamate receptors (Benz et al., 2004). It appears that the agonist stimulation led to intracellular Na⁺ load through ionotropic glutamate receptors, which then activated Na⁺/Ca²⁺ exchangers to extrude Na⁺ from astrocytes while importing Ca²⁺ to the cytosol, resulting in a rise of [Ca²⁺]_cyt. This cascade of events was inferred from the observation that agonist-induced HCA release was blocked in the presence of benzamil, a Na⁺/Ca²⁺ exchanger blocker. While some have argued that the reverse mode of this exchanger only occurs at pathophysiological conditions, it is worth investigating what this exchanger contributes to physiological Ca²⁺ entry and consequential exocytoctic glutamate release from astrocytes.

(ii) Mitochondria

In astrocytes, mitochondria form filamentous reticula, and are in perpetual fusion and fission as they are in other cell types (Westermann, 2002). Mitochondria congregate perinuclearly, but can be located at the cell periphery including processes (Collins et al., 2002). To achieve translocation, mitochondria travel on microtubules and actin filaments. While the actin cytoskeleton dictates the distribution of mitochondria, it appears that the intermediate filament vimentin may be involved as well (Tang et al., 2008). Interestingly, in neurons mitochondria display activity-dependent motility at the synapses (Li et al., 2004; Chang et al., 2006; Mironov, 2006).

Mitochondria are efficient Ca²⁺ sinks during periods when rapid increases in [Ca²⁺]_cyt occur in a cell. They possess a mitochondrial Ca²⁺ uniporter (MCU) that can transport Ca²⁺ into the mitochondrial matrix powered by a negative membrane potential (between −140 mV and −180 mV) across the inner mitochondrial membrane generated by the electron transport chain. In contrast to the SERCA and PMCA which have higher affinities for Ca²⁺ than the MCU, Ca²⁺ is taken up by MCU only at [Ca²⁺]_cyt greater than ~ 0.5 μM (Miyata et al., 1991; Simpson and Russell, 1998) and dominates Ca²⁺ transport at ~10 μM and above (Alonso et al., 2006). While such high concentrations are normally not reached in the bulk cytosol, they exist at high Ca²⁺ loci known as microdomains (Schneggenburger and Neher, 2005; Oheim et al., 2006). Indeed, Rizzuto et al. (1998) showed that mitochondria are juxtaposed to IP₃ channels of the ER and are exposed to higher concentrations of Ca²⁺ than what would exist at the bulk cytosol. Indeed, Ca²⁺ released from the IP₃-sensitve intracellular stores in mucosal mast cells caused local [Ca²⁺]_cyt increases of over 16 μM (Csordas et al., 2006). Thus, it seems reasonable to assume that similar localized [Ca²⁺]_i increases would occur in astrocytes.

It should be noted that as conditions in a cell are not homogenous i.e. ion concentration, mitochondrial Ca²⁺ handling is not homogenous either. In T lymphocytes, subsets of mitochondria modulates the entry of Ca²⁺ through the SOC channels by translocation (Hoth et al., 1997; Quintana et al., 2007). This is a likely function of some mitochondria in astrocytes as well. Kolikova et al. (2006) demonstrated using total internal reflection fluorescence microscopy that at least a subset of mitochondria become trapped near the plasma membrane when cultured astrocytes are challenged with glutamate or ATP. Indeed, numerous studies of various cells types report that different subsets of mitochondria take up Ca²⁺ depending whether stimulated cells released ER Ca²⁺ or whether they allowed Ca²⁺ entry from the extracellular space (Simpson and Russell, 1996; Collins et al., 2001).

As sinks to trap free Ca²⁺ released from the ER, mitochondria are positioned to affect transient Ca²⁺ increases required for vesicle fusion, and exocytosis in secretory cells. Reyes and Parpura (2008) demonstrated that mitochondria can modulate [Ca²⁺]_cyt and glutamate release in cortical astrocytes challenged with mechanical stimuli. Taking a pharmacological approach, we showed in parallel experiments that blocking the MCU with ruthenium 360 (Ru60) increased the cytosolic Ca²⁺ accumulation, and glutamate release in cortical astrocytes. In contrast to Ru360 treatment, decreasing mitochondrial Ca²⁺ efflux by blocking the mitochondrial Na⁺/Ca²⁺ exchanger with CGP37157, or increasing mitochondrial Ca²⁺ load by inhibiting formation of the mitochondrial permeability transition pore with cyclosporin A, decreased cytosolic Ca²⁺ accumulation, and glutamate release in cortical astrocytes. Hence, these data suggest that the transient increase in [Ca²⁺]_cyt is correlative to the amount of glutamate released in cortical astrocytes, and that mitochondria have the capacity to modulate the degree of this release. Taken together, astrocytic mitochondria may be active modulators of Ca²⁺-dependent glutamate release from astrocytes. They could act as temporary holding compartments of excess Ca²⁺ at the fine astrocytic processes surrounding the tripartite synapse, in a similar manner as has previously been shown in neuronal terminals at the neuromuscular junction (Tang and Zucker, 1997).

Concluding Remarks

The purpose of this review was to summarize the Ca²⁺ sources for exocytotic glutamate release from astrocytes. There are three major Ca²⁺ sources that contribute to this process: (i) the predominant source from internal ER stores, (ii) the SOC entry from the extracellular space and (iii) mitochondria, as outlined in Figure 1. There are several issues that remain to be resolved. As indicated earlier, the activation of ionotropic receptors expressed by astrocytes can lead to increases in [Ca²⁺]_cyt. Whether this excitation is coupled to the Ca²⁺-dependent exocytotic glutamate release from these cells awaits further experimentation. Furthermore, it will be necessary to determine whether the same Ca²⁺ sources operate under physiological and pathological conditions or whether there are specific Ca²⁺ sources operating under particular conditions. For example, acute oxidative stress in astrocytes causes influx of extracellular Ca²⁺ through L-type VGCCs (Bond and Greenfield, 2007). Whether this Ca²⁺ source plays a role in exocytotic glutamate release from astrocytes and whether such release occur in various neurodegenerative diseases causatively linked to oxidative stress remains to be assessed. Alternatively, all possible Ca²⁺ sources could operate together in astrocytes at all times, but each contributing a different portion to the total amount of glutamate being released. It appears that addressing these questions in cell culture systems depends on application of sufficient human time and effort, since many of the specific reagents and methods, some of which are described earlier, for teasing apart the contributions from individual Ca²⁺ sources have already been developed. Although such an approach would require a labor intensive effort of information collection and analysis, it is a pre-requisite before taking experiments out of culture dishes and rigorously testing hypotheses on freely-moving animals and humans.