In this study we report the direct evidence of anion channel-mediated glutamate release mechanism. We provide a series of evidence for Ca2+-activated, Best1-mediated glutamate release from astrocytes. For example, the selective gene silencing of astrocytic Best1 channels significantly reduced GPCR-induced and Ca2+-dependent glutamate release from astrocytes as measured by HPLC detection of glutamate from cultured astrocytes (Figure 3), and by FRET glutamate sensor in cultured astrocytes (Figure 4). The FRET based glutamate sensor has been well-characterized in the previous report . In that report it has been demonstrated that FRET change is quite specific for glutamate because the sensor shows binding kinetics and high selectivity toward glutamate with no apparent change in ratio with APV, NBQX, NMDA, KA, AMPA, etc. . We performed a calibration experiment by measuring glutamate EC50 in our imaging system and experimental conditions. Our calibration results showed that the FRET-based glutamate sensor in cultured astrocyte can be useful in detecting micromolar and submicromolar concentrations of released extracellular glutamate from a single astrocyte.
Until now, it has been proposed that glutamate could be released from astrocytes through multiple routes, including Ca2+- and SNARE-dependent vesicular exocytosis, the reversal action of glutamate transporters, transportation by cystine-glutamate antiporter, and permeation through channels or receptors, such as P2X7 receptor, volume regulated anion channel (VRAC), or gap junction hemichannel . Among these, the molecular mechanism of Ca2+-dependent glutamate release has been extensively studied mainly because the generation of astrocytic Ca2+ transient responding to neuronal activity is one of the most important physiological readout of neural activity [3, 5, 7, 10]. Numerous studies have proposed the SNARE-dependent exocytosis of glutamate as a potential route for astrocytic Ca2+-dependent glutamate release. These studies demonstrated that astrocytic glutamate release was sensitive to exocytosis blockers or the expression of the essential machineries for regulating Ca2+-dependent exocytosis . In addition to the vesicular mechanism, there is multiple lines of evidence suggesting that astrocytes express a Ca2+-dependent but non-vesicular glutamate release machinery . Despite its unclear molecular identity, it has been reported that an increase in Ca2+ by GPCR activation induces glutamate release from astrocytes through osmolyte-permeable anion channels [10, 28], which are activated by intracellular Ca2+ or increase in cell volume . In our previous reports, we observed that Best1-mediated Ca2+-activated anion current could be induced by a host of GPCR agonists such as mGluR1/5, P2Y, B2, S1P, and LPAR in astrocytes , and that activation of those receptors elicits Ca2+-dependent glutamate release . Finally, Best1 was demonstrated to be permeable to GABA and mediate tonic GABA release in cerebellar glial cells . Recent study showed that GABA is abundant in cerebellar glial cells to be released tonically via Best1 to cause tonic inhibition . However, in hippocampus, GABA is not present in glial cells; rather, Best1 might release glutamate near synapse. Therefore, it is plausible that Best1, at least in part, could be the downstream target of various Gq-coupled GPCRs that signal through intracellular Ca2+ to release glutamate in hippocampal astrocytes.
In addition to Best1 channel, other anion channels might participate in anion channel-mediated glutamate release as previously suggested, such as Ca2+-induced but volume regulated osmolyte-permeable channels . In fact, there are several studies showing that bestrophin channel could be opened by both cytosolic Ca2+ and cell volume increase [17, 35], raising a possibility that Best1-mediated glutamate release is also triggered by Ca2+-dependent cell swelling and subsequent activation of VRAC. However, in our previous study we demonstrated that PAR-1 activation does not induce significant swelling  indicating that Ca2+ directly initiates Best1 channel-mediated glutamate release from astrocytes. The PAR1-induced astrocytic glutamate release was not completely inhibited by Best1 knock-down (Figure 4). Still, vesicular release mechanism was not involved because Best1 silencing did not affect the expression level of known genes involved in vesicular release (Figure 5A) and glutamate release from astrocytes was not reduced by treatment with Conconomycin A or Tetanus toxin (Figure 5B). There is a possibility that other ion channels that are independent of Best1 channel might participate in PAR1-induced glutamate release from astrocytes.
We utilized the native GPCR, PAR1, which has been extensively used in numerous studies to selectively activate astrocytes. Even though PAR1 is expressed in a subset of dentate granule cells , PAR1 has been shown to be expressed exclusively in astrocytes in human and rodent CA1 hippocampus [7, 38, 39], as well as in the nucleus of solitary tract  to mediate neuron-glia interaction. Therefore, the source of PAR1-induced glutamate is most likely astrocyte, as a direct consequence of increase [Ca2+i. Although there is no direct evidence of how PAR1 is activated in the physiological situation up to now, the recent study demonstrated that tPA-plasmin pathway is an endogenous PAR1 agonist , suggesting that physiological PAR1 activation is initiated by the activation of tPA-plasmin pathway in physiological condition such as synaptic plasticity [42, 43].
In summary, we reveal a novel anion channel-mediated glutamate release mechanism in cultured astrocytes. The ideas and tools developed in this study should prove to be helpful in understanding the physiological role of glutamate release mechanism and its functional significances.