The signaling role for chloride in the bidirectional communication between neurons and astrocytes

Abstract

It is well known that the electrical signaling in neuronal networks is modulated by chloride (Cl⁻) fluxes via the inhibitory GABA_A and glycine receptors. Here, we discuss the putative contribution of Cl⁻ fluxes and intracellular Cl⁻ to other forms of information transfer in the CNS, namely the bidirectional communication between neurons and astrocytes. The manuscript (i) summarizes the generic functions of Cl⁻ in cellular physiology, (ii) recaps molecular identities and properties of Cl⁻ transporters and channels in neurons and astrocytes, and (iii) analyzes emerging studies implicating Cl⁻ in the modulation of neuroglial communication. The existing literature suggests that neurons can alter astrocytic Cl⁻ levels in a number of ways; via (a) the release of neurotransmitters and activation of glial transporters that have intrinsic Cl⁻ conductance, (b) the metabotropic receptor-driven changes in activity of the electroneutral cation-Cl⁻ cotransporter NKCC1, and (c) transient, activity-dependent changes in glial cell volume which open the volume-regulated Cl⁻/anion channel VRAC. Reciprocally, astrocytes are thought to alter neuronal [Cl⁻]_i through either (a) VRAC-mediated release of the inhibitory gliotransmitters, GABA and taurine, which open neuronal GABA_A and glycine receptor/Cl⁻ channels, or (b) the gliotransmitter-driven stimulation of NKCC1. The most important recent developments in this area are the identification of the molecular composition and functional heterogeneity of brain VRAC channels, and the discovery of a new cytosolic [Cl⁻] sensor – the Wnk family protein kinases. With new work in the field, our understanding of the role of Cl⁻ in information processing within the CNS is expected to be significantly updated.

1. Introduction: the physiological significance of transmembrane Cl⁻ gradients

Chloride (Cl⁻) is the most abundant physiological anion. Yet, it receives little attention outside of the fields studying excitable cells and acid-base homeostasis. With the exception of kidney tissue and some secretory epithelia, the extracellular Cl⁻ levels are kept in the narrow range of 98–106 mM. In contrast, the intracellular Cl⁻ concentrations can vary from as low as 5 mM in neuronal cells to 65 mM or higher in certain epithelial cells (see for example [1;2]). Intracellular Cl⁻ levels are largely determined by the transmembrane potential and presence of secondary active Cl⁻ transporters. In this review, we propose an important role for [Cl⁻]_i in bidirectional neuron-astrocyte communication. But before focusing on the significance of Cl⁻ in the CNS, it is helpful to briefly recapitulate the general physiological functions of this anion. Fundamentally, Cl⁻ serves as a counterion for movement of the major cations, Na⁺, K⁺, Ca²⁺, and H⁺ (for comprehensive reviews see [3–5]). The plasmalemmal cation transport in excitable cells and ion-transporting epithelia is assisted by Cl⁻ conductivity pathways, such as voltage-gated Cl⁻ channels. Within the cell, electrogenic proton movement in acidic organelles is facilitated by Cl⁻/H⁺ exchangers. Other, equally important Cl⁻ functions include (i) cell volume homeostasis, (ii) regulation of membrane potential, and (iii) cell proliferation and initiation of apoptosis. These are briefly introduced below and summarized in Fig. 1.

Fig. 1. Diverse roles of Cl⁻ in cellular physiology.

A, The presence of membrane-impermeable organic anions (O⁻) and uncharged molecules (O⁰) makes cells accumulate positively charged ions, leading to the phenomenon known as Donnan cell swelling. Passive extrusion of Cl⁻ allows cells to maintain electroneutrality across their semi-permeable surface membrane and prevent Donnan swelling. B, When intracellular [Cl⁻] deviates from its electrochemical equilibrium, activation of the ligand-gated Cl⁻ channels [GABA_A and/or glycine (GlyR) receptors] leads to a depolarizing or hyperpolarizing current. C, During apoptosis, cells undergo apoptotic volume decrease as a result of the loss of intracellular K⁺ and Cl⁻. D, Cell volume homeostasis relies on the balance of inward and outward ion fluxes, which determine the movement of osmotically obligated water. Left panel, Swollen cells undergo regulatory volume decrease (RVD) driven by loss of KCl and osmotically obligated H₂O. KCl is lost due to activation of the functionally coupled K⁺ channels and volume-regulated anion channels (VRAC), as well as the electroneutral K⁺,Cl⁻ cotransporters (KCC). Right panel, Shrunken cells engage the process of regulatory volume increase (RVI) that is powered by the accumulation of NaCl. NaCl uptake is mediated by the net activity of the electroneutral Na⁺,K⁺,2Cl⁻ cotransporters (NKCC), Na⁺/H⁺ exchangers, and Cl⁻/HCO₃⁻ anion exchangers.

1.1 Cl⁻ in the Gibbs-Donnan equilibrium and cell volume regulation

Participation in cellular volume homeostasis represents a very important and evolutionarily conserved role of Cl⁻ in animal cells, which is rarely covered outside of the “specialized” reviews. Cl⁻ contributes to cell volume control via two separate mechanisms.

Under steady-state conditions, low intracellular [Cl⁻] compensates for the presence of impermeable large-molecular-weight organic molecules, which in the intracellular milieu carry net-negative charge (illustrated in Fig. 1A). Due to the presence of organic anions, cells tend to accumulate positively charged particles in their cytosol and have a propensity to swell, a phenomenon termed Donnan cell swelling [6]. Extrusion of Cl⁻ allows for the maintenance of electroneutrality across semi-permeable cell surface membranes, meaning that the sum of positive and negative charges is the same inside and outside the cell, irrespective of the type of charged particle (Gibbs-Donnan equilibrium). In living cells, colloidal accumulation of water is not completely compensated via this mechanism. Therefore, the Gibbs-Donnan equilibrium is, in reality, a quasi-equilibrium, which is maintained by the constant osmogenic and electrogenic activity of the Na⁺/K⁺-ATPase (2 K⁺ in in exchange of 3 Na⁺ out, discussed in [7]).

Besides the constant pressure of Donnan swelling, cells experience frequent transient changes in their volume due to a variety of factors, such as (a) changes in extracellular osmolarity, (b) intracellular catabolic processes, which lead to the accumulation of low molecular weight osmolytes, or (c) net accumulation or extrusion of ions and osmolytes due to activation of diverse plasmalemmal transporters and channels. Swollen or shrunken cells engage “emergency” mechanisms, involving a variety of volume-sensitive transporters. These work collectively to extrude or accumulate inorganic ions and/or small organic molecules (see Fig. 1D). The net transfer of ions and other osmolytes drives the movement of osmotically obligated water, and powers cell volume normalization. Swollen cells undergo the process of regulatory volume decrease (RVD), which depends on the cooperative activity of swelling-activated K⁺ channels, Cl⁻ channels, and/or electroneutral K⁺-Cl⁻ cotransporters. Shrunken cells restore their volume via regulatory volume increase (RVI) by engaging Na⁺/K⁺/Cl⁻ cotransporters and/or the functionally coupled activity of Na⁺/H⁺ and Cl⁻/HCO₃⁻ exchangers. The molecular origins and functional properties of the relevant Cl⁻ transporters are summarized in Section 2. This topic can be further explored by reading any of the relevant comprehensive reviews [7–11]. In general terms, RVD is mediated by the net-loss of KCl with the additional contribution of organic osmolytes. RVI is largely the result of net-accumulation of NaCl.

1.2 Regulation of the membrane potential

As the dominant anion in the extracellular milieu, Cl⁻ has a high capacity for modulating the plasma membrane potential. Background (unstimulated) Cl⁻ conductance contributes to setting the resting membrane potential. This aspect of Cl⁻ physiology has been best studied in skeletal muscle cells, in which Cl⁻ currents are responsible for up to 80% of the resting membrane conductance, stabilize the membrane potential, and limit muscle excitability [12]. Accordingly, mutations in ClC-1, the principal voltage-gated Cl⁻ channel in skeletal muscle cells, have been linked to many cases of the autosomal recessive generalized myotonia and the autosomal dominant myotonia congenita, two diseases in which cell hyperexcitability causes skeletal muscle stiffness (see for example [13–15]).

Unlike skeletal muscles, cells in the CNS have comparatively low resting plasmalemmal Cl⁻ conductance and maintain their [Cl⁻]_i away from the levels predicted by the Nernst equilibrium (see Section 2.1). Consequently, neuronal cells utilize the ligand-gated Cl⁻ channels, such as GABA_A and glycine receptors, for modulating membrane potential and excitability (see Fig. 1B). Activation of these channels can cause depolarizing or hyperpolarizing currents based on the established transmembrane Cl⁻ gradients (for detailed discussion see Section 2.3). The voltage-gated Cl⁻ channels, e.g. ClC-2, can also regulate intracellular [Cl⁻] and excitability in neuronal cells and contribute to the buffering of extracellular [Cl⁻] by astrocytes (for references and discussion see Section 2.4).

1.3 Cell proliferation, migration, and initiation of apoptosis

There is a very extensive literature that links Cl⁻ channels, principally, the volume-sensitive Cl⁻ channels, to proliferation, migration, and apoptosis in numerous cell types (for review see [7;16]). Outside of brain development, the significance of cell proliferation in the CNS is limited. Therefore, we do not discuss the potential role of Cl⁻ in this process. Migration is also not a major feature of normal brain physiology, with the notable exception of the innate immune cells of the CNS, microglia. Under normal physiological conditions microglial cell bodies are immobile, but these cells actively relocate their branched processes in search of pro-inflammatory stimuli and signs of neuronal hyperexcitation. In response to tissue damage or inflammation, microglia retract their processes and move towards the pathological foci, where they engage in phagocytosis and tissue remodeling [17;18]. In microglia, broad spectrum Cl⁻ channel blockers reversibly inhibit the formation of ramified processes and lamellipodia, limit the cytokine-stimulated cell migration, and block proliferation and phagocytosis [19–24]. Outside of the brain, contributions of Cl⁻ channels to cell motility have been demonstrated in many other cell types (for review see [7;25]).

It is also important to acknowledge that Cl⁻ transporters may be important in programmed cell death. One of the hallmarks of apoptosis is a marked cell shrinkage, which has been termed apoptotic volume decrease or AVD [26;27] (illustrated in Fig. 1C). In a number of cell types, AVD depends on the activity of the volume-sensitive channel VRAC (see Section 2.5), is associated with the loss of intracellular Cl⁻, and can be induced by reductions in extracellular Cl⁻ levels (see for example [26;28–30] and review [7]). In the CNS, activity of VRAC has been directly implicated in the excitotoxic cell death of neurons, and pathological release of excitatory amino acids from astrocytes (see [31–34] and review [35]). Therefore, in the brain, volume-sensitive Cl⁻ channels may contribute to tissue damage, both directly – by promoting apoptosis, and indirectly – via the release of excitotoxic neurotransmitters and downstream activation of cell death pathways.

2. Cl⁻ transporters and channels which govern intracellular Cl⁻ levels

As briefly outlined in the Introduction, plasmalemmal Cl⁻ gradients are established by several secondary active ion transporters, and allow neurons and astrocytes to maintain intracellular Cl⁻ levels either above or below the electrochemical equilibrium for this ion. When [Cl⁻]_i deviates from equilibrium, transient changes in Cl⁻ membrane permeability via voltage-, volume-, and ligand-gated ion channels produce net charge transfer and influence many cellular functions. For neurons, the impact of Cl⁻ currents on cellular excitability is well established and has been covered in several comprehensive reviews (see for example [36–38]). In contrast, the influence of Cl⁻ fluxes on astrocytic properties is less understood, and will be discussed in the present manuscript in the context of bidirectional neuron-astrocyte communication. In this section, we briefly describe the molecular nature of diverse Cl⁻ transporters and ion channels in neural cells, and discuss their differential impact on intracellular Cl⁻ levels (summarized in Fig. 2).

Fig. 2. Overview of ion transporters and channels, which are responsible for setting intracellular Cl⁻ levels in neural cells.

The transmembrane gradients of Na⁺ and K⁺ are established by the Na⁺/K⁺-ATPases (NKA). The negative charge on the plasmalemmal membrane passively drives Cl⁻ outside of the cell via a plethora of mechanisms, which are collectively termed Cl⁻ “leak”. However, in neurons and astrocytes [Cl⁻]_i deviates from the levels predicted by the electrochemical equilibrium due to the activity of the Na⁺,K⁺,2Cl⁻ cotransporter (NKCC1) and/or the K⁺,Cl⁻ cotransporters (KCC1-4). Opening of voltage-gated Cl⁻ channels (VGCC), ligand-gated Cl⁻ channels (LGCC), or volume-regulated anion channels (VRAC) moves Cl⁻ in or out of the cell, depending on the electrochemical driving force. Additionally, Cl⁻ movement also occurs via the Cl⁻/HCO₃⁻ anion exchangers (AE3 and others).

2.1 Na⁺, K⁺, 2Cl⁻ cotransporters (NKCC)

NKCCs are two closely related proteins – NKCC1 (SLC12A2) and NKCC2 (SLC12A1), which belong to the large SLC12A superfamily of cation-chloride cotransporters. NKKC1 is ubiquitous, and can be found at various levels in all types of brain cells (reviewed in [39]). NKCC2 is the kidney-specific isoform [40] and, therefore, is not discussed in this manuscript. Both NKCCs mediate the bidirectional, electroneutral symport of one Na⁺, one K⁺, and two Cl⁻ in or out of the cell. Because the combined electrochemical gradients for Na⁺, K⁺, and Cl⁻ favor inward transport, NKCCs facilitate Cl⁻ accumulation above the level of its electrochemical equilibrium [39].

In neuronal physiology, the ability of NKCC1 to drive net-accumulation of intracellular Cl⁻ is functionally significant. Early in development, neural precursors and immature neurons express high levels of NKCC1 and consequently have elevated intracellular [Cl⁻] [36]. In these cells, GABA and glycine act as excitatory neurotransmitters [41]. Throughout the maturation process, neurons downregulate NKCC1 and replace it with electroneutral KCC transporters (see Section 2.2). Such a change leads to the precipitous drop in intracellular [Cl⁻], and this transition converts the actions of GABA and glycine from excitatory to inhibitory (reviewed in [36;38]). Based on the data derived from gramicidin-perforated patch recordings in brain slices, resting [Cl⁻]_i drops from 37 mM in E16 neural precursors to 12 mM in mature cortical neurons at P16 [42]. In the adult brain, [Cl⁻]_i was measured as low as 5 mM in certain neuronal populations [1].

Unlike neurons, astrocytes preserve high activity of NKCC1 throughout development and, therefore, have higher intracellular [Cl⁻]. This aspect of astroglial physiology has been thoroughly explored in cell cultures. In cultured astrocytes, the intracellular [Cl⁻] has been estimated at 20 to 50 mM. These estimates are derived using various experimental techniques, including steady-state isotope distribution (31–50 mM, [43]), sharp electrodes (20–40 mM [44]), and electrophysiological recordings with the K⁺ ionophore gramicidin (29 mM [45]). The data on astrocytic Cl⁻ levels in vivo and in situ are scarce. An early in vivo study used ³⁶Cl⁻ distribution to estimate glial (largely astrocytic) [Cl⁻] to be 46 mM in the cortex and 36 mM in the cerebellum [46]. More recently, Untiet et al. measured [Cl⁻]_i of 52 and 35 mM in immature and mature cerebellar Bergman glial cells, respectively, in brain slices using a chloride-sensitive MQAE FLIM signal [47]. Consistently, a number of studies in brain slices collected indirect evidence for high astrocytic [Cl⁻]_i, including the hyperpolarizing effects of GABA receptor agonists and the depolarizing effects of Cl⁻ channel blockers [48;49].

While discussing the impact of NKCC1 activity on cellular functions, it is important to remember that this transporter is potently stimulated by protein phosphorylation, low [Cl⁻]_i, and cell shrinkage (reviewed in [37;38]). Early studies in non-neural cells found that NKCC1 is activated by numerous agonists linked to the cAMP-dependent PKA, Ca²⁺/DAG-dependent PKCs, c-Jun N-terminal kinase (JNK), Erk1/2, and others (e.g. [50–54]). Yet, NKCC1 stimulation by various protein kinases is cell type-specific; in some cases, the same signaling cascades have been reported to produce opposite functional effects (discussed in [39;55]). For example, in the majority of cell types PKCs activate NKCC1, while in the kidney MDCK cell line it is completely inhibited by the PKC agonist PMA [56]. Based on this information and due to the lack of relevant phosphorylation motifs in the protein stricture, it has been concluded that many protein kinases modulate NKCC1 indirectly [39;55]. Recently, the critical role for the WNT/SPAK/OSR1 signaling cascade in the regulation of NKCCs and other cation-chloride cotransporters has been established (reviewed in [57–59], and discussion in Section 3.2). This signaling axis may be the actual target for indirect effects of the previously implicated protein kinases (see for example [60]).

2.2 K⁺, Cl⁻ cotransporters (KCC)

The electroneutral KCC cotransporters encompass four additional members of the SLC12A family and include KCC1 (SLC12A4), KCC2 (SLC12A5), KCC3 (SLC12A6), KCC4 (SLC12A7). All KCCs mediate the bidirectional transfer of one K⁺ and one Cl⁻ in or out of the cell (reviewed in [61;62]). In the context of [Cl⁻]_i homeostasis, KCCs are the functional opposites of NKCCs because they thermodynamically favor the efflux of intracellular Cl⁻, and drive [Cl⁻]_i below the level of its electrochemical equilibrium. KCC1 is expressed in a nearly ubiquitous manner, while KCC2 is neuron specific [62–64]. KCC3 and KCC4 are abundant in the brain, but present at diverse levels in various neuronal and astroglial populations [65–67].

In the CNS, the neuron-specific KCC2 has been particularly well-studied. Functional upregulation of KCC2 during neuronal maturation is responsible for the switch in the actions of GABA and glycine receptors from excitatory to inhibitory [38;41;68]. This isoform differs from other KCCs because it is constitutively active. Unlike KCC2, the “classical” KCC isoforms (1, 3, and 4) are largely silent under isoosmotic conditions, but become strongly activated in response to cell swelling [38;61]. Deletions and mutations in KCC3 and KCC4 produce severe central and peripheral phenotypes, and the relevant changes have been considered in the context of cell volume regulation in either neuronal or glial cells [69;70].

Although the essential role of protein phosphorylation/dephosphorylation in the regulation of KCCs has been known for many years, the underlying molecular mechanisms have been elucidated only recently. All KCCs are inhibited by direct phosphorylation by SPAK and OSR1, two closely related kinases which belong to the WNK signaling cascade [71;72] (see Section 3.2 for detailed description of signaling pathway). In contrast, dephosphorylation of KCCs by the serine-threonine phosphatases (PP) type 1 and 2A leads to their activation. This process was first explored for KCC1 in red blood cells [61;73–75]. It appears that the membrane-bound PP1 responds to changes in ionic strength, while the membrane-bound PP2A responds to cell swelling [75]. More recent model studies further elaborated on the mechanisms for phosphatase actions, and established both direct interactions with KCC proteins and involvement of the scaffolding protein, apoptosis-associated tyrosine kinase AATYK1 [76;77].

2.3 The ligand-gated GABA and glycine receptors-channels

The most recognized Cl⁻ permeability pathways in the brain are the ionotropic receptors for two inhibitory amino acid neurotransmitters, γ-aminobutyric acid (GABA) and glycine. These two families of receptors are structurally related and belong to the superfamily of Cys-loop receptor proteins [78;79].

GABA_A channels are pentameric ligand-gated Cl⁻ channels assembled from nineteen diverse members of the GABA_A family (GABR genes, for comprehensive reviews see [80;81]). In mature neurons, GABA_A channels are located in either the postsynaptic membrane, where they mediate fast neuronal inhibition, or at extrasynaptic sites, where they respond to ambient extracellular GABA levels and produce long-term inhibition. As already mentioned, the functional impact of neuronal GABA_A receptors is determined by chloride-cation cotransporters, mainly by the opposing work of NKCC1 and KCC2 (see Sections 2.1 and 2.2).

Astrocytes also express GABA_A receptors [48;82;83], although the receptor arrangement in these cells is less clear. Unlike their actions in mature neurons, opening of GABA_A channels in astrocytes causes outward Cl⁻ movement and membrane depolarization (see for example [48;84]). The functional role of GABA_A currents in astroglia may be questioned because these cells have a highly clamped membrane potential, which is stabilized by the electrical connectivity within the astrocytic syncytium [85]. Yet, electrophysiological experiments in primary astrocyte cultures suggest that GABA_A receptors are physiologically relevant. In hippocampal slices GABA_A-mediated depolarization activates astrocytic voltage-gated Ca²⁺ channels and causes cytosolic [Ca²⁺] transients [83]. Another brain slice study reported modulation of the activity of astrocytic K⁺ channels by GABA_A agonists [45].

The structurally related glycine receptor (GlyR) family has five distinct isoforms, GlyRα1-4 (GLRA1-4) and GlyRβ (GLRB), which assemble in either homo- or heteropentameric channels, all of which are activated by glycine [86;87]. Various GlyR subunits are expressed throughout the CNS, with the highest abundance in the spinal cord, brainstem nuclei, and retina [87]. In neurons, GlyRs play a role, which is similar to GABA_A, and in the adult brain mediate hyperpolarization due to Cl⁻ movement into the cell [86;87]. Very little information is available on GlyRs in astrocytes. To the best of our knowledge, only one study has identified functional astroglial GlyR Cl⁻ currents in situ, in spinal cord astrocytes [88]. Based on available expression data, it seems unlikely that GlyRs play a dominant role in astroglia.

In the context of receptor pharmacology and physiology, it is important to note that both GABA_A and glycine receptors can be activated by the atypical aminosulfonic acid taurine, which in the brain serves as the endogenous agonist for these receptor-channels (reviewed in [89]).

2.4 Voltage-gated Cl⁻ channels (ClC family)

The voltage-gated Cl⁻ channels belong to the evolutionarily conserved ClC family (encoded by the CLCN1 through CLCN7, CLCNKA and CLCNKB genes). These have been discovered based on their homology with ClC-0 channels from the electric organ of ray Torpedo marmorata [90]. In mammals, there are nine diverse ClC proteins, four of which (ClC-1, ClC-2, ClC-Ka and ClC-Kb) form plasmalemmal Cl⁻ channels, while five others (ClC-3 through ClC-7) function as intracellular Cl⁻/H⁺ exchangers (reviewed in [3]). As already briefly mentioned in Section 1.2, voltage-gated Cl⁻ channels have a high capacity for modulating resting membrane potential and membrane conductance. Because neurons utilize ligand-gated Cl⁻ fluxes to control their excitability, these cells tend to have low expression of voltage-gated Cl⁻ channels. The same applies to astrocytes, whose membrane permeability is dominated by K⁺ currents. Yet, even the limited expression of voltage-gated Cl⁻ channels in neural cells has a strong impact on their physiology.

ClC-2 (CLCN2) has been detected in both neurons and astrocytes, but the expression of this channel strongly varies depending on the brain region and cell subtype [91;92]. It seems that ClC-2 conductance counteracts GABA_A function, because overexpression of CLC-2 in dorsal root ganglia neurons blunted GABA effects on neuronal excitability [93]. In hippocampal slices prepared from ClC-2-null animals, pyramidal neurons had lowered resting Cl⁻ conductance and associated increases in their excitability [94]. The effect of ClC-2-deletion on electrical activity in neuronal networks was more complex, producing a net inhibition, which has been explained by the elevated excitability and activity of inhibitory neurons [94]. That being said, the overall impact of ClC-2 on neuronal excitability is far from clear. Mutations in ClC-2 have been proposed as a susceptibility factor in idiopathic generalized epilepsy, although the direct link between ClC-2 and epilepsy remains tenuous [95]. ClC-2 knock-out mice have no changes in seizure thresholds [96].

In astrocytes, immunoreactivity of ClC-2 is region-specific and highly polarized, with a high abundance reported next to GABAergic neurons [91]. The latter findings led to the suggestion that ClC-2 may mediate Cl⁻ delivery to the neuronal populations with intense GABAergic activity [91]. Whether ClC-2 function significantly impacts membrane potential and electrical properties of glial cells is still under debate. Deletion of ClC-2 does not dramatically change [Cl⁻]_i levels in Bergman glial cells, suggesting its minimal contribution to resting membrane conductance [47]. Nonetheless, the ClC-2 knockout animals have severe glial phenotypes, particularly in their white matter. ClC-2^−/− mice develop age-dependent leukodystrophy, manifesting as widespread myelin vacuolation in the brain and spinal cord [96]. In humans, similar leukodystrophy was subsequently identified in three adult and three pediatric patients carrying ClC-2 mutations [97]. It is unknown in which cell type, astrocytes or oligodendrocytes, the loss ClC-2 causes myelin deficiencies. Nevertheless, the growing consensus in the field is that glial ClC-2 is critical for ion and water homeostasis in the brain. ClC-2 functions are likely additionally modulated by two auxiliary proteins, GlialCAM and MLC1, mutations of which cause leukodystrophies with phenotypes resembling those observed upon ClC-2 deletion (see [98–101]).

The widely distributed ClC-3 (CLCN3) is an intracellular Cl⁻/H⁺ exchanger that was initially cloned from the brain [3;102]. There, it is expressed in acidic intracellular organelles, particularly in synaptic vesicles [103]. Deletion of ClC-3 limits synaptic vesicle acidification and neurotransmitter uptake, and causes postnatal degeneration of the hippocampus and retina [103]. A number of publications expressed an alternative view that ClC-3, or some of its splice variants, can also mediate plasmalemmal Cl⁻ conductance (e.g. [104–106]). Nevertheless, the bulk of the ClC-3 work and the recent papers, which extensively characterized the heteroexpressed ClC-3 variants, all strongly indicate that the previously reported plasmalemmal “ClC-3” conductances have likely been mediated by other types of endogenous anion channels [3;107;108]. Other intracellular Cl⁻/H⁺ exchangers, ClC-4 through ClC-7, have also been detected in neurons and astrocytes, with ClC-6 showing a neuron-specific pattern of expression [109–112].

2.5 Volume-regulated anion channels (LRRC8 family)

There is one Cl⁻ channel that can be detected in virtually every type of mammalian cell: namely, the volume-regulated anion channel (VRAC). Initially, VRACs were identified only on a functional level, as the robust Cl⁻ currents induced by cellular swelling upon exposure to hypoosmotic media. These volume-sensitive Cl⁻ currents displayed very similar biophysical properties across many cell types, and thus, were ascribed to be mediated by the same channel or group of highly related channels (reviewed in [16;113;114]). Based on their unique biophysical profile, VRAC channel has been alternatively referred to as volume-sensitive outwardly rectifying Cl⁻ channel (VSOR), or volume-sensitive organic osmolyte-anion channel (VSOAC) [16;113;114]. The VSOAC name highlights the channel permeability to both inorganic anions (such as Cl⁻ and HCO₃⁻) and a variety of small, organic osmolytes, either charged or uncharged. This latter feature of VRAC is highly important for understanding its roles in the brain.

The molecular nature of VRAC remained elusive for several decades. Many candidate proteins were proposed to mediate swelling-activated Cl⁻ conductance, but all of them were eventually rejected (discussed in [115;116]). Recently, two laboratories used genome-wide siRNA screens to independently identify VRAC as the heteromeric product of proteins belonging to the leucine repeat-rich containing family 8 (LRRC8) [117;118]). The LRRC8 family includes five related proteins, LRRC8A-E. One of these, LRRC8A, is indispensable for ion conductance, but must heteromerize with at least one additional subunit to produce functional VRAC channels [117;119;120]. RNAseq data and quantitative PCR indicate that all five LRRC8 proteins are expressed in brain cells, including neurons and astrocytes, however the expression levels of LRRC8E are very low, at least 10-fold lower compared to other members of the LRRC8 family [121;122]. The precise assembly of the partnering subunits determines the cell-type specific biophysical properties of VRAC [117;119;120], including its signature channel inactivation at positive potentials [117;123], and the selectivity for organic osmolytes passing through the VRAC pore [124–126].

The main function of VRAC is in cell volume regulation (see Section 1.1). VRAC activity drives regulatory cell volume decrease via two mechanisms: (i) it provides a route for Cl⁻ and HCO₃⁻ release, and in such a way assists in the electroneutrality of swelling-activated fluxes of K⁺ via separate but functionally coupled K⁺ channels, (ii) it creates a pathway for the release of small organic osmolytes. Altogether, VRAC directly or indirectly facilitates the movement of numerous osmolytes, and in such a way drives the efflux of osmotically obligated water (reviewed in [7–9]). Physiological stimuli that lead to VRAC activation are not limited to cell swelling. Stimulation of G_q-coupled GPCRs or oxidative stress have been found to produce limited VRAC activation and act synergistically even with small degrees of cellular swelling in various cell types, including glial cells [127–131]. The underlying mechanisms for GPCR actions are incompletely understood, but likely involve multiple intracellular signaling cascades. The Ca²⁺-dependent PKC isoforms, PKCα and β, have been most frequently implicated in the receptor-stimulated VRAC opening, however many other protein kinases may also contribute (see [132–135] and reviews [116;136;137]). This information is helpful for understanding VRAC behavior under physiological conditions and in pathological states.

Once activated, VRAC acts much like the ligand-gated Cl⁻ channels and moves Cl⁻ toward its electrochemical equilibrium, with the consequences that have been already discussed in Section 2.4. What makes VRAC unique among other Cl⁻ channels is its ability to facilitate the efflux of various neuroactive substances, including the excitatory amino acids glutamate and aspartate, the inhibitory neurotransmitters GABA and taurine, and perhaps others [119;122;125]. The latter aspect of VRAC physiology is highly important for bidirectional neuron-astrocyte signaling as reviewed by one of us in [35], and further discussed in Section 4.2. Recent findings suggest that certain cells, including brain astrocytes, express several distinct LRRC8-containing VRAC heteromers; one of which is preferentially responsible for the movement of Cl⁻ and anionic amino acids, while the other favors small uncharged molecules, including GABA, taurine, and glutamine [124–126].

Because of their recent discovery, the cell type-specific functions of the LRRC8 proteins, including LRRC8A, are yet to be explored using the powerful tools of molecular genetics. Global deletion of LRRC8A produces a severe phenotype, with significant embryonic and postnatal lethality, and defects in numerous organs and tissues [138]. The tissue-specific LRRC8A knockouts are now being developed, but none have been characterized thus far.

2.6 Anion exchangers

In addition to Cl⁻ transporters and permeability pathways, which are discussed in the prior sections, brain cells express a variety of Cl⁻/bicarbonate transporters belonging to the large SLC4 transporter family [139]. These proteins are important for the exchange of products of cellular metabolism (CO₂) and regulation of pH_i; however, they also deserve brief mentioning due to their direct or indirect impact on [Cl⁻]_i.

Two members of the Cl⁻-transporting SLC4 family, which are most abundant in the CNS, include the electroneutral Cl⁻/HCO₃⁻ anion exchanger 3 (AE3, encoded by SLC4A3), and the Na⁺-activated Cl⁻/HCO₃⁻ exchanger, (NDCBE, product of SLC4A8) (reviewed in [139]). These two exchangers perform functionally opposite roles in the context of Cl⁻ transport and pH regulation. AE3 is stimulated by alkalosis and exports metabolically produced bicarbonate in exchange for extracellular Cl⁻, thus serving as an acid and Cl⁻ loader. In contrast, the NDCBE exchanger responds to cytosolic acidification by taking up one Na⁺ and 2 HCO₃⁻ ions in exchange for one Cl⁻, and thus works as an acid and Cl⁻ extruder [140]. As mentioned in Section 1.1, the coordinated work of AE anion exchangers and Na⁺/H⁺ exchangers allows for regulatory volume increase in shrunken cells, as their functionally coupled work accumulates cytosolic NaCl (for comprehensive review see [141]).

3. Cl⁻ as an intracellular signaling ion

As discussed in the previous two sections, the impact of Cl⁻ on cellular functions is attributed to the charge transfer across plasmalemmal or intracellular membranes, without considering the potential intracellular signaling properties for this anion (but see review [142]). Yet, older reports and recently accumulated evidence suggest that this notion may be overly simplistic, and that alternative mechanisms of Cl⁻ actions within the cell also exist.

3.1 Direct regulation of ion channels and transporters by [Cl⁻]_i

The intracellular Cl⁻ levels can directly modulate activities of some ion channels and transporters. In the simplest case, [Cl⁻]_i regulates biophysical properties of the voltage-gated Cl⁻ channels, ClC-0 and ClC-2, by acting as the gating particle within the channel pore [143;144]. More intriguingly, intracellular Cl⁻ can allosterically modify activities of other ion channels, which do not conduct this anion. In neural and muscle cells of C. elegans, increases in [Cl⁻]_i and [Ca²⁺]_i additively activate the high conductance K⁺ channel cSLO-2 [145]. Binding of these two regulatory ions occurs at adjacent C-terminal sites [145]. In mammals, the cSLO-2 orthologues, mSLO2.1 (also known as Slick, the product of KCNT2 gene) and mSLO2.2 (Slack, encoded by KCNT1) produce K⁺ channels, which are activated by increases in [Cl⁻]_i and [Na⁺]_i, and inhibited by physiological levels of ATP_i [146;147]. Together, this unusual pattern of regulation makes SLO2 channels unique polymodal sensors for metabolic stress and possibly hypoxia. Regulation by [Cl⁻]_i is not restricted to K⁺ channels. The nonselective cation channel TRPM7 (Transient Receptor Potential subfamily M member 7) is inhibited by high intracellular Cl⁻ via anion biding to a poorly characterized domain within the protein [148].

Besides ion channels, direct sensitivity to [Cl⁻]_i was established for two splice variants of the electrogenic Na⁺/HCO₃⁻ cotransporter, NBCe1 (SLC4A4) [149]. NBCe1 and other related HCO₃⁻ transporters from the SLC4 family were initially discovered in the renal tissue [150], but soon found elsewhere, including in astrocytes and neurons [151;152]. The activity of the NBCe1-B and NBCe1-C isoforms is strongly inhibited in the physiological range of [Cl⁻]_i (5–60 mM) via interaction with one or more Cl⁻-binding GXXXP motifs [149]. The Cl⁻-dependent changes in transporter activity modulate acid-base homeostasis, and, in such a way, alter a variety of cellular functions. There are data indicating that [Cl⁻]_i can also modulate the activity of the Na⁺/H⁺ exchanger NHE-1, in a manner dependent on the C-terminal domain of the protein [153].

3.2 Cl⁻_i is a modulator of intracellular signaling cascades

A number of early experimental observations in 1980s and early 1990s, led to the hypothesis that Cl⁻ can regulate intracellular signaling cascades. Thus, in purified enzyme assays, activity of Gα_i/o was strongly inhibited by Cl⁻ and Br⁻ with the half-maximal effects at 3–20 mM, while other anions showed no effect [154]. In permeabilized neutrophils, lowering [Cl⁻]_i elicited robust protein phosphorylation and an oxidative burst, likely via the activation of small GTP-binding proteins [155]. In salivary epithelial cells, a [Cl⁻]_i- and pertussis toxin-sensitive G-protein has been found to regulate the activity of Na⁺ channels [156]. Along the same lines, in the airway epithelium, [Cl⁻]_i regulates protein phosphorylation via changes in the activity of nucleoside diphosphate kinase [157]. Surprisingly, these and other early findings gained little traction. More recently the idea of [Cl⁻]_i sensing has been firmly linked to serine/threonine protein kinases belonging to the WNK (With No lysine [K]) family [59;158;159].

WNK refers to the unique catalytic site structure of the four kinases in this family, which lack one of the conserved lysine residues responsible for the coordination of ATP within their active center [158]. After their initial cloning, WNK1 and WNK4 genes were soon linked to familial cases of hypertension, hyperkalemia, and hyperchloremia [160], and later to pathological changes in activity of the thiazide-sensitive Na⁺,Cl⁻ cotransporter (NCC) and the bumetanide-sensitive NKCC2 (reviewed in [59]). Subsequent work established that WNKs do not phosphorylate these transporters directly, but rather act via two closely related downstream protein kinases: SPAK (Sterile20-related Proline-Alanine-rich Kinase) and OSR1 (Oxidative Stress Responsive kinase 1) [161–164]. In addition to the activation of NKCC1/2, WNK-SPAK/OSR1 signaling cascade reciprocally regulates (inhibits) the activity of all four KCC transporters, thus providing coordinated control over the critical Cl⁻ influx and efflux pathways [71;72] Among WNK family members, WNK3 appears to be the most abundant and functionally significant enzyme in the CNS, but other WNK proteins are also expressed [165;166].

The idea that one or more WNK kinases serve as intracellular Cl⁻ sensors had been around for a while and was indirectly supported by findings of [Cl⁻]_i-sensitivity of cation-chloride cotransporters (for early discussion see [167]). However, the direct experimental support for this hypothesis has been collected in very recent studies. Several groups identified the structural basis for direct modulation of WNK1, WNK3, and WNK4 by [Cl⁻]_i [168;169]. Cl⁻ (or Br⁻) binds to the DLG motif in the kinase domain and N-terminal activation loop of the WNK enzymes. This anion binding inhibits protein autophosphorylation, and prevents enzyme activation [168]. For WNK1 and WNK3, the inhibitory actions of Cl⁻ develop in the concentration range of 5–20 mM. WNK4 is apparently more sensitive to [Cl⁻]_i and is potently inhibited in the concentration range of 5–10 mM [169]. The proposed mechanism of WNK-mediated [Cl⁻]_i sensing and the downstream effects on cation-Cl⁻ cotransporters are depicted in a simplified form in Fig. 3.

Fig. 3. Cl⁻ as an intracellular signaling ion.

Recent findings identified the With No lysine [K] (WNK) protein kinases as intracellular [Cl⁻] sensors. Binding of Cl⁻, which occurs in the N-terminal activation loop of WNK1-4, inhibits autophosphorylation and activity of these enzymes. Reductions in [Cl⁻]_i promote WNK1-4 autophosphorylation and activation. Active WNKs phosphorylate and stimulate two closely related protein kinases SPAK/OSR1, which in turn phosphorylate NKKC1 and KCC1-4. The functional effects of phosphorylation on cation-Cl⁻ cotransporters are opposite: activation of NKCC1 and inhibition of KCC1-4. When WNKs are inactive, NKCC1 and KCC1-4 are dephosphorylated by the serine/threonine protein phosphatases PP1 and PP2A. See text for additional details.

Although the cell volume- and Cl⁻-sensitive changes in activity of the WNK cascade have been mainly considered in the context of regulation of SLC12A transporters, the relevant kinases have multiple targets, both membrane and intracellular. For example, in C. elegans the WNK signaling axis, specifically the SPAK-like kinase, inhibits the activity of the Cl⁻ channels belonging to the ClC family [170;171]. There were also reports about direct interactions between WNKs and serum- and glucocorticoid-induced protein kinase 1 (SGK1) [172], WNK1-dependent regulation of the Erk cascade proteins MEKK2/3 [173], and WNK-dependent phosphorylation of claudins [174], etc.

In the context of astroglial physiology, it may be relevant to mention the putative role of [Cl⁻]_i in regulating exocytosis. Astrocytes are the secretory cells of the CNS, which, among other mechanisms, utilize vesicular secretion for release of bioactive molecules [175]. In the past, intra-cellular Cl⁻ has been shown to be important, or even obligatory for sustaining exocytosis in neurohypophysial nerve endings, cultured brain pituitary cells, and beta cells of the pancreas (e.g. [176–178]). This is likely related to the function of intracellular Cl⁻/H⁺ exchangers from the ClC family, which support vesicular acidification [178], but may also involve activation of small GTPases [179]. Interestingly, the effects of [Cl⁻]_i on exocytosis can be bell-shaped, with non-physiologically high Cl⁻_i suppressing this process [176]. It remains to be explored if the newly discovered [Cl⁻]_i sensors WNK kinases are involved in these phenomena.

4. Potential role of Cl⁻ in the crosstalk between neurons and astrocytes

Over the last two decades, the field of astroglial physiology has undergone a dramatic transformation. Rapidly accumulating evidence supports the existence of extensive and reciprocal signaling between neurons and astrocytes, which is largely mediated by neuro-transmitters and gliotransmitters (see [180–183]). Activity-dependent fluctuations in extracellular ion levels, such as K⁺ and Na⁺, and intracellular transients in [Ca²⁺] and [Na⁺], represent important elements of intercellular information transfer [181;184;185]. In this section, we make an argument that Cl⁻ fluxes also play a role in neuron-astrocyte signaling (see Fig. 4).

Fig. 4. The role of Cl⁻ in the bidirectional astrocyte-neuron communication.

The transfer of information from neurons to astrocytes, and the reverse process, are schematically separated by a dashed line. Top, Neurons alter astrocytic [Cl⁻]_i through the release of both excitatory and inhibitory neurotransmitters. Glutamate uptake through the Na⁺/K⁺-dependent excitatory amino acid transporters EAAT1 and EAAT2, gates the Cl⁻ permeability pore. Additionally, glutamate, ATP and several other signaling molecules activate GPCR pathways (such as mGluR receptors) and stimulate Cl⁻ uptake via NKCC1. Uptake of the inhibitory transmitter GABA through the GABA transporters, GAT-1 or GAT-3, is stoichiometrically associated with the symport of 2 Na⁺ and one Cl⁻. Bottom, Astrocytes directly modify neuronal [Cl⁻]_i via release of the gliotransmitters, GABA and taurine, which subsequently activate Cl⁻ fluxes through the neuronal GABA_A and glycine receptor (GlyR) channels. The main pathway for astrocytic GABA and taurine release is thought to be the volume-regulated anion channel (VRAC), but other release mechanisms also exist. See text for additional details.

4.1 Do neurons modify astrocytic Cl⁻ and is it of functional consequence?

A well-characterized role of astroglia is maintenance of an optimal environment for neuronal activity, which is accomplished via uptake of neurotransmitters, export of metabolites, and regulation of ionic composition of the extracellular milieu, particularly extracellular K⁺ levels (reviewed in [186–188]). These processes are predominantly concentrated in astrocytic endfeet engulfing neuronal synapses and brain microvessels. Many of the relevant astrocytic transporters directly or indirectly impact the transmembrane Cl⁻ fluxes.

Astrocytes express two main Na⁺/K⁺-dependent excitatory amino acid transporters; namely, EAAT1 (encoded by SLC1A3) and EAAT2 (SLC1A2) in humans, or GLAST and GLT-1, respectively, in rodents (reviewed in [189;190]). These transporters concentrate glutamate inside the cell using the energy of established gradients for monovalent cations. Each electrogenic transport cycle takes one glutamate molecule inside the cell together with 3 Na⁺ and 1 H⁺, in exchange for 1 K⁺ (reviewed in [191]). What is important for the present discussion – the activity of plasmalemmal glutamate transporters is also associated with Cl⁻ currents [192–195]. EAATs’ Cl⁻ permeability is uncoupled from the neurotransmitter transport, and, therefore, Cl⁻ movement is governed by its electrochemical gradient. Accordingly, when gated by glutamate, EAATs behave like small-conductance ligand-gated Cl⁻ channels (e.g. [195] and review [196]). In astrocytes, stimulation of the EAATs would be expected to decrease intracellular [Cl⁻]. In fact, this is exactly what has been confirmed in a recent study by the Fahlke laboratory. In slice recordings in Bergman glia, they found that developmental upregulation in the activity of GLAST and GLT-1 is associated with a drop in the intracellular [Cl⁻] from 52 to 35 mM [47]. As would be predicted, EAAT inhibitors increased the intracellular [Cl⁻] in Bergman glial by more than 15 mM [47]. Remarkably, the related neuronal EAAT4 and EAAT5 seem to function predominantly as ligand-gated inhibitory Cl⁻ channels, rather than glutamate transporters per se (discussed in [190;196]).

The inhibitory amino acid transmitter GABA is taken inside the cells by three Na⁺- and Cl⁻-dependent GAT transporters, GAT-1 (SLC6A1), GAT-2 (SLC6A13), and GAT-3 (SLC6A11), and the betaine-GABA transporter BGT-1 (SLC6A12) (reviewed in [190;197]). Among these, BGT1 is probably of minor importance because BGT-1-null mice have normal development and no high seizure susceptibility phenotype, indicating normal GABA signaling [198]. GAT-1 is found in both neurons and astroglia, GAT-2 levels are very low to undetectable in the brain, and the expression of GAT-3 appears to be restricted to astrocytes [190]. Activity of all GABA transporters depends on the gradients of Na⁺ and Cl⁻. Although the transport stoichiometry of these transporters is not fixed, normally each working cycle transfers 2 Na⁺ and 1 Cl⁻ with one neurotransmitter molecule [199;200]. While GABA uptake is dependent on and associated with Cl⁻ movement, much like for the EAATs, there is some evidence that GABA transporters also act as Cl⁻/anion channels [201]. The last caveat notwithstanding, GABA uptake by astrocytes is expected to lead to elevation in [Cl⁻]_i. Besides the possible modulation by neurotransmitter transporters, shifts in [Cl⁻]_i may also occur as a result of activation of GABA_A receptors, which has been already discussed in Section 2.3. An overview of astrocytic GABA receptors and GABA responses can be found in the comprehensive review by Porter and McCarthy [202].

Another significant mechanism through which neuronal activity can be coupled to changes in astrocytic [Cl⁻]_i is activation of NKCC1, which can be driven by metabotropic neurotransmitter receptors. The majority of such receptors on astrocytes are linked to either PLC-dependent increases in the intracellular [Ca²⁺] or cAMP signaling (reviewed in [202;203]). As outlined in Sections 2.1 and 3.2, NKCC1 activity is potently modulated by numerous Ca²⁺ and cAMP-dependent protein kinases, most likely via their downstream effects on the WNK signaling cascade. The physiological consequences of NKCC1 activation would be two-fold: (i) elevation of intracellular [Cl⁻]_i and (ii) cellular swelling, with the latter leading to opening of the swelling-activated channel, VRAC. In addition, metabotropic receptors for glutamate, ATP, and adenosine, and few other signaling molecules have been found to lead to limited, Ca²⁺-dependent VRAC opening, even in the absence of cell swelling (see for example [128;135;204]). Regardless of the stimuli, VRAC opening will reduce [Cl⁻]_i.

Altogether, the impact of neuronal activity on astrocytic [Cl⁻]_i is not uniform, and likely context-dependent. Activation of membrane Cl⁻ conductance during uptake of the excitatory neurotransmitters glutamate and aspartate, opening of GABA_A receptor-channels, opening of VRAC, or hyperpolarization-induced ClC-2 activity, are all likely to reduce intracellular [Cl⁻]. On the other hand, the activity of the Na⁺,Cl⁻-dependent transporters for GABA and several other neurotransmitters, or activation of NKCC1 should lead to elevations in [Cl⁻]_i. As discussed in the preceding text, we do not think that Cl⁻ movement causes significant shifts in the astrocytic membrane potential. Instead, [Cl⁻]_i may modulate astrocytic functions via changes in the activity of the Cl⁻-sensitive WNK protein kinases.

4.2 How astrocytes modify neuronal Cl⁻ and why is it functionally significant?

As a part of a “tripartite synapse”, astrocytes respond to and modulate neuronal synaptic activity via a Ca²⁺-dependent release of gliotransmitters, most notably glutamate and ATP, but also a variety of other substances [180–183]. Although the idea of vesicular gliotransmitter release dominates the field, several alternative mechanisms have also been considered, including the activation of VRAC, and others [128;205;206]. In this subsection, we discuss only the processes which rely on Cl⁻ signaling within neuronal cells.

The most obvious way in which astrocytes can modulate neuronal [Cl⁻]_i is the activation of neuronal GABA and glycine receptor-channels. It has been proposed that tonic GABA release in the healthy and pathological brain is largely of astrocytic origin, and mediated by the Bestrophin-1 Cl⁻/anion channels [207–209]. We speculate that the LRRC8-containing VRACs may also be involved in this process, because they are permeable to GABA, and, much like Bestrophins, can be activated in a Ca²⁺-dependent fashion [125;128]. In support of potential VRAC involvement, it has been known for years that activity of magnocellular neurons in supraoptic and paraventricular nuclei of the hypothalamus is modulated by gliotransmitter release by a volume-sensitive anion channel, likely VRAC. Hypothalamic astrocytes and specialized pituicytes in the neurohypophysis contain high levels of the atypical aminosulfonic acid taurine, and release it to the extracellular space both tonically, and in an osmolarity-dependent manner [210;211]. Taurine strongly modulates the secretion of vasopressin and oxytocin by the supraoptic magnocellular neurons, by acting as an agonist at their glycine receptors and modulating their membrane potential via Cl⁻ fluxes (see [212;213] and review [214]).

It is important to recall that neuronal [Cl⁻]_i and the polarity of GABA actions are determined by the electroneutral cation-Cl⁻ cotransporters. In neurons, the net activity of NKCC1 can be regulated at the expression level or via transporter phosphorylation. In cellular models, activation of neuronal NMDA, AMPA, or metabotropic mGluR1 and mGluR5 receptors leads to the Ca²⁺-dependent increase in NKCC1 activity [215;216]. Such activation may be driven at least in part by the release of excitatory gliotransmitters. The role of NKCC1 in setting normal neuronal [Cl⁻]_i and determining GABA actions during development has been already discussed (see Section 2.1). In a pathological context, NKCC1 can promote excitation and tissue damage, for example in epileptogenesis and traumatic brain injury [217;218]. Whether astrocytes can modify the activity of the functionally opposite neuronal KCCs is less clear, yet, any potential effects cannot be overlooked. Deletion of the neuron-specific KCC2 or neuron- and astrocyte expressed KCC3 and KCC4 lead to early postnatal mortality or severe central and peripheral phenotypes (reviewed in [38;70]). Finally, modulation of neuronal [Cl⁻]_i, by astrocytes or otherwise, can regulate the activity of WNK kinases, and in such a way provides regulatory feedback to KCCs and NKCC1, but also mediates other effects via phosphorylation of alternative targets (see Section 3.2).

5. Perspectives and challenges

The regulatory role of Cl⁻ conductance in neuronal signaling is well recognized. Therefore, it is very intuitive that astrocytic release of the inhibitory neuro/gliotransmitters, GABA and taurine, can provide a mechanism for the regulation of neuronal excitation. Substantial evidence in support of this notion already exists, at least for select brain areas (see Section 4.2). What remains uncertain is how the inhibitory gliotransmitters are being released from astrocytes. A vesicular release mechanism has been proposed for a number of gliotransmitters, and is still very much a part of the discussion. However, astrocytes express low-to undetectable levels of the vesicular GABA transporter vGAT (SLC32A1) [121]. Consistently, unlike synaptic vesicles in neurons, astrocytic vesicles do not contain measurable levels of GABA or glycine [219]. We are not aware of any vesicular transporters, which would accommodate taurine. Based on these considerations, alternative mechanisms for glial GABA and taurine release are being pursued, including plasmalemmal amino acid transporters, as well as VRAC and Best1 anion channels (see Sections 2.5 and 4.2). This area is still very much in flux.

Unlike the clarity of the role of Cl⁻ in regulation of neural function, we have a long way to go toward elucidating whether this anion plays signaling functions in astrocytes. The fact that astrocytic intracellular [Cl⁻] is kept above the level of its electrochemical equilibrium, indirectly implies the significance of Cl⁻ in astroglial physiology. However, modulation of membrane potential by Cl⁻fluxes is likely less important in astrocytes as compared to neurons. Activation of Cl⁻ permeability is expected to produce a moderate hyperpolarization of the already highly negative cell that is electrically coupled to its neighbors via an astrocytic syncytium. Perhaps the recent discovery that the WNK kinase family members can act as [Cl⁻]_i sensors and modulate the activity of other Cl⁻ permeability pathways points to new, unexpected signaling properties for this anion. Additionally, there are a few reports that intracellular Cl⁻ levels may impact exocytosis and vesicular release processes. Because astrocytes are considered to be the secretory cells of the CNS (see Section 3.2), exploring the putative connection between [Cl⁻]_i and gliotransmitter release may yield new, unexpected information.

The field faces a number of significant technical and conceptual challenges in testing specific and distinct roles for Cl⁻ fluxes and intracellular Cl⁻ levels in neuron-astrocyte communication. The biggest obstacle is in separating the impact of intracellular Cl⁻ signaling from the effects of changes in the membrane potential, because it is difficult to clamp intracellular [Cl⁻] without changing membrane polarization. Perhaps, new, specifically devised electrophysiological approaches will be of further assistance. Other barriers in the field are represented by the lack of Cl⁻ sensors with a good dynamic range, which would match physiological Cl⁻ levels, and the deficit of selective tools for manipulation of [Cl⁻]_i in vivo. The latter two obstacles appear to be surmountable with the recent advances in the development of protein-based Cl⁻ sensors and optogenetics tools. We think that introducing the Cl⁻-conducting channelrhodopsins into glial cells may be as instructive as recent studies in neurons. The relevant experiments will shed a light on already expected and, perhaps, as-yet-unanticipated Cl⁻ functions in glial physiology, and strengthen the idea of Cl⁻ as a signaling ion between neurons and astrocytes.

Highlights.

• The transmembrane chloride fluxes via GABA_A and glycine receptors are well-known to regulate excitability and communication within neuronal networks. Here we make an argument that chloride signaling is an important element in bidirectional neuron-astrocyte communication.

• Neurons modulate intracellular chloride levels in astrocytes as a result of (i) chloride fluxes associated with neurotransmitter uptake, (ii) modulation of astrocytic cation-chloride cotransporters, particularly NKCC1, (iii) activity-dependent astrocytic swelling and opening of chloride/anion channel VRAC.

• Astrocytes can modify neuronal chloride levels and excitability via the release of GABA and taurine, two endogenous agonists of inhibitory GABA_A and glycine receptors. Astrocytic release of inhibitory gliotransmitters involves VRAC and perhaps other release pathways.

• Finally, the recent discovery that the WNK family protein kinases are regulated by chloride opens the possibility that changes in cytosolic levels of this anion may act as an intracellular signal, both in neurons and astrocytes.

Major Abbreviations

AE

anion exchanger

AVD

apoptotic volume decrease

EAAT

excitatory amino acid transporter

GABA

γ-aminobutyric acid

KCC

K⁺-Cl⁻ cotransporter

NKCC

Na⁺-K⁺-2Cl⁻ cotransporter

RVD

regulatory volume decrease

RVI

regulatory volume increase

VRAC

volume-regulated anion channel