Physiology of Astroglia

Abstract

Astrocytes are neural cells of ectodermal, neuroepithelial origin that provide for homeostasis and defense of the central nervous system (CNS). Astrocytes are highly heterogeneous in morphological appearance; they express a multitude of receptors, channels, and membrane transporters. This complement underlies their remarkable adaptive plasticity that defines the functional maintenance of the CNS in development and aging. Astrocytes are tightly integrated into neural networks and act within the context of neural tissue; astrocytes control homeostasis of the CNS at all levels of organization from molecular to the whole organ.

I. DEFINITION AND OVERVIEW OF FUNCTION

Astroglia are a class of neural cells (also known as astrocytes) of ectodermal, neuroepithelial origin that sustain homeostasis and provide for defense of the central nervous system (CNS) (FIGURE 1). Astrocytes are highly heterogeneous in form and function and demonstrate remarkable adaptive plasticity that defines the functional maintenance of the CNS in development and aging. Astrocytes are tightly integrated into neural networks and act within the context of neural tissue; astrocytes control homeostasis of the CNS at all levels of organization from molecular to the whole organ. Astrocytes maintain molecular homeostasis of the CNS by transporting major ions and protons, by removing and catabolizing neurotransmitters, and by releasing neurotransmitter precursors and scavengers of reactive oxygen species. Astrocytes sustain neurotransmission by supplying neurons with neurotransmitter precursors and control cellular homeostasis through embryonic neurogenesis (that occurs from radial glia) and adult neurogenesis (which involves stem astrocytes of neurogenic niches). Astrocytes regulate metabolic homeostasis through synthesizing glycogen and supplying neurons with energy substrates. Astrocytes define the cytoarchitecture of the grey matter by tiling the latter and by forming contacts with the vasculature by vascular endfeet and by glial sheets at all surfaces of the brain. The vascular endfeet, which plaster along the entire vasculature, release vasoactive substances thus contributing to functional hyperemia. Astrocytes in the guise of glia limitans form the pial cover of the CNS, control blood-brain barrier and act as chemosensors, thus contributing to systemic homeostasis (regulation of energy balance, blood pH and Na⁺ concentration). Finally, through mounting reactive response, astrocytes (together with microglia) represent the main defensive system of the CNS (we shall not discuss astrogliopathology in the present paper, instead recommending recent comprehensive reviews (257, 258, 1329, 1352, 1353, 1637, 1638, 1815, 1818). These numerous functions of astrocytes are of vital importance for all aspects of CNS operation, including its development, experience-dependent adaptation and aging.

FIGURE 1.

Homeostatic functions of astroglia.

II. HISTORIC PROLOGUE

Rudolph Virchow introduced the concept of neuroglia¹ (1826, 1827) as true connective tissue of the brain, with little considerations for its cellular nature. Virchow referred to neuroglia as a Zwischenmasse or in-between tissue, into “which the nervous system elements are embedded” (1827). The very first account of neural cell that was subsequently classified as glia was, however, produced some while before Virchow's seminal deliberation. This was a radial-like glial cell of the retina, the Müller cell, described by Heinrich Müller in 1851 (1165). These cells were thereafter characterized in most minute details by Max Schulze (1579). In 1857 Karl Bergmann (155) discovered radial-like glial cells of the cerebellum, today known as Bergmann glial cells. Parenchymal glia received much attention by 19th century neuroscientists and numerous detailed descriptions of these cells, under many different names, have been published (FIGURE 2). The parenchymal neuroglia were named Bindesubstanzzelle (binding substance cells or connective cells) by Otto Deiters (398) or Fasernetz sternförmiger Zellen (fiber network stellate cells) by Leopold Besser (164). Carl Frommann (536) was the first to introduce connotation of the glue by naming glia Leim erfüllten Interstitien (glue-filled interstitium); Albert von Kölliker (894) called glial cells Sternförmige Zellen (star-form cells), Eduard Rindfleisch (1469) called them Stützcelle or Neurogliazellen (supportive cells or neuroglial cells), Victor Butzke (271) called them Gliakörperchen (glial bodies), Moritz Jastrowitz (787, 788) called them spinnenähnliche Gliazellen or Spinnezellen (spider glial cells or spider cells), Carl Ludwig Schleich (1565) called them Mooszellen (moss cells), and Gustaf Magnus Retzius (1458) called them asteroide Gliäcyten or Sternzellen (starlike gliocytes or star cells). Camillo Golgi (who always used the term neuroglia) was the first to demonstrate that glia represent a cellular population distinct from nerve cells, although he also believed that glial cells and neurons may transform into each other. Golgi identified glia as round cells with numerous fine processes extended in all directions; many of these processes are directed towards blood vessels (585). Using the silver-chromate staining technique (reazione nera), Golgi described a remarkable diversity of glial cells in the brain, reported glial networks, and identified glial endfeet plastering blood vessels (583–585). The silver-chromate impregnation technique was also instrumental in visualizing and identifying radial glia (145).

FIGURE 2.

Historic images of astrocytes. A: Müller glial cell of the sheep retina drawn by Max Schulze using a microscope from Amici. y, Brushlike fibrils extending from the outer Müller fiber in the outer granular layer; x, internal limiting membrane; a, opening in the limiting membrane; b, very delicate network of fenestrated membranes similar in the ganglion cell layer; c, network in the so-called molecular layer; d, nuclei as part of the Müller fibers; ee, cavity in which the nuclei or the cells of the internal granular layer are located. [From Schulze (1579). Image has been kindly provided by Prof. Helmut Kettenmann, Max Delbruck Centre for Molecular Medicine, Berlin.] B: cortical astrocytes drawn by Albert von Kolliker (894). C: Camillo Golgi’s drawings of astrocytes contacting blood vessels (583). D: the “Spinnenzellen” of Moritz Jastrowitz (787). E: morphological diversity of neuroglia in human fetal cortex (1458). F: close interactions between neuroglial (red; both interlaminar and protoplasmic astrocytes are clearly presented) and neuronal (black) networks in human brain. [From Schleich (1565).]

The term astrocyte (αστρον κψτοσ; astron, star and kytos, a hollow vessel, later cell, i.e., starlike cell) was introduced by Michael von Lenhossék in 1895 (969); prophetically Lenhossék proposed to call all parenchymal glia spongiocytes with astrocytes being a subtype.² Slightly earlier Albert von Kölliker and William Lloyd Andriezen distinguished grey and white matter glia; Andriezen named glial cells in grey matter protoplasmic and those in white matter fibrous. Andriezen believed these two cell types had different ontogeny, the protoplasmic cells being of mesoblastic origin, while the fibrous cells being ectodermal. He also contemplated the complexity of protoplasmic processes, indicating that they have “shaggy granular contour, as if a fine moss constituted the protoplasmic processes” (44). The term astrocytes for denoting parenchymal neuroglia was much popularized by Santiago Ramón y Cajal (FIGURE 3), who developed an astroglia-specific gold and mercury chloride-sublimate staining technique (550), which labeled glial fibrillary acidic protein (GFAP); this staining allowed Cajal to confirm the origin of astrocytes from radial glia (1429, 1430). Most of 19th and early 20th century neuroscientists [with singular exception of Carl Weigert who thought that glia were needed only to fill the gaps between neurons (1871)], assigned numerous functions to astroglia. Golgi, for example, contemplated glia as distributors of nutritive materials (583, 584). Ernesto Lugaro envisaged thin glial processes that infiltrate the synapses and metabolize neuroactive substances (1019). The active role of astrocytes in controlling information flow in the brain was suggested by Carl Ludwig Schleich, who postulated that astroglial processes may (through swelling and shrinking) control synaptic transmission (1565). Similar ideas were entertained by Ramon y Cajal, who thought that retraction of astroglial processes allows information flow during wakefulness, whereas expansion of astroglial processes halts interneuronal connectivity, thus inducing sleep (1427). Cajal also suggested the central role of astrocytes in controlling the vasculature of the brain and mediating functional hyperemia: contraction/relaxation of astroglial perivascular processes could increase or decrease the diameter of brain capillaries, thus regulating the blood flow (1427). Fernando De Castro, a pupil of Cajal, proposed that neuroglial cells may release neuroactive substances and directly participate in neural transmission (389), whereas Robert Galambos considered neuroglia as a central element for higher brain functions while neurons “merely execute the instructions glia give them” (543). The theme of glia being the primary element of information processing, memory, cognition, and consciousness is regularly resurfacing (132, 299, 1368, 1369, 1473); this stimulating conjecture only lacks credible experimental support. Physiological examination of neuroglia began in late 1950s when these cells were probed with electrophysiological and radiotracer techniques applied to in situ and in vivo preparations from vertebrates and mammals, and the first data on dynamic interactions between neurons and glia have emerged (682, 697, 919, 1276, 1731, 1862). In the late 1980s, Jean de Villis established purified cultures of neuroglial cells, which allowed direct examination of physiology of astrocytes at the single-cell level (1155).

FIGURE 3.

Images of astroglia drawn by Santiago Ramon y Cajal. A: Golgi impregnated glia from human cortex (2-mo-old child) in the plexiform layer (A–D), second and third layers (E–H and K, R, respectively), and perivascular glia (I and J). V, blood vessel. B: perivascular astrocytes. [These images are part of the collection of the Cajal Legacy at the Cajal Institute of the Spanish Research Council (CSIC), Madrid, Spain. Images have been kindly provided by Professor Ricardo Martínez Murillo.]

III. EVOLUTION OF ASTROGLIA

The evolutionary emergence of glia (natural history of which is exceedingly difficult to follow because cells do not leave many trails in the fossils) most likely began with a transition from diffuse nervous system (present in Ctenophora and Cnidarians) to a centralized nervous system when first neuronal ganglia emerged. The diffuse nervous system (that evolved from the ectoderm) is composed from multipolar and unipolar neurons. These are integrated by chemical synapses into several semi-independent networks; there is no firm evidence for these networks containing any type of supportive or associated cells. Detection of glia in phylogenetically lower taxa is not a trivial task, primarily because of absence of any specific markers; rather, morphological criteria are used for cell type identification. Probably the main criterion is a close association with neurons and coverage of neuronal structures, i.e., the feature that is fundamental for glial cells. Despite remarkable similarity of glial functions between invertebrates and vertebrates, it is likely that neuroglia appeared in phylogeny on several occasions and evolved every time in a distinct way using disparate genetic associations. In insects, for example, development of glia is controlled by the gene glial cells missing (gsm; Refs. 736, 799); whereas in mammals the gsm homolog is not even expressed in the CNS (860).

Centralization of the nervous system emerged together with bilateral symmetry, and hence the origin of glia should be sought at the base of bilateralia. Indeed, some of the most ancient bilateralia, the Acoelomorpha (1181), already have a nervous system with the frontally localized “bilobed” (i.e., bilobar) brain with a cellular cortex and a dense internal neuropile (7). Electron microscopy of the brains of Symsagittifera roscoffensis, Convoluta psammophila, Amphiscolops sp., and Otocelis rubropunctata (free living Acoela worms) found non-neuronal cells with electron-dense cell bodies in which nuclei occupy most of the cytosol, and lamellar processes extend into neuropil and surround groups of neurites (120, 162). The compact anterior “brain” is also present in Platyzoa (Rotifera and platyhelmintes), although not all representatives of this superphylum possess glial cells. Neuroglia seem to be absent in Rotifera and in tubellarian flatworms such as Catenulida or Macrostomida or Rhabdocoela (653). Supportive glial cells however were identified in higher platyhelmintes, the polyclads and triclads, in which they insulate and support nerve cords (590). In round worms, glial cells are mostly associated with sensory organs, although several glial cells seem to be specialized for neuronal support in the CNS and can be therefore considered to be proto-astrocytes (see below). Neuroglia are well developed in molluscs, in Annelida, and even more developed and diverse in Arthropoda, in insects and crustaceans, with some cells being quite similar to astrocytes; some of these cells express typical markers such as GFAP (653). The radial glia replace parenchymal glia in Echinodermata (the sister phylum of chordata); similarly, radial glia are the main feature in the brains of lower chordata, and radial glia are associated with the emergence of a layered nervous system. An increase in the thickness of the brain triggered a wave of astroglia diversification, which progressed in vertebrates.

A. Proto-astrocytes in Caenorhabditis elegans

The nervous system of Caenorhabditis elegans is composed of 302 neurons, 50 supportive (glial) cells of ectodermal origin, and 6 supportive cells derived from mesoderm (1258, 1861, see also http://www.wormatlas.org/hermaphrodite/neuronalsupport/Neurosupportframeset.html). None of these glial cells expresses classical astroglial markers. The CNS of C. elegans is represented by the nerve ring located in the frontal part of the body; the nervous ring receives processes of sensory neurons that are located in the periphery. The central ring also contains cephalic and motor neurons, which send efferent signals through the ventral and dorsal nerve cords. Majority (46) of C. elegans glia are associated with the sensory system. These cells are subdivided into 26 socket cells and 20 sheath cells that (together with neuronal processes) compose the worm's sensory organs known as sensillas (1371). The remaining four glial cells known as cephalic sheath (CEPsh) cells are associated with the neural ring. These CEPsh cells are bipolar; the anterior processes enwrap cephalic neuronal dendrites and form the corresponding sensilla in the lips of the animal. Posterior processes of CEPsh cell have lamellar morphology; they ensheath the nerve ring and send processes to the neuropil, where they contact synapses (1258, 1683). Consequently, these CEPsh cells can be defined as proto-astrocytes (FIGURE 4). The C. elegans also contains six mesoderma-derived supportive cells (known as GLR cells), which are located around the nerve ring. These cells (rather uniquely) make gap junctions with neurons and muscle cells and possibly contribute to neuronal-muscular communications (1258). The function of proto-astrocytes in the round worms are yet to be fully characterized. Arguably they control ion homeostasis in perisynaptic regions and are involved in neuronal development and morphogenesis. Although artificial ablation of glial cells (by either exposure to laser beam, or by expressing the diphtheria toxin A gene under control of glia-specific promoter) renders a complex of morphological, developmental, sensory, and behavioral deficits, it is, nonetheless, compatible with survival of the worm (82). Incidentally, C. elegans glia display some intermediate neuronal/glial physiology; for example, they generate Ca²⁺ signals through activation of voltage-gated channels and do not have functional intracellular Ca²⁺ stores (1682).

FIGURE 4.

Proto-astrocytes (cephalic sheath, or CEPsh glial cells) of roundworm C. elegans. A: a cartoon of an adult worm showing the four CEPsh glial cells (green) positioned in the anterior of the worm (inset). The CEPsh cell bodies with their velate processes are positioned around the central nerve ring (red), which they enwrap along with the proximal section of the ventral nerve cord. Additionally, each CEPsh glial cell possesses a long anterior process, projecting to the anterior sensory tip, which closely interacts with the dendritic extension of a nearby cephalic neuron (blue). Arrows indicate the dorsal (red arrow) and ventral (orange arrow) side of the worm. B: a confocal image showing green fluorescent protein expression driven by the hlh-17 promoter to visualize the four CEPsh glial cells (worm strain VPR839). The anterior (head) of a juvenile (larval stage 4) worm is shown; the worm is turned ~45 degrees from “upright” such that all four CEP sheath cells are visible. The sheath portion of the cells that form a tube around the dendritic endings of the CEP neurons are seen at the left of the image. The dorsal (red arrow) and ventral (orange arrow) CEPsh cell bodies are seen. The thin sheetlike extensions that surround and invade the nerve ring are seen in the rightmost part of the image. Scale bar, 20 µm. [From Stout et al. (1683).]

B. Homeostatic Glia in Annelida

The medicinal leech Hirudo medicinalis was employed for studying glial physiology in pioneering experiments of Stephen Kuffler and David Potter in the mid 1960s (920). The nervous system of the leech is composed of the anterior and posterior brains and the chain of 21 ganglia that lie in between. The anterior brain comprises six ganglia fused into two neuronal masses, while seven fused ganglia in the aft form the posterior brain. Somatic ganglia innervate corresponding segments of the leech body (342, 403). Every ganglion is composed of 400 neurons (with exception of the 5th and 6th ganglia innervating the reproductive system, which have ~700 neurons) and 10 glial cells. Each ganglion contains two connective glial cells, which ensheath axons, six packet cells covering neuronal cell bodies, and two giant glial cells (403). All three types of glia are connected through gap junctions formed by innexins [of which leech expresses 21 types (819), with the Hm-inx2 type seemingly being specific for glia (463)], creating a panglial syncytium (1005). The nervous system of the leech additionally contains some amount of microglial cells that proliferate in response to lesions (954). Both packet glia and giant glial cells perform homeostatic functions and hence resemble astrocytes. The packet glial cells contribute to the regulation of extracellular K⁺, especially at high extracellular K⁺ concentrations (1214, 1547). Somata of giant glial cells are 80−100 μm in diameter and are localized in the center of the ganglion; extensively branched processes of these cells extend (for 300−350 μm) through the entire neuropil and contact neuronal dendrites (1172). Processes of giant glial cells partition neuropil, thus segregating neuronal ensembles into functional domains; sometimes glial processes invaginate into neuronal somata creating a structure known as “trophospongium“ (717). Giant glial cells are characterized by large K⁺ permeability and relatively hyperpolarized resting potential (approimately −75 mV). They express multiple types of neurotransmitter receptors including ionotropic glutamate, acetylcholine, and serotonin receptors as well as metabotropic receptors to glutamate, serotonin, myomodulin, and possibly P_2Y-like purinoceptors and A₁-like adenosine receptors (403, 1166). Giant glial cells participate in homeostatic responses, such as regulation of pH involving several plasmalemmal Na⁺-HCO₃⁻ cotransporter, Na⁺-H⁺ and Cl⁻-HCO₃⁻ exchangers (399, 404, 405), as well as in regulation of neurotransmitters turnover through Na⁺-dependent glutamate and Na⁺-dependent choline transporters (407, 705, 1898). Giant glial cells respond to neuronal activity and to evoked behaviors by changes in membrane potential (400) as well as by generation of cytosolic Ca²⁺ signals that occur in both somata and processes, often in compartmentalised manner (1004). In contrast to mammalian glia, the main source of Ca²⁺ signal generation in leech glia is through opening of plasmalemmal Ca²⁺ channels. Termination of Ca²⁺ signal termination is mediated by plasmalemmal Ca²⁺ pump and Na/Ca²⁺ exchanger; the intracellular Ca²⁺ stores, although present, seem to play a minor role (408, 1004).

C. Astroglia-like Cells in Arthropods: The Case of Drosophila

The Arthropods (of which insects and crustaceans are by far the most numerous representatives) have well-defined CNS. The brain of Arthropods is divided into protocerebrum (mainly receiving visual information), deuterocerebrum (receiving sensory input), and tritocerebrum (which acts as an integrating center). The neuroglia is elaborated and highly diversified; neuroglial cells account for ~10% of the total number of cells (~90,000 cells) in Drosophila CNS. Neuroglia of Drosophila are classified (for details and further reading, see Refs. 27, 467, 528, 652, 1325) into the following classes: 1) wrapping glia of the peripheral nervous system; 2) surface glia, which makes the brain-hemolymph barrier. It is subdivided into perineural glia (relatively small cells lying on the ganglionic surface) and subperineural or basal glia (represented by large sheetlike cells connected with septate junction that forms the actual barrier). 3) Cortex glia contact neuronal cells somata in the CNS; each glial cell establishes contacts with many neurons. 4) Neuropil glia are located in the neuropil and cover axons and synapses. The neuropil glia are further subdivided into ensheathing or fibrous and astrocyte-like glia, which forms a perisynaptic glial cover. 5) Finally, tract glial cells cover axonal tracts connecting different neuropils (1751). Within these classes glial cells could be further subdivided into multiple subtypes with distinct morphology and function (282, 467, 1751); for example, the glia of the lamina (neuropil) of the optic lobe is classified into fenestrated glia, pseudocartridge glia, distal and proximal satellite glia, epithelial glia, and marginal glia. In the deep optic lobe glial cells are represented by giant optic chiasm glia, small outer optic chiasm glia, medulla satellite glia, and medulla neuropil glia; similarly, specific glial types were described in olfactory system and in antennal lobes. Recent cell mapping using glia-specific GAL4 drivers allowed for precise numerical calculation of various cell types in the Drosophila CNS. The perineural glia accounted for ~17%, subperineural for 2%, cortex glia for 20%, astrocyte-like glia for 34%, and ensheathing glia for 27% of total glial cells (903).

A similar degree of complexity is also found in other insects, for example, in housefly Musca domestica or in tobacco hornworm Manduca sexta in which some other types of glial cells, such as the “chandelier glia,” have been described (288, 1528). Drosophila glial cells that belong to classes ii to iv (i.e., surface glia, cortex glia, and neuropil glia) serve functions characteristic for astrocytes in mammalian systems; however, the degree of specialization seems greater with distinct cell types responsible for a distinct set of functions. The coverage of neuronal bodies is provided by highly ramified cortex glia that do not have much semblance to mammalian astrocytes [although they originate from bipolar cells somewhat similar to radial glia (462)], whereas the brain-hemolymph barrier is provided by similarly distinct surface glial cells. The barrier is sealed by septate junctions composed from several proteins (567, 1578) of which the most important are gliotactin, neuroglian, and α and β subunits of Na⁺-K⁺-ATPase (also known as Nervana 2). The glial barrier in the insects is functionally analogous to the endothelial blood-brain barrier in vertebrates. It selectively limits the exchange between hemolymph and the CNS, for example, protecting the brain from substantial fluctuations in K⁺ that occur after feeding (467). Glial cells in the insect CNS also plaster tracheoles, which deliver oxygen to the nerve tissue (1367). The neuropil glia similarly demonstrate specialization: the flat ensheathing glial cells line the borders of the neuropil, thus isolating it from the surrounding cortex. In addition, ensheathing glia delineate glomeruli in the antennae lobe, whereas the astroglia-like cells extend complex arborizations into neuropil and provide for synaptic coverage (441). The neuropil glia are connected through gap junctions formed by innexins 2 and 3 and optic ganglion reduced (ogre) protein (1662).

Glial cells in insects provide for a multitude of homeostatic functions, such as regulation of ionic balance in the CNS fluids and regulating clearance, recycling, and metabolism of neurotransmitters. These functions are fulfilled by a complement of pumps, transporters, and specific enzymes. In the retina, glial cells are the key element for recycling the principal neurotransmitter histamine. After being released by activated photoreceptors, histamine is (partially) accumulated by glial cells closely covering synaptic contacts (1099). In the glial cell histamine is converted into β-alanyl-histamine or carcinine by ebony (N-β-alanyl-biogenic amine synthetase)-catalyzed reaction (216), the ebony activity is further associated with black protein in a yet unknown way (1966). Subsequently, carcinin is transported back to photoreceptors through the specific solute carrier transporter carT from SLC22A family (1669). After entering photoreceptors, carcinin is hydrolyzed to histamine with catalytic help of the so-called Tan (acyltransferase) protein (1760), thus concluding the histamine-carcinine shuttle. Mutations in the components of this shuttle disrupt fly vision (311, 1966). The ebony amine synthetase expressed in Drosophila glia seemingly also controls circadian clock (1696). Genetic modifications of Drosophila glia that interferes with vesicle trafficking (by specific expression of temperature-sensitive dynamin) and ionic transport (by glia-specific expression of bacterial Na⁺ channel) affects circadian rhythmicity (1213).

The neuropil glia in the insects express excitatory amino acid transporters dEAAT1 and dEAAT2 (847, 1586) and glutamine synthetase (392) responsible for transport and recycling of glutamate. Glutamate transporters are preferentially localized to glial perisynaptic processes (1472), and their loss instigates neurotoxicity leading to neuropil degeneration and decreased lifespan (985). Glutamate homeostasis mediated by glial transporters contributes to insect behaviors, in particular to sexual behavior and courtship. This seems to be associated with glial-specific cystine-glutamate transporter, which regulates ambient glutamate concentration and hence the strength of glutamatergic transmission. Loss of function mutation of these transporters (known as genderblind, gb) results in homosexual courtship (610). Insect glial cells have been also found to take up γ-aminobutyric acid (GABA), thus regulating GABA levels in the neuropil (1259).

Glia in the insects are imperative for trophic and metabolic support of neurons and for neuroprotection; loss of glial cells by targeted ablation results in massive neuronal death (212). Similarly, neurodegeneration and neuronal loss occur in several mutants with malfunctional glia (such as drop dead, swiss cheese, and repo) (251, 905, 1900). In the retina of the honeybee, glial cells supply photoreceptors with alanine, which subsequently is converted to pyruvate for use in the Krebs cycle (1761). Finally, glial cells in insects contribute to CNS defense by mounting reactive gliosis (1037) and phagocytosis both in development and in the adulthood (441, 928).

D. Astrocytes in Chordata and Low Vertebrates

In the CNS of early Deuterostomia (which include Chordata, Hemichordata, and Echinodermata), the parenchymal glia common to invertebrates are substituted by radial glial cells. The radial glia, although being present at some developmental stages in the insects (see above) and being identified in some protostomes (in Annelida and Scalidophora; A. Reichenbach, personal communication), are mainly associated with vertebrates (1424), with their neuroepithelial, layered CNS contrasting the fused ganglia brain in the majority of invertebrates. In the Echinodermata (represented by sea urchin, star fishes, or sea cucumber), which are placed at the very base of Chordata, radial glial cells are the only glia in the CNS (the latter appears in the form of a circumoral nerve ring connected to radial nerve cords). These radial glial cells produce and secrete the Reissner's substance (1076, 1821), mainly comprising the glycoprotein known as SCO-spondin, which probably acts as cell adhesion modulator (580). The Reissner's substance is found in radial glia throughout Chordata from cephalochordates to Homo sapiens. Glia of Echinodermata have a characteristic radial morphology with elongated shape, long processes spanning the whole thickness of the neural parenchyma, perpendicular orientation to the surface of the neuroepithelium, and high level of expression of intermediate filaments in the cytoplasm (1075, 1076). Functional properties of these glial cells remain unknown.

Similarly, in the CNS of many early vertebrates, radial glia represent the main type of neuroglia, with almost complete absence of parenchymal glial cells. This in particular is characteristic for the brains with thin parenchyma. In elasmobranchii (chondrichthian fish, such as sharks and rays), the brains are subclassified according to their gross morphology. In type I, or “laminar” brain, neurons mainly concentrate in the periventricular zone; the brain walls are thin and ventricles are large. In the type II or “elaborate” brains, neurons migrate away from the periventricular zone and form nuclei; these brains have thicker walls and the ventricles are relatively small (62, 266). In the type I brains, radial glia dominate, whereas the elaborated brains of type II contain numerous well-developed astrocyte-like parenchymal glia (62). Emergence of parenchymal astrocytes in elaborate brains probably reflects an increased homeostatic challenge of the enlarged nervous tissue that cannot be met by radial glia. This constrains homeostatic capabilities of the radial glia and hence prompts an increase in numbers and complexity of parenchymal astrocytes (1452). Another possible factor could be associated with an increase in complexity of vascularization, which requires perivascular glial support that cannot be provided by radial glia (1879). In low vertebrates, parenchymal glia are the main component of the blood-brain barrier; in sharks for example, astrocytes completely enwrap vessels making them endocellular capillaries (254, 255). The radial glia seem to be the predominant glial cell type in all anamniotes (that is in fishes and amphibians). In zebra fish (the teleost), radial glial cells express GFAP and possess glutamine synthetase (indicating their possible role in glutamate homeostasis and metabolism) and aquaporin-4 (indicating their role in water homeostasis) (611, 1025). In contrast to elasmobranchii, the blood-brain barrier in teleosts is made by vascular endothelial cells, the arrangement common to most of the vertebrates (254). Similarly to higher vertebrates in the zebra fish and in amphibians, radial glia endfeet enwrap blood vessels (1025). Radial glial cells in zebra fish do not mount reactive response to brain injury; instead, they increase neurogenesis thus closing the lesions without scar formation (113).

The radial glia dominate the brains of reptiles (8, 814), although parenchymal cells bearing astrocytic morphology have been identified in the CNS of some snakes and lizards (1148, 1271). Parenchymal astroglia-like cells frequently appear in the brains of Cayman crocodiles where they intermingle with radial glia, the latter still being the predominant type (815). In adult birds where the astrocytes proper became a predominant cell type, the radial glia remains, although radial glial presence diminishes in postnatal development (30, 816). In mammals, the radial glia disappear almost completely with the exception of radial-like cells in retina (Müller glia), in cerebellum (Bergmann glia), in hypothalamus (tanycytes), and in subventricular zone.

E. Recapitulation

The astroglia (i.e., the parenchymal homeostatic supportive cells of the CNS) evolved several times in independent manner into many distinct phenotypes. This very much agrees with our view on astrocytes as highly opportunistic supportive cells that tailor their form and function to match the demands of progressively changing nervous tissue. In this context, the CNS evolved through division of functions between cell types: the neurons become mostly responsible for rapid propagation of signals associated with action potential and chemical synapses, whereas neuroglia assumed the responsibility for homeostasis and defense (1813).

Although astroglial cells (or astroglia-like cells at the earlier evolutionary stages) generally perform similar functions, their appearance differs substantially at different phylogenetic stages. The most primitive astrocytes operational in C. elegans combine the features of peripheral and central glia by contributing to formation of sensillas with their centrifugal processes and covering neurons by their centripetal ones. Further evolution resulted in multiplying forms and diversification. In annelids, the astroglial functions are divided between packet cells, which support neuronal cell bodies, and highly idiosyncratic giant glial cells that control homeostasis of the neuropil. Even more diversification was achieved in insects in which specific cell types solely responsible for barrier function, for supporting neurons, for isolating neuropil, and for homeostatic control over neuropil have emerged. The evolution of astroglia erupted anew in Chordata with the ascendance of radial glia, which became universal neural precursors, provided scaffold for neuronal migration, and in anamniotes assumed general control over CNS homeostasis. Increase in the size and thickness of the brain instigated an appearance of true astrocytes that combined all supportive functions in one cell type, which however is remarkably plastic and can modify the phenotype according to tissue demands. At the same time, many key homeostatic molecules such as ion pumps and neurotransmitter transporters are conserved through phylogeny. Despite the independent and multiple evolution, the main trend in astroglia development remains the same, and it is represented by gradual yet unabating shift of homeostatic and defensive functions from neurons to neuroglia. Together, increased intricacy of astroglial process-synaptic complexes contributed to a progressive increase in the computation power of the mammalian brain.

IV. ASTROGLIAL ORIGINS AND DEVELOPMENT

A. Astrogliogenesis in Mammals

Astrocytes are classical neural cells which (as neurons and oligodendroglial cells) originate from the neuroepithelium-derived radial glia, that are the universal neural progenitor (906). Development of astrocytes, however, differs from that for neurons. The latter originate from dedicated progenitor cells which derive from radial glial through asymmetric division. Astrocytes, in contrast, can develop from glial intermediate progenitors, through transformation of radial glia, from proliferation of differentiated astrocytes and even from NG2 glial cells (1140, 1564). The first stage of development of the nervous system is solely neuronogenic; the asymmetric division of radial glia produces either neuron-restricted intermediate progenitors or immature neurons. The neuronogenic period is promoted by pro-neuronogenic factors such as neurogenin 1, which simultaneously suppresses gliogenic route. The gliogenic switch in rodents occurs around embryonic day 12.5 in the spinal cord and around E16–18 in the cortex; the onset is controlled by the transcription factors nuclear factor I-A (NFIA) (410) and Sox 9 that acts in association with NFIA (827, 1675), by Notch signaling, or by the transcriptional repressor N-CoR. Many of these pathways are converging on JAK/STAT signaling cascade (527, 665). Activated STAT3 together with the co-activator complex p300/CBP bind to promoter of astroglial genes, thus triggering expression of astrocyte-specific genes, such as for example the gene of GFAP (830, 1779). Inhibition of this cascade is critical for suppression of the gliogenesis, and this is the mechanism of action of neurogenin 1, which binds to the p300/CBP complex and thus stops gliogenesis; other proneuronal factors such as Ngn2, NeuroD1, and Mash1 may also act in the same way. This inhibition is an important part of neuronal developmental sequence, allowing massive wave of neuronogenesis to populate neuronal layers and create early connectome. The gliogenic switch is regulated by several factors, most notably by cytokines of IL-6 family, which include ciliary neurotrophic factor (CNTF), leukemia inhibitor factor (LIF), and cardiotrophin-1 (CT-1) that all activate JAK/STAT cascade through signal-transducing co-receptors LIFRβ and gp130 (1191). In the in vivo context the relevant agonist of LIFRβ and gp130 is cardioytropin-1, which is secreted by newborn neurons. This arguably represents the timing mechanism that coordinates neurono- and gliogenesis (101, 1634). It seems that subpopulations of neuronogenic and gliogenic radial glia operate together, although it is impossible to exclude the third population of multipurpose radial glial cells (114). Asymmetric division of radial glia (which occurs in the subventricular zone) produces intermediate glial progenitor cells, which further develop to immature proliferative astrocytes; these latter migrate through cortical layers while continuing to proliferate. A distinct class of neural progenitors (which produce neurons, oligodendrocytes, and astrocytes) has been also identified in the layer 1 on embryonic and neonatal cortex (also known as marginal zone); these progenitors apparently colonize ventral and dorsal telencephalic ventricular zones (231, 364). The marginal zone progenitors may give rise to astrocytes of superficial layers (1–4), which could be one of the explanations for morphological differences between superficial and deep astroglia.

The embryonic astrogliogenesis, however, is responsible for only a fraction of astrocytes populating the CNS: the major peak of glial generation occurs during the second and third postnatal weeks (in rodents) when the number of nonneuronal (i.e., mostly glial) cells increases from ~4 million to over 140 million (96). In the cat, gliogenesis seems to be more prolonged and glia-to-neuron ratio increases from ~0.8 in young kittens (weighting 0.5 kg and being ~60 days old) to ~1.48 in mature animals of 3 kg weight (236). We may therefore assume that in humans the glial development similarly spans well into the adulthood.

The main part of postnatal astrogenesis, which accounts for ~50% of all astrocytes, is associated with symmetric division of differentiated astrocytes that occurs throughout the CNS (561). Incidentally this mechanism was proposed by Cajal a century ago, when he found pairs of astrocytes joined by their soma; these pairs Cajal defined as twin astrocytes or “astrocitos gemelos” (1428). Another source of astrocytes is associated with direct transformation of radial glia, which, at around birth, lose their apical processes and redesign themselves into protoplasmic astrocytes (395, 1566, 1830); this metamorphosis may account for 10–15% of mature astroglia (1140, 1511). Conceivably, astrocytes directly transformed from radial glial cells become stem cells of neurogenic niches in adulthood (1108). Thus astrogliogenesis proceeds through two distinct phases: in the early embryogenesis it occurs through intermediate glial progenitors that originate, through asymmetric division, from radial glia. These precursors migrate through the neural tissue, proliferate, and transform into astrocytes. The postnatal wave of astrogliogenesis (which is substantially larger) occurs through direct transformation of radial glia and symmetric division of differentiated astrocytes. Another possible source of astroglia may be associated with NG2 glial cells (1227), which can generate protoplasmic astrocytes that stay strictly in the ventral forebrain and do not migrate to other areas; furthermore, this astrogenic route seems to be temporally limited (748, 1959, 1960).

Morphological and functional heterogeneity of astrocytes, at least in part, is associated with their place of birth. It seems that diverse astroglial phenotypes are somehow linked to the site of their origin in the ventricular zone and are associated with the subfamily of progenitors, which were born from the spatially restricted set of radial glia and which migrated together along the processes of their parental cells (781, 1140, 1762). This regionalization of astrocytes proceeds together with regionalization of neurons, thus creating columnar structures that contain both types of cells (547, 1045). Recently developed optogenetic cell fate tracing technique (551), which uses GFAP promoter controlled expression of six fluorescent proteins, showed that astroglial clones disperse radially and remain within the confines of a single column (very much like neurons originated from the same radial glia; see Refs. 547, 1045). Thus neurons and astrocytes that are born together do grow together and acquire regional specificity. Furthermore, it seems that regional maturation of neurons and astrocytes is regulated by similar molecular factors. In the spinal cord, both cell types are under control of the bHLH factor scl/tal1 that regulates development of V2b interneurons and ventral astrocytes (410, 1176). At the same time astroglial phenotypes are shaped by their neuronal neighbors. In the cerebellum, for example, this involves the sonic hedgehog system that differentially regulates gene expression in Bergmann glia and velate astrocytes (499). In the adult brain, astroglial proliferation was found in many regions including cerebral cortex, corpus callosum, striatum, hypothalamus, and septum, although the rate of division is very low ranging between 0.05 and 0.45% (560).

B. Recapitulation

Astroglial development is fundamentally different from that of neurons: at birth neurons are generally postmitotic, whereas astrocytes retain proliferative capacity. Moreover, epigenetic regulation of astroglia seems to be similarly different: the glial methylome remains more fetal and hence labile during adulthood (989). In summary, a combination of genetic, developmental, and epigenetic factors make astrocytes truly opportunistic cells and define life-long adaptive plasticity of astroglia.

V. IDENTIFICATION AND MORPHOLOGY OF ASTROGLIA

A. Identification

Visualization and identification of astrocytes, especially in the in situ preparations and in the in vivo brain, are tasks far from trivial. The difficulties are in remarkable morphological heterogeneity and in the absence of a universal marker that may label all cells of astroglial lineage. Existing techniques include classical histological staining and immunocytochemistry (performed on fixed tissues), genetically controlled expression of astroglia-specific fluorescent markers, incubation with fluorescent probes with some preferential glial affinity, or intraglial injection of fluorescent dyes. Classical histological techniques (95) are represented by 1) Golgi staining (the silver nitrate impregnation technique) that exists in several modifications and in skillful hands may deliver detailed images of astrocytes with primary and secondary and even fine processes, when used in combination with electron microscopy (32, 1251, 1270); 2) a sublimated gold-chloride staining of Cajal, which labels astroglial filaments and endfeet (550, 1193); and 3) Hortega's silver impregnation method, which, with some modifications, has been occasionally used to label astrocytes for light and electron microscopy (885).

1. Genetic profiling of astrocytes

The mRNA expression levels of genes characterize the transcriptional state of the cell, thus providing insight into its function, activity, and developmental state as well as (in the case of disease) degree of pathological remodeling. The transcriptomic profile of human and mouse brain astrocytes have been characterized by microarray and RNA sequencing technologies in combination with cell sorting techniques such as fluorescence-activated cell sorting (FACS) or immunopanning of immuno- or transgenically labeled astrocytes. These studies have revealed novel functions of astrocytes, identified new cell specific markers, and characterized the molecular profile of reactive astrocyte in disease models, as well as highlighted differences and similarities between mouse and human astrocytes and between developing and mature astroglia.

a) major findings from transcriptional studies.

The first transcriptional study of astroglial cells isolated in vivo from adult mice cortex used FACS of astrocytes expressing green fluorescent protein (GFP) under control of either GFAP or glutamate transporter 1 (GLT-1) promoters (1010). It has been found that the majority of astrocyte-enriched genes (~34%) are involved in cellular metabolism and that genes encoding enzymes of the tricarboxylic acid cycle were expressed to a higher degree in astrocytes compared with neurons. Moreover, subsequent functional assays indicated that astrocytes exhibit prominent oxidative metabolism in the intact adult brain and hence can significantly contribute to functional brain imaging, including blood-oxygen level-dependent (BOLD) MRI signals (1010).

The first study to characterize the transcriptional profile of astrocytes isolated from the developing postnatal mouse cortex was performed by Ben A. Barres group (275). Cell suspensions from the cortices of 1- to 8-day-old and 17- to 30-day-old transgenic mice expressing EGFP under control of S100B promoter were first depleted of oligodendrocytes by immunopanning, and subsequently S100B/EGFP positive astrocytes were isolated by FACS. A comparison between immature (days 7–8) and mature (days 17–30) astrocytes showed that the genes highly expressed in the immature astrocytes are involved in general cell proliferation and development (e.g., cell cycle genes) and are not astrocyte specific. Conversely, most genes enriched in mature astrocytes are astroglia specific (the same result was found for oligodendrocytes). Astrocyte-enriched genes, which increase their expression during maturation, included genes encoding secreted proteins such as ApoE, ApoJ/clusterin, Pla2g7, Sparc, Sparcl1, and Mfge8 and genes implicated in psychiatric diseases such as Npas3, Mcl1, Lgi1/4, and Gpr56. Because expression of phagocytic genes in astrocytes increased with the brain maturation, the authors suggested that an important role for astrocytes in the mature brain is to clear apoptotic cells and amyloid deposits and to eliminate synapses. Consequently, aberrant regulation of synaptic formation by astrocytes during development and in the mature brain may contribute to neurodevelopmental, psychiatric, and neurological diseases (275). That astrocytes are capable of synapse elimination in vitro and in vivo was confirmed by the same group in a later study (334), and it has been hypothesized that disruption of the ability of astrocytes and microglia to eliminate and regulate synapse formation contributes to the pathogenesis of psychiatric disorders (335).

Reactive astrogliosis is a common response of astrocytes to brain injury and disease. The transcriptional profile of reactive astroglia was studied on FACS-isolated astrocytes from adult Aldh1l1-EGFP transgenic mice following brain injury (either ischemic stroke by MCAO causing cell death, or LPS injection causing neuroinflammation) (1933). Reactive astrocytes were found to markedly alter their transcriptional profile. Major changes have been detected in the genes encoding extracellular matrix genes, reflecting the ability of reactive astrocytes to modify the extracellular matrix when forming glial scar. Similarly, genes encoding intermediate filament proteins (GFAP, vimentin, and nestin) were highly upregulated in reactive astrocytes, reflecting morphological changes that occur upon activation. Other groups of genes, expression of which increased in the reactive astrocytes, encoded cytokines, proteins of the antigen presentation, and complement pathways, suggesting a role for astrocytes in communicating with the immune system upon injury. Even though reactive astrocytes induced by the two models of injury had similarities in their expression profiles, there was still a large number of genes that differed, highlighting that the response of astrocytes to injury is not uniform; instead, it is more complex and depends on pathology. Finally, this study found that the transcriptional profile of reactive astrocytes is highly similar to the transcriptional profile of cultured neonatal astrocytes produced by the McCarthy-de-Vellis method (1933). This finding was important since it stresses the significance of working with in vivo models instead of astrocytes in the dish when studying their physiological functions.

Morphological studies have shown that human astrocytes are much larger and more complex than mouse astrocytes (1248, 1249), and that mice transplanted with human astrocyte progenitor cells perform better in learning and memory tests (637). The first RNA-sequencing study of the transcriptional differences between cortical human and mouse astrocytes was published recently (1946). The purified fetal and mature/adult astrocyte populations from mice and human were obtained by immunopanning with an antibody against the astrocyte-enriched HepaCam after depletion of other cell populations. Comparing the fetal and mature human astrocytes, the authors found that fetal astrocytes had higher expression of proliferative and cell-cycle genes, which is similar to the astrocytes of the developing mice discussed above. With maturation, human astrocytes showed increased expression of genes involved in nerve impulse-transmission, cell-cell signaling, fatty acid metabolism, cell adhesion, and ion homeostasis. When human and mice astrocytes were compared, it appeared that 52% of mouse astrocyte-associated genes were enriched in human astrocytes, whereas only 30% of the human astrocyte-associated genes were enriched in mouse astrocytes. The same differences were found for the other cell types analyzed (1946). This suggests that in general human cells are more specialized than cells in mice, which makes sense evolutionarily, and that the transcriptional differences between human and mice astrocytes are similar to the species differences for other cell types. Unfortunately, the authors do not have a more detailed analysis and discussion of the differences between human and mouse astrocytes; however, their data are available online (http://web.stanford.edu/group/barres_lab/brainseq2/brainseq2.html) being thus available for all interested to do their own analysis.

b) molecular heterogeneity of astrocytes.

To date, no genomic studies have characterized the expected molecular heterogeneity of astrocytes because most studies were performed on cell populations of astrocytes identified by a marker gene or protein. Despite several comparisons of transcriptomes from astrocyte populations expressing specific cells markers, e.g., GFAP versus GLT1 positive astrocytes (1010), ALDH1L1 versus GLT1 positive astrocytes (1916), and ALDH1L1 versus HEPACAM positive astrocytes (1946), no major differences between these cell populations have been found, indicating that the markers are coexpressed in the majority of astrocytes. One way to identify the molecular heterogeneity of astrocytes would be to use unbiased single-cell transcriptomics. A few studies have collected transcriptome information of single astrocytes (384, 1732, 1938). However, these studies do not include further description of astrocyte heterogeneity possibly because this information could not be extracted from the data due to a relatively small sample size [only Zeisel et al. (1938) were able to identify two astrocyte subclasses, GFAP (type 1) vs. MFGE8 (type 2) expressing astrocytes].

2. Immunocytochemistry

The universal marker that may stain and reveal all astrocytes in the CNS does not exist. Remarkable morphological heterogeneity of astrocytes coincides with a substantial diversity in expression of different molecules and hence antibodies against them label subpopulations of astroglial cells with substantial regional differences (TABLE 1).

Table 1.

Markers of astrocytes

Molecule/Antigen	Detection Agent/Technique	Properties and Functional Relevance	Reference Nos.
Glial fibrillary acidic protein, GFAP	Monoclonal and polyclonal antibodies	Intermediate filament protein, expressed in many cells outside the nervous system; in the CNS expressed in a subpopulation of astrocytes with substantial region variability. Generally, GFAP expression is upregulated in reactive astroglia.	477, 715, 1355, 1880
Vimentin	Monoclonal and polyclonal antibodies	Intermediate filament protein; expressed in immature astrocytes, in subpopulations of protoplasmic and fibrous astrocytes, in Bergmann glia, and in tanycytes. Vimentin expression is upregulated in reactive astrocytes.	394, 597, 1350, 1351, 1568
S100B protein	Monoclonal antibodies	Ca2+-binding proteins, which act as Ca2+ buffers as well as Ca2+ sensors. Antibodies against S100B stain more astrocytes than GFAP in the grey as well as in the white matter.	444, 1251, 1549
Glutamate transporters: EAAT-1 (GLAST), EAAT-2 (GLT-1)	Monoclonal antibodies	Astroglia-specific glutamate transporters; show regional variability: EAAT1 is predominantly expresed in cerebellum; in other regfions EAAT2 is the main transporter type.	110, 807, 1570, 1606, 1882
Glutamine synthetase	Monoclonal and polyclonal antibodies	Astroglia-specific enzyme converting ammonia and glutamate into glutamine. Expressed in the majority of astrocytes. Immunostaining reveals full structure of the cell due to cytosolic localization of the enzyme.	49, 417, 1237, 1920
Aldehyde dehydrogenase 1 family, member L1 (ALDH1L1)	ALDH1L11-specific polyclonal antibody	ALDH1L1 is a key enzyme in folate metabolism contributing to nucleotide biosynthesis and cell division. Proposed as a specific astroglial markers with a reach substantially broader than GFAP. ALDH1L1 expression however changes with age, and it was also detected in a subpopulation of oligodendrocytes.	275, 1916
Connexins: Cx43, Cx30	Monoclonal and polyclonal antibodies	Both Cx43 and Cx30 are expressed exclusively in astrocytes; the Cx30 is expressed mostly in grey matter (being particularly concentrated in astroglial endfeet) and is absent in astrocytes from white matter.	413, 1189
Aquaporin: AQP4	Monoclonal antibodies	AQP4 in the CNS is expressed exclusively in astrocytes and ependymocytes. In healthy astrocytes, AQP4 is preferentially located in the endfeet and hence stains this structure.	1184
Transcriptional factor SOX9	Polyclonal antibodies	Specifically labels nuclei of astrocytes outside the neurogenic niches.	1701

a) GFAP.

GFAP was discovered in early 1970s (478, 1781); its exclusive expression in astrocytes in the CNS was soon noted (180, 1018) and has been well documented since (715). The GFAP, of which astrocytes express 10 different isoforms, belongs to an extended family of intermediate filaments and, together with vimentin, nestin and, occasionally, synemin, forms astroglial cytoskeleton (715, 1355). Genetic deletion of GFAP produces rather subtle physiological phenotypes; in GFAP^−/− mice, however, reactive astrogliosis is substantially impaired, whereas double deletion of GFAP and vimentin disrupts reactivity even more, which in turn exacerbates neuropathology (1351, 1880).

Staining with GFAP antibodies visualizes only a fraction of astrocytes with a substantial regional (and probably developmental) heterogeneity. This applies to mammals and birds, with some regions (striatum and tectum in mammals and neostriatum, paleostriatum augmentatum, and the superficial zone of tectum in birds) almost completely devoid of GFAP immunoreactivity (813). Astrocytes in cell cultures are almost invariably GFAP positive, whereas subpopulations of GFAP-labeled cells in situ and in vivo are substantially smaller (1842). The largest subpopulation of GFAP-positive astrocytes is present in juvenile hippocampus, with ~80% (or even more) of all cells being labeled with appropriate antibodies (264, 1251); similarly all Bergmann glia in cerebellum are GFAP immunoreactive (46, 1236). At the same time, the majority of astrocytes in other regions of healthy brain are not stained with GFAP antibodies (866, 1549, 1847). Morphology of GFAP-positive profiles is somewhat limited (FIGURE 5, A–C), because the immunolabeling of the cytoskeleton reveals only major processes, with finer parts of the cell remaining unstained (353). Thus GFAP reveals neither peripheral and perisynaptic processes (1451), nor endfeet plastering small (<8 μm) blood vessels (1622). Out of 47 cell-specific molecular signatures identified with single-cell mRNA sequencing of juvenile primary somato-sensory cortex and hippocampus of mice (1938), only 2 belonged to astrocytes. The gfap was being mainly limited to layer 1 astrocytes and glia limitans, and Mfge8 (encoding globule-EGF factor 8 protein or lactadherin) was expressed in the remaining parenchymal astrocytes, again pointing to the limited presence of GFAP in a substantial population of astroglial cells. Despite all these limitations, GFAP labeling is considered, by some, as the best for laser capture microdissection approach (1836, 1837).

FIGURE 5.

Visualization of rodent astrocytes with immunostaining against canonic markers GFAP, S100B, and glutamine synthetase. A–C: GFAP-stained astrocytes in entorhinal cortex (A), prefrontal cortex (B), and CA1 area of hippocampus (C). [A–C from Rodriguez et al. (1483). Reprinted with permission from Eureka Science Ltd.] D: astrocyte stained with antibody against S100B in the dentate gyrus of hippocampus. E: hippocampal astrcytes stained with anti-glutamine synthetase antibody. [D and E from Rodriguez-Arellano et al. (1480), with permission from Elsevier; and Rodriguez et al. (1486), with permission from Elsevier.] F: cortical tissue preparation with astrocytes labeled in green (EGFP expressed under EAAT2 promoter), astroglial nuclei labeled in pink were stained with antibodies against SOX9, and neurons (in red) were stained with antibodies against NeuroN. (From Sun and Nedergaard, unpublished observation.)

b) S100B protein.

The glycoprotein S100B is one of 24 S100 Ca²⁺-binding proteins, which are expressed only in vertebrates and act as Ca²⁺ buffers as well as transducers (Ca²⁺ sensors) for the intracellular Ca²⁺ signaling (444). In the CNS, S100B regulates various aspects of cell proliferation and differentiations, and it is known as an inhibitor of apoptosis (444, 622, 1436). There is some evidence that in astrocytes S100B may also contribute to shaping Ca²⁺ signals (1901). The S100B is also linked to regulation of the assembly of intermediate filaments by inhibiting GFAP polymerization in the presence of micromolar Ca²⁺ (172). Astrocytes produce and secrete (by a yet unidentified mechanism) S100B, which has (depending on concentration) either neurotrophic/neuroprotective or neurotoxic effects, stimulates astroglial proliferation and contributes (in higher concentrations) to astroglial reactivity and positively regulates microglial activation (10, 171, 956, 1788, 1824). There are also some indications that S100B acts as a regulator of synaptic plasticity and long-term potentiation (1229). Overall, S100B is engaged in intercellular signaling and may act as an extracellular messenger (445). In pathological conditions, expression of S100B substantially changes, and increased levels of this protein in serum and cerebrospinal fluids may have certain diagnostic relevance (444).

Because of high level of expression, S100B is universally used as a marker for astrocytes (FIGURE 5D), both in physiology and in pathology; astrogliotic response being associated with upregulation of S100B. In rodent hippocampi, S100B as a rule stains more astrocytes than GFAP; only ~80% of cells stained with S100B were also GFAP positive (1251). In the whole rat brain, the antibody against S100B stained approximately three times more astrocytes than GFAP; incidentally, the GFAP stained more cells in white versus grey matter, and very few GFAP-positive profiles were visualized in the cortex and in the brain stem (1549). The cell specificity of the S100B is, however, substantially less than that for GFAP. In the CNS it is expressed not only in astroglia but also in oligodendrocytes, in ependymal cells, in the choroid plexus epithelium, in vascular endothelial cells, in lymphocytes, and in some neurons (1664), in particular in neurons in the brain stem, cerebellum, forebrain, and the limbic system (1466).

c) glutamate transporters and glutamine synthetase.

Glutamate transporters and glutamine synthetase are key molecules behind glutamate turnover in the CNS (see sect. XIID). Astroglial glutamate transporters EAAT-1 (GLAST) and EAAT2 (GLT-1) are expressed almost exclusively in astrocytes (1570). The EAAT-1 is the most widespread, and respective antibodies label radial glia, fibrous and protoplasmic astrocytes, cerebellar Bergmann glia, retinal Müller glia, radial stem glia in the dentate gyrus and subventricular zone in both developing and adult CNS (110, 1606, 1882). The specific monoclonal antibody ACSA-1 against extracellular epitopes of EAAT-1 labeled most of protoplasmic and fibrous astrocytes as well as Bergmann and Müller glia (807). There are some issues with EAAT-1/2 specificity; the splice variant of EAAT-1 (which is rarely expressed in astrocytes) has been found in some neurons, in oligodendrocytes, ependymal cells, and epithelial cells of the plexus choroideus (1569). The EAAT2, in its turn, shows transient neuronal expression (including cerebral cortex and basal ganglia) during fetal development (1241).

Glutamine synthetase (GS) stains virtually all types of astrocytes (FIGURE 5E), including radial glia, Bergmann glia, retinal Müller glia, tanycytes, and ependymal cells; importantly, anti-GS antibodies label astrocytes in many regions with weak GFAP immunoreactivity (49). In the mouse entorhinal cortex, double stained with anti-GS and anti-GFAP antibodies, 78% of all labeled glial cells were solely GS positive, 12% GFAP positive, and only 10% were positive for both GS and GFAP (1920). Similarly, in the hippocampus the separate population of GS-positive astrocytes was identified; double staining showed that only 60% of these cells were positive for GFAP (1847). Glutamine synthetase is a cytosolic enzyme and hence staining with appropriate antibodies reveals the whole extent of the cytoplasm including fine perisynaptic processes (417, 1237). There are some indications that GS in astrocytes in vitro may be associated with vesicular structures (50). Several sporadic reports about GS expression in oligodendrocytes (281, 372, 1132) and in neurons (1476) remain debatable and unconfirmed. Arguably, GS may be considered as the most inclusive astrocyte marker.

d) other markers.

Several proteins, which are more or less exclusively expressed in astrocytes, have been identified and can be used as markers with varying success (TABLE 1). Vimentin (similarly to GFAP) is a member of the extended family of intermediate filaments present in mesenchymal cells. It is involved in many cellular functions and in particular in regulation of cell differentiation, adhesion, migration, regeneration stress, and cellular signaling (774). In the CNS, vimentin is primarily expressed in the astroglia, and particularly in the immature astrocytes. After birth, expression of vimentin steadily decreases, although it is still present at immunocytochemically detectable levels in protoplasmic and fibrous astrocytes in hippocampus and corpus callosum as well as in Bergmann glia and in tanycytes where it is coexpressed with GFAP (394, 1351). Vimentin also seems to be present in adult neural stem cells with astroglia-like phenotype in the neurogenic niches, and vimentin expression is upregulated in reactive astroglia (440, 597).

The water channel aquaporin 4 (AQP4) in the CNS is present in astrocytes and ependymocytes (534, 1217). In astrocytes, expression of AQP4 is highly polarized with the highest concentration in the endfeet (1184); as a result, labeling with AQP4 antibodies preferentially reveals endfeet structures. Astrocytes throughout the brain express connexins, with predominant expression of Cx43 and, to a lesser extent, of Cx30 (413, 1189). Staining the brain tissues with antibodies against both connexins shows punctate patterns; the Cx30 is expressed mostly in grey matter astrocytes (being particularly concentrated in their endfeet) and is absent in astrocytes from white matter (1189). The key enzyme of foliate metabolism, aldehyde dehydrogenase 1 family, member L1 (ALDH1L1) was found to be specifically expressed in astrocytes (1212) and was recently suggested as a specific antigenic marker. Polyclonal antibodies against ALDH1L1 stained more astrocytes than GFAP; at the cellular level ALDH1L1-staining revealed soma and fine processes (275). However, later analysis showed that ALDH1L1 expression changes with age, and ALDH1L1 is also expressed in a subpopulation of oligodendrocytes (1916). Furthermore, ALDH1L1 was reported to label mainly cortical astroglia with rather little staining of white matter astrocytes (1837). Another enzyme, the brain-specific form of fructose-1,6-bisphosphate aldolase, known also as aldolase C, was found to be expressed preferentially in astrocytes, although it was also detected in Purkinje neurons (1841). Astrocytes in mouse and human brain are enriched with transcription factor SOX9. Immunostaining with specific antibodies further identified an overlap between SOX9 nuclear staining and EAAT2 immunoreactivity, indicating specific astroglial labeling (1701). Antibodies against SOX9 stain the nucleus and hence do not reveal full astroglial profiles (FIGURE 5F), although they were successfully used for FACS isolation of astrocytes and for isotopic fractionation (1701).

Interlaminar astrocytes as well as fibrous astrocytes in the human brain are readily labeled with antibodies against CD44, a receptor for extracellular matrix molecules (179, 1655). Protoplasmic astrocytes are, as a rule, CD44 negative, although they may acquire this protein at later ages and in pathology (1655).

3. Injection of fluorescent dyes

Astrocytes can be visualized by fluorescent dyes, which can be injected (by microelectrodes) or perfused into (through the patch pipette) cells of interest (FIGURE 6A). For such labeling membrane impermeable probes, such as Lucifer yellow, Alexa dyes or biocytin are usually employed (264, 1251, 1839). Alternatively, astrocytes can be loaded with fluorescent probes by diolistic labeling in which Gene Gun technology is used to deliver gold or tungsten particles coated with lipophilic dyes into cells in slices (FIGURE 6B). Combining probes with different spectral properties (for example, red DiI and green DiD) allows selective visualization of adjacent cells (1248).

FIGURE 6.

Protoplasmic astrocytes. A: cortical protoplasmic astrocytes filled/injected with fluorescent dye. [Image courtesy of Prof. Milos Pekny and Dr. Ulrika Wilhelmsson (University of Göteborg).] B: protoplasmic astrocytes diolistically labeled with spectrally distinct probes. C: the EGFP-expressing cortical protoplasmic astrocytes with endfeet plastering the blood vessel.

4. Gliophilic fluorescent probes

The ability of astrocytes to preferentially accumulate fluorescent Ca²⁺ probes in acetoxymethyl (AM) form was recognized in early experiments on brain slices (883, 1599) and was often utilized to monitor Ca²⁺ signals specifically in astroglia (158, 1003). Subsequently, high propensity of astrocytes to accumulate AM calcium probes was noted during in vivo brain imaging when bolus loading was implemented to load neural cells with the indicator (FIGURE 7). As a rule, intracellular concentration of the Ca²⁺ indicator was much (~4−5 times) higher in astrocytes when compared with their neighboring neurons (701, 1680). Specific glial accumulation of Ca²⁺ probes is possibly associated with low glial expression of ABC cassette transporters that mediate extrusion of these dyes (1061). Of note, the bulk loading with Ca²⁺ indicators stains mainly astroglial soma, and hence it does not allow dynamic Ca²⁺ recordings from peripheral processes (1445).

FIGURE 7.

Two photon imaging of astrocytes in vivo. A: experimental setup. Exposed somatosensory cortex was loaded with the specific astrocyte marker sulforhodamine 101 (SR101) and Ca²⁺ indicator fluo 4-AM. Coverslip and 1% agarose were mounted on top of cranial window to minimize brain pulsation. Recording electrode was loaded with Texas red-dextran (red) and inserted into cortical layer 2 (100–150 μm below pial surface). B: example images showing cortical layer 2 astrocytes double labeled with SR101 and Fluo 4-AM. Only SR101-positive astrocytes also labeled with Fluo 4-AM (white arrowhead). Neurons appeared as dark round shape area (red arrowhead). Scale bar, 30 μm. [A and B from Tian et al. (1750).] C: overview side projection of an SR101-stained area (revealing astrocytes) in mouse neocortex ~30 min after dye application. The image is a maximum-intensity side-projection from a stack of fluorescence images taken through cranial window on an anesthetized mouse. [C from Nimmerjahn et al. (1224). Reprinted by permission from Macmillan Publishers Inc.] D: cortical astrocytes loaded with SR101 and imaged (using two photon confocal system) through the cranial window on an anesthetized mouse. (Image kindly provided by Dr. Hajime Hirose, RIKEN, Japan.)

Another gliophilic fluorescent probe that received much popularity in the imaging experiments is a cationic dye sulforhodamine 101 and its analogs sulforhodamine B or G (1224). Sulforhodamine is selectively taken up by astrocytes and, because of cytoplasmic localization, it reveals detailed cellular structure (FIGURE 7). The sulforhodamines readily penetrate blood-brain barrier (1803) and hence astroglial staining could be achieved by intravenous injections of sulforhodamine B (20 mg/kg). Astrocytes become fluorescent in ~40 min after injection, and the staining persists for up to 5 h (54). The accumulation of sulforhodamine 101 into astrocytes seems to be mediated by organic anion transporters, which are differentially expressed in astrocytes from different brain regions. As a result, sulforhodamine 101 readily stains hippocampal astrocytes but does not accumulate in astrocytes in the ventrolateral medulla (1572). Furthermore, sulforhodamine 101 labels only a subpopulation of astrocytes, which increases in postnatal development and generally coincides with a subpopulation of GFAP-positive mature astroglia (811). Of note, sulforhodamine 101 (used in the concentrations commonly utilized for astroglia staining) affects neuronal excitability; in particular, it induces seizure activity in slices (824) and in vivo (1440).

5. Astroglia-specific expression of genetically encoded markers

Astrocytes can be visualized in tissue preparations and in the in vivo CNS with the aid of fluorescent proteinacious probes selectively expressed in astroglia under the control of cell-specific promoters (FIGURE 6C). The first animal models of “fluorescent” astrocytes employed GFP or its enhanced analog (GFP or EGFP) expressed under the control of human GFAP promoter (1236, 1707, 1964) or by murine S100β gene promoter (1829, 1972). The pool of available astroglial promoters is continuously increasing, and several fluorescent protein probes with distinct spectral characteristics are now in use (703, 999). Astrocytes can also be visualized through expression of genetically encoded Ca²⁺ indicators (976) such as yellow Cameleon-Nano 50 (YC-Nano50) (821) or green GCaMP (fused GFP, calmodulin and a peptide sequence from myosin light chain kinase) (1190). Several variants of GCaMP have been tested in astroglia, including GCaMP2 (722), GCaMP 3 (1609), GCaMP5 (16), and red RCaMPs (16). Alternatively, astrocytes in vivo or in situ can be transfected with fluorescent markers using the lentiviral system. With the use of this approach, neurons and astrocytes were spectrally separated through neuronal expression of the red fluorescent protein tdTomato and astroglial expression of EGFP (1580).

B. Heterogeneity and Main Types

The class of astroglia embraces many cell subpopulations with radically distinct morphology and function; hence, the matter of astrocytic classification and definition according to structure and function has been always under debate (866). In this respect, we take a rather simplistic approach by recognizing all the cells of neuroepithelial origin that are responsible for regulation of any aspect of CNS homeostasis as astrocytes. According to this logic, the neural cells, the main function of which is myelination, belong to oligodendroglia (including mature oligodendrocytes and their precursors also known as NG2 glia), while microglia are clearly distinct in their myeloid origin.

1. Astroglial numbers

For a long time, there has been a significant confusion about the total number of glial cells and of astrocytes in the CNS of various species, most notably in the human brain. Estimates of (total) glia to neuron ratio in the humans varied wildly from 1:1 to 50:1 (116, 698, 1832), while the belief that neuroglia outnumber neurons by a factor of 10 remains popular in the literature. From this reckoning, another, even more popular, presumption that astrocytes are the most numerous cells in the brain is repeatedly proclaimed (e.g., Refs. 744, 1352, 1491, 1762).³ The morphometry, however, comes with rather different numbers. In rodents, astrocytes account for 10–20% of total cells in the brain (1701). A recently developed method of isotopic fractionation [which was validated by unbiased stereological techniques for humans and macaque monkeys (85)]shows that total numbers of neurons and glia in the human brain are roughly the same, with substantial variations between different brain regions (79, 676, 970). The ratio between nonneuronal cells and neurons varied between 11:1 for brain stem, 3.7:1 in the cortical regions including corpus callosum, and 0.2:1 in the cerebellum (for details of controversial history of glial numbers, see Ref. 1832). Similar figures were obtained with stereology: for example, in the neocortex the average number of neurons was 21.4 billion in females and 26.3 billion in males, whereas the mean number of glial cells was 27.9 billion in females and 38.9 billion in males, which gives an overall glia-to-neuron ratio of ~1.3 (1361). The glial-neuronal ratio (excluding microglia) in the grey matter of the human cortex was estimated at 1.65 (1603). How many of those glial cells are astrocytes? Again, the precise numbers are difficult to obtain due to the limitations of the universality of glial markers. On the basis of morphological criteria, in the neocortex astrocytes accounted for ~20−40%, oligodendrocytes for 50−75% and microglia for 5−10% of the total glial population (197, 1361). Other numerical estimations set microglia at ~10−15% of all glia in the human brain (1130), whereas NG-2 glia account for another 5−10% (386), so both oligodendrocytes and astrocytes account for ~75% of all neuroglia. Assuming substantial predominance of oligodendroglia in the white matter, which occupies more than a half of the human brain (1835), we have to conclude that astrocytes are not, in all likelihood, the major glial population and possibly account for 20−40% of all glial cells.

There is a well-documented increase in the glial density and hence in the number of astrocytes throughout the evolutionary ladder, with glia-to-neuron ratio increasing from ~0.05/0.1 in invertebrates to much higher numbers in mammals. In the cortex the glia-to-neuron ratio is 0.3−0.4 in rodents and rabbit, ~1.1 in cat; ~1.2 in horse, 0.5−1.0 in rhesus monkey, somewhere between 1.5 and >2 in humans, and 4−7.5 in elephants and whales (331, 443, 482, 531, 664, 984, 1300, 1448). Incidentally in singing birds with their remarkable intelligence and computing abilities reflected by an extreme packing density of neurons, the glia-to-neuron ratio is similar to rodents ranging between 0.4 and 0.6 (1262). In primates, the steady increase in astrocyte-to-neuron ratio from 0.6 for Saki monkey to 1.2 in Gorilla gorilla and to 1.65 in Homo sapiens was quantified (1603). An increase in glial-neuron ratio in mammalian evolution most likely reflects an increase in neuronal energy expenditure and hence a need for more support provided by glia; increased synaptic transmission similarly accounts for higher demand for homeostatic clearance as well as maintenance of balance of neurotransmitters and ions.

2. Main types of astrocytes in mammalian brain

a) protoplasmic astrocytes.

Protoplasmic astroglia represent the major population of astrocytes in the grey matter of the brain and of the spinal cord. Rodent protoplasmic astrocytes (FIGURES 6 AND 7) are characterized by a small round somata (~10 μm in diameter) from which 5−10 primary processes (~50 μm long) emanate; these processes branch to form highly elaborated and dense peripheral arborization that underlies distinctive spongiform morphology (264). The overall morphological appearances of protoplasmic astrocytes differ not only between but also within anatomical structure. In the CA1 hippocampal region, fusiform, elongated, and spherical cells were distinguished (264, 1232); many more morphological appearances are observed throughout the brain (476, 1270). Similarly, there is a degree of functional heterogeneity. For example, hippocampal astrocytes have been categorized into two subtypes as GluR cells (which express ionotropic glutamate receptors) and GluT cells (which lack glutamate receptors, but express glutamate transporters) (1088). These subtypes have distinct morphology and membrane properties (1665, 1839). It should be noted, however, that hippocampal GluT-astrocytes express metabotropic GluRs (168, 169, 277, 1335), and there is also evidence that they do express some ionotropic GluRs (154, 1952). Protoplasmic astrocytes in the grey matter contact each other only by their finest processes, thus occupying nonoverlapping territorial domains (264, 1251).

The density of protoplasmic astrocytes ranges between 10,000 and 30,000 per mm³ in different brain areas of rodents, whereas the volume of a single protoplasmic astrocyte lies within a range of 50,000−80,000 μm³. The surface area of protoplasmic astrocyte could be as large as 80,000 μm² with absolutely the major part of this surface area being associated with peripheral processes (264, 1251, 1451). These peripheral processes are rather short (2−10 μm) and are ultrathin (with thickness of perisynaptic processes being in a range of 100−200 nm). These fine processes exhibit morphological plasticity (672). At least one of the processes of the protoplasmic astrocyte contacts the blood vessel where it forms perivascular endfeet. Protoplasmic astrocytes also send processes to the pial surface, where they form subpial endfeet, which contribute to glia limitans. A single protoplasmic astrocyte in the rodent cortex contacts 4−8 neurons, surrounds ~300−600 neuronal dendrites, and provides cover for up to 20,000−120,000 synapses residing within its domain (264, 627).

b) fibrous astrocytes.

These cells populate white matter of the brain and of the spinal cord, the optic nerve, and the nerve fiber layer of the retina. The somata of fibrous astrocytes are organized in rows between the axonal bundles. Fibrous astrocytes project long (up to 100 μm) processes, which are radially oriented in the direction of the axon bundles (1022, 1248). Fibrous astrocytes have much fewer terminal fine processes as compared with protoplasmic ones; in addition, processes of fibrous astrocytes overlap reflecting the absence of domain organization characteristic for protoplasmic cells. The processes of fibrous astrocytes establish several perivascular or subpial endfeet and send numerous extensions (perinodal processes) that contact axons at nodes of Ranvier. Fibrous astrocytes show diverse morphology; for example, in rodent optic nerve, fibrous astrocytes are subdivided into transverse, random, and longitudinal depending on the orientation of processes with respect to the long axis of the nerve (268). The monoclonal antibody Mab 6.17 raised against homogenates of embryonic rat muscle (which recognizes yet unknown proteins possibly synthesized by Schwann cells and concentrated in the cleft of neuromuscular junctions) was reported to specifically stain fibrous but not protoplasmic astrocytes in the brains of rats (1467).

c) surface-associated astrocytes.

These are represented by a specific subpopulation of astrocytes associated with the cortical surface in the posterior prefrontal and amygdaloid cortex. The somata of these cells are positioned at the cortical surface. The surface-associated astrocytes send two types of processes: one descending to the layer I and the other extending to surround pial vessels (500).

d) velate astrocytes.

Velate astrocytes represent a variation of protoplasmic astrocytes dwelling in the regions of the brain that are densely packed with small neurons, for example, in the olfactory bulb or in granular layer of cerebellar cortex (305). Velate astrocytes have a small soma and relatively short leaflike processes with a very high surface-to-volume ratio (20−30 μm⁻¹). In cerebellum, these processes ensheath several granule neurons (as if forming envelopes of vellum, hence their name). These processes envelop groups of granule neurons and the glomeruli comprising mossy fiber rosettes, Golgi neuron boutons, and granule cell dendrites. Possibly this morphological arrangement allows for isolating synaptic structures and separate groups of mossy fibers conveying different types of information (252, 721, 1301). Cerebellar velate astrocytes show GFAP immunoreactivity (884).

e) pituicytes.

The major type of astroglia of the neurohypophysis is known as pituicytes (661). They express canonical astroglial markers GFAP and S100B. Pituicytes show a remarkable heterogeneity and are subclassified into major, dark, ependymal, oncocytic, and granular pituicytes based on morphological criteria (1723). The neurohypophysis does not contain neuronal cell bodies, but only terminals of magnocellular neurons. These terminals are surrounded by pituicytes, which provide for osmosensitivity and regulate neuropeptides secretion through release of taurine and through dynamic modulation of terminal coverage (754, 1260, 1506). In contrast to other astroglia, pituicytes receive synaptic contacts, and they are sensitive to several neurotransmitters and neurohormones (660).

f) gomori astrocytes.

In the arcuate nucleus of the hypothalamus, the specific set of astrocytes rich in iron and positive for Gomori's chrome alum hematoxylin staining has been recognized (1926). Some Gomori astrocytes have been also identified in the hippocampus (1924). It has been suggested that Gomori astrocytes may supply heme to neurons for the synthesis of heme enzymes (1563). These astrocytes have unusually high glucose metabolism and express high-capacity GLUT2 glucose transporters, which make them comply with the specific metabolic needs of hypothalamic neurons (1925).

g) perivascular and marginal astrocytes.

Perivascular astrocytes are localized close to the pia mater, where they form numerous endfeet with blood vessels. As a rule, they do not contact neurons, and their main function is in establishing the pial and perivascular glia limitans barrier, which assists in isolating the brain parenchyma from the vascular and subarachnoid compartments (996).

h) ependymocytes, choroid plexus cells, and retinal pigment epithelial cells.

These cells line up the ventricles and the subretinal space. The choroid plexus cells produce the cerebrospinal fluid, which fills the brain ventricles, spinal canal, and the subarachnoid space. Ependymocytes are endowed with numerous very small movable processes (microvilli and kinocilia), which by rhythmic movements produce a stream of cerebrospinal fluid (CSF) (1449).

i) radial glia.

Radial glia can be defined as bipolar cells that extend through the whole thickness of the neural tube. These cells form one basal endfoot and one or several pial endfeet. The radial glia was first described by Camillo Golgi as epithelial cells (584) and named radial glia by Giuseppe Magini (1046). The neuroblast potential of these cells was suggested by von Lenhossék and Ramon y Cajal (969, 1426; for more historic details, see Refs. 1424, 1619). The radial glia are thus universal neural precursor cells.

J) radial astrocytes.

The true (primary) radial glia exist only in the developing brain. In the adult mammalian brain several cell types with radial-like morphology populate different regions. These cells are referred to as adult radial glia, radial glialike cells, or radial astrocytes. Considering that their function is invariably associated with local homeostasis we believe that the latter term is more appropriate. In addition, it clearly distinguishes between true radial glia (which primarily have neurogenic function) from adult parenchymal glia with radial morphology.

Retinal Müller cells combine radial morphology with fine processes characteristic of protoplasmic astrocytes (1449). The cell bodies of Müller glia are located in the inner nuclear layer of the retina, and their main processes extend to the subretinal space (these processes are equipped with microvilli) and to the vitreal space (the terminal part of these processes is packed with multivesicular bodies related to secretion). The side processes of Müller cells contact and enwrap neurons (some of these processes have velate morphology), cover synapses (by fine perisynaptic processes), and form endfeet, which plaster blood vessels (1447, 1450). In the human retina, a single Müller cell covers and supports ~16 neurons.

The cerebellar radial astrocytes are represented by Bergmann glia, which have small cell bodies (~15 μm in diameter), are located in the Purkinje cell layer, and extend three to six processes to the pia. These processes send highly elaborated branches with high surface to volume ratio (~20 μm⁻¹) that cover synapses formed by terminals of granule neurons. A single Bergmann glial cell ensheathes 6,000–8,000 synapses (608, 609). Several (~8 in rodents) Bergmann glial cells surround a single Purkinje neuron with Bergmann glial processes forming a “tunnel” around the dendritic arborisation of Purkinje neurons.

Radial astrocytes present in the supraoptic nucleus have cell bodies located at the ventral borders of the nuclei in the ventral glia lamina. These cells send several thick processes in dorsoventral direction often spanning the entire nucleus (211, 773).

The neurogenic niche of the subventricular zone contains a subpopulation of radial glia-like neural stem cells, which express markers of astroglia and of stem cells. These radial stem astrocytes are not only participating in adult neurogenesis but provide for local homeostasis by extending processes to plaster blood vessels and by forming fine perisynaptic processes. These latter mostly cover asymmetric synapses localized within territorial reach of these astrocytes (1124, 1158). Some radial astrocytes were also found in the olfactory bulb and are possibly associated with neurogenesis and migration of young neurons.

An extended class of radial astrocytes found in the periventricular organs, in the hypothalamus, in the hypophysis, and in the raphe part of the spinal cord is represented by tanycytes (1482). Tanycytes [their name coming from Greek verb tanyo (τανυω), which means to stretch or to elongate] are bipolar cells. Their processes contact the ventricular wall and portal capillaries. In the periventricular organs, tanycytes establish the brain-blood barrier by forming tight junctions between their somata. Tanycytes show a degree of heterogeneity in expression of various markers, morphological appearance, and function. Tanycytes localized near to dorsomedial and ventromedial hypothalamic nuclei are classified as α1 and α2 tanycytes, while those localized close to the arcuate nucleus and median eminence are defined as β1 and β2 tanycytes (18, 362, 1474, 1482). Hypothalamic tanycytes can, potentially, be stem cells underlying adult neurogenesis; tanycytes have been found to express several markers of stem cells including nestin (108), doublecortin-like proteins (1523), bromodeoxyuridine (BrdU), and Sox2 (957).

3. Human astroglia

a) protoplasmic and fibrous astrocytes.

Human parenchymal astrocytes are much larger and far more complex than astroglial cells in lesser vertebrates (FIGURE 8). The average diameter of the domain belonging to a human protoplasmic astrocyte is ~2.5 times larger than the domain formed by a rat astrocyte (142 vs. 56 μm). The volume of the human protoplasmic astrocyte domain is ~16.5 times larger than that of the corresponding domain in a rat brain. Furthermore, human protoplasmic astrocytes have ~10 times more primary processes emanating from their somata, and correspondingly much more complex processes arborisation (1248). As a result, human protoplasmic astrocytes contact and integrate around 2 million of synapses residing in their territorial domains, whereas rodent astrocytes cover ~20,000–120,000 synaptic contacts (264, 1248). Likewise, the fibrous astrocytes, populating the white matter, are ~2.2 times larger in humans than in rodents.

FIGURE 8.

Comparison of rodent and human protoplasmic astrocytes. A: typical mouse protoplasmic astrocyte. GFAP, white. Scale bar, 20 μm. B: typical human protoplasmic astrocyte in the same scale. Scale bar, 20 μm. C and D: human protoplasmic astrocytes are 2.55-fold larger and have 10-fold more main GFAP processes than mouse astrocytes (human, n = 50 cells from 7 patients; mouse, n = 65 cells from 6 mice; means ± SE; *P < 0.005, t-test). E: mouse protoplasmic astrocyte diolistically labeled with DiI (white) and sytox (blue) revealing the full structure of the astrocyte including its numerous fine processes. Scale bar, 20 μm. F: human astrocyte diolistically labeled demonstrates the highly complicated network of fine process that defines the human protoplasmic astrocyte. Scale bar, 20 μm. Inset: human protoplasmic astrocyte diolistically labeled as well as immunolabelled for GFAP (green) demonstrating colocalization. Scale bar, 20 μm. [From Oberheim et al. (1248).]

b) interlaminar astrocytes.

Interlaminar astrocytes [the name introduced by Jorge Colombo (350)] have been originally described by Carlo Martinotti, William Lloyd Andriezen (who named them “caudate neuroglia fibre cells”), and Gustav Retzius as cells with small cell bodies, residing in the upper cortical layer and sending long processes towards deeper layers (44, 1074, 1458). Many decades later it was found that these astroglial cells exist only in the brains of higher primates including old world monkeys, apes, and humans (347, 349, 350). Human interlaminar astrocytes (FIGURE 9) have a small spheroid cell body (~10 μm) localized in the supragranular layer (or layer I) of the cortex and several short and one or two very long (up to 1 mm) processes, which penetrate through the cortex, and end in layers II to IV (1248). The density of interlaminar astrocytes in human brain is quite high and is higher than in chimpanzee (1248). Processes of interlaminar astrocytes often demonstrate a spiriform or corkscrew pattern. The processes terminate in the neuropil and occasionally on the vasculature (1248); terminal portions of these processes end up with specific boutonlike structures also known as terminal masses or end bulbs, which usually contain mitochondria. The processes of interlaminar astrocytes run parallel to each other forming a “palisade.” The phylogenetic origin of interlaminar astrocytes remains unknown, although they were found in brown lemur, Eulemur fulvus (348), indicating that they could have appeared ~30 million years ago in prosimians. The interlaminar astrocytes appear postnatally and are thought to originate from some astroglial precursors and not directly from radial glia (349). The interlaminar astrocytes are specifically stained with CD44 antibodies; they have high immunoreactivity for GFAP and S100B and have low expression of glutamine synthetase and glutamate transporters (1655). Long processes of interlaminar astrocytes contact several blood vessels as well as the pia and traverse territorial domains of protoplasmic astrocytes. Electrophysiologically interlaminar astrocytes displayed passive currents similar to other types of astroglia (1655). The specific function of these cells remains unknown, although they might be instrumental in connecting distant cells and integrating cellular groups into wider structures.

FIGURE 9.

Interlaminar and varicose projection astrocytes in human cortex. A: interlaminar astrocytes as seen by William Lloyd Andriezen (44). B: pial surface and layers 1–2 of human cortex. GFAP staining is in white; DAPI is in blue. Scale bar, 100 μm. Yellow line indicates border between layer 1 and 2. C: interlaminar astrocyte processes. Scale bar, 10 μm. [B and C from Oberheim et al. (1247), with permission of Springer.] D: varicose projection astrocytes reside in layers 5–6 and extend long processes characterized by evenly spaced varicosities. Inset: varicose projection astrocyte from chimpanzee cortex. GFAP, white; MAP2, red; DAPI, blue. Yellow arrowheads indicate varicose projections. Scale bar, 50 μm. E: diolistic labeling (white) of a varicose projection astrocyte whose long process terminates in the neuropil. Sytox, blue. Scale bar, 20 μm. F: high-power image of the yellow box in B, highlighting the varicosities seen along the processes. Scale bar, 10 μm. G–I: individual z-sections of the astrocyte in E, demonstrating long processes, straighter fine processes, and association with the vasculature. [D–I from Oberheim et al. (1248).]

c) polarized astrocytes.

These cells are similarly present only in the brains of primates. Their somata are positioned in the deep cortical layers very near to the white matter. These cells have one or two long (up to 1 mm in length) processes that penetrate into superficial cortical layers (1248).

d) varicose projection astrocytes.

The varicose projection astrocytes exist only in the brains of humans (FIGURE 9). These cells are characterized by several (up to 5) very long (up to 1 mm) unbranched processes that extend in all directions through the deep cortical layers. These processes are endowed with evenly spaced varicosities (1248). These cells are similar to atypical protoplasmic astrocytes, described in adult human brains. These cells acquire immunopositivity to CD44, while their numbers vary considerably between individual samples; they are never observed in neonatal brains (1655). Morphologically approximately one-third of these CD44-positive cells were similar to regular protoplasmic astrocytes; some (~10%) had decreased arborization, while ~60% had long and thin processes (1655). The atypical astrocytes phenotype was suggested to reflect individual age-dependent, possibly adaptive, changes.

e) human astrocytes and intelligence.

The possible contribution of astrocytes to the information processing remains a debatable question. It has been hypothesized that astrocytes represent the smallest computing unit of the brain, the microchip, for integration of synaptic input. This hypothesis is based on the observation that astrocytes in visual cortex generate Ca²⁺ signals to a shorter tuning range of light than neurons. This indicates that astrocytes are more selective than neurons to the visual input and that astrocytic Ca²⁺ spikes represent processed second level information (1581). Since the volume of human astrocytes is 15- to 20-fold larger than in the rodents, a single astrocytic domain covers many more synapses in the human brain. It is therefore tantalizing to ask whether integration of information by astrocytes constitutes an essential part of processing power and in part explains human intelligence. To start addressing this question, the humanized mice pups were developed by intra-brain implantation of human glial cell progenitors at postnatal day 1 (637). The human glial cells expanded, populated large parts of the chimerized mice brain, and replaced mouse glial progenitor cells (FIGURE 10). Further analysis demonstrated that the threshold for induction of long-term potentiation (LTP) in hippocampus was reduced, in a pathway partly driven by tumor necrosis factor (TNF)-α, but not by release of d-serine or adenosine. The humanized chimeric mice displayed enhanced memory and ability to learn in cognitive tests, including novel object recognition or auditory fear conditioning. Several parallel lines of work have similarly shown that human glial progenitors, independently of whether they are generated from embryonic tissue or from induced stem cells, exhibit a growth advantage and slowly replace the host murine glial progenitor cells and mature astrocytes (582, 1885).

FIGURE 10.

Human astrocytes replace host glia in mice engrafted with human glial progenitors. A: schematic outlining the procedure for magnetic cell sort-based isolation (MACS) of human glial progenitors, tagging with EGFP and xenografting at P1. The chimeric mice brains were analyzed in 0.5- to 20-mo-old chimeric mice. B: representative dot map showing the distribution of human nuclear antigen (hNuclei)⁺ cells in 3 coronal sections from a 10-mo-old human chimeric mouse. C: the complex fine structure of human astrocytes in chimeric brain replicates the classical star-shaped appearance of human astrocytes labeled with hGFAP in situ. Most cells in the field are EGFP⁺/hNuclei⁺/hGFAP⁺ (hGFAP, red). D: at 5 mo, EGFP⁺ cells typically infiltrated corpus callosum and cortical layers V and VI. All EGFP⁺ cells labeled with an antibody directed against human nuclear antigen (hNuclei) and most of the human cells were also labeled with an antibody directed against human GFAP (hGFAP, red). E: at 11 mo, many areas of cortex were infiltrated by evenly distributed EGFP+ /hNuclei⁺ cells. F: the hippocampus was also populated with EGFP+/hNuclei⁺ cells in a 14-mo-old animal, with the highest density in the dentate. G: human EGFP⁺/hNuclei⁺/GFAP⁺ cells (green arrow) were significantly larger than host murine astrocytes (red arrow). The anti-GFAP antibody cross-reacted with both human and mouse GFAP (red). Inset shows same field in lower magnification. H: histogram compares the diameter of mouse cortical astrocytes to human cortical astrocytes in situ (freshly resected surgical samples) and xenografted human astrocytes in cortex of chimeric mouse brain. The maximal diameter of mouse and human astrocytes (in situ and in chimeric mice) was determined in sections stained with an anti-GFAP antibody that labels both human and mouse GFAP. In B–F, n = 50–65; **P < 0.01, Bonferroni t-test. EGFP, green; hNuclei, white and white arrow; DAPI, blue. Scale bars: 50 μm (C); 100 μm (D–F); 10 μm (G). Data graphed as means ± SE. [From Han et al. (637).]

C. Gender Differences

There are multiple indications for sexual dimorphism of astroglia (6). Gonadal hormones regulate the expression of GFAP, astroglial numbers, and morphology. Steroid deficiency causes astroglial atrophy in hippocampus, whereas treatment of organotypic cultures with estradiol and testosterone increases astrocyte complexity and extension of their processes (409). In young rats for example, the total number of GFAP-positive hippocampal astrocytes was higher in males, whereas feminization or masculinization of these animals reversed this difference (351). In adult rats, the GFAP immunoreactivity and astroglial numbers were highest in proestrus females compared with males or diestrus females; moreover, male astrocytes displayed more stellate, whereas female astrocytes had more reactive morphological appearance (63). In the somatosensory cortex, GFAP immunoreactivity and astroglial processes density are maximal in both proestrus and diestrus, while they decrease significantly during estrus (1685). In medial amygdala of adult rats, the number and complexity of astrocytes were higher in males, when compared with females or with testicular feminized mutant males lacking functional androgen receptors (798). Similarly, male astrocytes are more complex and have longer and more numerous processes in preoptic area (33), in nucleus arcuatus (1143), and in supraoptic nucleus (1692). In the cerebellum, Bergmann glial cells in males demonstrated more immunoreactivity for GFAP, whereas in females they showed more immunoreactivity for vimentin (1693). Sexual dimorphism was also extended to cellular functions; for example, female astrocytes in culture demonstrated more efficient glutamate uptake (1153).

D. Astroglial Networks or Syncytia

Macroglia (that is astrocytes and oligodendrocytes) represent the reticular portion of the brain being intracellularly connected into functional syncytia by gap junctions. The latter are ubiquitous structures responsible for intercellular integration in many tissues connecting, for example, epithelial cells of the gastrointestinal tract and kidney, providing metabolic coupling in the liver, electrical coupling in the heart, intercellular signaling in endocrine tissues, and defining cochlear physiology and hence hearing (217, 638, 1216, 1488). Gap junctions are specialized areas where two apposing membranes of adjacent cells come very close together so that the intercellular cleft is reduced to a width of ~2–3 nm (493). Within these areas, each gap junction is made of many hundreds of intercellular channels or connexons, which provide for intercellular transport of ions, second messengers, and other biologically active molecules with smaller than 1,000 Da (see sect. VIIH). In the grey matter pairs of astrocytes are connected with ~230 gap junctions on average, and injection of Lucifer yellow or biocytin into a single astrocyte results in staining of ~50–100 adjacent astroglial cells.

The earlier concept of panglial syncytium that connects all macroglia into single functional network, although being confirmed for invertebrates (1163), does not fully apply to mammalian CNS. First, the gap junctions between oligodendrocytes are rather restricted (1438). Second, in many brain regions anatomically segregated astroglial networks follow anatomical structures; for example, astroglial syncytia are confined to individual barrels of somatosensory cortex or to individual glomeruli in the olfactory bulb (571, 739, 1509). Coupling between adjacent astrocytes is also not ubiquitous; some (probably up to 15–20%) of the neighboring astrocytes are uncoupled as judged by either dye diffusion or by double patch-clamp recordings from the pairs of closely apposed cells (740, 1103). Thus astroglial coupling is defined not only by spatial proximity but also by some other factors, and astroglial networks may represent a wiring system parallel to that formed by neurons. The gap junctional connectivity may also represent a specialized “analog” intercellular signaling system (alternative to “binary” interneuronal communication), which by utilizing intercellular diffusion of multiple molecules, can provide a second level of information processing in the CNS.

Panglial syncytia, which connect astrocytes and oligodendrocytes, have been visualized in thalamus, neocortex, and hippocampus (604). Gap junction plaques between oligodendrocytes and astrocytes, localized on oligodendroglial somata, processes and outer (abaxonal) layer of the myelin sheath have been visualized with freeze-fracture electron microscopy (1079, 1437). Coupling between astrocytes and oligodendrocytes was also confirmed electrophysiologically (in pair recordings) and by the dye coupling (270, 856, 1337, 1477). The functional role of panglial connectivity remains largely unknown; an obvious suggestion links this to K⁺ homeostasis (1105). Mutations in oligodendroglial/astroglial connexins cause severe deficits in myelination in white matter (1281). Mutation in the gene GJB1, which encodes Cx32, underlies Charcot-Marie-Tooth disease associated with CNS white matter lesions (963, 1736), whereas mutations in GJC2 (encoding Cx47) are linked to leucoencephalopathies (253) and hypomyelinating leukodystrophies known as Pelizaeus-Merzbacher disease (1774). In mice, genetic deletion of either Cx32 or Cx47 does not result in an obvious phenotype; elimination of both genes however causes severe white matter demyelination (1104).

E. Recapitulation

Astrocytes display a morphological heterogeneity comparable to that of neurons; therefore, detailed mapping of gene expression and functional idiosyncrasies of astrocytes in different parts of the brain is of paramount importance. This task is far from being trivial as no universal marker of astroglial cells has yet been discovered; neither the ideal astroglia-specific promoter allowing expression of reporter proteins has been identified. Astroglial syncytia with their anatomical segregation add another level of complexity providing anatomically predetermined communication channels. Most intriguing is the evolution of astrocytes in higher primates and in humans; these astrocytes are distinguished by their complexity, size, and specific cell types. The idiosyncracies of human glia are cell autonomous and they persist even after grafting human fetal astrocytes into a non-human host brain; moreover, human cells outcompete astrocytes of the host, populate the brain, and confer improved cognitive performance.

VI. ASTROGLIAL MEMBRANE PHYSIOLOGY: ION DISTRIBUTION AND MEMBRANE POTENTIAL

A. Ion Distribution

Similarly to all living cells there is a disparity between cytosolic and extracellular ion concentrations in astrocytes (FIGURE 11). Conceptually cytosolic ion concentrations are defined by respective membrane permeability, by energy-dependent active transport and, in case of Ca²⁺, by cytosolic specific buffers. Intra-astrocytic concentration of K⁺ lies between 120 and 140 mM, while K⁺ concentration in the CSF and in the interstitial fluid (ISF) is ~3 mM, being substantially lower than in the plasma (639, 801). This sets the equilibrium potential for K⁺ (E_K) at −98 mV (at 37°C). Concentration of cytosolic Na⁺ in astrocytes (15−20 mM) is generally higher than in the majority of neurons (8−10 mM), although in some neurons [Na⁺]_i may also be as high as ~15 mM (1499). Only few actual measurements of resting [Na⁺]_i in astroglia exist, and they show substantial variability ranging from 9−19 mM in cultured cells (851, 1461, 1496, 1498, 1500) to 15−20 mM in brain slices (940, 1777); in vivo recordings are still pending. With Na⁺ concentration in the CSF around145−155 mM, the corresponding E_Na ranges between +55 and +60 mV. It has to be noted that [Na⁺]_CSF may undergo rhythmic and substantial (by 10−40 mM) fluctuations (648). Concentration of ionized Ca²⁺ in the cytoplasm of astrocytes ranges between 50 and 150 nM, being somewhat higher than in neurons (1950). With the assumption of [Ca²⁺]_CSF in adult rodents around 1.4 mM (800), the E_Ca lies between +120 and +140 mV. Astroglial distribution of Mg²⁺, the second major divalent cation controlling multiple cellular functions, has been poorly investigated. Cytosolic concentration of free Mg²⁺ in cultured astrocytes measured with fluorescent probe Mag-fura 2 was around 125 μM (81); the CSF Mg²⁺ has been determined at ~0.9 mM [thus being substantially less than in serum (1699)] and hence E_Mg is ~25 mV.

FIGURE 11.

Ion distribution and corresponding values of equilibrium potentials for different ions between the cerebrospinal fluid, interstitial space, and cytoplasm of astrocytes and neurons.

It is generally believed that astrocytes contain relatively high concentration of Cl⁻ ions (when compared with 5 mM in neurons), which varies between 30 and 60 mM, with E_Cl being set around −35 mV (the [Cl⁻]_o is ~120 mM). High [Cl⁻]_i was deduced from membrane potential recordings, in which activation of GABA_A receptors (that are essentially Cl⁻ channels) always depolarized cultured astrocytes (857), and the apparent E_Cl was calculated at approximately −35 mV. With the use of gramicidin perforated patch clamp with different Cl⁻ concentrations in the pipette the [Cl⁻]_i was estimated at ~29 mM (126). The radioactive Cl⁻ extrusion assay estimated resting [Cl⁻]_i between 31 and 43 mM (865), whereas ion-sensitive microelectrode measurements found [Cl⁻]_i to be between 20 and 40 mM (854). In Bergmann glial cells in situ, [Cl⁻]_i (measured with chloride-sensitive probe MQAE) was around 50 mM in young (P5 to P6) mice and ~35 mM in animals aged between P13 and P100 (1778). High astroglial Cl⁻ is supposedly maintained by activity of various Cl⁻ transporters, such as for example Na⁺/K⁺/2Cl⁻ (NKCC1) cotransporters that move 2 Cl⁻ into the cell in exchange for 1 K⁺ and 1 Na⁺. It has to be noted, however, that high NKCC1 activity in cultured astrocytes contrasts to a rather low performance of this transporter in the in situ hipocampal preparation (942). Hence, direct measurements of astroglial [Cl⁻]_i in situ or in vivo in different types of astroglia are of paramount importance; these measurements, however, are yet to be done. The concentration of protons in astroglial cytoplasm is ~63 nM (pH 7.2), which, assuming the extracellular H⁺ concentration to be ~40 nM (pH 7.4) sets the E_H at approximately −12 mV. The cytoplasm of astrocytes is rich in CO₂ (~1.2 mM) and HCO₃⁻ (~17 mM).

B. Membrane Potential

The most characteristic electrophysiological signature of mature astrocytes is their hyperpolarized resting potential that is set close to E_K (approximately −80 mV) and low input resistance (5−20 MΩ) indicative of high resting membrane permeability for K⁺ (379, 1125, 1126). This is also reflected by nearly linear current to voltage relationship, which is the most characteristic electrophysiological signature of astroglia (FIGURE 12; Refs. 11, 337, 436, 455, 772, 811, 1012, 1265). Incidentally neonatal astrocytes are even more hyperpolarized (V_m around −85 mV) (1951). Some variability of resting V_m (between −85 and −25 mV) reported for astrocytes in culture, in organotypic and acute slices, and in the optic nerve (208, 917, 1096), most likely reflect preparation artifact. These membrane properties are intrinsic for astrocytes as isolated cells are very similar to syncytially connected cells in the nervous tissue (454); astrocytes in the syncytia are isopotential (1027). Fluctuations of astroglial V_m generally reflect changes in extracellular K⁺ concentration (36, 379).

FIGURE 12.

Passive membrane properties of astrocytes. A: voltage-clamp recordings from astrocytes freshly isolated from the cortex of transgenic mice expressing EGFP under control of the GFAP promoter. Astrocytes were identified by specific fluorescence; whole-cell currents were recorded in response to hyperpolarizing and depolarizing steps from −120 to +60 mV (step interval, 20 mV). To construct the current-voltage relationship, amplitudes of currents were normalized to the value measured at 0 mV; every point is mean ± SD for 20 cells. [From Lalo et al. (935).] B: voltage-clamp recordings from astrocytes in acute slices obtained from 3- and 20-mo-old GFAP-EGFP mice; astrocytes were identified by fluorescence. [From Lalo et al. (932).] C: voltage-clamp recordings from human astrocytes grafted into mouse brain. [From Han et al. (637).]

C. Recapitulation

Despite prominent morphological heterogeneity, astrocytes show remarkably similar electrophysiological signatures with hyperpolarized membrane potential and predominant passive K⁺ permeability. The negative and stable resting membrane potential of astrocytes influences their homeostatic capabilities through defining the electrodriving force for numerous membrane transporters.

VII. ION CHANNELS

A. Potassium Channels

1. Inward rectifier potassium channels, K_ir

Inward rectifying K⁺ channels, which pass K⁺ ions more easily into the cell (the inward direction) than out of the cell (the outward direction) have two transmembrane domain topology, assemble as homo- or heteromeric tetramers, and are represented by 16 subtypes encoded by distinct genes (KCNJ1−KCNJ18) classified into 7 families from K_ir1.x to K_ir7.x (695, 914). The K_ir1.x, K_ir4.x, K_ir5.x, and K_ir7.x channels are weakly rectifying channels; K_ir2.x are strongly rectifying constitutively active channels; K_ir3.x are G protein-coupled channels; whereas K_ir6 subunits assemble with sulfonylurea receptors (SUR, of which 4 isoforms exist) to form ATP-gated K⁺ channels. Channel rectification reflects a voltage-dependent block of these channels by Mg²⁺ and polyamines (1016). Inward rectifying channels in general have high open probability at potentials more negative than resting V_m; this open probability decreases with depolarization.

The K_ir4.1 (encoded by KCNJ10 gene) is the main type of astroglial inwardly rectifying K⁺ channel. It is functionally expressed in parenchymal astrocytes and radial astrocytes including retinal Müller cells and cerebellar Bergmann glia (TABLE 2 and Refs. 269, 696, 770, 817, 1646). Inward rectifying K⁺ currents were recorded from cultured and freshly isolated astrocytes from hippocampus, neocortex, optic nerve, cerebellum, spinal cord, and retina; the early experiments revealed their characteristic voltage dependence, sensitivity to extracellular K⁺, and inhibitory effects of extracellular Cs⁺ and Ba²⁺ (107, 1242, 1434, 1647, 1763). Astroglial inwardly rectifying currents are also modulated by extracellular Na⁺: reducing the latter significantly diminished K_ir amplitude (1435). The K_ir4.1 channels are present in forebrain, midbrain, and hindbrain with predominant astroglial localization (696), although distribution of K_ir4.1 channels has regional variability; for example, immunoreactivity for K_ir4.1 is prominent in many hippocampal astrocytes, in astroglial cells in the cerebral cortex, in the deep cerebellar nuclei, in Bergmann glia, and in Müller cells but not in astrocytes in white matter (1397). This latter finding is somewhat controversial as K_ir4.1 immunoreactivity was detected in white matter astrocytes in the optic nerve (817). In the spinal cord, expression of K_ir4.1 channels and density of inward rectifier K⁺ currents are the highest in astrocytes from ventral horn and the lowest in astrocytes from the apex of the dorsal horn (1264). At the cellular level, K_ir4.1 channels are usually concentrated in perivascular endfeet, in perisynaptic processes of astrocytes (696), and in the radial processes of Bergmann glia (229). The anatomical segregation of K_ir4.1 was also demonstrated by double staining with dystrophin (specifically localized in the membranes of perivascular endfeet) and with tomato lectin, which labels vascular endothelial cells (1154). Not all perisynaptic processes express K_ir4.1 however: in olfactory bulb for example more than a half of astroglial processes surrounding excitatory synapses were immunopositive for K_ir4.1 (696). Concentration of K_ir4.1 channels in the processes is developmentally regulated and gradual shift of K_ir4.1 from the soma to the periphery has been observed postnatally between P0 and P60 (1154). The overall density of K_ir4.1 channels substantially (severalfold) increases during postnatal development (1588). Studies on cellular and subcellular localization of K_ir4.1 channels have to be treated with care, however, because of the lack of perfectly specific antibodies. The K_ir4.1 channels significantly contribute to the resting astroglial membrane conductance; inhibition of K_ir currents with Ba²⁺ increases input resistance up to 20-fold and depolarizes astroglia (1434). Genetic deletion of K_ir4.1 channels in vivo or inhibition of their transcription in vitro increases the input resistance and depolarizes resting potential by 15−20 mV in protoplasmic astrocytes (1265, 1588) and in Müller glia (891). In uncoupled (by 30 min incubation with gap junction blocker carbenoxolone), astrocytes in slices 100 μM Ba²⁺ reduced the membrane conductance by ~50% and depolarized the membrane from −70 to −60 mV (1588). The degree of rectification of K_ir4.1 channels can be regulated by endogenous polyamines such as spermine and spermidine (1627). An elevation in extracellular K⁺ concentration increases the inward K⁺ currents proportionally to the square root of [K⁺]_o (1269, 1434).

Table 2.

Channels in astrocytes

Current or Channel Type/Subunit	Experimental Preparation/Technique	Pharmacology	Biophysical Properties and Functional Relevance	Reference Nos.
Voltage-gated sodium channels
INa	Rat/cultured primary astroglia/whole cell voltage clamp	TTX (KD ~0.5 mM), STX (KD ~30 nM)	First detection of fast voltage-activated Na+ currents in astrocytes; INa amplitudes were ~1 nA. Authors suggested that astrocytes synthesize Na+ (and K+) channels for subsequent insertion into the neighboring axon.	166
INa	Rat/cultured primary astroglia/optic nerve/whole cell voltage clamp	TTX (KD ~2.8 nM)	INa was detected in the majority of type 2 astrocytes and in many type 1 cells; respective current amplitudes were 400–3,100 pA and 10–300 pA. Glial INa had slower kinetics when compared with neuronal currents. In type 1 astrocytes, the steady-state inactivation curve was significantly shifted into hyperpolarizing direction with h∞ ~-80 mV.	105, 106
INa fast, INa slow	Rat/cultured primary astroglia/optic nerve, hippocampus/whole cell voltage clamp		Two types of INa with distinct (fast and slow) kinetics and steady-state inactivation were dissected. With increasing time in culture, slow current became dominant in hippocampal cultures.	1649, 1650
INa TTX sensitive, INa TTX resistant	Rat/cultured primary astroglia/optic nerve, hippocampus/whole cell voltage clamp	TTX (KD ~5.7 nM), TTX (KD ~1,007 nM)	The TTX-sensitive INa was confined to stellate astrocytes, whereas TTX-resistant INa dominated in fibroblast-like astrocytes.	1647, 1652
INa slow	Rat/tissue prints/optic nerve, hippocampus/whole cell voltage clamp	TTX (10 μM)	In tissue prints of astrocytes at P2, INa was not detected; in contrast, at P10, all astrocytes demonstrated slow INa.	107
INa	Rat/acute slices/hippocampus/whole cell voltage clamp		INa was recorded from a minor subpopulation (5 of 40) of GFAP-positive astrocytes.	1651
INa	Rat/acute slices/spinal cord/whole cell voltage clamp		INa with amplitudes of 20–55 pA was recorded from a subpopulation of cells with small somata and long processes with faint GFAP staining.	337
Nav1.5, Nav1.2, Nav1.3	Rat/cultured primary astroglia/spinal cord/RT-PCR, in situ hybridization, immunocytochemistry		The Nav1.5 identified as the predominant type of astroglial Na+ channels.	185, 189
Nav1.6	Rat/cultured primary astroglia/spinal cord/immunocytochemistry		The Nav1.6 expression was detected in embryonic cultures; immunoreactivity was confined to stellate cells only.	1444
Non-voltage-gated sodium channels
Nax	Rat		The Nax channel is expressed in astrocytes and ependymocytes in circumventricular organs. The channel is activated by an increase in [Na+]o above 150 mM; the activation threshold is reduced to ~140 mM in the presence of endothelin-3 acting through ETBRs.	1234, 1613, 1864
ENaC	Rat/tissue section/immunocytochemistry		The ENaC γ-subunits were found in GFAP-positive astrocytes using specific antibodies.	1116
Voltage-gated calcium channels
L-type Ca2+ currents	Rat/cultured primary astroglia/cortex/two-electrode patch clamp	Co2+ (2–10 mM), Cd2+ (1 mM), nifedipine (10 μM)	Ba2+-dependent action potentials or Ba2+ current were measured from individual astrocytes; at 10 mM Ba2+, maximal amplitudes were at nA range. Norepinephrine or dbt-cAMP enhanced or induced Ba2+ currents.	1038, 1041
L-type Ca2+ currents	Rat/cultured primary astroglia/cortex/whole cell patch clamp	Nifedipine (100 μM)	Ca2+ currents with maximal amplitudes ~100–200 pA were recorded only after cell preincubation for 1–2 h with forskolin or 8-bromo-cAMP.	104
L- and T-type of Ca2+ currents	Rat/cultured primary astroglia/acutely isolated astrocytes/tissue prints/optic nerve/whole cell patch clamp		High (L) and low (T) Ca2+ currents with maximal amplitudes of 50–200 pA were recorded from the majority of neonatal astrocytes. With age (at P30), T currents disappeared and L currents were detected only in 37% of cells.	106, 107
L-type Ca2+ currents	Rat/cultured primary astroglia/acutely isolated astrocytes/hippocampus/ [Ca2+]i microfluorimetry	Co2+ (1 mM), verapamil (>30 μM), nifedipine (100 μM), BAY K 8644 (10 μM)	High K+ depolarization triggered transient elevations in [Ca2+]i reaching ~1 mM; these were strictly dependent on [Ca2+]o, where inhibited by Ca2+ blockers and potentiated by BAY K 8644.	460, 1040
L-, N-, and R-type Ca2+ currents	Rat/cultured primary astroglia/cortex/whole cell or perforated patch clamp	Nifedipine (5 μM), ω-conotoxin GVIA (3 μM), SNX-482 (100 nM)	Ba2+ whole cell and single-channel currents showed complex pharmacological behavior. Whole cell currents were partially sensitive to L-, N-, and R-type Ca2+ blockers. RT-PCR revealed expression of relevant α1 subunits.	374
L-type Ca2+ currents	Rat/cultured primary astroglia/hippocampus/ [Ca2+]i microfluorimetry	Cd2+ (100 μM)	High K+ depolarization triggered transient elevations in [Ca2+]i, which instigated vesicular release of glutamate. RT-PCR revealed expression of mRNAs for α1B, α1C, α1D, and α1E VGCC subunits.	1909
Cav1.2, Cv1.3 channels	Mouse/primary culture/cortex/[Ca2+]i microfluorimetry	Verapamil (5 μM), nifedipine (5 μM)	High K+ depolarization induced [Ca2+]i transients, which disappeared after removal of extracellular Ca2+. Expression of mRNA for Cav1.2 and Cav1.3 was detected at mRNA and protein levels. In Cav1.2 KO mice, [Ca2+]i transients were reduced by 80%.	314
L-type Ca2+ channels	Rat/acute slices/ventrobasal thalamus/[Ca2+]i microfluorimetry	Nifedipine (1 μM), BAY K 8644	Spontaneous astroglial [Ca2+]i oscillations were potentiated by BAY K 8644 and inhibited by nifedipine.	1331, 1332
Ca2+ release-activated Ca2+ channels (ICRAC/ORAI)
ORAI1	Rat/primary culture/cortex/[Ca2+]i microfluorimetry, RT-PCR, immunocytochemistry		Expression of ORAI1 and STIM1 was detected at mRNA and protein level; overexpression of ORA1/STIM1 increased, whereas siRNA KO decreased thrombin-induced SOCE.	1150
ORAI3, ORAI2	Rat/primary culture/hippocampus/[Ca2+]i microfluorimetry, RT-PCR	Pyr6 (5μM)	Expression of ORAI3 mRNA was predominant, exceeding expression of ORAI2 6 times; ORAI1 was not detected at all.	1492
TRP channels
TRPA1	Rat/tissue sections/trigeminal caudal nucleus/immunogold electron microscopy		TRPA1 immunoreactivity was detected in the somata and processes of astrocytes; immunoreactivity increased significantly in the fine process of astrocyte in rats with experimental inflammation of the temporomandibular joint.	965
TRPA1	Rat/primary cultures, acute slices/hippocampus/[Ca2+]i microfluorimetry, whole cell voltage clamp	Gd3+, Li3+, TRP antagonist HC030031 (40 μM), TRPA1 agonist AITC (1 μM)	Constitutive activity of TRPA1 channels was found to trigger “spotty” near-membrane Ca2+ microdomains and contributed to resting [Ca2+]i.	1611
TRPV1	Rat/brain, spinal cord/immunocytochemistry, immuno-EM		TRPV1 channels were identified in astrocytes throughout the CNS. TRPV1 protein was preferentially localized in astroglial processes.	442, 1755
TRPV1	Rat/primary culture/whole cell patch clamp, immunocytochemistry	Capsazepine (30 μM)	TRPV1 channels were detected immunocytochemically; ~90% of all cell demonstrated TRPV1-mediated currents.	746
TRPV4	Rat/primary culture/whole cell patch clamp/[Ca2+]i microfluorimetry, immunocytochemistry, RT-PCR	TRPV4 agonist 4α-phorbol 12,13-didecanoate, 4αPDD (3 μM)	TPPV4 were detected at transcript and protein levels. Astrocytes generated TRPV4-mediated currents and [Ca2+]i transients.	135
TRPV4	Rat/hippocampus/acute slices, primary cultures/whole cell patch clamp/[Ca2+]i microfluorimetry, immunocytochemistry, Q-PCR	TRPV4 agonist 4αPDD (5 μM)	Postischemic astrogliosis was associated with an increase in TRPV4 expression, TRPV4-mediated currents, and [Ca2+]i transients	265
TRPC1, TRPC4, TRPC5,	Mouse embryo/primary culture, acutely isolated astrocytes/cortex/[Ca2+]i microfluorimetry, immunocytochemistry, RT-PCR		Members of TRPC family were detected at transcript and protein level. Treatment with anti-TRPC1 inhibiting antibody suppressed astroglial [Ca2+]i transients, likely due to an inhibition of SOCE	587, 606, 1055, 1393, 1462
TRPC6	Mouse/primary culture/[Ca2+]i microfluorimetry		In vitro knockdown of TPRC6 reduced receptor-induced Ca2+ entry. Treatment of astrocytes with IL-1β for 24–48 h increased TRPC6 expression and disrupted Ca2+ homeostasis.	163
Acid-sensitive ion channels, ASICs
ASIC1, ASIC 2a, ASIC3	Rat/primary culture/cortex/[Ca2+]i microfluorimetry, whole cell voltage clamp		ASICs expression was confirmed with immunocytochemistry and Western blotting; ASICs were mainly localized to the nuclei.	746
Inward rectifying potassium channels
Inward rectifying K+ current	Rat/acutely isolated astrocytes/hippocampus/whole cell patch clamp	Ba2+ (10 mM)	The inwardly rectifying K+ currents were isolated after inhibiting other K+ conductances with a mixture of TEA and 4-AP.	1763
Inward rectifying K+ current	Rat/primary cultures/spinal cord/whole cell patch clamp, cell-attached patch clamp	Cs+ (KD 189 μM), Ba2+ (KD 3.5 μM), TEA (90% block at 10 mM)	Two types of single-channel currents with conductances of 28 and 80 pS were detected. The open probability was largest at EK.	1434
Kir4.1 channels	Rat/primary culture/acute slices/whole cell patch clamp; RT-PCR, immunocytochemistry	Ba2+ (100 μM)	The Kir4.1 channels (detected at mRNA and protein levels) were principally responsible for resting K+ conductance. Genetic deletion of Kir4.1 protein resulted in depolarizing shift in resting Vm and in a 4.5 times increase in the input resistance.	1265
Kir4.1 homomeric, Kir4.1/Kir5.1 heteromeric channels	Mouse/brain sections/cortex, hippocampus, olfactory bulb, thalamus/immunocytochemistry/immunoelectron microscopy		Both Kir4.1 and Kir5.1 are expressed specifically in astrocytes; the Kir4.1/Kir5.1 colocalization was detected in cortex, hippocampus, and pontine nucleus. The Kir4.1/Kir5.1 heteromers were mainly confined to perivascular and perisynaptic processes.	694
Kir2.1, Kir2.2, Kir 2.3	Mouse/acute slices/hippocampus/whole cell patch clamp, RT-PCR	Ba2+ (100 μM)	Strongly rectifying Kir currents reflected the expression (at mRNA level) of Kir2.1, Kir2.2, Kir2.3 channel isoforms. These current were inhibited following activation of AMPA receptors; inhibition was due to an increased Na+ influx.	1576
Kir6.1	Rat, mouse/brain sections, acute slices/cerebellum	Tolbutamide (100 μM)	Kir6.1 expression was found in astrocytes from hippocampus, cortex, and cerebellum; Kir6.2 was detected in neurons only. Whole cell recordings revealed tolbutamide-sensitive KATP currents in Bergmann glial cells.	1745
Voltage-independent two-pore domain (K2P) K+ channels
TREK-1,TWIK-1	Rat/Acute slices/Hippocampus/Whole-cell patch clamp, Immunocytochemistry	Quinine (200 μM)	TREK-1,TWIK-1 channels were detected immunocytochemically and whole-cell currents sensitive to quinine were considered to reflect their operation.	1955
TREK-1,TREK-2,TWIK-1	Mouse/freshly isolated astrocytes/hippocampus/whole cell voltage clamp, singe-cell RT-PCR	Quinine (200 μM), bupivacaine (200 μM)	Transcripts for TREK-1,TREK-2, and TWIK-1 were identified in single astrocytes. K2P currents (with almost linear I–V relationship) were isolated as quinine- or bupivacaine-sensitive components after inhibition of Kir and A-currents with Ba2+ and 4-AP.	1588
TWIK-1/TREK-1 heterodimer	Mouse/primary cultures/brain sections/whole cell voltage clamp, immunocytochemistry		The TREK1/K2P2.1(KCNK2) and TWIK1/K2P1.1(KCNK1) channels were localized in hippocampal astrocytes by immunocytochemistry. Acidification increased currents mediated by TREK1 channels. Further analysis in cultured cortical astrocytes and in astrocytes in hippocampal slices identified TWIK-1/TREK-1 heterodimer (formed by disulfide bridge between cysteine-cysteine residuals of both subunits) as the predominant form of astroglial channel.	756, 1874, 1955
Voltage-gated delayed rectifying K+ currents
KSI (slow inactivating), KD, KA	Rat/primary cultures/spinal cord/whole cell patch clamp	4-AP (100 μM to 8 mM)	At 100 μM, 4-AP inhibited slowly inactivating outward current KSI; at 2 mM, KA and KD; at 8 mM, KIR.	214
Kv1.5	Rat/primary cultures, acute slices/hippocampus, cerebellum, spinal cord/immunocytochemistry, whole cell patch clamp	TEA (0.4–40 mM)	Specific Kv1.5 antibodies revealed widespread staining of GFAP-positive astrocytes, with particularly high immunoreactivity in perivascular endfeet. Treatment of astrocytes with Kv1.5 siRNA halved the amplitudes of delayed rectifying whole cell currents.	1513
Kv1.4	Rat/tissue sections/spinal cord/immunocytochemistry, Western blot, in situ hybridization		Both mRNA and protein for Kv1.4 were identified in astrocytes; Kv1.4 expression increased 6 wk after spinal cord injury	466
Kv11.1, or ERG1 (ether-à-go-go-related gene)	Rat/acute slices/hippocampus/immunocytochemistry, whole cell patch clamp	Dofetilide (100–1,000 nM), E4031 (100 nM)	ERG1 channels were localized to astrocytes by immunocytochemistry and characterized electrophysiologically	475
Rapidly inactivated K+ current (A-current)
A-current	Rat/hippocampus/acutely isolated astrocytes/whole cell patch clamp	4-AP (1 mM)	The A-current has a characteristic fast kinetics and steady-state inactivation with V0.5 at −60 mV	1763
Kv4.3, Kv3.2, Kv1	Rat/primary cultures/hippocampus/whole cell patch clamp, RT-PCR, immunocytochemistry	4-AP (4 mM), TEA (10 mM), ETYA (nonhydrolyzable analog of arachidonic acid, 10 μM)	Pharmacological analysis indicated that Kv4, Kv3, and Kv1 account for 70, 10, and <5% of A-current, respectively.	124
Ca2+-activated K+ channels
BK, IK	Rat/primary culture/cortex, hippocampus/cell-attached and whole cell patch clamp	TEA (1 mM)	Two types of unitary KCa currents with conductances of 71 ± 6 and 161 ± 11 pS were detected. KCa currents were potentiated by activation of mGluRs.	562
SK3 (KCa2.3)	Rat, mice/supraoptic nucleus/sections/immunocytochemistry		Immunoreactivity for SK3(KCa2.3) was mainly confined to astroglial processes.	65
BK	Rat/cortex/acute slices/perforated whole cell patch clamp, cell-attached patch clamp	Iberiotoxin, IbTX (200 nM), TEA (1 mM)	In astroglial endfeet, both whole cell and single-channel (unitary conductance 225 pS) BK currents were observed.	511
KCa3.1	Mouse/acute slices/cortex, whole cell patch clamp, immunocytochemistry, RT-PCR	KCa3.1 blocker TRAM-34 (1 mM), KCa2.1–2.3 and KCa3.1 agonist NS309	Astrocytes express KCa2.3 and KCa3.1 mRNA; immunoreactivity of KCa3.1 was restricted to processes. KCa2.1/KCa3/1 opener increased, whereas KCa3.1 inhibitor decreased whole cell currents.	1006
Intracellular Ca2+ channels
InsP3R type 2	Mouse, rat/primary cultures, brain sections/RT-PCR, immunocytochemistry, transgenic techniques		The InsP3R2 are considered to be the main type of InsP3 receptors in astroglia. Genetic deletion of InsP3 R2 leads to an almost complete disappearance of global astrocytic Ca2+ signals.	678, 718, 820, 1381, 1598, 1602, 1817
InsP3R type 1, InsP3R type 2	Mouse/primary cultures/organotypic and freshly isolated hipocampal slices/RT-PCR, immunocytochemistry, /[Ca2+]i microfluorimetry		InsP3 R1 transcripts and proteins were detected in astrocytes cultured from hippocampus and entorhinal cortex. Evidence for InsP3R1/2-mediated Ca2+ release was obtained in imaging experiments.	607, 986, 1604
RyR3	Mouse/primary cultures/[Ca2+]i microfluorimetry	Ryanodine (200 mM), chloro-m-cresol (4 mM)	RYR agonist chloro-m-cresol triggered [Ca2+]i elevation.	1089
TPC1, TPC2	Rat/primary culture/cortex/[Ca2+]i microfluorimetry, RT-PCR	NED-19 (TPC channel blocker, 10 μM), bafilomycin 1A (100 nM)	Injection of 100 nM NAADP triggered [Ca2+]i transients, sensitive to NED-19 and preincubation with bafilomycin 1A. RT-PCR revealed transcripts for TPC1 and TPC2 channels.	1370
Hyperpolarization-activated cyclic nucleotide-gated (HCN) channels
Hyperpolarization-activated cyclic nucleotide-gated channels, HCN1–4	Mouse/freshly isolated, FACS sorted astrocytes, acute slices/RT-qPCR, whole cell patch clamp		NCN 1–4 were detected at mRNA level; the whole cell HNC-mediated currents have been characterized.	719, 1520
Anion channels
Voltage-gated Cl− channels, ClC-1, ClC-2, ClC3	Mouse/primary cultures, acute slices/whole cell voltage clamp		ClC channels were detected at mRNA and protein levels. Inwardly rectifying Cl− currents (i.e., promoting Cl− efflux) have been characterized.	456, 1054, 1326, 1489
Best1	Mouse/primary culture, acute slices/cortex, hippocampus/[Ca2+]i microfluorimetry, whole cell perforated patch clamp, RT-PCR, immunocytochemistry	Niflumic acid (100 μM), flufenamic acid (100 μM), NPPB (100 μM)	Best1 was detected at mRNA (single-cell RT-PCR) and protein levels. Whole cell currents were characterized in isolated astrocytes and in astrocytes in slices. Silencing of Best1 gene with shRNA significantly suppressed Cl− currents.	1324
Volume-regulated anion channels, VRAC/SWELL1	Mouse/primary culture, acute slices/whole cell voltage clamp		Channels have been mainly characterized in vitro; VRAC channels have been shown to mediate release of excitatory amino acids, which have been detected in vitro and in situ. Increase in [Na+]i suppressed VRAC activity.	5, 138, 867, 868, 1122, 1333
Aquaporins
Aquaporin: AQP1, AQP4, AQP9	Mouse, rat/primary culture, brain sections/RT-PCR, immunocytochemostry		The main astroglial type is AQP4, which is primarily concentrated in the endfeet; are important water pathways, form the molecular basis for the glymphatic system.	1184, 1749
Connexons and pannexons
Cx43	Mouse/primary culture/Northern blot, immunocytochemistry, double-cell patch clamp		The Cx43 expression was found at cRNA and protein level with immunoreactivity; mainly concentrated at intercellular contact areas. Single-channel conductance was determined at ~56 pS.	570
Cx26	Rat, mouse/brain/tissue sections/Northern blot, Western blot, immunocytochemistry, EM		Expression of Cx26 was detected in astroglial gap junctions.	1187
Cx30	Rat/whole brain/primary cultures, tissue sections, acute brain slices/immunocytochemistry, EM		Cx30 protein was detected in astrocytes in situ and in vitro; Cx30 expression increased postnatally attaining maximum at 4 wk after birth. Cx30 was colocalized with Cx43 at gap junctions.	923, 1189
Cx26-Cx32 or Cx43/32-Cx47	Mouse/whole brain/brain sections/immunocytochemistry		Heteromeric Cx26-Cx32 or Cx43/32-Cx47 gap junctional channels were found to connect astrocytes with oligodendrocytes.	28
Panx1	Mouse/primary cultures/RT-PCR, whole cell perforated patch clamp	CBX (50 mM), MFQ (100 nM)	Detected at mRNA level and by voltage clamp on cultured embryonic astrocytes.	749, 758, 1443

The K_ir4.1 may coassemble with K_ir5.1 subunits forming a heteromeric channels with high sensitivity to pH (1728). These K_ir4.1./K_ir5.1 heteromers were identified in astrocytes in the retrotrapezoid nucleus and were suggested to contribute to chemoception (1164). In the Müller glia, K_ir4.1/K_ir5.1 heteromeric channels are concentrated in the cell body and in the distal part of the cell, whereas K_ir4.1 homomers are localized in the endfeet and in the perivascular processes; this distribution possibly reflects specific roles played by each channel type (769). In protoplasmic astrocytes, K_ir4.1/K_ir5.1 heteromeric channels were identified in the olfactory bulb and neocortex, whereas hippocampal astrocytes and cerebellar Bergmann glial cells expressed solely K_ir4.1 homomers (229, 694). At the cellular level, K_ir4.1/K_ir5.1 heteromeric channels mostly occur in perisynaptic and perivascular processes and in the processes close to pia mater (694). In the optic nerve, the K_ir4.1/K_ir5.1 heteromers were rather homogeneously distributed between somata and processes (229).

Some hippocampal astrocytes also express mRNA for K_ir2.1 (KCNJ2), K_ir2.2 (KCNJ12), and K_ir2.3 (KCNJ4) channels, thus generating strongly rectifying whole cell currents, compatible with K_ir2.x functional expression. These currents were inhibited by Na⁺ influx through AMPA receptors (1576). The K_ir2.1 immunoreactivity was also detected in some hippocampal astrocytes (1677). The K_ir2.2 channels were localized (by immunostaining) in Bergmann glial cells and in cerebellar astrocytes (971). Transcripts for K_ir2.1, K_ir2.2, and K_ir2.4 were detected in Müller glia (1419), while K_ir2.1 protein was also identified in these cells by immunocytochemistry (890). In the Müller glia, the K_ir2.1 channels are localized in membranes facing neuronal structures, whereas K_ir4.1 channels are mostly associated with perivascular endfeet; it was therefore hypothesized that (in the context of K⁺ siphoning) the former channels mediate K⁺ influx, while the latter mediate K⁺ efflux (890).

Little is known about astroglial expression of K_ir3.x (encoded by KCNJ3/5/6/9 genes) subunits that assemble into G protein-activated K⁺ channels. The mRNA specific for K_ir3.1 and K_ir3.2 was detected in Müller glia (1419), although this was not confirmed by immunocytochemistry (1629), suggesting the absence of functional protein. There is sporadic evidence about expression of K_ir3.1 protein (detected by Western blot) in cultured astrocytes from cortex and spinal cord (1268). In contrast, astroglial expression of ATP-sensitive K⁺ channels assembled from K_ir6.x and SUR1/2 is well documented. The K⁺-ATP channels open when ATP concentration decreases, and hence they are possibly involved in cell protection in conditions of metabolic stress; additionally, K⁺-ATP channels were suggested to regulate gap junctional permeability in astroglial syncytia (1745, 1897). Müller glial cells express mRNA for K_ir6.1 and K_ir6.2 (1419) and are immunopositive for K_ir6.1 and SUR1 (465, 1628, 1629). The GFAP-positive astrocytes from the brain stem express high levels of mRNA specific for K_ir6.1, and much lower levels for SUR1 (837), whereas K_ir6.2 mRNA immunoreactivity has been found in astrocytes in the corpus callosum and in cerebellar white matter (1954). In the adult rat brain, however, only K_ir6.1 (with K_ir6.2 being confined to neurons) were detected by immunocytochemistry in hippocampal, cortical, and cerebellar astrocytes as well as in Bergmann glia; in the latter, the K⁺-ATP currents (isolated as a tolbutamide-sensitive component of whole cell K⁺ conductance) were recorded (1745). The K⁺-ATP currents (defined as diazoxide-activated and tolbutamide-inhibited currents) were also detected in astrocytes from hippocampus and brain stem (237, 1936); the proportion of astrocytes expressing these currents substantially decreased during postnatal development [from 57% at P8−11 to only 8% at P16−19 (237)]. Ultrastructurally, K_ir6.1 channels in the cerebellum are concentrated in perisynaptic and peridendritic processes (1745).

2. Voltage-independent K⁺ channels

The two-pore-domain potassium channels (K_2P) family is encoded by 15 KCNK genes (480). The K_2P channels are open at (and hence contribute to) the resting membrane potential. Single-cell PCR performed on freshly isolated hippocampal astrocytes revealed a somewhat mosaic expression of TREK1/K_2P2.1(KCNK2), TREK2/K_2P10.1(KCNK10), and TWIK1/K_2P1.1(KCNK1) channels, with some cells expressing all three subtypes (1588). Electrophysiological isolation of K_2P current employing quinine and bupivacaine was not exactly straightforward because these agents also inhibit K_ir channels and A-currents. Hence, quinine- and bupivacaine-sensitive currents were recorded in the presence of Ba²⁺ and 4-aminopyridine (4-AP). These currents reversed at E_K and had almost linear current-voltage relationships (1588). The TREK1/K_2P2.1(KCNK2) and TWIK1/K_2P1.1(KCNK1) channels were also localized in hippocampal astrocytes by immunocytochemistry (1955). Acidification was shown to increase astroglial currents mediated by TREK1 channels (1874). Further analysis of cultured cortical astrocytes and astrocytes in hippocampal slices identified TWIK-1/TREK-1 heterodimer (formed by disulfide bridge between cysteine-cysteine residuals of both subunits) as the predominant channel type (756).

Evidence for the contribution of K_2P channels to astroglial resting conductance remains controversial; in hippocampal astrocytes it was reported to be quite considerable [~58% in astrocytes in hippocampal slices (1955); and even ~70% in cultured astroglia (756)]. Conversely, a rather low (~10%) contribution of K_2P channels to astroglial K⁺ permeability was found in freshly isolated hippocampal astrocytes (1588). In hippocampal slices isolated from TWIK-1/TREK-1 single and double knockout mice, neither astrocytic resting membrane potential nor resting conductance were affected (454). Subcellular protein fractionation of hippocampal tissue (using GFAP as a marker) demonstrated predominant retention of TWIK-1 channels in the cytosol with a rather minor (~5%) plasmalemmal presence (1857), further indicating a relatively minor role of these channels in astroglial K⁺ conductance. The TWIK-1 channels were shown to poorly discriminate between cations (308); furthermore, TWIK-1 channels have been shown to be permeable to NH₄⁺ over K⁺ (1029), and hence they may provide for astroglial accumulation of ammonium (1266) released during physiological synaptic transmission. When [K⁺]_o decreases below 3 mM, TWIK-1 channels become permeable for Na⁺ (1030) and, hypothetically, can generate Na⁺ influx into astrocytes even in physiological conditions when [K⁺]_o may transiently fall below 2 mM following bursts of neuronal activity (668).

3. Voltage-gated K⁺ channels, K_v

Voltage-gated K⁺ channels are represented by 12 subfamilies (K_v1 to K_v12) of which delayed rectifying (K_D) and transient (K_A) channels are operative in astroglia. These channels are inactive at the resting membrane potential, and their opening requires cell depolarization.

a) delayed rectifier.

Delayed rectifying K⁺ currents have been recorded from astrocytes in culture and in situ; these currents were monitored from astroglial cells in various regions of the CNS including cortex, hippocampus, cerebellum, and spinal cord (215). The threshold for K_D activation was at about −20 mV, and current amplitudes varied between 2 and 6 nA. Not much is known about the molecular diversity of astroglial delayed rectifier channels. The K_v1.5 (KCNA5) channel immunoreactivity has been identified in primary cultured rat spinal cord astrocytes as well as in astrocytes in adult rat hippocampal and cerebellar slices; immunoreactivity was particularly high in perivascular astrocytes and their endfeet. Treatment of these astrocytes with K_v1.4 antisense mRNA reduced current amplitude by ~50% (1513). Specific antibodies and in situ hybridization also revealed transcripts and proteins of K_v1.4 (KCNA4) in rat spinal cord astrocytes (466). Finally, the K_v11.1/ERG1 (KCNH2) channels were identified in hippocampal astrocytes by immunocytochemistry and characterized electrophysiologically (475).

b) A-type currents.

The rapidly activating and inactivating A-type currents have been recorded from cultured astrocytes and astrocytes in situ; their full activation requires conditioning hyperpolarization to −110 or −120 mV to remove strong steady-state inactivation characteristic for A channels. Their activation threshold lies around −70 to −50 mV, and these currents demonstrate rapid time-dependent inactivation (124, 215). In primary hippocampal astrocytes from rats, pharmacological, RT-PCR, and immunocytochemical analysis showed that 70, 10, and < 5% of total A currents are mediated by K_v4, K_v3, and K_v1 channels, respectively (124).

4. Ca²⁺-dependent K⁺ channels, K_Ca

The Ca²⁺-dependent K⁺ channels (K_Ca), which are structurally similar to voltage-gated K⁺ channels, are represented by three types (1869), designated as channels with big conductance (BK, or K_Ca1.1/KCNMA1), intermediate conductance (IK, or K_Ca3.1/KCNN4), and small conductance (SK, or K_Ca2.1−2.3/KCNN1−KCNN3). The K_Ca channels have a dual gating mechanism being controlled by cytosolic Ca²⁺ concentration as well as by V_m: without Ca²⁺ binding, the channel cannot be activated by membrane depolarization (1869).

Astrocytes express several types of K_Ca channels. In rat cultured cortical and hippocampal astrocytes, two types of unitary K_Ca currents with conductances of 70 pS and 160 pS (reflecting IK and BK channels) were detected; these channels, however, were blocked only by TEA and were insensitive to classical K_Ca blockers such as charybdotoxin, iberiotoxin, or apamin (562). Transcripts for SK (K_Ca 2.3) and IK (K_Ca3.1) channels were identified in mouse cortical astrocytes in acutely isolated slices (1006). The K_Ca3.1 immunoreactivity was confined to astroglial perivascular processes and endfeet; electrophysiology revealed only currents mediated by K_Ca3.1 [as judged by their sensitivity to the selective K_Ca3.1 blocker TRAM-34 and insensitivity to the K_Ca2.3 blocker apamin (1006)]. Similarly, the K_Ca2.3 protein has been immunocytochemically identified in astrocytic processes in the rat supraoptic nucleus (65). The BK channels (K_Ca1.1) proteins were identified by immunocytochemistry in the perivascular astroglial endfeet in hippocampus and cerebellum (1409), whereas patch-clamping of endfeet revealed large-conductance (225 pS) BK-single channel currents (511).

5. Multiple K⁺ channels underlie passive conductance of astroglial cells

Simultaneous expression of several types of K⁺ channels in astroglial cells ensures their passive properties across wide range of membrane potentials: at all levels of V_m, current-to-voltage (I–V) relation of the mature astroglial membrane is linear in vitro and in situ in all species (FIGURE 12) (455, 637, 1937, 1953). Passive membrane conductance of astrocytes is not associated with gap junctional coupling: neither pharmacological inhibition nor genetic deletion of connexins affect I–V linearity (455, 935, 1573, 1838), with the exception of the olfactory bulb (1453). At least three types of K⁺ channels, the inwardly rectifiers, the two-pore domain channels, and possibly Ca²⁺-activated K⁺ channels are responsible for hyperpolarized resting V_m. The relative contribution of each type of K⁺ channel probably varies in different astrocytes; in hippocampus, K_ir4.1 channels account for ~50% of overall K⁺ conductance (1028, 1588). A high resting K⁺ conductance and highly negative V_m define the homeostatic functions of astrocytes (1244) assisting transmembrane movement of ions and providing electrical driving force for membrane transporters.

B. Sodium Channels

1. Voltage-gated sodium channels

Physiological diversity of voltage-gated Na⁺ channels (Na_v) arises from nine isoforms (Na_v1.1 to Na_v1.9) of pore-forming α-subunit (encoded by SCN1A−11A genes) and four isoforms of auxiliary β-subunits encoded by SCN1B−4B genes (225, 297). Sodium channel proteins and sodium currents have been detected in astroglia in vitro and in situ (TABLE 2). The very first description of voltage-activated Na⁺ currents in cultured astrocytes was made in 1985 (166); in these experiments relatively large (~1 nA) tetrodotoxin (TTX)- and saxitoxin (STX)-sensitive Na⁺ currents were recorded. The relatively high densities of Na⁺ channels were somewhat incompatible with the nonexcitable nature of astrocytes, and hence the hypothesis of glial cells as donors of Na⁺ channels for neighboring neuronal elements was considered. Subsequently, Na⁺ currents were characterized in cultured astroglial cells (105, 106, 1647–1649, 1652). These currents were rather substantial varying between 100 and 300 pA in amplitude in the so-called type-1 astrocytes⁴ (105) and reaching 3 nA in type 2 astrocytes from the optic nerve; these latter astrocytes could even be forced to generate an action potential. Two types of Na⁺ currents with different kinetics (fast and slow) were subsequently described in cultured astrocytes from the optic nerve (1649) and hippocampus (1650). In astrocytes from the spinal cord, two types of I_Na distinct in their sensitivity to TTX (TTX sensitive and TTX resistant) were also characterized (1647, 1652).

Sodium currents were also found in more physiological preparations; first they were detected in nonenzymatically (“tissue prints”) isolated astrocytes from the optic nerve (107), and subsequently in astrocytes from hippocampal (1651) and spinal cord (187, 337) slices. In the slice preparations, I_Na were never found in all astrocytes: in the hippocampus (at P5 to P24) only 5 of 40 GFAP-positive cells had relatively small (~300 pA) Na⁺ current, whereas in the spinal cord (at P1 to P19) I_Na was confined to a population of cells with small somata and longish processes, with some faint GFAP staining. No Na⁺ currents were detected in neonatal astrocytes from stratum radiatum in acute hippocampal slices (1951). In contrast, Na⁺ currents were recorded from sulforhodamin101-negative astrocytes in juvenile (P3 to P15) hippocampal slices and were absent in more mature astroglial cells (811). At the molecular level, the predominate type of astroglial Na⁺ channels is represented by Na_v1.5 (which corresponds to the TTX-resistant or cardiac channel type) detected both in vitro and in situ at mRNA and protein levels (185, 186, 1316), although Na_v1.2, Na_v1.3 (189), and Na_v1.6 (1444, 1556) were also identified. Expression of Nav1.6 channels was found to be substantially upregulated in reactive astroglia (1957).

Despite substantial amount of data identifying voltage-gated Na channels in astrocytes from different regions of the brain, the exact localization and mechanisms of activation of these channels by physiological stimuli remain to be elucidated. Indeed, direct electrophysiological recordings of Na⁺ currents from mature astrocytes in situ remain scarce, while recordings from astrocytes in vivo are yet to be obtained. Nevertheless, the Na_v channels may carry several physiological roles, and they seem to contribute to pathological responses of astroglia (see Refs. 188, 1315, 1317 for recent review and literature). Strategic localization of Na_v channels in perisynaptic processes, which can possibly experience depolarization following activation of ionotropic receptors or Na⁺-dependent glutamate transporters, may contribute to generation of local Na⁺ signals (see sect. XB), this may in turn regulate numerous glial homeostatic cascades (881, 1499). It was also suggested that Na_v-mediated influx of Na⁺ is needed for sustained activity of Na⁺/K⁺ pumps, and inhibition of Na_v channels with TTX instigates cell death because of the pump failure (1648).

2. Na⁺_o-regulated Na⁺ channels, Na_x

Astroglia (astrocytes and ependymocytes) of the subfornical organ and organum vasculosum of the lamina terminalis (both are parts of the circumventricular organs surrounding ventricles) express a peculiar type of Na⁺ channel that is activated by an increase in extracellular Na⁺ concentration ([Na⁺]_o) (1235, 1865). These channels are known as Na_x and they were originally cloned from rat astrocytes (558) and from human cardiac muscle (568). The Na_x primary sequence is distinct from the Na_v channel family; likewise, the Na_x is very different from voltage-gated Na⁺ channels in its voltage dependence as well as kinetic and gating mechanism. Most likely, the Na_x channel belongs to a distinct class of membrane channels (1234). The Na_x channels in vitro are activated by an increase in concentration of extracellular Na⁺ above 150 mM; in vivo, in the presence of endothelin-3, which activates ET_B receptors selectively expressed in astroglia, the threshold for Na_x activation is shifted to 140 mM of [Na⁺]_o (706). Peak amplitudes of Na_x currents reached ~1.5 pA/pF. The Na_x channels act as molecular sensors for [Na⁺] in the circulation (see sect. XIIJ).

3. Epithelial sodium channel, ENaC

Epithelial Na⁺ channels constitute, together with acid-sensitive ion channels, an ENaC/degenerin superfamily; the ENaCs are non-voltage-gated, amiloride-sensitive Na⁺ channels encoded by four genes classified (in humans) as SCNN1A, SCNN1B, SCNN1G, and SCNN1D (643). The ENaCs are expressed in the CNS, and they have been being identified in neurons, ependymal cells of the choroid plexus, endothelial cells, and astrocytes (34, 1117, 1834). Strong immunoreactivity for γ-subunit of ENaC was detected in GFAP-positive astrocytes in circumventricular organs, white matter, and pia mater (1116); incidentally, no α- or β- ENaC immunoreactivity was detected in astrocytes, possibly indicating a unique homotrimeric channel assembly. Specific expression of ENaC in astrocytes from circumventricular organs may indicate their role in regulation of systemic Na⁺ homeostasis.

C. Calcium Channels

1. Voltage-gated Ca²⁺ channels

Three main families of voltage-gated calcium channels (VGCCs) are present in the nervous system; they are classified as Ca_v1.1–1.4 or L-type Ca²⁺ channels; Ca_v2.1–2.3 or P/Q/N/R-type channels, and Ca_v3.1–3.3 or T-type Ca²⁺ channels (296). Evidence for functional expression of VGCCs in astrocytes in living brain remains, however, controversial. Voltage-activated calcium currents were recorded in cultured astroglia (TABLE 2); the L-type currents were first contemplated from the recordings of Ba²⁺-dependent action potentials that can be generated after inhibition of K⁺ conductances in the presence of elevated intracellular cAMP (1038). Subsequently Ba²⁺ currents were detected in voltage-clamp experiments on cultured cortical astrocytes; these currents were blocked by Ca²⁺ channel blockers Co²⁺, Cd²⁺, and nifedipine and were much enhanced (or even induced) by norepinephrine and dibutyryl-cAMP (104, 333, 1041). Depolarization-evoked elevations in [Ca²⁺]_i which were inhibited by Co²⁺, verapamil, and nifedipine, while being potentiated by BAY K 8644, have been monitored in primary cultured and freshly isolated astrocytes (460, 1040).

Expression of various VGCCs in astrocytes has been documented at the molecular level. The Ca_v1.2 and Ca_v1.3 channels were found in the transcriptome of rodent cortical astrocytes (275, 1945). In neonatal cultured cortical astrocytes, expression of α1B (N-type), α1C (L-type), α1D (L-type), α1E (R-type), and α1G (T-type) VGCCs was detected at mRNA and protein levels (948). Expression of transcripts for α1C (L-type), α1B (N-type), and α1E (R-type) was further corroborated by electrophysiological recordings of ion currents with relevant pharmacology (374). The Ca_v2.2 N-type and Ca_v2.3 R-type VGCCs were immunocytochemically detected in pituicytes in situ, whereas the same cells in vitro expressed (at mRNA and protein levels) Ca_v1.2 L-type, Ca_v2.1 P/Q-type, Ca_v2.2 N-type, Ca_v2.3 R-type, and Ca_v3.1 T-type subunits (1851).

Much less evidence for functional Ca²⁺ channels originate from experiments in situ. Occasional recordings of T- and L-type Ca²⁺ currents from cells in young hippocampal slices (19) were, most likely, obtained from NG2 glia, which, in contrast to astrocytes, has more “excitable” phenotype (1228). Spontaneous [Ca²⁺]_i oscillations in astrocytes from the ventrobasal thalamus slices were somewhat potentiated by Bay K 8644 (1331) and were markedly blocked by 1 μM nifedipine (1332), these being indirect evidence for functional VGCCs. Another indirect evidence for functional astroglial VGCCs derives from the analysis of the presynaptic plasticity; in these experiments intracellular perfusion of astrocytes with membrane-impermeable VGCC inhibitor D890 had certain effects on synaptic strength (972). At the same time, experiments specifically dedicated to detecting voltage-gated Ca²⁺ currents or [Ca²⁺]_i transients in slices failed to find them (289, 1848).

There are however some indications of upregulation of astrocytic VGCCs expression and their specific role in pathological conditions. For example, expression of L-type Ca²⁺ channels increased in reactive astrocytes following mechanic or ischemic brain injury (1876). Activation of astrocytes by LPS induced upregulation of Ca_v1.2 channels, whereas genetic deletion of these channels prevented astrogliotic response (314). Similarly, expression of Ca_v1.2 channels in astrocytes in vitro and in the brain is substantially upregulated by exposure to ammonium (in culture) or in animal models of hyperammonemia (1852).

2. Ca²⁺ release activated Ca²⁺ channels of Orai family

The Ca²⁺-release activated Ca²⁺ current (I_CRAC) is the principal component of the store-operated Ca²⁺ entry (SOCE) in the majority of nonexcitable cells (1320). The molecular substrate of I_CRAC is represented by a family of three Orai1,2,3⁵ proteins, gating of which is controlled by the stromal interacting molecules, STIM1 and STIM2 localized in the membrane of the endoplasmic reticulum (1636). The nature of SOCE in astrocytes is complex with a documented role for TRPC1 channels (see below and sect. XA), while the astroglial expression and function of ORAI channels is yet to be fully characterized.

Orai proteins have been found only in astrocytes in vitro, in primary cultures. The Orai1 and STIM1 were detected by immunocytochemistry in rat cortical astroglia; overexpression of Orai1 increased the amplitude of SOCE in response to activation of PAR receptors with thrombin, whereas it was decreased by siRNA knockout (1150). In hippocampal rat astrocytes, the mRNA expression profile was different: the predominant isoform was Orai3; Orai 2 expression was six times lower, whereas Orai1 was not detected at all (1492). At the same time siRNA silencing of ORAI1 reduced SOCE in primary cultured cortical astrocytes. The only electrophysiological recording of I_CRAC was hitherto obtained from acutely dissociated Müller cells; the contribution of Orai channels was deduced from sensitivity to the specific blocker Synta 66 (1139).

3. Intracellular Ca²⁺ channels

Intracellular Ca²⁺ channels are represented by three distinct families of 1) inositol trisphosphate receptors, or InsP₃Rs (1052); 2) Ca²⁺-gated Ca²⁺ release channels, generally known as ryanodine receptors, or RyRs (1931); and 3) two-pore channels or TPCs activated by nicotinic acid adenine dinucleotide phosphate or NAADP (1339). The first two types of intracellular channels dwell in the endomembrane of the endoplasmic reticulum and mediate Ca²⁺ mobilization from the latter (1805), whereas TPCs are localized to the acidic Ca²⁺ stores (278).

The InsP₃ receptor family comprises three members: the InsP₃R1, InsP₃R2, and InsP₃R3, which all are activated by InsP₃ and modulated by ionized Ca²⁺. The InsP₃R1 shows specific “bell-shaped” regulation by [Ca²⁺]_i: channel open probability increases with an increase in [Ca²⁺]_i between the resting level (~50−100 nM) and ~1 μM, and inhibited at higher [Ca²⁺]_i. Activity of InsP₃R2 and InsP₃R3 linearly increases with an increase in [Ca²⁺]_i. The InsP₃R2 seems to be a predominant intracellular Ca²⁺ channel in astrocytes (678, 718, 1598, 1602, 1817). Genetic deletion of this receptor completely abolishes or strongly reduces endoplasmic reticulum Ca²⁺ release and Ca²⁺ signaling in hippocampal and cortical astrocytes (820, 1381). This, however, is not a universal finding, and several experiments in vivo demonstrated residual Ca²⁺ signals in astrocytes from InsP₃ R2^−/− mice (e.g., Ref. 663; for details, see sect. XA). These may indicate functional expression of other InsP₃ receptors, and indeed, the InsP₃R1 receptors were detected in hippocampal and cortical astrocytes at mRNA and protein levels (607, 1492). Direct comparison of [Ca²⁺]_i dynamics in astrocytes from InsP₃R2 and InsP₃R2/3 knockouts revealed functional Ca²⁺ release mediated by both InsP₃R1 and InsP₃R2 receptors (1604).

The RyRs are also represented by three types (633): the RyR1 (also known as “skeletal”), RyR2 (or “heart”), and RyR3 (or “brain”). The RyRs are activated by cytosolic Ca²⁺ and therefore act as an amplifier of Ca²⁺ signals; they also can be stimulated by the naturally occurring intracellular second messenger cyclic ADP-ribose. Expression of ryanodine receptors in cultured astrocytes was visualized by using fluorescent ryanodine (1819) and with immunocytochemistry (1623). In primary cultured and acutely dissociated cerebellar astrocytes, RyR3 mRNA and protein have been detected (1089), whereas immunocytochemistry revealed RyR3 in hippocampal astrocytes in situ (678). The functional role for RyRs in astrocytes remains unclear; caffeine-induced Ca²⁺ signals (presumably associated with RyR-mediated Ca²⁺ release) were observed in astroglial cultures (1623) and in astrocytes from thalamus (1331), but not from hippocampus (118). In primary astroglial cultures another agonist of RyRs, chloro-m-cresol (with EC₅₀ 1.5 mM) also triggered a [Ca²⁺]_i increase (1089).

Not much is known about expression of TPCs in astroglia; intracellular injection of 100 nM NAADP triggered Ca²⁺ release from acidic organelles in cultured rat cortical astrocytes. These NAADP-induced Ca²⁺ transients were sensitive to TPC antagonist NED-19 and bafilomycin 1A, known to disrupt endolysosomal system (1370). The RT-PCR demonstrated the expression of mRNA for TPC1 and TPC2; when overexpressed both channel subtypes were colocalized with endosomal and lysosomal organelles. Overexpression of TPC1 and TPC2 potentiated NAADP-induced [Ca²⁺]_i transients and facilitated autophagy (1370). The NAADP receptors were also visualized in cortical astrocytes in vitro with fluorescent probe NED-19, which revealed punctate staining with a high degree of colocalization with lysotraker (lysosomal fluorescent marker). Treatment of these cells with membrane-permeable NAADP-AM evoked [Ca²⁺]_i transients (99). Furthermore, TPCs were also suggested to contribute to astrocytic Ca²⁺ signaling induced by ATP and endothelin-1 (99).

4. Transient receptor potential or TRP channels

The extended family of transient receptor potential⁶ (TRP) channels with 27 members present in humans (1221), contributes to numerous and widely diverse physiological functions, being particularly important for all types of sensing including thermal sensation, nociception, chemoception, equilibrioception, and taste sensing (1219). The TRP channels (which are subclassified into 6 groups) are cation channels permeable to Na⁺, Ca²⁺, and K⁺ with great heterogeneity of permeation properties (1291). In the CNS, all TRPs members are expressed at different levels, with particularly high expression of TRPV, TRPC, and TRPM channels (1218). Several types of TRP channels are operational in astroglia (1816).

a) trpa1 channels.

The TRPA1 (where “A” stands for ankyrin) channels have high unitary conductance (~110 pS) and substantial Ca²⁺ permeability [P_Ca/P_monovalent ~5.9−7.9, with fractional Ca²⁺ current up to 23% (1221)]. These channels are known to be activated by noxious cold (below 17°C), by pungent substances derived from plants, by growth factors (via G protein-coupled receptors), and by proinflammatory agents (1220, 1221). The TRPA1 channels were found in somata and processes of astrocytes in the brain stem, in the rat trigeminal caudal nucleus using immunogold electron microscopy (965). Transcripts for TRPA1 were also detected in hippocampal astrocytes (1608). Functional expression of TRPA1 was monitored in a subpopulation of hippocampal astrocytes cocultured with neurons and in astrocytes in acute slices (1611). The operation of TRPA1 was deduced from monitoring spontaneous near-membrane (“spotty”) Ca²⁺ transients, which were suppressed by Gd³⁺, La³⁺, broad-spectrum TRP channel antagonist HC 030031, and following treatment with anti-TRPA1 siRNA. At the same time, TRPA1 specific agonist allyl isothiocyanate (AITC) potentiated near-membrane [Ca²⁺]_i transients and induced whole cell currents (1611). Inhibition of TRPA1 channels with Gd³⁺, La³⁺, and HC 030031 resulted in a significant (from ~120 to ~50 nM) decrease in basal [Ca²⁺]_i, this decrease was completely absent in TRPA1^−/− mice (1608). Cytosolic [Ca²⁺]_i dynamics controlled by TRPA1 was suggested to regulate astroglial uptake of GABA (1611) and release of d-serine (1608).

b) trpc channels.

The TRPC (where “C” stands for “canonical”) channels were detected in freshly isolated and in primary cultured astrocytes. Embryonic astrocytes in culture expressed mRNA encoding TRPC1 to TRPC6 proteins; these TRP channels were implicated in the generation of [Ca²⁺]_i oscillations induced by diacylglycerol analog oleyl-acetyl-glycerol, stimulation of metabotropic glutamate receptors or depletion of endoplasmic reticulum Ca²⁺ store (606, 1393). Transcripts for TRPC1, 2, 3, 4, and 6 were detected in astrocytes cultured from the spinal cord (1133). At the protein level, TRPC1, TRPC4, TRPC5, and TRPC6 were identified in cultured and freshly isolated embryonic mouse astrocytes (163). In the same embryonic cultured astrocytes, TRPC1 channels coimmunoprecipitated with InsP₃ receptors and SERCA2b endoplasmic reticulum Ca²⁺-ATPases (587), indicating intimate functional relations between the endoplasmic reticulum and plasmalemmal TRPC1-containing channels, compatible with the role of TRPC channels in astroglial store-operated Ca²⁺ entry (see sect. XA). Similar coimmunoprecipitation of TRPC1 channels, InsP₃ receptors, and Homer proteins was found in cortical astrocytes cultured from 3- to 5-day-old rats (1868), whereas colocalization of TRPC4 channels with ZO-1 scaffolding proteins was detected in cultured fetal human astrocytes (1644). In primary astrocytes cultured from visual cortices of newborn rats or freshly isolated from the same region of 1-, 8-, and 55-days old rats, expression of TRPC1, TRPC4, and TRPC5 channels was detected in Western blots and their cellular localization was mapped with immune labeling showing that TRPC1 channels were predominantly localized to the plasma membrane (1055). Expression of TRPC channels increased with age: at 1 day of age 47, 7, and 70% of astrocytes studied expressed TRPC1, TRPC4, and TRPC5 proteins, respectively, whereas at 55 days of age, all astrocytes expressed all three isoforms (1055).

Coassembly of TRPC1, which is a channel forming subunit, with ancillary TRPC4 and TRPC5 subunits forms a functional cationic channel (713, 1684) with substantial (P_Ca/P_monovalent varying between 1 and 9) Ca²⁺ permeability (1291). Activation of astroglial TRPC channels contributes to Ca²⁺ signals evoked by purinergic and glutamatergic (1055), as well as by mechanical (1055, 1461, 1462) stimulation. Treatment of astrocytes with an anti-TRPC blocking antibody substantially reduced [Ca²⁺]_i transients (1055, 1461, 1816). In primary cultured cortical astrocytes TRPC channels were claimed to be activated by hyposmotic shock, with ensuing Ca²⁺ signaling reportedly instigating trafficking of aquaporin-1 water channels to the plasma membrane, hence increasing water transport (352). The Na⁺/Ca²⁺ selectivity of astrocytic TRPC channels seem to be controlled independently: treatment with anti-TRPC1 antibodies increased Na⁺ influx while decreasing Ca²⁺ entry (1461). The TRPC6 channels have been shown to mediate Ca²⁺ influx into embryonic astrocytes acutely treated with interleukin-1β (163).

c) trpv channels.

The family of TRPV (“V” for vanilloid) channels incorporates six members, which are activated by various chemical (e.g., pungent agents such as capsaicin and allyl isothiocyanate), thermal and noxious stimuli; TRPV4 channels are also sensitive to osmotic pressure. All TRPV channels are Ca²⁺ permeable with P_Ca/P_monovalent between 1 and 10 for TRPV1–4, and P_Ca/P_monovalent >100 for TRPV5,6 (1221, 1291).

Immunoelectron microscopy and immunocytochemistry revealed wide expression of TRPV1 channels in rodent astrocytes in several brain regions and in the spinal cord, with preferential localization in the astrocytic endfeet (442, 1755). In cultured astrocytes, TRPV1 channels carried cationic currents activated by extracellular acidification (746). The TRPV1 channels may contribute to regulation or initiation of astrogliosis in the spinal cord (319), whereas activation of TRPV1 apparently contributes to Ca²⁺-dependent rearrangement of the cytoskeleton (707). The TRPV1 channels were also suggested to regulate production of ciliary neurotrophic factor with neuroprotective features (1192). Expression of TRPV1 channels was detected at transcript and protein levels in astrocytes in circumventricular organs including the organum vasculosum of the lamina terminalis, subfornical organ, and area postrema, which all represent chemosensitive areas of the brain. These channels were concentrated in the thick perivascular astroglial processes and were suggested to contribute to chemosensing. The blood-borne signals indeed may be recognized by astroglial processes: blood infusion of the TRPV1-selective agonist resiniferatoxin triggered expression of immediate early gene c-Fos in astrocytes from circumventricular organs (1059).

Another member of the TRPV family, TRPV4 channel, was detected in cortical and hippocampal astrocytes (86, 135, 265, 994); the TRPV4 channels also showed the tendency to be concentrated mainly in the processes. In primary cortical astrocytes, TRPV4 channels can be activated by hyposmotic stress and by cell swelling; their activation triggers cytosolic Ca²⁺ signals sensitive to the TRPV inhibitor ruthenium red, while specific agonist 4α-phorbol 12,13-didecanoate induces cationic currents (135). Similarly, TRPV4-mediated currents and [Ca²⁺]_i transients (sensitive to ruthenium red and the TRPV4 selective inhibitor RN1734) were recorded from astrocytes in hippocampal slices (265). It has been postulated that TRPV4 acting in concert with AQP-4 contributes to the development of regulatory volume decrease induced by hyposmotic shock (136, 137).

D. Hyperpolarization-Activated Cyclic Nucleotide-Gated (HCN) Channels

The Na⁺/K⁺ HCN channels (HCN1−4, with Na⁺/K⁺ permeability ratio of 1: 4) have been detected in healthy astrocytes in situ; however, their expression increased significantly after ischemia (719, 1520). At transcript level, all four mRNA for HCN 1−4 were identified, whereas only HCN1−3 proteins have been detected in reactive post-stroke astrocytes; in these cells, the whole cell HNC-mediated currents have been characterized (719, 1520).

E. Acid-Sensitive Ion Channels, or ASICs

The acid-sensitive ion channels are rather ubiquitously expressed in the nervous system; ASIC1, ASIC2a, and ASIC3 were localized with immunocytochemistry in cultured rat cortical astrocytes. All three isoforms were predominantly localized in the nuclei and were not associated with membrane currents (746). In freshly isolated mouse hippocampal astrocytes, however, single-cell RT-PCR revealed transcripts neither for ASICs, nor for TRPV1 channels (1874). Expression of ASIC1a was reported to be increased in reactive astrocytes in the context of chronic epilepsy, and their activity may contribute to seizure generation (1911). To conclude, the functional importance of ASICs for astroglial physiology in vivo remains unknown.

F. Anion Channels

The whole cell outwardly rectifying voltage-dependent Cl⁻ currents were initially observed in cultured astrocytes depolarized to potentials more positive than −50 mV (166, 602). These findings, however, were not replicated in other studies, which demonstrated almost complete absence of anion conductance in astroglia at rest (868). Changes of astrocytic morphology (due to osmotic stress or treatment with cAMP, which induced cell stellation), however, activated the dormant Cl⁻ conductance (945). Apparently, Cl⁻ channels in astroglia are controlled by direct link to cytoskeletal actin (946). The actin connection may also explain earlier findings that single-channel Cl⁻ currents could be detected only in excised patches (106).

Several types of chloride (anion) channels [represented by the ClC channel family, by cystic fibrosis transmembrane conductance regulator (CFTR) channels, by voltage-dependent anion-selective channels (VDAC), by Ca²⁺-dependent Cl⁻ channels, and by volume-regulated anion channels (VRAC)] have been identified in astroglia (868). In cultured astrocytes, ClC-1, -2, and -3 channels were detected at transcript and protein levels (1326, 1944), while ClC-mediated currents were characterized in astrocytes in vitro as well as in hippocampal slices (1054, 1326). The immunoreactivity for ClC-2 was also found in Bergmann glia in cerebellum (192). Of note, ClC-2 channels were clustered in astroglial processes in the vicinity of GABAergic synapses, which prompted the hypothesis of astroglia-dependent regulation of intracleft Cl⁻ concentration and hence of GABAergic transmission (1618). The ClC channels are inwardly rectifying (i.e., promoting Cl⁻ efflux at potentials below E_Cl) and hence they mediate Cl⁻ secretion from the astrocyte. Furthermore, ClC-2 channel open probability is high at astroglial resting potential and is modulated by change in the cell volume. Thus these channels facilitate release of Cl⁻ in response to cell swelling (456, 1489).

The VRAC are fundamental for cell volume control and contribute to regulation of cell proliferation, cell motility, and apoptosis. Molecular identity of VRAC was discovered recently after identification of a plasma membrane spanning protein leucine-rich repeat containing 8A (LRRC8A) that was dubbed SWELL1 (1416, 1833). In astrocytes, VRAC are specifically important for regulatory volume decrease that invariably follows astroglial swelling due to osmotic disbalance (868). The VRAC channel currents induced by osmotic stress have been characterized in astrocytes in vitro (5, 138, 1326). The VRAC channels were also reported to form a conduit for excitatory amino acids; this was initially suggested by monitoring release of radioalbeled taurine, glutamate, and aspartate (867, 1333). Glutamate release, mediated by VRACs, has been subsequently observed in astroglial primary cultures (506, 991) and in astrocytes in situ (111, 1717). The activity of VRAC is claimed to be regulated by cytosolic Na⁺: increase in [Na⁺]_i suppressed VRAC activity (1122). The LRRC8A/SWELL1 are critical for swelling-activated release of glutamate and taurine from cultured rat astrocytes (757).

Astrocytes also express an anion channel of Bestrophin (Best) family. The Best gene, responsible for a dominantly inherited, juvenile-onset form of macular degeneration called Best vitelliform macular dystrophy, encodes a Ca²⁺-activated anion channel (655). In the brain, Best1 channels were reported to be largely expressed in astrocytes (1255, 1324), being concentrated in perisynaptic regions (1322, 1893). These Best1 channels were activated by relatively moderate increases in [Ca²⁺]_i at V_m −70 mV (i.e., close to astroglial resting potential), while Best1-mediated Cl⁻ currents could be blocked by broad-spectrum antagonists of anion channels such as niflumic acid, NPPB, and flufenamic acid (1324). Astroglial Best1 channels are suggested to form a conduit for secretion of glutamate and GABA (see sect. XIE).

G. Aquaporins

Astrocytes express three members of aquaporin family, the AQP1, AQP4, and AQP9 (84, 1546), although AQP4 has by far the largest presence (1184). The AQP4 is composed from eight intramembrane helical segments M1−M8 and two translation initiation sites designated as Met1 and Met23, which in turn define two AQP4 isoforms, M1 and M23 (1314). Hypothalamic tanycytes express solely AQP9 (474). The highest expression of AQP4 has been indentified in the cerebellum with lower AQP4 presence in the hippocampus, diencephalons, and cortex (750). Astroglial AQP4 channels are concentrated in the perivascular and subpial endfeet, where their density is at least 10 times higher than in other parts of the cell. In the endfoot membrane, the AQP4 channels are assembled into orthogonal arrays of particles (also known as square arrays) that are observed under electron microscopy of freeze-fracture replicas (411, 1439, 1889). High densities of AQP4 in the endfeet mediate direct interactions with pericytes (618), with endothelial cells and with extracellular matrix with a specific role for proteoglycan agrin (280).

Physiological roles of AQP4 are subject of intense investigations, as it seems to contribute to a wide variety of CNS functions. Rather surprisingly, at the time of generation, the AQP4 knockout mice did not demonstrate any obvious phenotype (1031). Subsequently, the gross alterations of nervous function have been revealed as AQP4^−/− mice displayed seriously impaired olfaction (1014), and even more seriously affected hearing: nearly all mice lacking AQP4 are deaf (1014, 1111). More detailed studies revealed a sevenfold decrease in astroglial water permeability (1641), deficient K⁺ buffering, compromised regulation of extracellular space (181), impaired hippocampal LTP/LTD without any changes in background synaptic transmission (1557, 1632), and altered spatial memory (1632, 1941). The AQP4 channels were found to be linked to astroglial Ca²⁺ signaling induced by osmotic stress (1749). Astroglial AQP4 channels also seem critically important for operation of the CNS glymphatic system (see sect. XIIH). Finally, genetic deletion of AQP4 significantly modifies various neuropathologies (1184).

H. Connexons

Connexons are fundamental for integrating astrocytes into interconnected syncytia. They form gap junctional channels spanning through the membranes of adjacent cells. Gap junctional channels are composed from two precisely aligned connexons from the two adjacent coupled cells (414, 1524). Assemblies of 100s of connexons into large cristalline-like structures make gap junction plaques between adjacent cells. When connexons fail to align with a congruent proteins from a neighboring cell, they may act as a stand-alone gated pores, known as hemichannels (490).

Connexons are assembled from six subunits, known as connexins (Cx), that represent a multigene family (with 21 members in human genome) which differ in molecular mass between 26 and 62 kDa; this underlies the nomenclature (e.g., Cx26 or Cx43). All connexins are similar in topology with four membrane-spanning domains. Functional connexons have a high conductance reflecting the pore with diameter between 6.5 and 15 Å; this pore is permeable not only for ions but also for hydrophilic molecules with molecular mass <1 kDa, including second messengers, nucleotides, glucose, various metabolites (397, 650, 1714), and even siRNA (1783). Of note, connexins have a high turnover rate with a half-life of several hours; degradation of connexins involves internalization into a specific double-membrane vacuoles known as annular junctions or connexosomes (888). Generally, connexons may assemble into homo- or heteromers, and gap junction channels can be made from identical connexons (making homotypic channel) or different (heterotypic channels). When the channel is composed from heteromeric connexons, it is classified as heteromeric. Finally, the gap junctions can connect the same type of cells (homocellular gap junction) or different type of cells (heterocellular gap junction). In astrocytes, connexons may also connect parts of the same cell through reflexive gap junctions (1890); these may create signaling and diffusional shortcuts in the highly arborized astroglia.

Astroglial expression of connexins is complex, and hence, astrocytes may contain multiple types of gap junctional channels. Astrocyte-astrocyte homocellular gap junctions are composed from Cx26, Cx30, and Cx43 (570, 923, 1187), of which Cx43 is the most abundant (1186). It seems that Cx30 and Cx43 may assemble as both homo- and heteromers, whereas Cx26 seems to make homomers only (28). Expression of Cx43 is ubiquitous, while the highest expression of Cx30 has been determined in thalamus and leptomeninges. Substantially lower expression of Cx30 is found in cortex and hippocampus, and no expression at all was detected in white matter (1186, 1640). Expression of Cx26 is restricted to astrocytes in subcortical areas, such as hypothalamus as well as reticular thalamic and the subtalamic nuclei (1188). Astrocyte-oligodendrocyte heterocellular gap junctions are composed from heterotypic channels which, in vitro, can be represented by four operational complexes: Cx47/Cx43, Cx47/Cx30, Cx32/Cx30, or Cx32/Cx26 (1049), although in situ the Cx32/Cx30, Cx47/Cx43 appear to dominate functional intercellular connectivity (28, 1282). Pairing Cx43 and Cx32 does not produce functional gap junctional channel (1878). There are some sporadic reports of astrocyte-neuronal heterocellular contacts (31, 1182, 1299), which are probably limited to developing brain and could not be considered as a significant pathway for glial-neuronal communication.

In cultured astrocytes obtained from Cx43 knockouts, expression of Cx26, Cx30, Cx40, Cx45, and Cx46 was detected (412, 1741), whereas Cx30 expression was almost doubled in Cx43-deficient animals (1742). These changes in expression explain why genetic deletion of Cx43 results only in partial (~50%) suppression of gap junctional astroglial connectivity (1742). Expression of connexins may be regulated by neuronal factors: coculturing neurons with astrocytes upregulates Cx43 and initiates Cx30 expression in the latter (899). Connexon-based gap junction channels are voltage dependent (being sensitive to trans-junctional potential, although some channels are also regulated by V_m); this voltage gating depends on the subunit composition (691). Conductance of Cx43-formed gap junctional channel in cultured astrocytes was around 50−60 pS (413, 570). Biophysical properties of connexons are regulated by multiple factors including pH, [Ca²⁺]_i, or phosphorylation state, which is controlled by protein kinases A, C, and G and by mitogen-activated protein kinases, etc. (473, 691). Connexons are also positively modulated by intracellular spermine and spermidine (134).

Unpaired connexons, the hemichannels, have been identified in astrocytes in vitro and in vivo; all three connexons expressed in astrocytes (Cx26, Cx30, and Cx43) can operate as hemichannels (572). As a rule, these hemichannels are closed in healthy resting cells, but can be activated by low external calcium concentration, by substantial depolarization, by some specific intracellular Ca²⁺ signals, or by exposure to proinflammatory agents (1273, 1274). The hemichannels are considered to provide conduits for astroglial secretion of various substances, including neurotransmitters and neuromodulators (see sect. XIC). Depolarization activates unpaired connexons, which generate voltage- and time-dependent transmembrane currents, biophysical parameters of which depend on channel subunit composition. In physiological conditions (i.e., at normal concentration of divalent cations), open probability of Cx channels is very low at resting membrane potential. Mouse Cx30 channels show prominent voltage dependence (1784). This is not the case for Cx43 (357), while human Cx26 hemichannels show bell-shaped voltage dependence (593). The unitary conductance for Cx26 hemichannels is ~320 pS (593), for Cx32 hemichannels ~90 pS (1535), and for Cx43 hemichannels ~220 pS (1457). Of note, the conductance of Cx43 gap junctional channel is about four times lower than that of unpaired one (413, 570), although theoretically it should be only half. This probably indicates some additional regulation of the channel in gap junctional configuration.

I. Pannexons

Pannexons have no homology with connexons and are classified as gap junction proteins because of some homology with innexons; pannexons never form intercellular channels and hence exist as stand-alone plasmalemmal gated pores (490).⁷ At the same time the overall topology (four transmembrane domains) and pharmacological properties of pannexins and connexins are rather similar, suggesting some common functional evolution and gating mechanisms (1002). Depending of the stimulation mode, pannexon opens as a large-conductance (~500 pS) pore or operates as ~50 pS anion channel (1032, 1856). Pannexon channels are assembled from six subunits, of which three subtypes pannexin1, -2, and -3 (or Panx1, -2,-3) are known. Transcripts for Panx1 were identified in astroglia in vitro and in situ (749, 1443), and Panx1 currents were electrophysiologically characterized in primary cortical astrocytes (758). Astroglial pannexons were activated by voltage and by stimulation of P2X₇ receptors; they were blocked by gap junctions blockers carbenoxolone (CBF) and mefloquine (MFQ) and were permeable to fluorescent tracer YoPro (758). The Panx1 channels in astrocytes could also be activated by an elevation of extracellular K⁺ concentration, which shifts activation potential of pannexons to more hyperpolarized values (1620). Astroglial pannexons were reported to be activated by fibroblast growth factor 1 (554). Pannexons, in high conductance state, can be permeable to molecules with molecular mass up to 1.5 kDa and hence may contribute to astroglial secretion (see sect. XIC). Panx1 in particular is considered as transmembrane conduit for ATP (375). The P2X₇-pannexon complex was also implicated in Ca²⁺-independent release of d-serine from cultured astrocytes (1306). The evidence for functional pannexons and their physiological role in astrocytes in vivo remain, at best, circumstantial.

J. Recapitulation

Astrocytes express a remarkable variety of ion channels, including several types of voltage-gated channels usually associated with excitable cells. Whether some of these channels (most notably voltage-gated Na⁺ and Ca²⁺ channels) contribute to astroglial physiology, or their expression is induced in special circumstances, remains to be clarified. The backbone of astroglial membrane permeability is formed by K⁺ channels and connexons. Connexons underlie functional integration of astrocytes into syncytia, fundamental for intercellular signaling and long-range diffusion of metabolites and second messengers. Several types of K⁺ channels underlie ohmic conductance at the wide range of membrane potentials essentially defining the passive electrophysiological signature of astrocyte. Large K⁺ permeability confers stability on astroglial membrane potential and hence secures many homeostatic functions; the variable complement of other channels allows for versatility and flexibility, hence contributing to astroglial adaptive capabilities.

VIII. RECEPTORS

Astrocytes are capable of expressing virtually any type of receptor found in the CNS (TABLE 3), which allows astroglia to perceive the neurochemical landscape of the nervous tissue. The first hints for astroglial expression of functional neurotransmitter receptors came from intracellular microelectrode recordings from pericruciate cortical cells of anesthetized cats. The cells were impaled blindly, and neurons and glia were distinguished by their excitability. Iontophoretic injections of GABA or acetylcholine (ACh) evoked glial depolarization, which was considered to reflect modulation of membrane ion pumps (908). Glial depolarization in response to GABA, glycine, β-alanine, and taurine was also observed in organotypic preparations of medulla oblongata, pons, and spinal cord and was considered to reflect K⁺ release from neurons (731). Direct electrophysiological recordings from cultured astrocytes, devoid of neuronal contamination, demonstrated functional expression of glutamate, GABA, and glycine receptors (220, 853). Subsequent experiments on astroglia in vitro demonstrated that astrocytes indeed express multiple receptor types.

Table 3.

Astroglial receptors to neurotransmitters and neuromodulators

Receptor Type/Subunits	Properties, Function, and Localization	Reference Nos.
Ionotropic receptors
AMPA glutamate receptors: GluA1, GluA2, GluA3, GluA4	Detected in hippocampus, cortex, cerebellum, white matter, Bergmann glial cells, immature astrocytes. Cationic Na+/K+ channels or Na+/K+/Ca2+ channels. Receptors lacking GluA2 subunit (predominantly localized in Bergmann glial cells) have moderate Ca2+ permeability (PCa/Pmonovalent ~1). Activation triggers cell depolarization and Ca2+ influx.	777, 932, 935, 1169, 1589
Glutamate NMDA receptors: GluN1, GluN2C, GluN2D, GluN3	Detected at mRNA and protein level in cortex, hippocampus, spinal cord, amygdala, locus coeruleus, and retinal Müller glia. NMDA receptor-mediated currents were characterized in astrocytes from cortex, spinal cord, and subpopulation of hippocampal astrocytes. Na+/K+/Ca2+ channels. Astroglial receptors display weak Mg2+ block and intermediate Ca2+ permeability (PCa/Pmonovalent ~3). Activation triggers inward current and Ca2+ entry.	52, 354, 464, 935, 972, 1304, 1305, 1562, 1593, 1965
P2X purinoceptors: P2X1/5, P2X7	At mRNA and protein levels, all seven subunits of P2X receptors were identified (in various combinations) in astrocytes from cortex, hippocampus, cerebellum, spinal cord, brain stem and retina. P2X1/5-mediated currents were recorded in cortical astrocytes; functional P2X7 receptors were found in cortex, hippocampus, and retina. Na+/K+/Ca2+ channels. Ca2+ permeability of P2X1/5 receptors is: PCa/Pmonovalent ~2. Ca2+ permeability of P2X7 receptors depends on the pore formation and can be very high (PCa/Pmonovalent >10). Activation triggers cationic current and Ca2+ entry.	435, 516, 520, 541, 763, 776, 829, 937, 1001, 1305, 1313
GABAA receptors: α2, γ1	Detected at mRNA, protein, and functional levels in hippocampus, cortex, cerebellum, optic nerve, spinal cord, and pituitary gland. Cl– channel. Activation triggers Cl− efflux and cell depolarization. Possibly play fundamental role in regulation of [Cl−] in the synaptic cleft to maintain inhibitory transmission.	339, 468, 773, 853, 854, 857, 1042, 1167, 1336, 1800
Glycine receptors: α1, β	Mainly present in the spinal cord, where they were detected at transcript, protein, and functional levels. Cl– channel. Activation triggers Cl− efflux and cell depolarization.	874, 1336
Nicotinic cholinoreceptors: α3, α4, α7; β2, β4	Detected in astrocytes in vitro and in hippocampal astrocytes in situ. Na+/K+/Ca2+ channels. Receptors containing α7 subunit display high Ca2+ permeability (PCa/Pmonovalent ~6).	601, 1257, 1597, 1726, 1737
Metabotropic receptors
Glutamate receptors: mGluR3, mGluR5	Detected in astrocytes throughout the CNS. The most abundant is mGluR3 receptor subtype which inhibits adenylyl cyclase; mGluR5 receptors associated with Ca2+ signaling are downregulated in first postnatal weeks.	277, 720, 879, 947, 1308, 1400, 1401, 1702
GABAB receptors: GABAB1a, GABAB1b, GABAB2	Detected in hippocampus, cortex, and spinal cord. Linked to PLC through Gi/o proteins; activation of GABAB receptors triggers Ca2+ release from the ER with associated Ca2+ signals and [Ca2+]i oscillations in vitro, in situ and in vivo.	822, 1008, 1066, 1800
Adenosine receptors: A1, A2A, A2B, A3	Detected at mRNA, protein, and functional levels in hippocampus, cortex, cerebellum, and spinal cord. All receptors were found to be linked to Ca2+ signaling and to cAMP cascades. A1 and A2 receptors regulate expression of glutamate and GABA transporters receptors; A2A receptors regulate Na+/K+ pump while A3 receptors mediate neuroprotection.	177, 178, 317, 338, 369, 438, 504, 796, 1085, 1230, 1399
P2Y purinoceptors: P2Y1,2,4,6,12,13,14	Detected at mRNA, protein, and functional levels throughout the CNS, in hippocampus, cortex, cerebellum, brain stem, retina, and spinal cord; the P2Y1,2,4 are being dominating types. Generally are linked to PLC/InsP3/Ca2+ signaling cascade. P2Y receptors are also coupled to various signaling pathways, including MAP kinases, extracellular signal-regulated kinase (ERKs), the stress-activated protein kinases (SAPKs), the JNKs, p38/MAPK, glycogen synthase kinase, and the Akt kinase.	159, 226, 248, 519, 541, 771, 795, 842, 880, 1202, 1347, 1348, 1390, 1808
Muscarinic cholinoreceptors, mAChR: M1, M2, M3	Detected in hippocampus and amygdala. Control PLC, InsP3 production, and Ca2+ release from the ER. M3 receptors are linked to neurogenesis by regulating astroglial expression and release of fibronectin and laminin-1.	56, 617, 1599
Adrenergic receptors: α1AR, α2AR; β1AR, β2AR, β3AR	Detected at mRNA, protein, and functional levels in hippocampus, cortex, cerebellum, optic nerve, and spinal cord. α-ARs are mainly linked to PLC/InsP3/Ca2+ signaling, whereas β-ARs mainly control cAMP production and glial glucose metabolism.	123, 431, 459, 684, 882, 1596
Serotonin receptors: 5-HT1A, 5-HT2A, 5-HT2B, 5-HT5A, 5-HT7	Detected at mRNA and protein level throughout the brain. The 5-HT2B is the predominant type; mainly 5-HT receptors are linked to PLC/InsP3/Ca2+ signaling. 5-HT2B receptors are also connected to ERK1/2, andcPLA2. 5-HT7 receptors (found in suprachiasmatic nucleus) stimulate adenylyl cyclase.	129, 651, 688, 895, 1536, 1561, 1943
Dopamine receptors: D1, D2, D3, D5	Detected at mRNA and protein levels in basal ganglia and in substantia nigra. Astroglial D2 receptors are found in neocortex where they account for 30% of all D2 binding sites; D2 receptors are also present in fibrous astrocytes of white matter. Mainly linked to PLC/InsP3/Ca2+ signaling.	88, 858, 1134, 1135, 1455, 1459, 1782, 1934
Histamine receptors: H1, H2, H3	Have been detected at mRNA and protein level and functionally characterized in hippocampus and cerebellum. Linked to PLC/InsP3/Ca2+ signaling, adenylyl cyclase, glucose metabolism, and regulation of expression of glutamate transporters. Regulate synthesis of cAMP.	60, 764, 809, 882, 987, 1100, 1599
Cannabinoid receptors: CB1	Detected in hippocampus in situ. Linked to Ca2+ signaling and to synaptic plasticity.	592, 634, 1197, 1198
Oxytocin and vasopressin (V1b) receptors	Detected in hypothalamus. Linked to PLC/InsP3/Ca2+ signaling. V1b receptors also control PKC, CaMKII, and ERK1/2 signaling cascades.	424, 662, 1709, 1949
PACAP/VIP receptors: PAC1, VPAC1, VPAC2	Have been detected throughout the brain. All receptors stimulate adenylyl cyclase; PAC1 receptors are linked to PLC/InsP3/Ca2+ signaling. May also regulate energy metabolism and glycogenolysis.	70, 605, 656, 803, 1077
Bradykinin receptors: B2	Identified in cultured astrocytes; linked to PLC/InsP3/Ca2+ signaling and Ca2+-activated Cl− currents.	17, 330, 990
Opioid receptors: μ, δ, κ	Mainly detected in astrocytes in vitro. Linked to various signaling cascades, regulate expression of glutamate transporters, and may affect growth and development.	128, 983, 1673

The complement of receptors expressed by astrocytes in situ and in vivo is restrictive and depends on the brain region. The modality of neurotransmitter receptors expressed by astroglia matches that of their neuronal neighbors and is most likely controlled by the local neurotransmitter environment (1814). For example, receptors expressed by Bergmann glial cells and its neuronal neighbor the Purkinje neuron are optimized to sense neurotransmitters released by neuronal afferents, which form synapses accessible by these cells. Similarly, astrocytes express glycine receptors in the spinal cord, where glycine acts as a main inhibitory mediator. Astroglial expression of dopamine receptors is prominent in basal ganglia, which utilize dopaminergic transmission, while astroglial expression of serotonin receptors is restricted to areas contacting serotonergic terminals (1806). Therefore, expression of astroglial receptors in vivo is regulated by neurochemical input, which makes astrocytes perceptive to signals specific for each particular region of the brain.

A. Purinoceptors

Purinergic signaling system that utilizes purines and pyrimidines as extracellular signaling molecules is arguably the most ubiquitous being present in every organ and system without any obvious anatomical segregation. Correspondingly most of the living cells possess at least one (and often several) type(s) of purinoceptors (263). These latter are classified into adenine (P0) receptors, adenosine (A or P1) receptors, and nucleotide (ATP, ADP, UTP) receptors of the P2X (ionotropic) and P2Y (metabotropic) varieties (262, 1239, 1807). In the CNS, purines and pyrimidines are released from neurons (mainly from their terminals) and from neuroglia, with Ca²⁺-regulated exocytosis being probably the most widespread (and certainly the most studied) mechanism (2). After being released ATP is rapidly degraded by specific extracellular enzymes, known as ectonucleotidases (1968) to ADP, AMP, and adenosine that all act as purinergic agonists, which often can have opposite effects on target cells. The absolute majority of neuroglial cells studied so far express some purinoceptors (except P0 receptors, astroglial presence of which has not been studied yet). Neuroglial cells are also capable of releasing ATP and adenosine, which makes them an important source of physiologically released purines in the CNS and places purinergic transmission at the fore of gliotransmission (267, 1810). Purinergic signaling system also contributes to neuropathology; in particular, ATP released from damaged cells acts as a universal “danger” signal (often defined as damage-associated molecular pattern, DAMP) that controls glial defensive reactions such as reactive astrogliosis and activation of microglia (520).

1. Adenosine receptors

Adenosine receptors, represented by four subtypes (A₁, A_2A, A_2B, and A₃) with distinct pharmacological and functional properties (525), are classical G protein-coupled seven-transmembrane-spanning metabotropic receptors. Different receptors subtypes may assemble as heteromers and even oligomerize with other metabotropic receptors (263). As a rule the A₁ and A₃ receptors exert an inhibitory effect on adenylyl cyclase (mediated through G_i/o proteins), whereas A_2A and A_2B receptors activate cAMP production via G_s proteins. All adenosine receptors may regulate phospholipase C (PLC) and thus inositol 1,4,5-trisphosphate (InsP₃) synthesis. In some cells A₁ receptors were reported to activate K⁺ and/or Ca²⁺ channels (3, 204, 383, 525).

All four types of adenosine receptors have been identified in astrocytes (TABLE 3) using various functional assays applied to both in vitro and in situ preparations (383). Transcriptome analysis reveals high expression of A_2B receptors in cortical astrocytes (275, 1010). The first report of adenosine receptors in glia described adenosine-induced hyperpolarization in a subpopulation of cultured spinal cord and cerebellar astrocytes; this effect was blocked by broad antagonist 8-phenyltheophylline (734). Shortly thereafter, adenosine receptors with pharmacological profiles corresponding to A₁ and A₂ types were identified in human fetal astrocytes (1894). The A₁ receptors suppressed, whereas A₂ receptors potentiated synthesis of cAMP. In astroglial cultures prepared from 2-day-old rats, the cAMP production was stimulated by activation of A_2B receptors in a dose-dependent manner (1345). In the same cultures stimulation of A₁ receptors in type 1 (but not in type 2) astrocytes inhibited production of cAMP (1344).

Expression of A₁ receptor-specific mRNA was demonstrated in rat cultured astroglia (176). Activation of A₁ receptors in rat cultured astrocytes was linked to PLC. Incidentally, PLC activation was observed only in cultures with high levels of A₁ receptor expression (177) and upregulation of A₁ receptors synthesis potentiated A₁-dependent PLC stimulation (176, 178). Activation of PLC is directly linked to Ca²⁺ mobilization from the endoplasmic reticulum, and indeed, stimulation of A₁ receptors in cultured neonatal astrocytes initiated intracellular Ca²⁺ release as well as Ca²⁺ entry and potentiated histamine induced Ca²⁺ mobilization (1343). Similarly, adenosine induced [Ca²⁺]_i elevation in the majority of astrocytes in acute rat hippocampal slices (1399). Other types of adenosine receptors are also connected to Ca²⁺ signaling. In astrocytes in olfactory bulb slices, adenosine, which occurred following enzymatic degradation of ATP released from olfactory nerve terminals, induced [Ca²⁺]_i elevation via activation of A_2A receptors (438). In acutely isolated cortical astrocytes, A_2B receptors (as judged by sensitivity to the selective A_2B antagonist alloxazine) triggered Ca²⁺ signals (1391). The sensitivity of acutely isolated cells to adenosine was much higher when compared with cultured cells, thus indicating modified adenosine receptor expression in the in vitro conditions (1391). The A₃ receptors mediated adenosine- and guanosine-evoked [Ca²⁺]_i transients in cultured mouse astrocytes (317). Adenosine, acting through A₁ or A_2B receptors, also modulates astroglial Ca²⁺ signals originated from stimulation of other metabotropic receptors and endoplasmic reticulum Ca²⁺ release (25, 504, 796, 1252, 1253, 1753). Similarly, activation of A₁ receptors suppressed sustained Ca²⁺ influx following opening of P2X₇ receptors in cultured cortical astrocytes (1233).

Adenosine receptors modulate astroglial uptake of neurotransmitters through A₁-dependent regulation of expression of EAAT2/GLT1 (1896) or regulation of GABA transport by A₁-A₂ heteromers; in this latter case, A₁ receptors signal through G_s proteins to increase, while A₂ receptors signal through G_i/o proteins to inhibit GAT1/3 GABA transporters (369). In hippocampal astrocytes, A_2A receptors inhibit EAAT2/GLT1 transporter (1085, 1230) and evoke glutamate release from astrocytes through [Ca²⁺]_i and a protein kinase A-dependent pathway (981, 1230). This in turn potentiated neuronal activity due to an increase in glutamate concentration in the synaptic zones (1230). In astrocytes (it has been argued) A_2A receptors directly (through physical association) interact with Na⁺-K⁺-ATPase providing negative modulation. Suppression of the activity of the Na⁺/K⁺ pump following A_2A receptor activation has been considered as a mechanism for subsequent decrease in glutamate transporter-dependent uptake due to decrease in the transmembrane Na⁺ gradient (1084). Selective deletion of A_2A receptors from astrocytes (using Gfa2 promoter) decreases glutamate uptake and triggers psychomotor and cognitive deficits that have some semblance to schizophrenic phenotype (1086). Adenosine receptors also exert trophic effects; for example, A₁ and A₃ receptors contribute to neuroprotection probably through stimulating the PI3K and ERK1/2 MAPK pathways (184, 338).

2. Ionotropic P2X purinoceptors

Ionotropic P2X purinoceptors are archetypal ligand-gated cationic (Na⁺/K⁺/Ca²⁺) channels (FIGURE 13), assembled as homo- or heterotrimers from seven different subunits encoded by distinct genes that are classified P2X₁ to P2X₇ according to historical order of cloning (1239). Receptors formed through homo- or heteromeric assembly of P2X₁ to P2X₆ subunits (hitherto homomeric composition was shown for P2X_1–5 subunits, P2X₆ subunits apparently cannot olygomerize; heteromeric compositions are represented by P2X_1/2, P2X_1/4, P2X_1/5, P2X_2/3, P2X_2/6, and P2X_4/6 channels) are activated by low (submicromolar and micromolar) ATP concentrations, while homomeric P2X₇ receptors require millimolar ATP for activation (1240, 1705).

FIGURE 13.

Main properties of ionotropic receptors functionally expressed in astroglia.

There is a substantial discrepancy between evidence for P2X subunits expression in astroglia and data on their functional manifestation. Primary cultured astrocytes contain transcripts for P2X_1–5 and P2X₇ receptors (435, 541). In tissue extracts from rat nucleus accumbens, RT-PCR revealed expression of all seven P2X mRNAs (516). Cortical astrocytes contained transcripts for P2X₁ and P2X₅ subunits (937). In acutely isolated rat retinal Müller cells, P2X₃, P2X₄, P2X₅ but not P2X₇ mRNAs were identified (776), whereas P2X₇ mRNA was detected in human Müller glia (1313). Immunocytochemistry also demonstrated expression of P2X subunits. In astrocytes from healthy nucleus accumbens, P2X_2,3,4 receptors proteins were detected (although all 7 subunits were present at mRNA level); mechanical lesion however triggered additional expression of P2X₁ and P2X₇ subunits (516). Immunoreactivity for P2X₁ and P2X₂ receptors was detected in astroglial cells in cerebellum (829, 1001), and P2X₂ receptors were found in spinal cord astrocytes (828), P2X₄ receptors were identified in astrocytes from the brain stem (69) and in Müller glia (708), whereas hippocampal astrocytes were immunoreactive for P2X_1–4, P2X₆, and P2X₇ receptors (922).

Electrophysiologically, P2X-mediated currents are much less characterized. In primary cultured astrocytes, ATP evoked inward currents and [Ca²⁺]_i rise, although subunit composition of the underlying receptors was not identified (1050, 1844). In hippocampal astrocytes, both acutely isolated or in slices, no P2X-mediated currents were found, contrasting RT-PCR and immunocytochemistry findings indicative of P2X subunits expression (778). No ATP-induced currents have been found in Bergmann glial cells in acute cerebellar slices (880). In astrocytes from acutely isolated optic nerves, ATP triggered large [Ca²⁺]_i transients sensitive to the P2X receptor antagonist NF023 (used at 100 μM concentration at which it blocks all P2X_1–4 receptors); the [Ca²⁺]_i response could be also mimicked by the broad P2X agonist α,β-methylene ATP (α,β-meATP) (631, 786). In cortical mouse astrocytes, expression of P2X_1/5 heteromeric receptor was reported (FIGURE 14) (937). These receptors are highly sensitive to ATP (EC₅₀ ~50 nM); they have specific biophysical properties represented by biphasic kinetics with distinct peak and steady-state components, activation of “rebound” tail current following the washout of the agonist, and a very little desensitization in response to the repetitive agonist applications (936, 937). The P2X_1/5 receptors contribute to “glial synaptic currents” triggered in astrocytes by stimulation of neuronal afferents in cortical slices (932, 936). The P2X_1/5 receptors also mediate spontaneous “miniature” postsynaptic currents in astrocytes in cortical slices (936). Astroglial P2X_1/5 have intermediate Ca²⁺ permeability (P_Ca/P_monovalent ~2.2), and their activation by endogenous agonists or by synaptically released ATP triggered transient cytoplasmic Ca²⁺ signals (1305). It is of course important to mention that most humans carry a single-nucleotide polymorphism at the 3′ splice site of exon 10 of the human P2X5 gene, which renders the P2X₅ subunit nonfunctional (898).

FIGURE 14.

P2X_1/5 receptor-mediated currents in cortical astrocytes. A: the family of ATP currents evoked by repetitive applications of the agonist. The currents show no apparent desensitization. Current traces have a complex kinetics comprising the peak, the steady-state component, and the “rebound” inward current recorded upon ATP washout as indicated on the graph. B: concentration dependence of ATP-induced currents in cortical astrocytes. Membrane currents recorded from a single cell in response to different ATP concentrations are shown on the left. The right panel shows the concentration-response curves constructed from 9 similar experiments; current amplitudes were measured at the initial peak and at the end of the current, as indicated on the graph. C: inhibition of ATP-induced currents by PPADS. Currents recorded at various concentrations of PPADS are shown on the left, and the concentration dependence of inhibition for peak and steady-state components constructed for 7 individual experiments is presented on the right. The peak component of the response was more sensitive to PPADS. Application of PPADS started 2 min before application of ATP. All recordings were made at holding potential of −80 mV. [From Lalo et al. (937).]

Astroglial P2X₇ receptors have received much attention, mostly because of their pathological significance (i.e., initiation of apoptosis and cell lysis, regulation of astrogliosis or controlling processing and release of cytokines; see Refs. 520, 763 for relevant literature). The P2X₇ receptors are unique in their low ATP sensitivity (in all probability it is ATP^4- that acts as a true agonist), almost complete absence of desensitization, and ability to produce large transmembrane pores upon intense stimulation (1239). The very first in situ hybridization mapping (346) failed to detect P2X₇ receptors in the brain. Subsequent studies revealed P2X₇ transcripts and proteins in many areas including cortex, hippocampus, medulla oblongata, cerebellum, thalamus, and amygdala (1658). The specificity of antibodies used for labeling P2X₇ receptors is, however, far from ideal (1621), thus making many observations contentious.

In primary cultured astroglia, expression of P2X₇ receptors was frequently documented at both the transcriptional and protein levels (435, 457, 541, 752, 797, 1194, 1309, 1850). Freshly isolated astrocytes as well as astrocytes in slice preparations similarly demonstrated immunoreactivity for P2X₇ receptors (922, 1261, 1309). At the same time, in-depth analysis of the cellular distribution of P2X₇ mRNA in the rat brain using isotopic in situ hybridization detected these receptors in neurons, oligodendrocytes, and microglia but failed to identify P2X₇ receptors transcripts in astrocytes (1930). Astroglial expression of P2X₇ receptors, as a rule, increases after brain injury of various etiology (516, 517, 1194).

In cultured astrocytes, both P2X₇-mediated (as judged by pharmacology, activation by Bz-ATP, and inhibition by oxATP) [Ca²⁺]_i transients (93, 541, 1233) and membrane currents (457, 1238, 1515) were characterized. The P2X₇ currents were also detected in Müller cells freshly isolated from human retina (1313). In isolated optic nerve P2X₇ receptors mediated Ca²⁺ signaling, which was absent in P2X₇^−/− mice (631). The P2X₇ membrane currents were recorded from rat and mouse cortical astrocytes in acute slices. These currents demonstrated all characteristic properties of P2X₇ receptors; they were activated by millimolar ATP and 100 μM Bz-ATP; the agonist sensitivity was increased by removal of extracellular divalent cations; currents were blocked by Brilliant Blue G, as well as by a rather specific agonist A 438079, and they were absent in P2X₇ receptor knockout animals (1261).

Contribution of P2X₇ receptors to astroglial physiology is multifaceted, and seemingly, it may involve pathways and mechanisms not related to opening of the cationic channel. Indeed, activation of P2X₇ receptors in cultured astrocytes is linked to release of glutamate, GABA, ATP, and other purines by exocytosis, through P2X₇-associated transmembrane pore or through Cl⁻/HCO₃⁻-dependent mechanism of an unidentified nature (93, 457, 458, 1690). Sustained release of glutamate was also observed in astrocytes in situ in hippocampal slices following prolonged activation of P2X₇ receptors (503). Activation of P2X₇ receptors increased (by ~60 times) synthesis of endocannabinoid 2-arachidonoylglycerol in cultured astrocytes (1840), modulated release of TNF-α (915), stimulated nitric oxide (NO) production (1175, 1194), induced phosphorylation of AKT (783) and p38MAPK/ERK1/ERK2 (565), prompted transmembrane transport of NADH (1015), increased production of lipid mediators of inflammation cysteinyl leukotrienes (91), and regulated NF-κB signaling (797). Evidence also exists indicating P2X₇-mediated regulation of expression of P2Y₂ receptors (370) and aquaporin-4 (960) in cultured rat astrocytes.

All these functional outcomes of P2X₇ receptors activation are mediated through diverse intracellular signaling pathways. The P2X₇ receptor has a uniquely extended COOH terminus that interacts with several proteins and hence activates intracellular signaling pathways such as extracellular signal-regulated protein kinases (ERKs), serine-threonine kinase Akt (Akt), c-Jun NH₂-terminal kinases (JNKs), mitogen-activated protein kinase (MAPK), and p38 kinase (520). This P2X₇-dependent signaling may be activated by low physiological concentrations of ATP, reflecting non-channel mode of receptor function (426). There are also occasional data on heterogeneity of P2X₇ signaling. Stimulation of P2X₇ receptors led to pore formation in cortical but not in hippocampal astrocytes (173). This heterogeneity may reflect, for instance, expression of different splice variants of the receptor: the splice variant P2X₇A is linked to pore formation and cell death, whereas the P2X₇B variant does not form the pore and stimulates cell growth (12).

3. Metabotropic P2Y receptors

Metabotropic P2Y purinoceptors are subclassified into the P2Y_1,2,4,6,11 and P2Y_12,13,14 groups based on phylogenetic similarity and G protein preference (1), with P2Y_1,2,4,6,11 receptors being coupled through G_q/G₁₁ proteins to PLC and hence to InsP₃-mediated Ca²⁺ release, while P2Y_12,13,14 receptors modulate ion channels and inhibit adenylyl cyclase via G_i/o proteins.

Metabotropic P2Y purinoceptors are widely expressed in astroglia (1810). Cortical astrocytes in culture express mRNA for P2Y_{1,2,4,6,12,13} and UDP-glucose P2Y₁₄ receptor (4, 141, 435, 541), whereas spinal cord astrocytes predominantly express transcripts for P2Y_1,2 receptors (495). In acutely isolated hippocampal CA1 astrocytes, P2Y₁ receptors were identified in ~50% of cells at both transcriptional and protein levels although some cells expressed P2Y_2,4 receptors. Proportion of cells bearing P2Y₂ receptors increased from ~5% at P8−P12 to ~38% at P25 (1961). Isolated rat and human Müller retinal glial cells expressed mRNA and proteins of P2Y_1,2,4,6 receptors (532, 533). Immunoreactivity of P2Y_1,4 receptors was detected in the nucleus accumbens; in cortex, astrocytes were positively stained for P2Y_1,2,4,6 receptors (518).

Stimulation of P2Y receptors of cultured astrocytes, as a rule, leads to PLC-dependent InsP₃ production and Ca²⁺ signals originating from InsP₃-induced endoplasmic reticulum Ca²⁺ release (159, 248, 771, 795, 842, 1347, 1348, 1808). The P2Y-mediated InsP₃-induced intracellular Ca²⁺ release was characterized in cultured spinal cord astrocytes (1532, 1533) and in retinal Müller glia (233). Activation of SOCE often contributes to P2Y-mediated astroglial Ca²⁺ signaling (893). Stimulation of P2Y receptors evoked [Ca²⁺]_i transients in pituitary glia and in rat neurohypophysial astrocytes (1758, 1773). Pharmacological profiling of astrocytic [Ca²⁺]_i dynamics in vitro and in situ reveals the leading role for P2Y₁ and P2Y₂ receptors (141, 489), although other receptors (especially P2Y₁₄) may also contribute (541). Propagating astroglial Ca²⁺ waves require P2Y₁ and P2Y₂ receptors (142, 223). The P2Y/PLC/InsP₃ cascade and resulting Ca²⁺ signaling were also characterized in detail in astrocytes in situ, in Bergmann glia (880, 1390), in stratum radiatum (221), in olfactory bulb (438), and in the optic nerve (631).

In addition to being closely linked to Ca²⁺ signaling, P2Y receptors are coupled with numerous intracellular signaling pathways, including MAP kinases [ERKs, the stress-activated protein kinases (SAPKs), the JNKs, and p38/MAPKs, (226, 1201, 1202)]. The P2Y receptors induce Akt phosphorylation in astrocytes in vitro (519), while P2Y_1,2,4 receptors are linked to glycogen synthase kinase (GSK)-3, a signaling molecule involved in cell survival, cell cycle regulation, proliferation, and differentiation (1200). The P2Y receptors in astrocytes are also coupled with α_vβ₃/β₅ integrin signaling pathways, which control cytoskeletal remodeling and motility (1872), while P2Y_2,4 receptors regulate ERK1/2 and the signal transducer and activator of transcription 3 (STAT3) signaling cascade (1863). Stimulation of P2Y₂ receptors induces the phosphorylation and hence transactivation of the epidermal growth factor receptor (EGFR), which regulates astroglial proliferation (1380).

B. Glutamate Receptors

1. Ionotropic glutamate receptors in astroglia

a) ampa receptors.

Functional expression of ionotropic glutamate receptors of the AMPA type in astroglia has been well documented in vitro and in situ. In particular, these receptors were electrophysiologically characterized in astrocytes in hippocampal (777, 1589) and cortical slices (FIGURE 15; Refs. 932, 935), as well as in cerebellar Bergmann glia (1169, 1522). All four main subunits of AMPA receptors (GluA1–GluA4) have been identified in astrocytes, although combinations may vary in different brain regions. In hippocampus, for example, GluA2 and GluA4 subunits are predominantly expressed, which is reflected by linear I–V relation and low Ca²⁺ permeability (1589). In cortical astrocytes, GluA1 and GluA4 subunits are the most abundant (355). The AMPA receptor complex in Bergmann glia is devoid of GluA2 subunit which stipulates double-rectifying I–V relationship and some (P_Ca/P_monovalent ~1) Ca²⁺ permeability (563, 1169). In the spinal cord, astroglia immunoreactivity for GluA4 subunit was concentrated in the perivascular processes, whereas somata were stained with GluR2/3 antibodies (228). Conditional deletion of AMPA 1/4 receptors from Bergmann glia led to a retraction of glial perisynaptic processes and deficient fine motor coordination (1522).

FIGURE 15.

AMPA and NMDA receptor-mediated currents in cortical astrocytes. A: NBQX inhibits the fast component of glutamate-induced current. Representative traces illustrate the current before, during, and after application of 30 μM NBQX (left panel) as well as the NBQX-sensitive current obtained by subtraction (right panel). The concentration dependence of the block of the fast component for four cells (IC₅₀ = 2.2 ± 0.4 μM, Hill coefficient = 1.9) is shown in the inset. B: d-AP5 inhibits the slow component of glutamate-induced current. Representative traces demonstrating the effect of 1 μM d-AP5 (left panel) and the d-AP5-sensitive component were obtained by subtraction (right panel). The concentration dependence of the block for five cells (IC₅₀ = 0.64 ± 0.1 μM, Hill coefficient = 1.6) is shown in the inset. C: NMDA-induced (2-s application) currents in a single astrocyte and concentration-response curve constructed from six such experiments (EC₅₀ 0.34 ± 0.06 μM, Hill coefficient = 1.5). D: glycine-dependent potentiation of astrocyte NMDA response. NMDA-induced currents in glycine-free normal extracellular solution are shown on the top; NMDA-induced currents in the presence of different glycine concentrations (30 nM and 1, 10, and 30 μM) are displayed on the bottom. The concentration-response curve (ΔI_norm represents the amplitudes of current increase normalized to the maximal increase at 30 μM glycine) constructed from seven experiments is shown on the right (EC₅₀ 1.1 ± 0.07 μM, Hill coefficient = 1.2). [From Lalo et al. (935).]

b) kainate receptors.

Functional expression of kainate receptors in astrocytes is yet to be identified, although the relevant subunits were detected at the mRNA and protein levels (228, 548). Expression of GluK1–5 was also found in reactive (but not in healthy) astrocytes in a model of temporal lobe epilepsy (1798).

c) nmda receptors.

Astroglial expression of the N-methyl-d-aspartate (NMDA) receptors was initially detected by EM immunocytochemistry of cortical preparations: the GluN1 and GluN2A/B labeling was found in peripheral astrocytic processes (52, 354). Transcripts for GluN1 and GluN2A/B subunits were identified in Bergmann glia (1023) and in cortical astroglia (1562). In human astrocytes, transcripts for all seven NMDA receptors subunits (i.e., GluN1, GluN2/A-D, and GluN3A,B) have been found (962). The GluN1 immunoreactivity was detected in adult rat astrocytes from the lateral and basal nuclei of the amygdala in the basolateral amygdala (497), in the bed nucleus of the stria terminalis (600), and in the locus coeruleus (1785). In humans, GluN1, GluN2A, and GluN2B proteins were detected in cortical astrocytes [where they predominantly concentrate in the processes (354)], in fetal astroglia (962), and in Müller glial cells (1414).

Exposure of slices to NMDA induced membrane currents and [Ca²⁺]_i transients in the cortical (935, 1261, 1562), in the spinal cord (1965), and in a subpopulation of hippocampal astrocytes (1400, 1593, 1666), whereas small NMDA-induced Na⁺ currents were observed in cerebellar Bergmann glial cells (1168). Evidence for functional NMDA receptors in hippocampal astrocytes remains controversial; several studies on slices or on acutely isolated cells failed to identify any NMDA receptor-mediated currents or [Ca²⁺]_i responses (276, 1589, 1600; see also Refs. 714, 1809). Cationic currents mediated by NMDA receptors were characterized in astrocytes acutely isolated (with nonenzymatic vibro-dissection procedure) from mice somatosensory cortex (FIGURE 15). These NMDA-induced currents were positively modulated by glycine and blocked by specific NMDA receptor antagonists MK-801 and d-2-amino-phosphonopentanoic acid (d-AP5) (935, 1809).

Astroglial NMDA receptors are assembled as heterotetramers from GluN1, GluN2 C or D, and GluN3 subunits. This composition defines their specific biophysical properties represented by weak Mg²⁺ block (which develops at about −120 mV) and relatively low Ca²⁺ permeability (P_Ca/P_monovalent ~3), as well as idiosyncratic pharmacology reflected by sensitivity to memantine and GluN2C/D subunit-selective antagonist UBP141 (464, 936, 1304, 1305). In cortical astrocytes NMDA receptors generate miniature spontaneous currents or they can be activated by neuronal synaptic inputs (932). Opening of astroglial NMDA receptors (in response to endogenous agonist application or synaptic stimulation) results in Ca²⁺ influx and in generation of Ca²⁺ signals (1305). In primary cultured adult rat astrocytes, NMDA receptors were found to be activated by mechanical stimulation delivered as fluid shear stress in the absence of agonists (1057). In cocultures of hippocampal astrocytes and neurons, application of NMDA-glycine mixture caused substantial (~30 mV) depolarization of astroglia; this depolarization was sensitive to MK 801 (972). In acute slices from astroglia-specific NR1 subunit knockout mice, application of NMDA with glycine caused much smaller astroglial depolarizations when compared with wild-type controls (972). Activation of astroglial NMDA receptors was also linked to antioxidant transcriptional activation nuclear factor-erythroid 2-related factor-2 (Nrf2), which stimulates release of glutathione precursors, thus boosting neuronal glutathione biosynthesis and providing for antioxidant protection (794).

2. Metabotropic glutamate receptors

The seven-transmembrane-spanning domain G protein-coupled metabotropic glutamate receptors are divided into three groups, according to their G protein coupling and signal transduction. Group I comprises mGluRs 1 and 5, group II includes mGluRs 2 and 3, and group III includes mGluRs 4, 6, 7, and 8. In general, group I receptors are coupled to PLC/InsP₃/Ca²⁺ release cascade, whereas receptors from groups II and III are coupled to adenylyl cyclase (1231).

The mGluR3 that inhibits adenylyl cyclase is the most abundant astroglial receptor throughout life span (1702). In early postnatal development astrocytes almost universally express mGluR5 receptors which, when activated, trigger Ca²⁺ signals and [Ca²⁺]_i oscillations both in vitro and in situ (361, 879, 947, 1400, 1401). The mGluR-mediated Ca²⁺ astroglial signals were claimed to accompany hippocampal synaptic transmission, hence providing astrocytes with a detector of the latter (720, 1308); these observations however were not universally confirmed (1700). Astroglial expression of mGluR5 was shown to be downregulated in postnatal development, and these receptors almost completely disappear in adult brain (277, 1702). In the developing brain mGluR5 receptors contribute to functional maturation of astrocytes; ablation of these receptors led to serious deficits in arborization of astroglial processes and expression of glutamate transporters (1149).

C. GABA and Glycine Receptors

1. GABA_A receptors

Ionic currents mediated by an opening of GABA_A receptors have been widely recorded and characterized in astrocytes in vitro, in acutely isolated cells, and in astrocytes in slices from various brain regions, including hippocampus, cerebellum, retina, hypothalamus, supraoptic nucleus, and spinal cord (339, 773, 853, 854, 857, 1042, 1167, 1336, 1800). These GABA_A currents were invariably inward (Cl⁻ efflux) at resting membrane potentials, with E_rev around −40 mV reflecting high [Cl⁻]_i (see sect. VIA). Unitary GABA_A currents have 30-pS single-channel conductance. The subunit composition of astroglial GABA_A receptors remains unknown; α1 and β1 subunits were detected in hippocampal astrocytes (523) and α2 and γ1 in Bergmann glia (1470). The pharmacology of astrocytic GABA_A receptors is complex: in Bergmann glia benzodiazepine was inactive [reflecting expression of γ1 subunit (1167)]; in cerebral cortical and fibrous astrocytes, benzodiazepine potentiated GABA_A currents, whereas methyl-4-ethyl-6,7-dimethoxy-beta-carboline-3-carboxylate (DMCM) reduced GABA_A currents in fibrous astrocytes while potentiating these currents in protoplasmic astroglia (1504). Astroglial GABA_A receptors are linked to shaping cell morphology, and their sustained activation (in vitro) leads to an increased complexity of astroglial processes (1087). Astroglial GABA_A receptors, however, may have a much more important role in regulation of extracellular Cl⁻ concentration (see sect. XIIA). In freshly isolated hippocampal astrocytes (523) and in astrocytes in hippocampal slices (1098), activation of GABA_A receptors triggers Ca²⁺ influx through voltage-gated Ca²⁺ channels and hence evokes Ca²⁺ signaling.

2. GABA_B receptors

Metabotropic GABA_B receptors are widespread in astrocytes, and their activation primarily results in endoplasmic reticulum Ca²⁺ release (822, 1098, 1222). Astrocytes express all three types (GABA_B1a, GABA_B1b, and GABA_B2) of GABA_B receptors, which, in the hippocampus, tend to be clustered in perisynaptic processes (307). Rather unexpectedly, astroglial GABA_B receptors are linked to the PLC/InsP₃ cascade through G_i/o proteins (1066), not through G_q/G₁₁ proteins typically required. Activation of GABA_B receptors triggers Ca²⁺ signals and [Ca²⁺]_i oscillations in astrocytes in vitro, in situ, and in vivo (822, 1008, 1066, 1800).

3. Glycine receptors

Glycine receptors, which, similarly to GABA_A receptors, belong to pentameric receptors and similarly to GABA_A receptors, mediate Cl⁻ fluxes, have been detected in astrocytes in spinal cord slices (1336). Single-cell RT-PCR performed on these astrocytes revealed expression of α1 and (in ~50% of cells) β-subunits of the receptor (874). Glycine is the major inhibitory neurotransmitter in the spinal cord, and, predictably, not only astrocytes but also oligodendroglia and oligodendroglial precursors in this part of the CNS express glycine receptors (1336).

D. Acetylcholine Receptors

1. Ionotropic (nicotinic) ACh receptors

Ionotropic ACh receptors, historically known as nicotinic receptors (nAChRs), are widely expressed throughout the CNS; the nAChRs are archetypal pentameric ligand-gated channels permeable to Na⁺, K⁺, and Ca²⁺. The first evidence for astroglial expression of nAChRs was obtained in autoradiographic studies (726); subsequently, nicotine was found to hyperpolarize cultured astroglia derived from several regions of the brain and the spinal cord (732, 735). Operational nAChRs were found in astroglial cells in cultures and in situ; in both types of preparations, their activation triggers [Ca²⁺]_i elevation resulting from Ca²⁺ influx through the receptor (1257, 1726) as well as from further amplification by Ca²⁺-induced Ca²⁺ release (1597). Precise subunit composition of astroglial nAChRs remains unknown, although receptor Ca²⁺ permeability infers the presence of α7 subunit [for α7 homomeric receptors P_Ca/P_monovalent is ~6 (1312)]. Analysis of mRNA expression in cortical astrocytes found α4, α7, and β2 subunits, whereas in human astrocytes from hippocampus and entorhinal cortex, the α3, α7, and β4 subunits were revealed by immunocytochemistry (601, 1737). Astroglial α7-nAChRs are glio- and neuroprotective. This neuroprotection is possibly mediated through the release of glial cell-derived neurotrophic factor, GDNF (998, 1719). The density of α7 nAChRs is claimed to greatly increase in Alzheimer's disease (1929); furthermore, β-amyloid may either activate or inhibit α7 receptors [depending on concentration (1293)] which in turn may affect astroglial pathological responses and even trigger release of neurotoxic factors (1726).

2. Metabotropic (muscarinic) ACh receptors

Functional M₁ and M₂ receptors linked to InsP₃ production were identified in cultured astroglia, whereas M₃ receptors were identified at mRNA level (42, 1179). In hippocampal slices, carbachol (100 μM) triggered [Ca²⁺]_i transients in ~35% of GFAP or S100B positive astrocytes; these responses were blocked by atropine and selective M₁ receptor inhibitor pirenzepine (1599). Furthermore, muscarinic ACh receptor-mediated Ca²⁺ responses were elicited by synaptically released ACh in astrocytes in hippocampal slices (56); arguably, these Ca²⁺ responses contribute to ACh-dependent synaptic plasticity in the hippocampus. In cortical astrocytes cocultured with hippocampal neurons, carbachol, acting through M₃ receptors, induced expression and release of fibronectin and laminin-1, which in turn accelerated neuritogenesis (616); incidentally, release of these factors (and consequently neuritogenesis) was inhibited by alcohol (617).

E. Receptors for Monoamines

1. Adrenergic receptors

The noradrenergic innervation of the brain is provided by numerous projections of neurons, the somata of which are localized in locus coeruleus. These projections do not form localized synaptic contacts; rather, norepinephrine is released from axonal varicosities into the neuropil and hence exerts neurohormonal action through volume transmission. Neurons from rostral and medium portions of locus coeruleus innervate the cerebral hemispheres and cerebellum; projections from the caudal part are sent to the spinal cord. Astrocytes throughout the brain abundantly express adrenergic receptors, which regulate metabolism, morphological plasticity, and physiological responses. The very first observation of catecholamine-mediated increase in cAMP in fetal rat cultured astrocytes was made in 1972 (577). The notion that astrocytes can be an important, if not the main, target of adrenergic innervation (FIGURE 16) appeared in late 1980s (1676) after discoveries that astroglial adrenoceptors control many aspects of cellular metabolism and trophic responses (for early studies, see Refs. 644, 689, 1530, 1757, 1786, 1787).

FIGURE 16.

Direct stimulation of locus coeruleus elicits Ca²⁺ transients in cortical astrocytes. A: the scheme of experimental setup. B: average fluorescence of the image field in C over time showing Ca²⁺ response to locus coeruleus stimulation. C: fluo 4-labeled astrocytes taken at time points indicated by numbers in B. Insets are blown-up view of boxed area in pictures. Scale bar, 50 μm. D: images from coronal sections through somatosensory cortex illustrating the effect of N-(2-chloroethyl)-N-ethyl-2-bromobenzylamine (DSP-4) on locus coeruleus projection neurons as labeled by antibodies to TH (in red) and MAP-2 (in green) with DAPI labeling of nuclei for orientation. Scale bar, 100 μm (20 μm in inset). E: consistent with the significant reduction in locus coeruleus projection neurons and despite almost twice the stimulation intensity (216 ± 33 vs. 400 ± 39 μA; P = 0.004, t-test), there is a significant reduction in the cortical Ca²⁺ response to locus coeruleus stimulation in DSP-4-treated animals (P = 0.013, t-test). [From Bekar et al. (123).]

Both α- and β-adrenoceptors have been found in astroglia in vitro, in slices, and in vivo at mRNA, protein, and functional levels (687). Ultrastructural analysis revealed α- and β-adrenergic receptors in astrocytic processes (51, 53), which are often closely associated with varicosities and terminals of axons of noradrenergic neurons (53, 344, 949). Freshly dissociated and FACS sorted murine cortical astrocytes were found to express mRNA for α_1A-, α_2A-_, β₁-, and (albeit at much lower level) β₂-adrenoceptors (687).

The α₁-adrenoceptors are linked to PLC and InsP₃ cascades, and their activation elicits Ca²⁺ release from the endoplasmic reticulum, an effect well documented in astrocytes in culture (247, 1531, 1596), in freshly dissociated cells (1747), in slices (459, 882), and in vivo (FIGURE 17) (431). These receptors are activated by endogenous adrenergic agonists as well as by norepinephrine released from neuronal varicosities following stimulation of locus coeruleus (431). In neocortical slices from 8- to 12-wk-old mice, as well as in freshly isolated astrocytes from the same slices, norepinephrine caused robust [Ca²⁺]_i increase in all astrocytes; on the contrary, the effect on neurons was rather minor. This signaling was mediated by α₁ adrenoceptors (Ca²⁺ responses were mimicked by specific agonist A61603 and inhibited by specific blocker terazosin); the sensitivity of astroglial receptors was high, with K_D for norepinephrine of ~360 nM (1311).

FIGURE 17.

Stimulation of α₁- and β-adrenoceptors triggers Ca²⁺ signaling in astrocytes in vivo. A: cortical astrocytes loaded with rhod 2-AM were imaged in layers I/II to detect changes in [Ca²⁺]_i. Agonists were injected using a microelectrode loaded with an artificail cerebrospinal fluid (aCSF) and the tracer Alexa 488. B: representative images of astrocytic [Ca²⁺]_i responses to the α₁-AR agonist methoxamine in awake Glt-1-EGFP transgenic mice. Astrocyte-specific loading of rhod 2-AM was confirmed by colocalization with Glt-1-EGFP. Scale bar = 100 μm. C: rhod 2-AM ΔF/F₀ traces from B, normalized to Glt-1-EGFP fluorescence. Representative aCSF-microinjection trace is shown at the bottom. D: astrocytic [Ca²⁺]_i responses to adrenergic and acetylcholinergic receptor agonists, measured by rhod 2-AM ΔF/F₀. Bar graph shows the percentage of astrocytes within the puff radius responding with a [Ca²⁺]_i increase. ***P < 0.001, one-way ANOVA, Bonferroni correction [n = 25 trials in 6 animals (methoxamine); 14 trials in 4 animals (isoproterenol); 24 trials in 5 animals (dexmedetomidine); 23 trials in 5 animals (carbachol)]. E and F: bar graphs showing average rhod 2-AM ΔF/F₀ and response duration from trials with responding cells. **P < 0.01; *P < 0.05, one-way ANOVA, Bonferroni correction [n = 25 trials (methoxamine), 13 trials (isoproterenol), 17 trials (dexmedetomidine), 20 trials (carbachol)]. Data are means ± SE. [From Ding et al. (431).]

The α₂-adrenoceptors in astrocytic processes have been identified immunochemically in the intact brain (53, 1118). Their functional expression has been corroborated in primary cultures of astrocytes by α₂-adrenoceptor binding (684), by mRNA and protein localization (1118, 1460), and by Ca²⁺ imaging (123, 1223, 1531).

All three types of β-adrenoceptors, i.e., β₁-, β₂-, and β₃-receptors have been identified in astrocytes in vitro and in vivo using radioligand binding, mRNA, and protein detection as well as immunocytochemistry (298, 997, 1060, 1596). The β₁-adrenoceptors are linked to regulation of glycogen synthesis and mediate cAMP-dependent inhibition of astrocytic K⁺ channels; β₂-adrenoceptors are mostly connected to adenylyl cyclase through G_s proteins and are implicated (together with β₃-adrenoceptors) in regulation of glucose uptake through modulation of GLUT1 glucose transporter and glucose metabolism through PKA signaling pathways (448, 684, 755, 1514). Individual astrocytes simultaneously express α- and β-adrenoceptors, and stimulation of both receptors triggers temporally distinct cellular events represented by Ca²⁺ oscillations developing within seconds and much slower tonic increase in cAMP, which saturates within 100s of seconds (723).

2. Serotonin receptors

Serotonin (or 5-hydroxytryptamine) is one of the ancient signaling molecules, being produced and released by bacteria, in which it regulates swimming behavior and growth (1215). Astroglial cells express several types of metabotropic 5-HT receptors, including 5-HT_1A, 5-HT_2A, 5-HT_2B, and 5-HT_5A subtypes, which all have been identified at transcript and protein level in glial cultures, in freshly isolated cells, and in the brain tissue (80, 293, 1536). These receptors tend to concentrate at the astroglial processes (293). In general, 5-HT₂ receptors are coupled to PLC with subsequent production of InsP₃ and DAG and Ca²⁺ signaling; serotonin-induced [Ca²⁺]_i transients were characterized in astroglia in vitro (1536) and in situ (651).

In astrocytes in vitro, the 5-HT_2B receptor seems to be predominant (895). Expression of mRNA of this receptor in astrocytes, freshly isolated and FACS sorted from the mouse brain, is about two times larger than in neurons from the same preparation (1943). There is mounting evidence that serotonin-specific reuptake inhibitors (major antidepressant drugs such as fluoxetine or sertraline) interact with and activate astroglial 5-HT_2B receptors (688, 1943) with subsequent Ca²⁺ signaling (1561) or phosphorylation of extracellular regulated kinases 1/2 (ERK1/2) or upregulation of calcium-dependent phospholipase A₂ (cPLA₂) (688, 1943). Interactions between these antidepressants and astroglial 5-HT_2B receptors may represent a pharmacologically relevant mechanism, while the receptor per se can be a legitimate target for new therapies (1364).

The 5-HT₁ and 5-HT₅ receptors are connected (negatively for the former and positively for the latter) with adenylyl cyclase. There is also evidence for astroglial expression of 5-HT₇ receptor that stimulates adenylyl cyclase: it was found in primary cultures of rat (704, 1614) and human (343) astrocytes and detected immunocytochemically in the mouse suprachiasmatic nucleus (129).

3. Dopamine receptors

Astroglial expression of dopamine receptors was discovered by autoradiography in the striatum (729). Using cell cultures and slices of basal ganglia, D₁, D₂, D₄, and D₅ receptors were identified at transcript and protein levels (1134). Similar experiments on striatal astroglial cultures revealed more restricted expression of D₁ (1934) or D₂ receptors (88). Astrocytic D₁ receptors were found in monkey and human cultured striatal astrocytes, with pharmacology somewhat different from rats (1820). Striatal astrocytes were also reported to express D₅ receptors (235). Subsequently, dopamine receptors were detected and characterized in cortical structures. Strong expression of D₂ receptors was found, at the ultrastructural level, in astroglial processes surrounding cortical interneurons; astroglial D₂ expression accounted for one-third of all D₂ binding sites in the neocortex from rodents, monkeys, and humans (858). Strong immunoreactivity for D₂ receptors was identified in human fibrous astrocytes in the white matter (1135). The D₁ receptor subtype was detected in mouse cortical astrocytes in vitro and in situ (1456).

Signaling cascades regulated by dopamine receptors are diverse; generally, activation of D₁ and D₂ receptors by dopamine or by specific agonists in astrocytes triggers [Ca²⁺]_i transients of variable amplitude and kinetics, and it is universally conceded that these Ca²⁺ responses originate from InsP₃-induced endoplasmic reticulum Ca²⁺ release (858, 1455, 1459, 1782). An alternative mechanism of dopamine-induced Ca²⁺ signaling, described in astrocytes cultured from midbrain, cortex, and hippocampus, proceeds through dopamine uptake, oxidation of dopamine by monoamine oxidase B, and release of reactive oxygen species, which induce lipid peroxidation and activation of PLC with subsequent opening of InsP₃Rs (1782). In hippocampal slices, dopamine induced a biphasic Ca²⁺ response comprising initial [Ca²⁺]_i elevation associated with Ca²⁺ release from the endoplasmic reticulum, which was followed by [Ca²⁺]_i dropping below the resting baseline. The [Ca²⁺]_i increase was mediated by D₁ and D₂ receptors, whereas [Ca²⁺]_i undershoot by D₂ receptors (792). In addition, D₁ receptors regulate glycolytic activity [manifested by an increase in cytosolic NADH (1455)]. The D₂ receptors were shown to link with αB-crystallin (1595), whereas D₅ receptors with phosphoinositide 3-kinase (235). Chronic stimulation of D₂ receptors in cultured astroglia with ropinirole stimulated NGF and GDNF secretion from cultured rat cortical astrocytes (1256), whereas in rat hippocampal slices D₂ receptors activated by apomorphine (a nonspecific dopamine receptor agonist) induced S100B secretion (1195).

4. Histamine receptors

Astrocytes express H₁, H₂, and H₃ histamine receptors, which have been identified by binding, expression, and functional assays. The first evidence for functional histamine receptors in astroglia derived from autoradiographic studies [which showed binding of H₁ and H₂ antagonists (725)] and electrophysiology (733): at high concentrations histamine caused depolarization (sensitive to H₁ antagonist pyrilamine), while at low concentrations it caused hyperpolarization (sensitive to H₂ antagonist cimetidine). These initial observations have been subsequently confirmed by binding and pharmacological assays (see Ref. 808 for references). Transcripts for H₁, H₂, and H₃ receptors have been detected in astrocytes in vitro (809), while H₃ receptors were also detected by immunocytochemistry in situ (1100); of note, astroglial H₃ receptors expression is rather minor. Functionally, H₁ receptors are linked to several signaling cascades including 1) Ca²⁺ signaling mediated by PLC/InsP₃ with subsequent Ca²⁺ release [this has been characterized both in vitro and in situ (764, 882, 1599)]; 2) regulation of glucose metabolism, the underlying mechanism possibly associated with histamine-induced Ca²⁺ signaling (60); 3) upregulation of EAAT2/GLT-1 glutamate transporter (496); and 4) stimulation of the release of neurotrophic factors GNGF and neurotropin-3 (809, 987). Activation of H₂ receptors stimulates cAMP production and glycogen breakdown (60, 913).

F. Bradykinin Receptors

Bradykinin receptors of the B₂ type (the G protein 7-transmembrane metabotropic receptors) have been identified in cultured astrocytes (330); activation of these receptors results in InsP₃ production and Ca²⁺ release from the endoplasmic reticulum (1328, 1672); the [Ca²⁺]_i increase was also concomitant with the activation of an inward current (578). This inward current likely reflects Cl⁻ efflux through Ca²⁺-activated anion channels, possibly of the VRAC variety (17, 990).

G. Cannabinoid Receptors

Cannabinoid receptors of CB₁ and CB₂ types are metabotropic receptors coupled to G_i/o proteins. The CB₁ receptors were identified in rat cultured astroglia (1667). They were also detected in brain tissue, in astrocytes, in the caudate putamen nucleus (1484), in the dorsal horn of the spinal cord (1529), and in olfactory and limbic structures where CB₁ immunoreactivity was concentrated in perivascular glia (1136).

Astroglial CB₁ receptors are linked to regulation of metabolism: their activation increases glucose oxidation and ketogenesis in vitro (196, 1534). In astroglial cultures stimulation of CB₁ receptors inhibits NO production (1137) and release of TNF-α and chemokines in LPS-activated astrocytes (1601). In astrocytes from mouse stratum radiatum pressure, application of 2-arachidonylglycerol (2AG), arachidonylethanolamide (AEA), (R)-(+)-methanandamide (MAEA), or (R)-(+)-WIN 55,212–2 (WIN) triggered [Ca²⁺]_i transients that were inhibited by 2 μM of the selective antagonist of CB₁ receptors AM251. The CB₁-mediated astroglial Ca²⁺ signaling was also reportedly triggered by depolarization-induced neuronal release of endocannabinoids (1197). Subsequently, it was shown that glutamate released from astrocytes acts on presynaptic mGluRs to increase neurotransmitter release in the CA1 and CA3 hippocampal regions (1198). A similar mechanism may also induce presynaptic LTP (592). At the same time, acute stimulation of the CB₁ receptors triggered LTD and impaired working memory in mice; these effects were eliminated in conditional astroglia-specific CB₁ knockout animals (634).

H. Neuropeptide Receptors

1. Vasopressin and oxytocin receptors

Vasopressin induced [Ca²⁺]_i transients in about two-thirds of cultured astroglia from circumventricular organs; these were sensitive to the broad V₁ receptor antagonist d(CH2)5[Tyr(Me)2]8-arginine-vasopressin (810). Similarly, V₁ receptors mediated vasopressin-induced Ca²⁺ signaling in pituicytes (662). There is some regional heterogeneity in astroglial receptors expression: hippocampal astrocytes in vitro were reported to predominantly express V_1b receptors, whereas cortical astrocytes express mainly V_1a receptor; activation of both subtypes resulted in [Ca²⁺]_i elevation and stimulated glutamate release (1709). Activation of V_1a receptors in embryonic cortical astrocytes also caused activation of PKC, CaMKII, and ERK1/2 signaling cascades (1949), while significantly decreasing gene expression of cytokines including IL-1β and TNF-α (1948). These effects were mediated through cAMP response element-binding protein (CREB) activation (1948).

Oxytocin receptors were initially detected in cultured astrocytes using autoradiography (425); subsequently, the functional link of these receptors to PLC/InsP₃/endoplasmic reticulum Ca²⁺ release has been directly demonstrated in rat embryonic hippocampal astrocytes in vitro (424). Similar oxytocin-induced Ca²⁺ signaling was demonstrated in cultures of female postpubertal hypothalamic astrocytes; incidentally, oxytocin-induced [Ca²⁺]_i transients were blocked by mGluR1 antagonist and potentiated by mGluR1 agonist, which led the authors to suggest that oxytocin receptors are linked to Ca²⁺ signaling through mGluRs (926). Expression of astroglial oxytocin receptors was controlled by neuronal factors in a rather peculiar way: neuronal conditioned medium increased (probably through transforming growth factor-β, TGF-β), whereas neuronal membranes fraction decreased the level of oxytocin receptors mRNA and number of oxytocin binding sites (1129).

2. Endothelin receptors

There are three endothelins (ET-1, ET-2, and ET-3) and two types of endothelin receptors, ET_A and ET_B, both belonging to a metabotropic 7-transmembrane domain G protein superfamily. The ET-1 is the preferential ligand for ET_A receptors, while all endothelins are equipotent at ET_B receptors. In the healthy brain, overall astroglial expression of endothelin receptors is relatively low (812, 1376, 1487). In primary astroglial cultures, ET_A and ET_B receptors have been identified by autoradiography (724, 1801). Activation of ET_A/B receptors triggered [Ca²⁺]_i transients in astroglia (200, 1068). In mouse cerebellar Bergmann glial cells, application of 100 nM of all three isoforms of endothelin triggered [Ca²⁺]_i transients originating from endoplasmic reticulum Ca²⁺ release; these [Ca²⁺]_i transients were inhibited by the selective ET_B receptor antagonist BQ-788 (1767).

Stimulation of ET_A/B receptors inhibits astroglial gap junctions due to dephosphorylation of Cx43. This inhibition was found in culture (200, 569) and in hippocampal slices (201), using the dye transfer technique as well as paired patch voltage-clamp recordings (1102). Blockade of gap junctions led to a subsequent inhibition of intercellular Ca²⁺ waves and intercellular diffusion of glucose (199). Inhibition of gap junctional connectivity occurred rapidly, with complete block of dye transfer in minutes; however, the connectivity recovered within ~1.5 h in the presence of ET-1 reflecting desensitization of ET_A receptors. Chronic activation of ET_A/B receptors with ET-1 downregulated Cx43 expression, increased astroglial glucose uptake, and boosted astroglial proliferation. Furthermore, activated endothelin receptors increased protein synthesis and changed astroglial morphology in vitro (658). It should be noted that in pathology astrocytes (which normally do not produce endothelin) start to express and release ET-1 and upregulate expression of endothelin receptors (737).

3. Pituitary adenylyl cyclase activating polypeptide and vasoactive intestinal peptide receptors

Pituitary adenylyl cyclase activating polypeptide (PACAP) is a member of the secretin/glucagon/vasoactive intestinal peptide (VIP) family acting through specific receptors classified as PACAP-selective PAC₁, VPAC₁, and VPAC₂ receptors (286). All three types of receptors were immunolocalized in astrocytes throughout the brain (803); in cultured astrocytes, however, predominate expression of mRNA for PAC₁ and VPAC₂ receptor was detected (70, 605; see also Ref. 1077) for additional references). All three receptor types are linked to multiple signaling cascades; all three activate adenylyl cyclase (656, 1077). PAC₁ receptors are, in addition, coupled to PLC and hence stimulate InsP₃ production and Ca²⁺ release from the endoplasmic reticulum (1733, 1734). Activation of PAC₁ receptors is also linked to phosphorylation of ERK1/2 and possibly to activation of PKA and PKC (1077). Exposure of cultured astrocytes to picomolar concentrations of PACAP significantly increases their proliferation (656), whereas inhibition of PACAP/VIP receptors suppressed neocortical astrogenesis (1973). Both PACAP and VIP stimulate astroglial glycogenolysis (1653) and induce morphological plasticity as manifested by increased stellation of cultured astroglia (1373).

4. Natriuretic peptide receptors

Atrial natriuretic peptide (ANP), brain natriuretic peptide, and C type natriuretic peptide (CNP) are present in the CNS (1405). They act through binding to natriuretic peptide receptors (NPRs). ANP binds preferentially to NPR-A, while brain natriuretic peptide and CNP bind to NPR-B receptors; all NPs bind with equal affinity to NPR-C (1017). The ANP receptors were first identified in cultured mouse astrocytes by measuring ¹²⁵I-ANP binding (1739). Subsequent experiments revealed astroglial expression of all three types of natriuretic peptide receptors: the NPR-A (213, 1698), NPR-B (1698, 1967), and NPR-C (1631, 1698). NPR-A and NPR-B are plasmalemma-bound guanylyl cyclase receptors, which mediate intracellular signaling by increasing intracellular cGMP. The NPR-C is a “clearance receptor” that removes peptides from the extracellular space, but does not itself possess guanylyl cyclase activity (1405).

5. Angiotensin II receptors

In primary cultures astrocytes mainly express AT₁ angiotensin II receptors (1698); albeit in the healthy brain the expression levels are quite low (1245). The AT₁ is a G protein-linked receptor that induces STAT1 and STAT2 phosphorylation with further recruitment of activation and mobilization of the transcription factor NF-κB. Signaling to astrocytes through AT₁ receptors seems to contribute to the regulation of permeability of the blood-brain barrier (538). Expression of AT₁ receptors is upregulated in various forms of pathology, such as heart failure (768) or lesioning of afferent axons with subsequent synaptic degeneration (538).

6. Somatostatin receptors

Somatostatin receptors have been identified in astrocytes in situ in the hippocampus, amygdala, and hypothalamus by colocalization of a somatostatin-gold conjugate and GFAP immunoreactivity (907, 1107). In the hippocampus, both parenchymal and perivascular astrocytes in the stratum oriens and stratum radiatum of CA1 and the dentate gyrus bind the somatostatin conjugate (1107), with receptors being present in both astrocytic processes and cell bodies. Expression of somatostatin transmembrane receptors (SSTR) was analyzed at the mRNA level, and it appeared that rat and human cortical astrocytes express SST1, SST2, and SST4 receptor subtypes (501); in hypothalamic astroglial cultures only the SST2 subtype was detected (1825).

7. Opioid receptors

First evidence for opioid receptors in astroglia derived from analyzing binding of the opioid ligand [³H]etorphine and κ-receptor ligand [³H]ethylketocyclazocine in cryosections of the pituitary gland; this binding increased significantly after sectioning the pituitary stalk that resulted in gliotic response. On the basis of this observation, the predominant localization of κ-opioid receptors in astrocytes (pituicytes) was advocated (256). Astroglial expression of opioid receptors was also suggested in experiments demonstrating that morphine inhibited norepinephrine-induced accumulation of glucose into rat cortical astrocytes in vitro (1346). Slightly later autoradiography (with [³H]naloxone) demonstrated opioid receptors in astrocytes cultured from embryonic chick forebrain (1043) and operational δ- and κ-opioid receptors were contemplated for cultured rat astrocytes from cortex, striatum, and brain stem on a basis of specific agonists/antagonists effects on cAMP production (483, 484). Selective stimulation of κ-opioid receptors induced [Ca²⁺]_i transients, which were blocked by nifedipine, although the exact mechanism remains unclear (485). Activation of astroglial δ-receptors, however, triggered [Ca²⁺]_i elevation associated with endoplasmic reticulum Ca²⁺ release (1746). Transcripts of δ-, κ-, and μ-opioid receptors were found in astroglial cultures from cortex, striatum, cerebellum, hippocampus, and hypothalamus, with overall prevalence of δ- and κ-receptors and minor expression of μ-type (1521). Similarly μ-receptors were not detected in astrocytes in situ in rat spinal cord (831). At the same time μ-receptors in immortalized rat cortical astrocytes regulate expression of isoforms of thrombospondins (1387). Activation of opioid receptors exerts various effects on astroglial physiology; for example, by regulating expression of glutamate transporters (983), or affecting astroglial growth (1673) or activating ERK/MAPK phosphorylation signaling cascade (128).

8. Tachykinin receptors

All three metabotropic tachykinin receptors, the NK1 sensitive to substance P, NK2, activated by neurokinin A and NK3, activated by neurokinin B are expressed in astroglia. In cultured astrocytes activation of NK1 receptors by substance P inhibits K⁺ channels while activating Cl⁻ channels, which leads to depolarization (83). In astrocytes in situ activation of NK1 receptors with 1 μM substance P triggered [Ca²⁺]_i transients in ~15% of all cells tested (651).

9. Bombesin receptors

Bombesin, originally discovered in the frog skin (38), is represented in mammals by neuromedin B, a gastrin-releasing peptide, and neuromedin C. These peptides and their receptors are found across the CNS. Bombesin binding sites were found in astrocytes in rat explant cultures from cortex, cerebellum, brain stem, and spinal cord (728), whereas bombesin BB2 receptors transcripts were identified in cultured human astrocytes (1078). Stimulation of these receptors with neuromedin C and neuromedin B (the latter being 50 times less potent than the former) induced [Ca²⁺]_i transients, membrane hyperpolarization, and (under voltage-clamp) outward (presumed K⁺) currents with E_rev approximately −80 mV (1078).

I. Receptors for Leptin and Insulin

Leptin is the “anorexic adipokine,” the cytokine produced by adipose cells and released into the circulation. Leptin acts on hypothalamic neural circuitry to curb food intake. The leptin receptors [LepR or Ob(ese)R; obese protein being another name for leptin] are membrane spanning receptors of class 1 cytokine receptors family linked to JAK/STAT, PI₃K-Akt, or ERK signaling cascades (1730). Probably the first observation of astrocyte-associated Lep-Rs was obtained by double anti-GFAP and anti-LepR immunostaining of the arcuate nucleus and median eminence (322). Subsequently, leptin receptors were immunolocalized in GFAP-positive astrocytes in the subcommissural organ of rabbits (377) and in GFAP-positive glia of the nucleus tractus solitarius (378). Strong astrocytic presence of leptin receptors (again revealed by specific immunoreactivity) was found in mice living on a high-fat diet (742) or in mutant mice predisposed to obesity (1307). The LepR mRNA was localized in astroglia in healthy rat hypothalamus with double-labeling fluorescent in situ hybridization; the LepR protein was colocalized with GFAP-positive astrocytic profiles (743). Incidentally, in adult-onset obesity hypothalamic astrocytes and not neurons show the most prominent increase in LepR expression (742, 1307), which possibly indicate a special role for astrocytes in regulation of food-related behaviors (see sect. XIIM). Leptin receptors contribute to regulation of astroglial glutamate transport and glucose uptake (539).

Insulin receptors (IRs) have been detected in rodent and human astrocytes in primary cultures (673, 1958). Astrocyte-specific (using GFAP and EAAT1 promoters) conditional deletion of astroglial IRs impaired brain glucose sensing, changed astroglial morphology, decreased astroglial coverage of hypothalamic neurons, and suppressed glucose transport into the brain (549).

J. Complement Receptors

Complement, which consists of ~20 plasma proteins and another ~10 membrane proteins, represents an important part of the humoral immune system. Complement activation produces C3 convertase, which in turn activates the third complement component (C3). Subsequent processes result in generation of small fragments with signaling properties, represented by C3a, C3b, and C5a molecules (1354). Receptors for C3a (557, 767, 1615) and C5a (555, 1196) have been detected in human fetal astrocytes and in rodent astroglia; the C5a-like receptor was also identified in rat astrocytes (559). The immunoreactivity for CR1 receptors, which regulate phagocytosis of various particles tagged with C3b, C4b, andC1q, has been found to preferentially colocalize with astroglia in the human brain (514, 556).

K. Platelet-Activating Factor Receptor

Platelet-activating factor (PAF) (1-O-alkyl-2-acetyl-sn-glycero-3-phosphocholine) is a potent biologically active molecule, discovered initially as an activator of platelets; it is operational in various tissues including CNS. Receptors for PAF were, for the first time, detected in cultured astrocytes using binding to specific ligand [³H]WEB 2086 (239). Subsequently, it was found that picomolar concentrations of PAF stimulate formation of InsP₃ in astroglia in vitro (1180, 1384) as well as promote concentration-dependent astroglial secretion of NGF (240) and release of prostaglandin E₂ from cortical cultured astroglia (1738).

L. Protease-Activated Receptors

Protease-activated receptors are G protein-coupled metabotropic receptors activated following cleavage of a part of extracellular domain by serine proteases; three members of this family (PAR-1, PAR-3, and PAR-4) are activated by thrombin, whereas PAR-2 is a trypsin receptor (1639). Existence of receptors for thrombin in astroglia was suggested following early observation of thrombin-induced stimulation of proliferation and morphological remodeling (599, 1374). Thrombin-activated PAR-1 and trypsin-activated PAR-2 receptors have been initially detected at mRNA level in primary cultured rat newborn astrocytes; activation of these receptors induced [Ca²⁺]_i transients reflecting endoplasmic reticulum Ca²⁺ release (1770, 1772). The astroglial PAR-1 receptor demonstrated a peculiar mode of desensitization, which involved cleavage of the NH₂ terminus (1771). Subsequent analysis revealed expression of transcripts and proteins for all four (PAR1−4) receptors in cultured astroglia, and all four receptors were linked to generation of Ca²⁺ signals (1855). Astroglial PAR receptors are linked to several intracellular signaling pathways, which include production of ROS and stabilization of hypoxia inducible factor-1α through ERK, JNK, and PI3K/Akt cascades (1962). Activation of PAR-1 receptors (for example, by selective peptide agonist TFLLR) is often used for selective stimulation of astroglia in situ (933, 1607), despite the fact that the very same TFLLR-sensitive receptors are known to trigger neuronal Ca²⁺ signaling (635).

M. Ephrin Receptors

Ephrins and their receptors (EphA and EphB) contribute to remodeling synaptic contacts (1174); ephrin receptors are tyrosine kinases activated by membrane-associated ephrin ligands. These Eph receptors (as well as ephrins) have been found in perisynaptic astrocyte processes (290, 510). Activation of EphA receptors induced formation of phylopodia and release of glutamate from astroglial cells in organotypic hippocampal slices (1207), while in cultured astrocytes activation of EphB3 and EphA4 receptors triggered secretion of d-serine (1963).

N. Succinate Receptors

The citric acid cycle intermediate succinic acid may rise quite high [from basal level of 5 μM up to 125 μM (709)] in tissues in physiology (for example during exercise) and in disease (for example in hypertension and metabolic diseases). Receptors for succinate known as SUCNR1 (GPCR91) are classical G protein-coupled metabotropic receptors, and they have been localized (at transcript level) in astrocytes (630). Stimulation of these receptors with succinate (with EC₅₀ ~50−60 μM) triggered [Ca²⁺]_i transients in ~15% of astrocytes from acute slices of nucleus accumbens (1138).

O. Toll-like Receptors

Toll-like receptors (TLRs), deeply involved in various defensive mechanisms associated with neuroglia, are type 1 transmembrane glycoproteins containing a leucine-rich repeat (LRR) domain and a Toll/IL-1 receptor (TIR) domain. These receptors are linked to multiple intracellular signaling systems through five intracellular “adaptor” signal transmitting proteins, which, for example, include myeloid differentiation factor 88 (MyD88) that links TLRs to NF-κB, TNF-α, or interleukin cascades (391, 1759). Although expression of TLRs 1−9 was found in cultured mouse astrocytes, the TLR3 (at least in physiological conditions) seems to be the most abundant (292, 1097). Astroglial expression of TLR3 was detected in vitro, in situ, and in vivo (498). In human astrocytes expression of a TLR1−5 and TLR9 has been reported (250, 779); of these TLR2−4 were identified also in vivo, especially in pathological conditions (250, 779, 1695). In general, expression of TLRs remarkably increases in diseased CNS and in reactive astroglia (1759). In the latter, TLR3 mediate production and secretion of numerous neuroprotective factors such as ciliary neurotrophic factor, neurotrophin-4, and vascular endothelial growth factor, consequently, activation of TLR-3 boosts neuronal survival in organotypic brain cultures (250).

P. Steroid Receptors

Classical nuclear estrogen receptors of α- and β-types (ERα or ERβ) have been found in astrocytes in primary and explant cultures from several brain regions (303, 730, 1545); ERβ receptors were localized by immunostaining in the processes and perikarya of astrocytes of hippocampal tissue (78). Estradiol triggered Ca²⁺ signaling in cultured astroglia (303) through transactivation of mGluR1 and subsequent InsP₃-induced Ca²⁺ release from the endoplasmic reticulum (925). The ERα and ERβ may localize not only intracellularly, but also at the plasma membrane; in addition specific membrane ERs, classified as G protein-coupled estrogen receptor (GPER or GPR30), the Gq-membrane ER (the Gq-mER) and the ER-X have been characterized (6). The GPER and Gq-mER have been detected in astrocytes (26, 924), where they were linked to generation of cytoplasmic Ca²⁺ signals (924) or modulation of expression of EAAT2 glutamate transporters (836). In cultured astrocytes activation of membrane α-ER receptors by estradiol and related compounds was shown to either inhibit glutamate uptake (1545) or stimulate it by increasing EAAT1/2 expression (836). Astrocytes also express nuclear progesterone receptors (PR), membrane progesterone receptors of the PR_A type, and progesterone receptor membrane component-1 (Pgrmc1) (930, 1866) as well as androgen receptors, which were found in primary cultured astroglia (806) and in minor populations of astrocytes in situ (446).

Q. Recapitulation

Numerous receptors expressed in astrocytes safeguard their ability to monitor neurochemical environment and tune homeostatic responses to the requirements of the nervous tissue. Astroglial receptors are linked to intracellular ionic signaling and astroglial metabolism through both ionotropic and metabotropic signaling cascades. Astrocytes follow synaptic transmission by a wide assortment of receptors to neurotransmitters and neuromodulators. The modality and types of these receptors vary between brain regions and show developmental plasticity; the immediate neurotransmitter environment most likely regulates expression of receptors. Notably in the adult awake behaving brain, astroglial calcium signaling is almost exclusively governed by the noradrenergic input from the locus coeruleus.

IX. ASTROGLIAL MEMBRANE TRANSPORTERS

A. ATP-Dependent Transporters

1. P-pumps

a) the Na⁺-K⁺-ATPase, or NKA.

The NKA is arguably the most prominent astroglial P-pump, which catalyzes transmembrane movement of Na⁺ and K⁺ with stoichiometry of 3Na⁺ (expelled from the cell):2K⁺ (imported into the cell). Unequal charge transfer across the membrane stipulates electrogeneity of NKA; as the pump removes 1 positive charge from the cell, the net outward current is generated resulting in a moderate hyperpolarization. The NKA is composed from several subunits: the principal catalytic α subunit, which binds ATP and the selective antagonist ouabain, assembles with β-subunit required for pump operation; the associated γ subunit may regulate the affinity of NKA to K⁺ and Na⁺ (832, 1697). The NKA subunits exist in several isoforms (α1−α4 and β1−β3) with distinct tissue and cell localization (191); in the CNS, the α1 subunit is expressed in all cell types, the α2 subunit is present in astrocytes, whereas the α3 subunit is preferentially localized in neurons (690, 805).

Cell specific expression of α isoforms of NKA defines the functional properties of neuronal and astroglial NKA. The affinity of the α2-containing pump to K⁺ is substantially lower compared with NKA composed of α1 and α3 isoforms (the [K⁺]_0.5 for α2β1 composition is ~3.6 mM, whereas [K⁺]_0.5 for α1β1, α1β2, α3β1, and α3β2 assemblies vary between 0.25 and 0.65 mM; Ref. 942). Consequently, astroglial NKA is activated by physiological rises in [K⁺]_o, whereas neuronal NKA, being fully saturated at physiological [K⁺]_o, is activated solely by an increase in [Na⁺]_i (625, 674, 686, 942, 1496). This distinct sensitivity to extracellular K⁺ defines the contribution of astroglial NKA to K⁺ buffering (see sect. XIIA). The astroglial NKA is stimulated by norepinephrine through β-adrenoceptors (625) and regulated by endogenous ouabain-like molecules (690). Loss of function mutation of α2 subunit of astroglial NKA causes deficits in K⁺ buffering and (through downregulation of glutamate transporters expression) in glutamate uptake, which underlies increased susceptibility to spreading depression and a subtype of migraine with aura, known as familial hemiplegic migraine type 2 (283).

b) Ca²⁺-ATPases.

Plasmalemmal Ca²⁺-ATPases (PMCAs) or Ca²⁺ pumps extrude excess of Ca²⁺ into the extracellular space thus contributing to the cell Ca²⁺ homeostasis (1814). The pump operates as a Ca²⁺-H⁺ exchanger with stoichiometry 1Ca²⁺:1/2H⁺; each transportation cycle is associated with a hydrolysis of one ATP molecule (234). Cultured cerebellar astrocytes were found to express (at both mRNA and protein levels) PMCA1, PMCA4 and to a lesser extend PMCA2 (530). Another Ca²⁺ pump, the sarco(endo)plasmic reticulum Ca²⁺-ATPase (SERCA) is operational in the membrane of the endoplasmic reticulum; it is solely responsible for maintaining high intra-endoplasmic reticulum Ca²⁺ concentration. Astrocytes predominantly express the SERCA2b subtype of the pump (1152).

2. Astroglial F- and V-type pumps

The F-type and V-type ion pumps are involved in proton transport. The V (vacuolar)-type H⁺ pumps are present in astrocytes in the plasmalemma (~35% of total amount), in lysosomes, and in secretory vesicles where they provide for an acidic intraluminal environment driving vesicular neurotransmitter transporters (1318, 1388). The F-type pumps are ubiquitously present in mitochondria being the fundamental part of ATP synthesis.

3. ABC binding cassette transporters

The ABC transporters represent an extended superfamily (in excess of 100 members) of ATP-dependent membrane transporters (654) subclassified into five groups (from A to G), which are expressed throughout the CNS. Astrocytes express several members of ATP cassette transporters including, for example, ABCA1, ABCA2, and ABCA3 transporters (863), P-glycoprotein (mdr1a, mdr1b), multidrug resistance-associated-protein (mrp1, mrp4, mrp5), etc. (92).

B. Secondary Transporters of Solute Carrier Family

1. Glutamate transporters

a) Na⁺-dependent glutamate transporters.

Astroglial glutamate transporters (known as excitatory amino acid transporters or EAATs) are members of SLC1 family of high-affinity glutamate and neutral amino acid transporters (see TABLE 4). Astrocytes specifically express EAAT1/SLC1A6 and EAAT2/SLC1A2, rodent versions of which (respectively) are often referred to as GLAST1 (glutamate-aspartate transporter 1; Ref. 1678) and GLT-1 (glutamate transporter 1; Ref. 1392). The EAAT2 transporter was the first to be identified in the rat brain (381); shortly after the EAAT1 was discovered (1727), and characterization of human variants followed (68). Both EAAT1 and EAAT2 are assembled as homotrimers, and several functional splice variants have been described (for a comprehensive account of the molecular structure and structure-function relations of glutamate transporters, see Ref. 1792). Astroglial glutamate transporters show differential expression in various brain regions: the EAAT1 is predominant in cerebellum (966), in the retina (1442), and in circumventricular organs (148); whereas EAAT2 prevails in all other parts of the brain. Expression of both transporters is rather high: the EAAT1 and EAAT2 account for 0.3 and 1.3% of total tissue protein in hippocampus and 1.8 and 0.3% for all tissue protein in cerebellum. The average density of EAAT1 is ~4,700/μm² in Bergmann glia and ~2,300/μm² in the CA1 region, whereas the density of EAAT2 is ~8,500/μm² in hippocampus and ~740/μm² in the cerebellum (966). The actual densities of both transporters are certainly much higher in perisynaptic membranes. The EM immunogold experiments clearly demonstrated that EAAT1 as well as EAAT2 are preferentially concentrated in astroglial membranes facing neuropil and their density is lower in the membrane portions facing neuronal somata or other astrocytes (313). Of note, ~10% of total EAAT2 is expressed in neurons (1956).

Table 4.

Astroglial SLC transporters

Transporter Type	SLC Classification	Stoichiometry	Properties and Function	Reference Nos.
Excitatory amino-acid transporters 1 and 2: EAAT1, EAAT2; in rodents known as GLAST and GLT-1	SLC1A3, SLC1A2	3Na+: 1H+: 1Glu (In): 1K+ (Out). Electrogenic	EAAT1/2 provide for ~80% of glutamate uptake in the CNS. Erev >10 mV; do not reverse in physiological conditions. Operation of EAATs results in substantial increase of [Na+]i.	380, 966, 1290, 1494, 1792, 1939
Sxc− cystine/glutamate antiporter	SCL7A11	1 Glu (Out): 1 Cys (In). Electroneutral	Exchanges glutamate for cystine; the latter is a main precursor for glutathione. Regulates extracellular glutamate in extrasynaptic regions. Electroneutral	23, 326, 844, 1162
Na+-coupled neutral amino acid transporters: SNAT 3, SNAT 5	SLC38A3, SLC38A5	1Na+ (In): 1Gln: 1 H+ (Out). Electroneutral	Exports glutamine from astrocytes.	680, 1552
GABA transporter: GAT-1/GAT-3	SLC6A12	2Na+:1Cl−: 1GABA (In), (?) 3Na+:1Cl−: 1GABA (In). Electrogenic	Normally removes extracellular GABA. Erev is close to resting Vm, can reverse in physiological conditions.	121, 356, 1044, 1121
Glycine transporter: GlyT1	SLC6A9	2Na+: 1Cl−: 1Gly (In). Electrogenic	Removes extracellular glycine. Erev is close to resting Vm, can reverse in physiological conditions.	491, 1932
Cationic amino acid transporter: Acs-1	SLC7A10	Electroneutral	Na+ -independent, high-affinity transport of small neutral d- and l-amino acids. In physiological conditions may transport glycine, l- and d-serine and l-cysteine.	470
Sodium-calcium exchanger: NCX1, NCX2, NCX3	SLC8A1, SLC8A2, SLC8A3	3Na+ (In): 1Ca2+ (Out). Electrogenic	Electrogenic exchanger; Erev close to Vm, operates in both forward and reverse modes.	195, 588, 878, 1303, 1461, 1725
Sodium-proton exchanger: NHE1	SLC9A1	1Na+ (In): 1H+ (Out). Electroneutral	Removes H+ from astrocytes; Erev >60 mV, hence cannot reverse in physiological conditions.	321, 402
Sodium-bicarbonate transporter: NBC	SLC4A4	1Na+: 2/3 HCO3− (In). Electrogenic	Erev close to Vm, operates in both forward and reverse modes.	244, 321, 402, 603, 1743
Sodium-potassium-chloride co-transporter: NKCC1	SLC12A2	1Na+: 1K+: 2Cl− (In). Electroneutral	Erev is determined by [Cl−]i, in physiological context; transport is always inward. May participate in clearing large K+ loads.	789, 818, 942, 1039
Mitochondrial sodium-lithium-calcium exchanger: NCLX	SLC8B1	3Na+ (In): 1Ca2+ (Out)	Main pathway for Ca2+ extrusion from mitochondria. In cultured astrocytes interacts with store-operated Ca2+ entry.	1327
Na+-dependent ascorbic acid transporter: SVCT2	SLC23A2	2Na+: 1Asc− (In). Electrogenic	Accumulates ascorbic acid into astrocytes; part of antioxidant defense.	149, 553, 896
Concentrative nucleoside transporter: CNT2, CNT3	SLC28A2, SLC28/A3	1 Na+: 1 nucleoside (In). Electrogenic	Accumulates adenosine into astrocytes. Part of adenosine homeostatic cycle.	975, 1363
Equilibrative nucleoside transporters: ENT-1, ENT-2, ENT-3, ENT-4	SLC29A1, SLC29A2, SLC29A3, SLC29A4	Electroneutral	Transports adenosine across plasmalemma according to its concentration gradient.	205
Na+-dependent dopamine transporter: DAT1	SLC6A3	1/2 Na+:1 DA: 1Cl− (In)	Uptake of dopamine; data on astroglial functional expression are controversial.	834, 886, 1722
Neutral amino acid transporter subtype ASCT2	SLC1A5	1Na+: 1d-Ser (In)	Transmembrane transport of d-serine.	1070
Norepinephrine transporter: NET/SLC6A2	SLC6A2	1NA: 2Na+ and 1 Cl−(In). Electrogenic	Uptake of norepinephrine and dopamine; has higher activity for dopamine.	765, 1577, 1722
Vesicular glutamate transporters: VGLUT1, VGLUT2, VGLUT3	SLC17A6, SLC17A7, SLC17A8	Unknown; is H+ and Cl− dependent	Glutamate accumulation in secretory vesicles.	153, 169, 368, 529, 902, 1144, 1279
Vesicular nucleotide transporter: VNUT	SLC17A9	Unknown	ATP accumulation in secretory vesicles.	1550
Vesicular polyamine transporter: VPAT	SLC18B1	Unknown	Spermine and spermidine accumulation in secretory vesicles.	693
Glucose transporters: GLUT1, GLUT2	SLC2A1, SLC2A2	Electroneutral	GLUT1 is the main astroglial glucose transporter.	22, 967, 1151
Sodium-dependent glucose transporter		1Glucose: 2 Na+ (In)	Detected only in astrocytes in vitro.	1799
Monocarboxylate transporters: MCT1, MCT4	SLC16A1, SLC16A3	Electroneutral	Lactate-exporting transporters.	151, 1422
Monocarboxylate transporter: MCT2	SLC16A7		Lactate-importing transporter. Was identified in astroglial endfeet in the brain stem and in the corpus callosum.	1356, 1422
Zink transporter: ZnT	SLC30		Main Zn2+ transporter.	1590

Astroglial glutamate transporters utilize the transmembrane Na⁺ gradient as the driving force. Translocation of a single glutamate molecule is coupled with cotransport of 3 Na⁺ and 1 H⁺ and with countertransport of 1 K⁺ (FIGURE 18; Refs. 1290, 1939). With the use of this stoichiometry, both the equilibrium potential for the transporter and glutamate concentrating capacity could be calculated using modified Nernst equation:

$$E_{\text{EAAT}}=RT/\hspace{0.17em}F\left[3\left(\text{Na}\right)+1\left(\text{H}\right)-1\left(\text{K}\right)-1\left(\text{Glu}\right)\right]\times \text{ln}\left\{3\text{ln}\left({\left[{\text{Na}}^{+}\right]}_{\text{o}}\hspace{0.17em}/\hspace{0.17em}{\left[{\text{Na}}^{+}\right]}_{\text{i}}\right)+\text{ln}\left({\left[{\text{H}}^{+}\right]}_{\text{o}}\hspace{0.17em}/\hspace{0.17em}{\left[{\text{H}}^{+}\right]}_{\text{i}}\right)+\text{ln}\left({\left[{\text{Glu}}^{-}\right]}_{\text{o}}\hspace{0.17em}/{\hspace{0.17em}\left[{\text{Glu}}^{-}\right]}_{\text{i}}\right)-\text{ln}\left({\left[{\text{K}}^{+}\right]}_{\text{o}}\hspace{0.17em}/\hspace{0.17em}{\left[{\text{K}}^{+}\right]}_{\text{i}}\right)\right\}$$

FIGURE 18.

Stoichiometry of astroglial solute carrier (SLC) neurotransmitter transporters. EAAT1,2 , excitatory amino acid transporters 1 and 2; Sxc, cystine-glutamate exchanger; GAT1,3, GABA transporters 1 and 3; GlyT1, glycine transporter; ENT, equilibrative nucleotide transporters; CNT, concentrative nucleotide transporters; NET, norepinephrine transporter.

The E_EAAT is defined by transmembrane ion gradients and by cytosolic and extracellular glutamate concentrations (877, 1939). The extracellular concentration of glutamate in resting conditions is around 25 nM (677), the cytosolic concentration of glutamate in neurons is usually assumed to be in a range of 1−10 mM. In astrocytes, however, glutamine synthetase catalyzed conversion of glutamate into glutamine ascertains much lower glutamate levels determined at around 0.3 mM (227). Hence, at the resting state, the E_EAAT lies at a positive membrane potential around ~9 mV, and reversal of the transporter is possible only in conditions of substantial increase in extracellular K⁺ and intracellular Na⁺. Such a reversal can be achieved in experimental conditions (1713), which however reflect a severe pathology associated with a loss of cellular ion homeostasis. In physiological conditions, the stoichiometry of the transporter is capable of supporting ~10⁶ transmembrane glutamate gradient and can operate as an uptake system even when transmembrane ion gradients are seriously perturbed. Transmembrane Na⁺ gradient contributes the most to setting E_EAAT: 10-fold changes in [Na⁺]_o shift the equilibrium potential by 100 mV, whereas 10-fold changes in [Glu⁻]_o, [H⁺]_o, and [K⁺]_o shift E_EAAT by 31, 25, and −31 mV, respectively (1939). Nonetheless, in the physiological context, the major role belongs to glutamate, the concentration of which in the synaptic cleft may increase by 5−6 orders in magnitude during synaptic activity. Increase in extracellular glutamate shifts E_EAAT toward positive values; at 1 mM of glutamate in the cleft, the transporter reverses at 145 mV. This is a strong safety factor that ensures sustained glutamate uptake regardless of ionic perturbations. Transporter stoichiometry also confers electrogeneity to the EAATs, which generate transmembrane current mainly carried by Na⁺ ions (877, 1792). This Na⁺ influx may elevate [Na⁺]_i by 10−20 mM (1777), thus contributing to cytoplasmic Na⁺ signals (see sect. XB).

Operation of astroglial glutamate transporters is also associated with thermodynamically uncoupled Cl⁻ flux through an aqueous pore (similar to a classical ion channel) that is an intrinsic part of the transporter molecule (1541, 1791). The Cl⁻ current reverses at E_Cl, and Cl⁻ flux is relatively minor for astroglial transporters, in contrast to EAAT4/5 which have a substantial Cl⁻ permeability. Nonetheless, activation of glutamate uptake was described as the major pathway for Cl⁻ efflux in cerebellar Bergmann glia (1778). The Cl⁻ flux pathway also allows movement of water that is accompanying operation of the transporter to compensate for redistribution of osmolytes (1790). It has been estimated that water flux through EAATs attains ~10% of that through aquaporins (1034, 1036).

Kinetics of the glutamate translocation by EAATs is relatively slow; both EAAT1 and EAAT2 transport ~30 molecules of glutamate per second (1286, 1956). The glutamate binding to the transporters (K_m ~20 μM) is however much faster, and hence glutamate transporters concentrated at the perisynaptic processes act as almost instant buffers for glutamate. The higher is the density of transporters, the higher is their buffering capacity (1792). The EAAT2 in cultured hippocampal astrocytes has a remarkable lateral mobility regulated by glutamate, possibly allowing a continuous exchange of glutamate-bound and unbound transporters to maintain high buffer capacity of perisynaptic zones (1177).

b) sxc⁻ cystine/glutamate antiporter.

The cystine/glutamate antiporter, in contrast to Na⁺-dependent glutamate transporters, exports glutamate from the cell in exchange for cystine (FIGURE 18). This exchanger, known as Sxc⁻, is a heteromeric amino acid transporter composed of xCT/SCL7A11 and 4F2hc/SLC3A2 proteins (243). In the CNS, Sxc⁻ is mainly expressed in astroglia (23, 326, 844), although it was also detected in some immature neurons, in microglia, in endothelial cells, in ependymal cells, in the choroid plexus, and in the leptomeninges (232). Expression of Sxc⁻ in cultured astrocytes is substantially (up to 7 times) upregulated by dibutyryl-cAMP (581, 1587) and by depletion of intracellular glutathione (1587). The Sxc⁻ antiporter provides cystine required for production of glutathione and contributes to regulation of extrasynaptic level of glutamate (see sect. XIID).

2. Glutamine transporters

Main astroglial transporters for glutamine belong to the SLC38 sodium-coupled neutral amino acid transporter (SNAT) family. Astroglial complement of transporters (also known as N system) is represented by SNAT3/SLC38A3 and SNAT5/SLC38A5; glutamine transport is coupled with cotransport of 1Na⁺ and countertransport of 1H⁺ being thus electroneutral (1552). Increase in [Na⁺]_i directly stimulates glutamine efflux (1752).These transporters are optimized for glutamine efflux; the neuronal system A (represented by SNAT1/SLC38A1, SNAT2/SLC38A2, and SNAT4/SLC38A4) is in contrast specialized for glutamine uptake, and it is electrogenic because it cotransports a single molecule of uncharged glutamine with 1Na⁺ (241). Incidentally, these electrogenic transporters depolarize neurons and hence glutamine may act as an extracellular messenger (1517). Interaction of N and A transporting systems underlie the operation of glutamine-glutamate(GABA) shuttle (see sect. XIID).

3. GABA transporters

Transporters for GABA belong to an extended family (~20 members) of SLC6 sodium- and chloride-dependent neurotransmitter transporters. The GAT1/SLC6A1, GAT3/SLC6A11, and betaine/GABA transporter-1 BGT-1/SLC6A12 have been identified in the CNS, with the first two members being the most important (1585, 1956). The GAT3 is considered to be the more or less specific astrocytic transporter; the GAT1 is predominantly neuronal, although it is also present in astroglia (356, 1044). There is evidence for expression of BGT1 transporter in cultured astrocytes (182, 1263); however, expression in situ has not been unequivocally confirmed. The GAT3 are almost exclusively (as judged by immunocytochemistry) concentrated in astroglial processes oriented towards neuropil which covers symmetric and asymmetric synapses close to neuronal somata, as well as basal and apical dendrites (1121, 1463). There are also some species differences: in rats, for example, GAT3 is localized to astrocytes, whereas in cats, monkeys, and humans substantial expression was found also in oligodendroglia (1406). In the cerebellum GAT3 is highly expressed in the perisynaptic processes of Bergmann glia, which are the main sink for GABA because Purkinje neurons lack GABA transporters (1121). Similarly, in the thalamus, GABA transporters (GAT1 and GAT3) are expressed only in astroglia (387); GAT1 is concentrated in membranes close to synapses, whereas GAT3 is localized more distantly, being thus responsible for extrasynaptic GABA transport (121).

Transport of GABA is thermodynamically coupled with symport of Na⁺ and Cl⁻ (FIGURE 18) with stoichiometry 1GABA:2Na⁺:1Cl⁻ (GABA being a zwitterion does not carry any charge); as the result, GABA translocation is associated with net influx of a positive charge (846, 1013, 1420). There is some evidence that Cl⁻ fluxes can be equilibrated, hence reducing the effective stoichiometry to 1GABA:2Na⁺, although this matter remains unresolved (1007). With the assumption of high cytosolic Cl⁻ content in astroglia, the contribution of Cl⁻ to transporter equilibrium may however be relevant. Very recently, the classical stoichiometry of GABA transporters was challenged by experiments on Xenopus oocytes expressing GAT1 and GAT3; shifts in reversal potentials evoked by 10-fold changes in [Na]_o, [Cl]_o, and [GABA]_o hinted for 1GABA:3Na⁺:1Cl⁻ coupling stoichiometry (1881).

The reversal potential of GABA transporter (as calculated by Nernst equation) is

$$E_{\text{GAT}}=RT/\hspace{0.17em}F\left[2\left(\text{Na}\right)-1\left(\text{Cl}\right)\right]\times \text{ln}\left\{2\text{ln}\left({\left[{\text{Na}}^{+}\right]}_{\text{o}}\hspace{0.17em}/\hspace{0.17em}{\left[{\text{Na}}^{+}\right]}_{\text{i}}\right)+\text{ln}\left({\left[\text{GABA}\right]}_{\text{o}}\hspace{0.17em}/\hspace{0.17em}{\left[\text{GABA}\right]}_{\text{i}}\right)+\text{ln}\left({\left[{\text{Cl}}^{-}\right]}_{\text{i}}\hspace{0.17em}/\hspace{0.17em}{\left[{\text{Cl}}^{-}\right]}_{\text{o}}\right)\right\}$$

E_GAT lies around −50 mV for 1GABA:2Na⁺:1Cl⁻ coupling; the extra Na⁺ in stoichiometry obviously shifts E_GAT to more positive (about −10 mV) values. This is functionally important because E_GAT at −50 mV implies that relatively small depolarization and/or an increase in [Na⁺]_i can turn the transporter into the reverse mode, which have been experimentally demonstrated in neocortical slices (1777). Extracellular concentration of GABA also substantially affects the E_GAT; increase in [GABA]_o shifts the equilibrium potential to positive values thus favoring GABA uptake. Conceptually, however, the probability for astroglial GABA release in vivo in physiological conditions remains questionable (see sect. XIE). To conclude, precise determination of stoichiometry of astroglial GABA transporter is ultimately needed. There is some evidence for lateral mobility of GABA transporters; however, whether it occurs in astrocytes and what is its functional significance remains a matter for speculation and modeling (1584). Finally, membrane presence of GABA transporters can be regulated by internalization (1585).

4. Glycine transporters

Sodium-chloride-dependent plasmalemmal glycine transporters are represented by two varieties, GlyT1/SLC6A9 and GlyT2/SLC6A5. It is generally accepted that the GlyT1 belongs to astroglia (1932), whereas GlyT2 operates in neurons, in glycinergic nerve terminals (491), and in glutamatergic neurons where it possibly regulates extracellular glycine needed for modulation of NMDA receptors (165). Cellular discrimination of receptors reflects different functional roles, with glial GlyT1 being mainly responsible for removal of glycine and termination of transmission and neuronal GlyT2 being responsible for replenishment of releasable pool of glycine in the terminal. There are, however, some indications for GlyT2 expression in astroglia: these transporters were found in the so-called gliosomes (subcellular particles) obtained from spinal cord astrocytes (1423), in cultured rat cortical cultured astrocytes (at mRNA and protein levels), and in brain slices by double immunostaining with anti-GlyT2 and anti-GFAP antibodies (67). Incidentally, there are also some hints for GlyT2 compensating the GlyT1, after the latter was specifically and conditionally knocked out in astroglia (492).

Glycine transporters have a distinct stoichiometry: the GlyT1 cotransports a single molecule of glycine together with 2Na⁺ and 1Cl⁻ (FIGURE 18), whereas GlyT2 has the stoichiometry 1Gly:3Na⁺:1Cl⁻ (1510). The equilibrium potential for GlyT1 is defined by extracellular concentration of glycine and intracellular concentration of Na⁺, and hence can be reversed at physiological membrane potentials. This reversal was demonstrated electrophysiologically in heterologous expression system (76). Release of glycine through reversed transporter was also shown on cultured cortical astrocytes (1605). This release was stimulated by dopamine which rapidly increased glycine synthesis; subsequent increase in cytoplasmic glycine concentration arguably led to a transporter reversal. Release of glycine following reversal of GlyT1 induced by 20-mV depolarization of Bergmann glial cells was found in cerebellar slices (747). The GlyT2 in contrast has a very positive reversal potential; in addition, having an extra Na⁺ molecule for coupled transport makes the driving force for glycine uptake for GlyT2 two orders of magnitude larger, which allows neuronal transporter to maintain intraterminal glycine concentration ~20−40 mM needed for low-affinity (K_D ~20 mM) vesicular glycine transporter (1704). Astrocytes in the brain stem and spinal cord also express high densities of Acs-1/ SLC7A10 amino acid transporters, which mediate efflux of glycine (470).

5. Adenosine transporters

Adenosine transport in astrocytes is mediated by two classes of transporters (FIGURE 18): the plasmalemmal equilibrative [i.e., controlled by adenosine transmembrane gradient (870)], nucleoside transporters ENT-1/SLC29A1, ENT-2/SLC29A2, ENT-3/SLC29A3, and ENT-4/SLC29A4 and Na⁺-dependent concentrative nucleoside transporters CNT2/SLC28A2 and CNT3/SLC28A3, which cotransport adenosine together with 1 molecule of Na⁺. These latter transporters have been indentified in cultured (1363) and freshly isolated astrocytes (975).

6. Transporters for monoamines

There are numerous reports indicating that astrocytes in vitro accumulate norepinephrine (642, 727), which uptake (as well as uptake of dopamine) occurs in Na⁺-dependent manner (1360, 1591), indicating the role for dedicated transporter. This most likely is represented by norepinephrine transporter NET/SLC6A2 that couples monoamine transport with 2Na⁺ and 1Cl⁻ (FIGURE 18). The NET transports both monoamines, with higher affinity for dopamine than norepinephrine (1294). This transporter was identified in astrocytes at mRNA, protein, immunostaining, and functional levels (765, 1577, 1722). Data on expression of dedicated dopamine transporter DAT/SLC6A3 remain controversial with evidence for (834, 1722) and against (886) its presence in astrocytes.

7. d-Serine transporters

Transmembrane transport of d-serine in astrocytes is mainly mediated by a neutral amino acid transporter subtype ASCT2 (SLC1A5), which is an alanine-, serine-, cysteine-preferring neutral amino acid transporter (1070). The ASCT2 is Na⁺ dependent, with Na⁺ to amino acid stoichiometry of 1:1 (1091).

8. Sodium-calcium exchangers

a) plasmalemmal sodium-calcium exchanger, or ncx.

Plasmalemmal Na⁺-Ca²⁺ exchanger (NCX) is a fundamental component of cellular Ca²⁺ homeostasis. The NCX utilizes transmembrane Na⁺ gradient for export of Ca²⁺ from the cells against concentration gradient; these exchangers are expressed in the majority of cell types (human erythrocytes being an exception). An important feature of the NCX mechanism is its ability to operate in both forward (Ca²⁺ extrusion) and reverse (Ca²⁺ influx) modes, the switch between these modes being determined by membrane potential and transmembrane Na⁺ gradient (195). It was indeed the reverse mode of NCX operation that was initially discovered in cultured astrocytes using cellular imaging (193) and ⁴⁵Ca²⁺ uptake (1724).

The NCX belongs to an SLC8 family (1026) represented by three gene products NCX1/SLC8A1, NCX2/ SLC8A2, and NCX3/SLC8A3, all of which are expressed in astroglia, with some evidence indicating higher expression of NCX1 (1317). Another type of the Na⁺-Ca²⁺ exchanger, the K⁺-dependent NCKX/SLC24, is expressed exclusively in neurons and has not been detected in astrocytes (29). The NCX proteins are mainly concentrated in astroglial peripheral and perisynaptic processes, where they are often colocalized with glutamate transporters and possibly with glutamate ionotropic receptors (194, 1120).

The stoichiometry of astroglial NCX is 3Na⁺:1Ca²⁺, and hence the equilibrium potential of the exchanger could be calculated from Nernst equation, which can be reduced to the simple formula E_NCX = (nE_Na – 2E_Ca)/(n − 2), where n is a stoichiometry of Na⁺, while E_Na and E_Ca are equilibrium potentials of Na⁺ and Ca²⁺, respectively. Assuming [Ca²⁺]_i of 50−80 nM and [Na⁺]_i of 15 mM, the E_NCX could be as negative as about −85 to −90 mV, being thus very close (or even slightly more negative) to the resting membrane potential of astrocytes. The E_NCX of course is profoundly influenced by fluctuations in [Ca²⁺]_i and [Na⁺] in the cytoplasm and in the extracellular space, and hence the operational mode of NCX seems to dynamically fluctuate between forward and reverse transport. Conceptually, depolarization and increase in intracellular Na⁺ concentration favor NCX operation in the reverse mode, whereas increase in [Ca²⁺]_i promotes the forward mode of the exchanger. The role of NCX in Ca²⁺ extrusion and shaping [Ca²⁺]_i transients has been directly demonstrated in vitro (588, 1725) and in situ (878); inhibition of NCX transport increased the amplitude and prolonged the recovery of [Ca²⁺]_i responses to metabotropic agonists. The reverse mode of NCX has been similarly documented (878, 1303, 1461); Ca²⁺ influx through the NCX could attain excitotoxic proportion and trigger astroglial death (209, 873). Finally, the reversal of NCX by artificial increase in [Na⁺]_i was demonstrated in astrocytes expressing channelrhodopsin2; activation of the latter by light increased cytosolic Na⁺, which in turn led to Ca²⁺ influx through the NCX (1912).

b) mitochondrial sodium-calcium exchanger, or nclx.

Movements of Ca²⁺ across mitochondrial membrane are fundamental for excitation-energy coupling because an increase in intramitochondrial Ca²⁺ stimulates energy production by regulating Krebs cycle enzymes (1712). During cellular activity accompanied with [Ca²⁺]_i increase, Ca²⁺ enters mitochondria through mitochondrial Ca²⁺ uniporter (MCU), which is, in essence, a highly selective Ca²⁺ channel (875). Termination of mitochondrial Ca²⁺ is supported by mitochondrial Na⁺/Ca²⁺ and sometimes H⁺/Ca²⁺ exchangers. The mitochondrial Na⁺-Ca²⁺ exchanger has been isolated and cloned and named NCLX/SLC8B1 [NCLX stands for Na⁺-Ca²⁺-Li⁺ because the latter can also be transported (1302)]. The NCLX is abundantly expressed in astroglial mitochondria and contributes to shaping of cytosolic and mitochondrial Ca²⁺ and Na⁺ signals (1327).

9. Sodium-proton exchanger, NHE

The Na⁺-H⁺ exchanger expressed in astrocytes is represented by NHE1/SLC9A1 transporter; this is the main system expelling H⁺ from astroglia (321, 402). The electroneutral stoichiometry [1Na⁺ in exchange for 1H⁺ (1278)] is reflected by a positive reversal potential, which ascertains that NHE1 operates as H⁺ extruder across the whole range of physiological membrane potentials. An important task for NHE1 (FIGURE 19) is to export protons generated by metabolism and acquired by astrocytes in the process of glutamate uptake (each glutamate brings a single H⁺ ion) and Ca²⁺ extrusion (PMCA exchanges 1 or 2 H⁺ for each Ca²⁺ ion expelled). Activity of astroglial NHE1 is positively modulated by phosphorylation mediated by several protein kinases associated with metabotropic signaling cascades (1454). Acidification of the cytosol by metabolic inhibition triggers substantial influx of Na⁺, which supports extrusion of protons; this in turn triggers Na⁺ elevation, Ca²⁺ entry through the reversed Na⁺/Ca²⁺ exchanger, aberrant signaling, abnormal release of glutamate, and cell death (209, 301, 872).

FIGURE 19.

Astrocytes and pH homeostasis. Astrocytes regulate extracellular pH by supplying H⁺ through operation of NHE, EAAT1/2, MCT, and possibly V-H⁺ pump, by removing H⁺ by PMCA and by supplying extracellular space with HCO₃⁻ through NBC. EAAT1,2, excitatory amino acid transporters 1 and 2; NHE, Na⁺-H⁺ exchanger, PMCA, plasmalemmal Ca²⁺-ATPase; VH⁺ pump, vacuolar H⁺ pump localized in plasmalemma; MCT, monocarboxylate transporter; NBC, Na⁺-bicarbonate exchanger.

10. Sodium-bicarbonate cotransporter, NBC

Astroglial sodium-bicarbonate transporter NBCe1/SLC4A4 [member of SLC4 family of bicarbonate transporters (1490)] provides for transmembrane flux of bicarbonate being, together with NHE, the main component of astroglial pH homeostatic system (321, 402). Operational NBCe1 was discovered in leech glial cells (405) and subsequently in mammalian cultured astroglia (1246) as well as in astrocytes in hippocampal slices (603). The NBC cotransports Na⁺ and bicarbonate with a stoichiometry 1Na⁺: 2HCO₃⁻ or 1Na⁺:3HCO₃⁻ [this latter stoichiometry for example was described in retinal Müller cells (1210)], resulting thus in the net influx of one or two negative charge(s), which defines the electrogenic effect of NBCe1 (FIGURE 19). In astrocytes, the NBCe1 operates in both forward mode (which results in an increase in [Na⁺]_i and [HCO₃⁻]_i) and reverse mode in which both Na⁺ and HCO₃⁻ are exported from the cell. The E_NBCe1 lies around −70 mV, and hence minor changes in astroglial membrane potential or transmembrane gradients of bicarbonate to sodium rapidly change the mode of transporter operation (244, 1743).

11. Sodium-potassium-chloride cotransporter, NKCC1

The Na⁺-K⁺-Cl⁻ cotransporter 1 NKCC1/SLC12A2 moves Na⁺, K⁺, and Cl⁻ into the cell with electroneutral stoichiometry of 1Na⁺:1K⁺:2Cl⁻ (1035). In physiological context, NKCC1 always operates in the forward (i.e., importing all three ions) mode and hence the NKCC1 is implicated in K⁺ buffering (see sect. XIA). In pathological conditions, excessive increase in [Na⁺]_i may reverse the transporter (850). Astroglial expression of NKCC1 was shown at mRNA level in Bergmann glia (818); NKCC1 was also immunolocalized in astrocytes from optic nerve (1039) and spinal cord (1410). Function of NKCC1 was mainly studied in astrocytes in vitro (1687, 1689) or in the isolated optic nerve (1039). In cultured astrocytes, NKCC1 was responsible for setting high resting Cl⁻ concentration, for cell swelling (789), for K⁺ uptake, and for the release of excitatory amino acids (1688). Similarly, functional NKCC1 responsible for high [Cl⁻]_i was described in Bergmann glial cells (1778). At the same time, experiments in situ in hippocampal slices questioned a functional role of NKCC1 in protoplasmic astrocytes (942).

12. Vesicular neurotransmitter transporters

Accumulation of neurotransmitters in specific secretory vesicles (see sect. XIB) is a task for three families of vesicular transporters. The SLC17 family contains vesicular transporters for glutamate (VGLUT) and nucleotides (VNUT); SLC 18 family embraces vesicular transporters for acetylcholine (VAChT), monoamines (VMAT), and polyamines (VPAT); while GABA/glycine vesicular transporter (VIAAT or VGAT) is the sole member of SLC32 family (190, 312, 782, 953, 1550). All vesicular transporters use the energy of trans(vesicular)membrane H⁺ gradient that is created by vacuolar H⁺-ATPases. The stoichiometry and precise mechanism of vesicular transporters operation is not yet entirely clear; for VGLUT for example, intravesicular H⁺ and Cl⁻ allosterically activate transport of glutamate (481); VNUT are similarly Cl⁻ dependent (1789).

All three types of vesicular glutamate transporters (VGLUT1–2/SLC17A6–8) have been identified in astrocytes in culture (169, 368, 529, 902, 1144). The data obtained in situ are less uniform. Western blotting, single vesicle imaging, and immunostaining of cortical, hippocampal, and cerebellar slices with subsequent quantitative image analysis failed to identify astroglial expression of all three VGLUTs (977). Similarly, mRNA for VGLUT1–3 was not detected in the RNA-sequencing transcriptome of mouse cortical astrocytes (1945). At the same time, immunogold electron microscopy has detected VGLUT1 associated with glutamate-containing synaptic-like microvesicles located in rat hippocampal astroglial processes (153); immunostaining and confocal microscopy revealed astroglial expression of VGLUT1 in the dentate-molecular layers, in the stratum radiatum of CA1 hippocampal area, in the frontal cortex, and in the striatum (1279). Immunostaining in combination with confocal microscopy as well as immunogold labeling in combination with electron microscopy identified expression of VGLUT3 in astrocytes in hippocampus, in frontal cortex, and in Bergmann glia (1280). Astrocytic expression of VNUT/SLC17A9 transporter has been discovered recently (1550); immunoreactivity for VNUT colocalized with vesicular markers in astrocytes freshly isolated from somatosensory cortex of adult mice (933). Astrocytes were found to express VPAT/SLC18B1 polyamine transporter that concentrates spermine and spermidine in secretory vesicles (693). There are sporadic data on astroglial expression (in vitro) of VMAT as well (1867), whereas VGATs so far have not been found in astroglia. Finally, there are predictions for specific d-serine vesicular transporter VSerT (1071), the molecular nature of which, however, remains unknown.

13. Glucose transporters

The main astroglial glucose transporter is GLUT1/SLC2A1 (22). Immunostaining revealed the presence of this transporter in grey matter astroglia (1151) with preferential localization in endfeet and perisynaptic processes. Exposure of astrocytes to glutamate rapidly (within seconds) and potently (up to 14 times) stimulates glucose uptake into astrocytes acting most likely through GLUT1 (1000). There is some evidence for astroglial expression (probably limited to some brain areas, for example, to hypothalamus) of low-affinity GLUT2/SLC2A2 glucose transporter (967). Astrocytes in culture were also reported to express Na⁺-dependent glucose transporter SGLT1/SLC5A1, which imports glucose together with 2 Na⁺ molecules (1799).

14. Monocarboxylate transporters, MCT

The main pathway for astroglial secretion of lactate is provided by plasmalemmal monocarboxylate transporters 1 and 4 (MCT1/SLC16A1, MCT4/SLC16A3) which are specialized on H⁺-linked transport of lactate, although they have different affinities (K_m for MCT1 ~5 mM, K_m for MCT4 ~34 mM; Ref. 629). Astroglial localization of MCT1 was confirmed with immunocytochemistry in situ; usually MCT1 immunoreactivity colocalizes with S100B (but not with GFAP) indicating their presence in distal processes (1389). Astrocytes in culture were found to specifically express only MCT1 at transcript and protein levels (396, 1356). Astrocytes in vivo, however, express both MCT1 and MCT4 (151, 1422); apparently MCT4 expression requires higher oxygen tension (1493). In the brain stem and in the corpus callosum of rats (but not mice), some perivascular endfeet of astrocytes were reported to express MCT2/SLC16A7 transporter (1356, 1422). The MCT transporters are equilibrative and may mediate either export or import of lactate depending on concentration gradeints for monocarboxylate and H⁺ (629).

15. Ascorbic acid transporters

l-Ascorbic acid is a powerful scavenger of reactive oxygen species and part of antioxidant system. Plasmalemmal Na⁺-dependent vitamin C transporter SVCT2/SLC23A2, which translocates the reduced form of the ascorbic acid (with a stoichiometry of 2 Na⁺ together with 1 ascorbate), has been identified in cultured astrocytes (149, 896), although it was not detected in situ (149). Focal ischemia following unilateral middle cerebral artery occlusion induced astroglial expression of SVCT2 (150). The SVCT2 transporters have been detected in hypothalamic tanycytes in vitro and in situ at transcript and protein levels (553).

16. Zinc transporter

In the CNS, Zn²⁺ released at glutamatergic synapses interacts with plasmalemmal receptors, in particular with NMDA receptors, and contributes to synaptic plasticity. Some astrocytes (for example, Bergmann glial cells; Ref. 1860) express high densities of several types of Zn²⁺ transporter ZnT (belonging to SLC30 family), although the precise role of astroglia in Zn²⁺ homeostasis and role of Zn²⁺ in glial signaling and glial-neuronal interactions remains to be characterized (1590).

C. Recapitulation

Membrane transporters are critical elements of astroglial homeostatic system. The genes for transporters are the most represented in the astroglial transcriptome, and similarly their expression level is the highest (275, 1010). The energy-dependent Na⁺/K⁺ pump maintains concentration gradients for these two ions, which is critical to sustain membrane potential and to generate the electrochemical driving force for operation of SLC transports. These latter translocate multiple molecules, from ions and neurotransmitters to energy substrates, across the membrane to maintain stable molecular environment of the brain and to support vital neuronal functions.

X. IONIC SIGNALING IN ASTROGLIA

Transient fluctuations in cytosolic ionic content represent the most ancient signaling system. The existence of transmembrane ionic gradients is imperative for life, and the very first cells already were in possession of these gradients. What defined the composition of the ancestral cytosol remains largely unknown, hypothetically, for example, the low [Ca²⁺]_i reflected similarly low Ca²⁺ concentration in the primeval ocean (848, 1395). Transmembrane ion gradients also developed by following Donnan forces (449) and of course ultimately required selective permeability of the cellular membrane. The resulting ionic disparity and cytosolic fluctuations in ion concentrations have been used for cell signaling needs ever since. Conceptually, changes in cytosolic concentration of any ion, di- or monovalent, can contribute to regulation of cellular events, and hence any ion can act as a second messenger (1277). In astrocytes, the intracellular signaling role for Ca²⁺ and Na⁺ is well documented, for Cl⁻ it has been proposed (125), and for K⁺ it has not yet been considered.

A. Calcium Signaling in Astroglia

The concept of glial Ca²⁺ excitability was instigated by a series of observations on primary cultures which demonstrated that chemical or mechanical stimulation of astrocytes almost invariably triggers changes in [Ca²⁺]_i either in the form of [Ca²⁺]_i transients or [Ca²⁺]_i oscillations (385, 479, 842, 864, 1093). Discovery of [Ca²⁺]_i waves traveling over astroglial syncytium (306, 360, 361, 382) following focal stimuli bestowed long-range communication capabilities on Ca²⁺ signals. Studying astroglial [Ca²⁺]_i become popular among gliobiologists with a host of papers published; for comprehensive list of references, we direct the reader to relevant reviews covering 25 years of glial calcium signaling (15, 115, 408, 512, 1518, 1610, 1808, 1814, 1817, 1831).

Calcium ions are arguably the most widespread and ubiquitous signaling molecules operating in all types of living cells (160, 284, 666, 1378, 1394). Calcium homeostasis and calcium signaling in astroglia (simialrly to other eukariotic cells) are maintained by coordinated activity of Ca²⁺ fluxes created by plasmalemmal and organellar channels, pumps, and transporters and Ca²⁺ binding to intracellular proteins which serve as Ca²⁺ buffers and Ca²⁺ sensors (FIGURE 20; for extended account on Ca²⁺ signaling, see Refs. 161, 260, 285, 929, 1377, 1379, 1661).

FIGURE 20.

Molecular pathways of astroglial Ca²⁺ signaling. Cytoplasmic calcium signaling is driven by the concentration gradients for Ca²⁺ ions between extracellular space and the cytosol and between the lumen of intracellular organelles (mainly endoplasmic reticulum) and the cytosol. Physiological stimulation opens plasmalemmal or intracellular Ca²⁺ channels, thus creating Ca²⁺ fluxes which translate into [Ca²⁺]_i dynamical changes in intracellular compartments. In astroglia, Ca²⁺ signals originate from both InsP₃-induced Ca²⁺ release from the ER calcium store (which can be also amplified by Ca²⁺ release through ryanodine receptors) and Ca²⁺ entry via plasmalemmal Ca²⁺ channels (see sect. VIIC for detailed description), ionotropic receptors (see sect. VIII), or reversed Na⁺-Ca²⁺ exchanger. In the cytosol, Ca²⁺ interacts with Ca²⁺-bindng proteins, which are represneted by Ca²⁺ buffers and Ca²⁺ sensors. The latter are Ca²⁺-regulated enzymes, which initiate various cellular reactions. Cytosolic Ca²⁺ signals are terminated by Ca²⁺ extrusion into extracellular space by Ca²⁺ pumps and Na⁺-Ca²⁺ exchanger and Ca²⁺ uptake into the ER (by SERCA pumps) and mitochondria (by mitochondrial uniporter). Mitochondrial Ca²⁺ entry acts as the main link between cell activity and energy production by regulating mitochondrial electron transport and ATP synthesis. CBP, Ca²⁺ binding proteins; ER, endoplasmic reticulum; GPCR, G protein-coupled metabotropic receptors; InsP₃R2, InsP₃ receptor type 2; NCX, Na⁺-Ca²⁺ exchanger; ORAI, Ca²⁺ release-activated Ca²⁺ channels; PMCA, plasmalemmal Ca²⁺-ATPase; RyR, ryanodine receptors; SERCA, sarco(endo)plasmic reticulum Ca²⁺-ATPase; SOCE, store-operated Ca²⁺ entry; TRP, transient receptor potential channels; VGCCs, voltage-gated calcium channels.

Recent advances in Ca²⁺ imaging revealed a highly variable spatiotemporal organization of Ca²⁺ signals across complex morphological landscape of astroglial cell. Introduction of genetically encoded Ca²⁺ indicators and new microscopic techniques demonstrated that astroglial compartments have specific mechanisms controlling [Ca²⁺]_i and hence are capable of generating distinct Ca²⁺ signals. In physiological (in situ or in vivo) settings, Ca²⁺ signals in astroglial processes often develop independently from somatic Ca²⁺ signals and demonstrate singular kinetic profiles and underlying mechanisms. In the distal processes, Ca²⁺ signals occur as localized microdomains either spontaneously without obvious link to neuronal activity (821, 1208, 1609) or in response to synaptic stimulation (423, 609, 821, 1225, 1660). Local Ca²⁺ signals in astroglial processes may be linked to specific morphological structures such as “appendages” in Bergmann glial cells (FIGURE 21; Ref. 609) or perisynaptic leaflets (1610). As a rule, local Ca²⁺ signals in astroglial processes developed independently from the somatic [Ca²⁺]_i transients and had distinct kinetics. Usually [Ca²⁺]_i transients were larger and longer in the soma; in the processes both frequent and relatively fast [Ca²⁺]_i fluctuations (423, 1308, 1831) as well as very slow (~40 s) events, defined as “Ca²⁺ twinkles” (821) were reported.

FIGURE 21.

Ca²⁺ microdomains in Bergmann glia. Stimulation of parallel fibers triggers local calcium signals in Bergmann glial cells. A: experimental protocol. Parallel fibers (PF) were stimulated via a pipette connected to a stimulator (STIM) while calcium-dependent fluorescence responses were recorded in a Bergmann glial cell (BG). PCL, Purkinje cell layer. B: confocal fluorescence intensity image of a patch-clamped Bergmann glial cell dialyzed with the calcium indicator Oregon green 488 BAPTA-1 (right). Two processes were distinguished (indicated as 1, 2). C: calcium signals in response to PF stimulation were measured independently for each process (left). Time of PF stimulation is marked by arrows. D: the responding process (1) was further subdivided into regions of interest, in which calcium signals were measured separately (right). Time of PF stimulation is marked by an arrow and a dotted line. (From Matyiash, Grosche, Reichenbach, Verkhratsky, and Kettenmann, unpublished data.)

Spatial heterogeneity of astroglial [Ca²⁺]_i dynamics reflects different molecular mechanisms of Ca²⁺ signal generation. Genetic deletion of InsP₃R type 2 (InsP₃R2^−/−) led to an almost complete disappearance of spontaneous and evoked somatic [Ca²⁺]_i responses, (14, 820, 1381, 1382), indicative of the predominant role for metabotropic receptors/PLC/InsP₃ signaling cascade. In hippocampal astroglial processes, the fastest (half time of recovery ~700 ms) and most localized (microdomain size ~4 μm) [Ca²⁺]_i transients coexisted with longer (recovery half time >3 s) and larger (~12 μm in size) [Ca²⁺]_i microdomains. Both types of localized Ca²⁺ dynamics in astroglial processes were associated with InsP₃R-mediated Ca²⁺ release; they were inhibited by intracellular injection of heparin and were almost completely eliminated in InsP₃R2^−/− mice (423). This observation was not universally confirmed, and in many experiments local evoked and spontaneous Ca²⁺ signals in distal astroglial processes were only partially reduced in the type 2 InsP₃R^−/− mice (663, 821, 1660). In hippocampal astrocytes studied in situ, spontaneous Ca²⁺ tranients confined to fine processes persisted in InsP₃R2^−/− KO mice and were entirely dependent on plasmalemmal Ca²⁺ influx (1516). Moreover, in type 2 InsP₃R^−/− animals, astroglial [Ca²⁺]_i transients evoked by metabotropic agonist endothelin were almost completely inhibited in the soma, while they were left unchanged in the processes (1660). Anatomical segregation of Ca²⁺ signaling suggests alternative routes for Ca²⁺ mobilization in different parts of the astrocyte. These alternative molecular cascades may involve Ca²⁺ entry through ionotropic receptors and plasmalemmal channels, or Ca²⁺ influx mediated by reverse mode of NCX (115, 1610, 1817, 1831). Reversal of NCX may readily occur in response to rise in [Na⁺]_i accompanying operation of major neurotransmitter transporters; such NCX-mediated [Ca²⁺]_i events have been, for example, observed in astrocytes in slices of the olfactory bulb (439). Heterogeneity of Ca²⁺ mobilization may also arise from alternative intracellular signaling associated, for example, with cAMP (663).

In depth analysis of resting [Ca²⁺]_i in hippocampal astrocytes in situ and cortical astrocytes in vivo with fluorescence-lifetime imaging microscopy revealed a [Ca²⁺]_i gradient between peripheral processes and soma (1950). Basal [Ca²⁺]_i levels in the distal processes of 21-day-old mice were ~125 nM, whereas [Ca²⁺]_i in the soma was ~100 nM. These gradients were even higher in cells from younger animals ([Ca²⁺]_i in processes and soma were ~200 and 130 nM, respectively; Ref. 1950). Incidentally, these experiments also pointed out that astroglial resting [Ca²⁺]_i was ~1.5−2 times higher compared with their neighboring neurons (1950). The gradient of basal [Ca²⁺]_i aimed from the distal processes to the soma highlights regional differences in [Ca²⁺]_i regulation and possibly further corroborates the importance of transmembrane Ca²⁺ entry in peripheral structures of astrocytes. Compartmentalization of Ca²⁺ signaling in fine astroglial processes may involve specific positioning of mitochondria which tend to immobilize themselves in sites of focal [Ca²⁺]_i spikes (423). Similarly, functional compartmentalization can occur around plasmalemma-endoplasmic reticulum junctions characterized in astrocytes in vitro; it appeared that key plasmalemmal (NCX1, NKA) and endoplasmic reticulum (SERCA2 and InsP₃Rs) molecules coimmunoprecipitated, indicating the link between plasmalemma and endomembrane (968). Finally, localized [Ca²⁺]_i dynamics may reflect specific distribution of various types of receptors or channels at different cellular locations (1831). The physiological relevance for heterogeneous basal [Ca²⁺]_i in astroglia remains to be explored; local differences of [Ca²⁺]_i homeostasis may reflect functional specialization of subcellular compartments.

Further probing of local Ca²⁺ signaling in astrocytes using membrane-tethered Ca²⁺ indicator revealed even smaller submembrane [Ca²⁺]_i transients (“spotty Ca²⁺ signals”) associated with Ca²⁺ influx through TRPA1 channels (1608). Importantly, Ca²⁺ mobilizing mechanisms differ between astrocytes from different brain regions: Ca²⁺ microdomains in Bergmann glia and in the main processes of hippocampal astrocytes were mediated solely by InsP₃ receptors (423, 879). The near-membrane [Ca²⁺]_i transients were mediated by TRPA1 in astrocytes in CA1 hippocampal area, whereas TRPA1 contribution was less prominent in stratum radiatum and absent in CA3 region (1608). In neocortical astrocytes, [Ca²⁺]_i transients may be amplified by RyR-mediated [Ca²⁺]_i-induced Ca²⁺ release (1311), which however seems to be absent in hippocampal astroglia (118). Sporadic evidence even suggests the role for voltage-dependent Ca²⁺ channels in generating astroglial [Ca²⁺]_i responses (972, 1332). In the in vivo settings, sensory stimulation triggered global [Ca²⁺]_i transients in astrocytes in somatosensory cortex. These Ca²⁺ signals were initiated in the processes and then spread towards the soma with a speed of 15 μm/s; genetic deletion of type 2 InsP₃R eliminated this type of signaling (821). In the olfactory bulb, however, physiological stimulation of the olfactory sensory neuron terminals triggered [Ca²⁺]_i transients only in the astroglial processes, and left the soma idle (1288). In cortical astrocytes from adult (7−20 wk) mice examined in culture, in slices, and in vivo, spontaneous Ca²⁺ microdomains in fine processes were linked to Ca²⁺ release from mitochondria mediated by brief openings of mitochondrial permeability transition pore (13).

Further adding to the complexity, astroglial [Ca²⁺]_i dynamics and Ca²⁺ signaling mechanism undergo developmental regulation (1702, 1950). In the brain of adult nonanesthetized animals, astrocytes in the CNS are under control of almost ubiquitous noradrenergic innervation; sensory stimulation, arousal, and locomotion are associated with release of norepinephrine from locus coeruleus projections which trigger large global [Ca²⁺]_i signals in astrocytes in multiple brain regions (FIGURE 22, Refs. 431, 1341). These global Ca²⁺ signals were mediated by α₁-adrenoreceptors, and inhibition of α₁-adrenoceptrs almost completely blocked all spontaneous Ca²⁺ activity in awake behaving mice (431). A similar role for monoaminergic input has been recently described in Drosophila glia, where Ca²⁺ signals in vivo were driven by norepinephrine analogs tyramine or octopamine (1033). Similar global astroglial signaling is observed in attention and vigilance state, when widespread astrocytic responses are evoked by acetylcholine release from projection of the nucleus basalis of Meynert and are mediated through mAChRs-InsP₃ signaling cascade (316, 1720). Finally, global astroglial [Ca²⁺]_i signaling that occurs throughout the entire cortex was observed in response to transcranial direct current stimulation; this type of Ca²⁺ signaling was again mediated solely through α₁-adrenoceptors (1141). How all these globalized Ca²⁺ signals interact with Ca²⁺ microdomains remains hitherto unknown.

FIGURE 22.

Startle stimulation triggers widespread astroglial Ca²⁺ signaling. A: astrocytic Ca²⁺ transients measured by rhod 2-AM were detected in response to 30 s of 3-Hz whisker stimulation. Representative local field potential LFP and electromyogram (EMG) recordings are shown. B: representative images of rhod 2-AM fluorescence increases during whisker stimulation in a Glt-1-EGFP animal. Scale bar = 100 μM. C: selected cells from B. Rhod 2 ΔF/F₀ was normalized to Glt-1-EGFP fluorescence. Electrocorticogram (ECoG) traces corresponding to rhod 2 ΔF/F₀ are shown below. D: air pulses were directed at the face or tail of the animal to elicit a startle response. Representative ECoG and EMG traces are shown with no apparent evoked ECoG response and strong EMG activity, indicative of a startle response. E and F: representative images and corresponding rhod 2 ΔF/F₀ traces show stable and repeatable astrocytic [Ca²⁺]_i transients after startle stimulations. Bottom: representative ECoG traces from startle stimulation are shown. G: bar graph showing the response rate of cortical astrocytes to whisker and startle stimulation, averaged across animals. *** P < 0.001, one-way ANOVA, Bonferroni correction [n = 11 animals (parietal cortex startle and whisker stimulation), 4 animals (frontal cortex startle)]. H: scatter diagram with superimposed mean and SE for the delay from the onset of whisker/startle stimulation to the beginning of astrocytic rhod 2 ΔF/F₀ responses. Average delay from each successful trial is shown. *** P < 0.001, one-way ANOVA, Bonferroni correction [n = 28 trials in 14 animals (whisker stimulation), 38 trials in 13 animals (parietal cortex startle), and 9 trials in 4 animals (frontal cortex startle)]. Data are means ± SE. [From Ding et al. (431).]

Another fundamental aspect of astroglial [Ca²⁺]_i signaling is linked to the propagating intercellular Ca²⁺ waves that have been widely considered as the substrate for long-range information transfer in astrocytic syncytia (388, 1553, 1741). Propagating Ca²⁺ waves have been characterized in detail in astrocytes in vitro (360, 361, 619, 1635, 1935) and in situ (649, 1560, 1622, 1873, 1883), and the mechanisms of propagation have been discerned. The spread of Ca²⁺ wave was found to be mediated either by intercellular diffusion of InsP₃ (or possibly also cyclic ADP-ribose, cADPR) through gap junctions (20, 712, 974) or through regenerative paracrine ATP-mediated signaling (40, 61, 366, 659) or through the combination of both (1553). Operation of these mechanisms seems to be anatomically segregated: for example, ATP-mediated propagation dominates in corpus callosum, whereas in the neocortex Ca²⁺ waves depend entirely on gap junctions (620).

Appearance, physiological role, and mechanisms of propagating astroglial Ca²⁺ waves in vivo are yet to be investigated and apprehended. Propagating Ca²⁺ waves were detected in cerebellum in syncytia of velate astrocytes; they were sensitive to purinoceptor antagonists and spread for ~50 μm at a speed of 4−11 μm/s (722). Spontaneous Ca²⁺ waves were also observed in Bergmann glia (1081). In contrast, Ca²⁺ waves have not been detected in the in vivo cortex in both awake and anesthetized animals. In cortical astrocytes Ca²⁺ signals seem to be restricted to individual cells or to a limited number of adjacent cells, without much spreading to neighboring regions. For example, stimulation of sensory inputs by mechanical displacement of whiskers in mice induced [Ca²⁺]_i transients in astrocytes in the barrel cortex; these Ca²⁺ signals were confined to individual cells and did not produce propagating Ca²⁺ waves (1858). Sensory stimulation of astrocytes in the visual cortex of ferrets evoked [Ca²⁺]_i transients in single astrocytes but did not trigger propagating Ca²⁺ waves (1581). Synchronized somatic [Ca²⁺]_i transients were similarly observed in cortical astrocytes imaged in vivo in response to focal application of α/β-adrenoceptors agonists (123). All in all, the extent of propagation and the velocity of astroglial Ca²⁺ waves seem to be substantially limited in vivo compared with experiments in the dish.

Propagating fast (at 60 μm/s) Ca²⁺ waves were found to spread through thousands of astrocytes in mouse hippocampus; these waves were called somewhat romantically “glissandi” from a musical term denoting a rapid series of ascending or descending notes on the musical scale (921). These Ca²⁺ waves were sensitive to inhibitors of gap junctions and purinoceptors, and somewhat unexpectedly, they were blocked by 2 μM TTX (921). These observations, however, were performed on mice heavily anaesthetised with urethane, while cortices were removed by aspiration to gain the access to hippocampi; consequently these unusually fast Ca²⁺ waves might reflect the traumatic or toxic injury of the brain tissue.

To conclude, hierarchy of astroglial Ca²⁺ signals ranges from local microdomains to propagating Ca²⁺ waves and synchronized global Ca²⁺ signals (FIGURE 23). Characterization of physiological targets and coding principles of astroglial Ca²⁺ signaling remain the ultimate and yet unanswered question. Fluctuations in [Ca²⁺]_i obviously impact on gene expression and on astroglial energy metabolism (through accumulation in mitochondria); these however are generic targets of Ca²⁺ signals in all eukaryotic life forms. Which astroglia-specific processes are governed by [Ca²⁺]_i? How do intracellular proteins decipher incoming Ca²⁺ signals? The Ca²⁺-regulated secretion is the most popular target, yet [Ca²⁺]_i may regulate many other astroglial mechanisms from expression of membrane transporters to regulation of K⁺ uptake (1853). The remarkable heterogeneity in parameters and mechanisms of astroglial Ca²⁺ signaling most likely reflects an extensive adaptive potential of astrocytes, which depending on the context may employ distinct Ca²⁺ signaling toolkits or rapidly remodel the existing ones to meet the homeostatic requirements of their immediate environment.

FIGURE 23.

Hierarchy of astroglial Ca²⁺ signaling.

B. Sodium Signaling in Astrocytes

Stimulation of astrocytes, either mechanical or chemical or synaptic, triggers [Na⁺]_i transients with distinct spatiotemporal parameters; these transients are the substrate of the cytosolic Na⁺ signaling (881, 1499). The [Na⁺]_i transients in response to physiological stimulations have been observed in astrocytes both in vitro (851, 1496, 1498) and in situ (877, 878, 940, 1495). Regulation of [Na⁺]_i has its own idiosyncrasies: 1) it relies solely on plasmalemmal Na⁺ transport because of absence of intracellular storage compartments and 2) cells do not have (or we are not yet aware) specific Na⁺ buffers, which may spatially organize cytosolic Na⁺ dynamics. The signaling function of Na⁺ is mediated through the transmembrane concentration gradients which drive multiple plasmalemmal transporters and through “Na⁺ sensors” represented by Na⁺-sensitive enzymes. These are by far less characterized compared with Ca²⁺ sensors, although some of them have been identified. Several metabotropic receptors including class A G protein-coupled receptors (to which adenosine or dopamine receptors belong) have a conserved binding site for Na⁺, which negatively modulates their affinity to agonists (845). Changes in Na⁺ promote the dissociation of trimeric G proteins, thus affecting the activation or inhibition of Gβγ-gated ion channels (202, 1471). The Na⁺-dependent K⁺ channels, which mediate slow afterhyperpolarization, operate in neurons (170), whereas in astrocytes increases in [Na⁺]_i facilitate spermine-induced inhibition of inward rectifier K_ir4.1 channels (916).

Astroglial Na⁺ homeostasis and Na⁺ signals are governed solely by transmembrane Na⁺ fluxes (FIGURE 24); Na⁺ ions enter the cytosol through plasmalemmal channels and Na⁺-coupled SLC transporters and are extruded mainly by NKA and possibly by NCX operating in the reverse mode (881, 1499). Resting [Na⁺]_i in astrocytes is relatively high (15−20 mM, see sect. VIA); it defines driving force and sets the equilibrium for many Na⁺-dependent transporters critical for homeostatic activity of astroglia. Inhibition of NKA with ouabain or by decreasing extracellular K⁺ results in rapid increase in [Na⁺]_i (586, 762, 869, 1496, 1645, 1846), which indicates significant constitutive Na⁺ influx at rest. This influx may reflect background channel activity or Na⁺ entry mediated by plasmalemmal transporters. The NKCC1 seems to contribute to setting resting [Na⁺]_i in astrocytes in vitro, as its inhibition by the diuretic bumetanide decreases resting [Na⁺]_i by several millimolar (851, 1689). The role of NKCC1 in regulation of astroglial ionic fluxes in vivo remains, however, questionable (942). Resting Na⁺ influx may be also mediated by Na⁺/bicarbonate (NBC) transporter (851).

FIGURE 24.

Na⁺ entry pathways. Astroglial Na⁺ signaling is generated by Na⁺ entry through plasmalemma; the Na⁺ influx occurs through Na⁺ channels, ionotropic receptors, and TRP channels. Na⁺ entry also accompanies operation of plasmalemmal SLC transporters and Na⁺-Ca²⁺ exchanger. Termination of Na⁺ signals is mainly mediated by Na⁺ extrusion via Na⁺-K⁺ pump. The TRPC channels are the molecular substrate of astroglial store-operated Ca²⁺ entry; as these channels pass both Na⁺ and Ca²⁺, they establish the link between metabotropic activation of Ca²⁺ release and Na⁺ signaling. AMPAR, AMPA receptors; EAAT1,2, excitatory amino acid transporters 1 and 2; ER, endoplasmic reticulum; GAT1,3, GABA transporters 1 and 3; GlyT1, glycine transporter; GPCR, G protein-coupled metabotropic receptors; NCX, Na⁺-Ca²⁺ exchanger; Na_v, voltage-gated Na⁺ channels; Na_x, Na channels gated by extracellular concentration of Na⁺; NKA, Na⁺-K⁺-ATPase; NMDAR, NMDA receptors; P2XR, P2X puriunoceptors; SOCE, store-operated Ca²⁺ entry; TRP, transient receptor potential channels.

Operation of glutamate transporters (which cotransport 3 Na⁺ with 1 glutamate) creates substantial Na⁺ entry, which may elevate [Na⁺] by 10−20 mM, both in response to endogenous glutamate or to synaptic stimulation (FIGURE 25). Glutamate transporter-associated [Na⁺]_i transients have been observed in cultured astrocytes (310), in situ in Bergmann glia (140, 877), in astrocytes from neocortex (938, 1776), and from calyx of Held (in these [Na⁺]_i increased by ~3.5 mM; Ref. 1780). Uptake of GABA produces smaller Na⁺ entry (reflecting the 2Na⁺: 1GABA stoichiometry of GAT) and may increase [Na⁺]_i by 4−6 mM (310, 1777). Sodium entry can also be mediated by ionotropic receptors (878, 940), by TRP channels (1461), or by NCX operating in the forward mode (878). Relative contribution of Na⁺ entry pathways differs between different astrocytes: in hippocampal astrocytes Na⁺ influx occurs solely through glutamate transporter, whereas in Bergmann glia the contributions of transporter and AMPA glutamate receptors are almost equal (940). In response to synaptic stimulation, [Na⁺]_i transients are directly proportional to the stimulation strength. At low stimulation intensity, [Na⁺]_i increased only in astroglial processes; increase in the intensity of stimulation resulted in larger [Na⁺]_i transients occupying the whole cell from processes to soma (940).

FIGURE 25.

Astroglial Na⁺ signaling associated with glutamate uptake in cerebellar Bergmann glial cells. A: voltage dependence of glutamate-transporter current as compared with kainate-induced (AMPA-mediated) currents. The current traces are shown on the top (transporter current on the left and receptor-mediated current on the right), while current-voltage curves for glutamate- and kainite-induced currents are displayed at the bottom. For each cell membrane, currents at different voltages were normalized to the amplitude of current recorded at the holding potential of −70 mV. Note that glutamate-induced responses do not reverse at +20 mV, whereas kainate-induced currents show a reversal potential close to 0 mV. *P < 0.05 and **P < 0.01, significant difference between glutamate- and kainite-induced currents. B: simultaneous recordings of glutamate-induced inward current and [Na⁺]_i elevation in normal conditions in Na⁺-free (Na⁺ substituted by NMDG⁺) solution. C: stimulation of parallel fibers activates membrane currents and transient [Na⁺]_i elevation in Bergmann glial cell. The [Na⁺]_i transient is not affected by inhibitors of ionotropic glutamate receptors. [From Kirischuk et al. (877).]

Astroglial [Na⁺] transients evoked by either chemical or synaptic stimulation are quite long-lasting with recovery times in a range of tens of seconds (310, 878, 940, 1780), which argues against free Na⁺ diffusion in the cytosol. Mechanisms shaping kinetics of [Na⁺]_i transients are enigmatic; possibly long-lasting [Na⁺]_i responses may reflect prolonged periods of Na⁺ entry or Na⁺ binding/unbinding to plasmalemmal transporters (1494). Similarly unknown is the possibility for generation of [Na⁺]_i microdomains, especially in small subcellular compartments; these diffusion-limiting regions may be associated with perisynaptic processes or with astroglial endfeet (1243). Astrocytic Na⁺ signals can propagate through individual cells and through astroglial syncytia in the form of spreading [Na⁺]_i waves with a velocity ~60 μm/s; in hippocampal slices, these Na⁺ waves expand through all neighboring astrocytes (941). Astroglial Na⁺ waves propagate mainly through Cx30 or Cx43 gap junctions (941).

The existence of spatially restricted localized [Na⁺]_i microdomains is indirectly supported by specific colocalization of Na⁺-dependent transporters, pumps, and channels in astroglial processes. Electron microscopy of cortical and hippocampal astrocytes revealed concentration of NCX1–3 in distal processes enwrapping excitatory synapses (1120), which coincided with the localization of NMDA receptors (354). In cultured astrocytes, NCX was colocalized with NKA at specific domains of plasma membrane directly opposing outposts of the endoplasmic reticulum (194); local Na⁺ signaling may occur in these compartments (589). The NKA are also spatially linked to glutamate transporters as indicated by colocalization and coimmunoprecipitation of α2 NKA subunit and EAAT1/2 in cortical astrocytes (329, 1475, 1501). Local Na⁺ signals associated with glutamate transport in astroglial perivascular endfeet in hippocampal slices increase ATP consumption possibly reflecting local homeostatic adaptation (939).

Astroglial Na⁺ executes the signaling function through several mechanisms (FIGURE 26). Changes in transmembrane Na⁺ gradient control multiple transporters responsible for neurotransmitter homeostasis, and relatively small increases in [Na⁺]_i can reverse for instance GABA and glycine transport. Cytosolic Na⁺ also regulates glutamine-glutamate(GABA) shuttle through direct action on glutamine synthetase (139) and regulation of glutamine transporters (1752). These regulatory mechanisms are of great potential importance in perisynaptic processes, where local [Na⁺]_i transients may control local homeostatic responses. Fluctuations of [Na⁺]_i contribute to regulation of K⁺ buffering through affecting NKA transport; similarly, [Na⁺]_i is fundamentally important for pH homeostasis by regulating NBC and NHE. By controlling reversal potential of NCX, astroglial [Na⁺]_i transients may contribute to Ca²⁺ signaling, for example, by initiating local Ca²⁺ influx in distal processes. Changes in [Na⁺]_i are directly linked to astroglial metabolism, through controlling glycolysis and lactate production and possibly regulating ATP synthesis (309). Cytosolic Na⁺ signals can invade mitochondria (156); moreover, astroglial mitochondria can generate spontaneous Na⁺ spikes (77). This mitochondrial connection may further implicate Na⁺ into regulation of astroglial energy metabolism, which however remains hypothetical (1494).

FIGURE 26.

Na⁺ signals regulate multiple molecules and multiple homeostatic functions (see text for further details). ASCT2, alanine-serine-cysteine transporter 2; ASIC, acid sensing ion channels; CNT2, concentrative nucleoside transporters; EAAT, excitatory amino acid transporters; ENaC, epithelial sodium channels; GAT, GABA transporters; GS, glutamine synthetase; GlyT1, glycine transporter; iGluRs, ionotropic glutamate receptors; Na_x, Na⁺ channels activated by extracellular Na⁺; NAAT, Na⁺-dependent ascorbic acid transporter; NBC, Na⁺/HCO₃⁻ (sodium-bicarbonate) cotransporter; NCX, Na⁺/Ca²⁺ exchanger; NCLX, mitochondrial Na⁺/Ca²⁺ exchanger; NHE, Na⁺/H⁺ exchanger; NKCC1, Na⁺/K⁺/Cl⁻ cotransporter; NET, norepinephrine transporter; MCT1, monocarboxylase transporter 1; P2XRs, ionotropic purinoceptors; SN1,2, sodium-coupled neutral amino acid transporters which underlie exit of glutamine; TRP, transient receptor potential channels; ROS, reactive oxygen species; VRAC, volume-regulated anion channels.

In summary, the sodium signaling system may be considered as the pathway for fast coordination of neuronal activity with “homeostatic” response of astroglia mediated through Na⁺-dependent transporters, concentrated in perisynaptic processes. Influx of Na⁺ into these perisynaptic processes, characterized by exceedingly high surface-to-volume ratio, may also trigger local depolarization, which will further regulate plasmalemmal transporters.

C. Recapitulation

Intracellular ionic signaling provides the substrate for astroglial excitability coupling external stimulation with functional cellular response. Spatiotemporal fluctuations in cytosolic concentration of two ions, Ca²⁺ and Na⁺, emerge in response to physiological stimulation. Astroglial calcium signaling has complex hierarchical organization ranging from highly localized near-membrane Ca²⁺ microdomains to global Ca²⁺ spikes and oscillations engulfing whole astrocyte; these global signals further develop into propagating Ca²⁺ waves in astroglial syncytia. Importantly, in the in vivo awake brain, Ca²⁺ waves seem to be much more restricted than in the ex vivo preparations. Complex organization of Ca²⁺ signals (FIGURE 23) reflects distinct mechanisms of generation associated either with production of InsP₃ and Ca²⁺ release from the endoplasmic reticulum (global Ca²⁺ signals and propagating Ca²⁺ waves) or with Ca²⁺ entry through plasmalemmal channels and reversed Na⁺/Ca²⁺ exchanger. Astroglial Na⁺ signaling relies on Na⁺ entry through plasmalemmal channels and transporters. The two systems of ionic signaling are mechanistically connected as many membrane channels are permeable to both Ca²⁺ and Na⁺, while depletion of endoplasmic reticulum stores triggers store-operated activation of TRPC channels that generate fluxes of both ions. Physiological outcome of ionic signals is defined either by direct interaction of Ca²⁺ with enzymatic cascades or by dynamic changes in transmembrane Na⁺ gradients that regulate the majority of SLC transporters. Ionic signals regulate several known astroglial functions (see sect. X), although precise pathways are still in need of in depth experimental analysis.

XI. ASTROCYTES AS SECRETORY CELLS OF THE CNS

A. The Concept of Astroglia as Gliocrine Cells

Astroglial secretion has been postulated at the beginning of 20th century. In 1909, Hans Held observed, using the molybdenum hematoxylin stain, granular inclusions in neuroglial processes, which he interpreted as a sign of active secretion (671). A year later, Jean Nageotte visualized secretory granules in parenchymal grey matter glia using the Altmann method of fucsin labeling (1185). These secretory granules (also known as gliosomes) have been often described afterwards, and astrocytes were linked to an endocrine role, secreting substances from endfeet to the blood (579, 1362). This endocrine role has never been confirmed; however, astrocytes were found to be capable of secreting a remarkable assortment of molecules [see TABLE 5; of note, astroglial secretome contains ~180 proteins (450, 849)] that contribute to the regulation of CNS development and homeostasis, synaptogenesis, and cognitive function. This stimulated the concept of glycrine system in which neuroglial cells (as microglia and to some extend oligodendrocytes are also capable of releasing biologically active agents) are regarded as elements of the brain-wide secretory network (1797, 1811). Astroglia-derived secretory substances include (TABLE 5) 1) classical neurotransmitters; 2) neuromodulators; 3) neurotransmitter precursors; 4) hormones and peptides; 5) eicosanoids; 6) metabolic substrates; 7) scavengers of ROS; 8) growth factors; 9) various “trophic” factors that, for example, regulate synaptogenesis and synaptic connectivity; and 10) pathologically relevant molecules such as inflammatory factors. Apart from these, astrocytes constantly secrete ions (most notably Cl⁻) and water. Secretion pathways operational in astrocytes include 1) vesicle-based exocytosis, 2) diffusion through plasmalemmal pores/channels, and 3) extrusion through plasmalemmal transporters (FIGURE 27). The same molecule can be released through different pathways.

Table 5.

Astroglial signaling molecules and mechanisms of their secretion

Secreted Substance	Function	Secretion Mechanisms
Neurotransmitters
Glutamate	Activates glutamate ionotropic (AMPA, KA, and NMDA) and metabotropic receptors in neurons and neuroglia. May regulate synaptic strength and neuronal excitability and affect synaptic plasticity.	Exocytosis (368, 745, 993, 1065, 1145, 1328, 1906, 1942); connexons (572, 1919); P2X7Rs (457, 503); Best-1 channels (1893); VRAC: (111, 506, 867, 991, 1717); cystine-glutamate antiporter Scx− (1080, 1162); reversed EAAT1/2, only in severe pathological conditions (1290, 1713).
		[H+]i-dependent glutamate release in Bergmann glia.
ATP	Activates neuronal and glial ionotropic P2X receptors (excitatory action) and metabotropic P2Y receptors (Ca2+ signaling; astroglial Ca2+ waves, trophic effects). Rapidly degrades to ADP, AMP, and adenosine by ectonucleotidases, adenosine acts through A1 (inhibitory effects), A2A, A2B, and A3 (metabotropic effects) receptors in neurons and neuroglia.	Exocytosis (513, 628, 933, 1334, 1795); lysosomes (785, 979, 1947); connexons (61, 324, 365, 823, 1663, 1681, 1754); Panx-1 (1691); P2X7R (1690)
GABA	Inhibitory neurotransmitter acting on neurons and on neuroglia.	Reversed GABA transporters, GAT1 or GAT3 (670, 1776, 1777); Best-1 chloride channels (964); VRAC (900)
Glycine	Inhibitory neurotransmitter in the spinal cord. Coagonist for NMDA receptors.	Reversed glycine transporters GlyT1 (492); VRAC (327)
Precursors for neurotransmitters and neuromodulators
Glutamine	Precursor for neuronal glutamate and GABA.	Glutamine transporters: sodium-coupled neutral amino acid transporters SNAT3/SLC38A3 and SNAT5/SLC38A5 (1552, 1752)
Pro-enkephalin	Precursor for enkephalins (112).	Unknown
l-Serine	Precursor for neuronal d-serine (1913).	Alanine-serine-cysteine transporter 1, ASCT1 and ASCT2 transporters
Neuromodulators
Taurine	Agonist to both glycine and of GABAA receptors. In supraoptic nucleus astroglia-derived taurine regulates inhibitory glycine receptor tone (327).	Connexins (1918); VRAC (327, 867, 1918)
l-Aspartate	Positive modulator of NMDA receptors.	VRAC (867); connexons (1919)
Kynurenic acid	Inhibitor of NMDA and acetylcholine receptors; astroglia-derived kynurenic acid affects glutamatergic (291, 1396), GABAergic (122), cholinergic (1969), and dopaminergic (1441) transmission. Aberrant production and synthesis can be associated with schizophrenia (1918).	Unknown
Lipoprotein receptor-related protein 4 (Lrp4)	Regulates presynaptic glutamate release (1703).	Unknown
Endozepines	Expressed at high levels in astrocytes in thalamus and in astrocytes and tanycytes in hypothalamus. Astroglia-derived endozepine acts as a positive modulator of synaptic inhibition in the thalamic reticular nucleus (332).	Unknown
Polyamines, spermine, and spermidine	Spermine and spermidine are predominantly localized in astrocytes; they can be released through Ca2+ dependent process and regulate various cellular functions, including modification, stimulation, or inhibition of AMPA, NMDA receptors, K+ channels, and gap junction.	Exocytosis (693)
Hormones and neuropeptides
Atrial natriuretic peptide (ANP)	Local vasodilatator. Contributes to control of systemic salt intake.	Exocytosis (911)
Endothelin-3	Local vasoactive hormone (471).	Unknown
Sphingosine 1-phosphate	Regulates cell proliferation and immune response.	ATP-binding cassette transporter A1 (1544)
Neuropeptide Y	Metabotropic neurotransmitter.	Exocytosis (1407, 1425)
Thyroid hormones thyroxine (T4) and triiodothyronine (T3)	Astrocytes exclusively express type 2 deiodinase (D2) that converts T4 to T3. Astrocytes accumulate T4 by organic anion transporting polypeptide 1C1 (OATP1C1/SLCO1C1), convert it to T3, which is then released to brain parenchyma.	L-type amino acid transporter 2 LAT2/SLC7A8 (1156)
Sphingosine 1-phosphate	Regulation of cell proliferation and immune response.	ATP-binding cassette transporter A1 (1544)
Eicosanoids
Arachidonic acid, prostanoids	Multiple; including intercellular signaling and control of innate immunity.	Direct release from membranes (1178, 1905, 1907)
Metabolic substrates
Lactate	Possible energy substrate in neurons. Possible activator of specific metabotropic receptors.	Monocarboxylate transporters MCT1, MCT4 (151, 1422); pannexons (?) (833); lactate-permeable channel (1656)
Citrate	Astrocyte synthesize and release citrate in physiological context (1877). In the extracellular space, citrate may possibly regulate (through chelation) Ca2+ and Mg2+.	NaDC/SLC13A (Na+-dependent dicarboxylate transporter) (1297); VRAC (?)
Growth factors
Neurotrophins: NGF, NT-3, BDNF	Multiple trophic effects including regulation of neuronal survival, growth, and regeneration.	Largely unknown. Pro-BDNF (secreted by neurons) is taken up into astrocytes by endocytosis, and after conversion into mature form is released by exocytosis (147)
Pro-inflammatory factors
TNF-α	Mediates AMPA (117) and GABA receptor trafficking (1408); affects and may even induce synaptic plasticity; impairs learning and memory (621).	Unknown
IL-1β, IL-6	Control of neuroinflammatory response; regulation of learning and memory, trophic effects (328, 488).	Unknown
C3a complement factor	Modulation of neuronal morphology and function (through C3a receptor); control of neuroinflammatory response.	Exocytosis, lysosomes(?) (931, 982, 1674)

Reference numbers are given in parentheses.

FIGURE 27.

Multiple astroglial secretory pathways.

B. Exocytosis

Exocytosis utilizes specific membranous organelles carrying heterogeneous cargo and capable of fusing with plasmalemma, hence providing the release of their luminal content. Exocytosis is an evolutionary conserved (appearing probably in Archea) and ubiquitous property of most eukaryotic cells (1657). Exocytotic vesicular release (that underlies both constitutive and regulated secretion) is regulated by an extended family of SNARE (soluble N-ethyl maleimide-sensitive fusion protein attachment protein receptor) proteins. Of these, R(arginine)-SNAREs (also known as vesicle-associated membrane proteins or VAMPs) are associated with the membranes of secretory vesicles, while Q(glutamine)-SNAREs are associated with plasmalemma (784, 839). In the process of excitation-secretion coupling, high [Ca²⁺]_i instigates formation of a ternary SNARE complex from R/Q-SNAREs, which mediates fusion of vesicles with plasma membrane (1694).

1. Astroglial secretory organelles

Astrocytes possess intracellular (synaptic-like microvesicles, dense-core vesicles, and lysosomes) and extracellular (ecto- and endosomes) secretory organelles (FIGURE 28). Intracellular vesicles are cellular organelles that may completely fuse with cellular membranes, whereas extracellular vesicles are membranous compartments released into the surrounding environment.

FIGURE 28.

Astroglial secretory organelles.

a) synaptic-like microvesicles, or SLMVs.

Astroglial SLMVs have a diameter of 30−100 nm; they usually appear in pairs or in groups (up to 15) of vesicles, thus being much less numerous than neuronal synaptic vesicles, 100−1,000 of which are clustered in presynaptic terminals (153, 169, 804). The SLMVs were identified in processes and somata of astrocytes from various brain regions (including hippocampus, cortex, and cerebellum) both in culture and in situ (153, 169, 1145, 1280). The role for large (1−3 μm) SLMVs identified in astroglial processes in hippocampal slices (825) remains unclear. Astroglial SLMVs contain vesicular neurotransmitter transporters VGLUT1–3 and VNUT (see sect. IXB12). In astrocytes in culture, SLMVs were found to store glutamate or costore glutamate and d-serine (1071, 1160). In contrast, analysis of astrocytes in situ revealed distinct populations of glutamate and d-serine-containing vesicles (153). Also, intracellular mobility of astroglial vesicles has its peculiarities: for example, glutamate-containing vesicle movement accelerates with an increase in [Ca²⁺]_i (1671), whereas peptidergic vesicles and endolysosomes slow down (1402, 1404); such differences have never been seen in neurons.

b) dense-core vesicles, or DCVs.

Astroglial DCVs usually have a diameter of 100−600 nm, and their population is substantially smaller than SLMVs (279, 368, 753, 1407), even smaller DCVs (~50 nm) were also reported (1404). In astrocytes in culture, DCVs were found to carry secretory proteins secretogranin II (279, 1295, 1407) and secretogranin III (1296), chromogranins (753), ANP (902, 1295), neuropeptide Y (1407, 1425), and ATP (341, 1310). The chromogranin and secretogranin II-containing DCVs were also found in human astrocytes from tissues obtained during surgery (753).

c) secretory lysosomes.

Lysosomes are the essential intracellular organelle of eukaryotes generally responsible for degradation and recycling of macromolecules and autophagy (909), although lysosomes may also participate in secretion. There are several reports indicating the role of lysosomes in astroglial secretion, in particular in Ca²⁺-dependent release of ATP in astrocytes in culture (785, 979, 1947). Astrocytic secretory lysosomes have diameters of 300−500 nm; they can be tagged with various fluorescent conjugated dextrans (1793), with FM-dyes or with [2'/3′-O- (N'-methylanthraniloyl)adenosine-5′-O-triphosphate], or MANT-ATP (1947); they coexist (and can be coreleased) with other types of secretory vesicles in the same cell (993). Astroglial secretory lysosomes are in possession of VNUT, which mediate accumulation of ATP (1292), and they express lysosomal specific markers such as cathepsin D, lysosomal-associated membrane protein 1 (LAMP1), and ras-related protein Rab7 (1947). Lysosomal exocytosis is mediated by tetanus toxin-insensitive vesicle-associated membrane protein VAMP7 (304), and downregulation of VAMP7 suppresses ATP release (1804). Lysosomal fusion was also reported to be triggered by distinct slow and localized Ca²⁺ signals (1804). Similarly to other cells, secretory lysosomes in astrocytes are likely to play a role in membrane repair (43).

d) extracellular vesicles.

The extracellular vesicles represented by exosomes and ectosomes carry a wide spectrum of bioactive substances including cytokines, signaling proteins, mRNA, and microRNA. Exosomes are formed intracellularly from multivesicular bodies (1083), whereas ectosomes are produced through direct budding of the plasma membrane (1744). Astroglial exosomal secretion has been observed in response to oxidative and heat stress (1735) or in pathology, for example, in astrocytes surrounding amyloid plaques in a mouse model of familial Alzheimer's disease (1854). Some astroglia-released exosomes were also containing mitochondrial DNA (613). Despite high expectations of exosome-mediated intercellular signaling in the CNS (537), the astroglial component of it remains largely uncharacterized.

Shedding of ectosomes from astroglial membrane is triggered following activation of P2X₇ purinoceptors. It occurs because of rapid activation of acid sphingomyelinase that moves to the plasma membrane outer leaflet and modifies membrane structure/fluidity leading to vesicle blebbing and shedding (174, 175). Astroglial ectosomes are heterogeneous with diameters between 100 and 1,000 nm (174, 1411). These ectosomes were reported to carry fibroblast growth factor-2 and vascular endothelial growth factor (1411), IL-1β (174), nucleoside triphosphate diphosphohydrolases (302), matrix metalloproteinases (1551), and acid sphingomyelinase and high levels of phosphatidylserine on their membrane outer leaflet (174).

2. Slow time course of astroglial exocytosis: molecular determinants

Kinetics of exocytosis differ very much between different cells, being determined by 1) number and spatial clustering of vesicles, 2) the nature of SNARE proteins, 3) number of R-SNAREs associated with the vesicle, and 4) the spatio-temporal organization of “trigger” [Ca²⁺]_i nanodomain (469, 839, 1716). Kinetics of astroglial exocytosis are fundamentally different that of neurons. In the latter, fusion is extremely fast, being typically fully accomplished in <0.5−1 ms (1694), whereas in astrocytes an exocytotic event develops in a time scale of seconds (1811). Indeed, time lapse video monitoring of fluorescently labeled VGLUT1/2-containing vesicles in single astrocytes showed the development of fusion within hundreds of milliseconds after the onset of [Ca²⁺]_i increase (1065, 1537). Even slower exocytosis (developing in seconds) was observed when imaging synapto-pHluorin-labeled vesicles (222, 993). Similar slow kinetics were observed using fluorescent synaptobrevin 2-labeled vesicles (1056). Exocytosis of peptidergic vesicles in astrocytes unfolded with about a minute delay after stimulation (1295, 1407, 1425); similarly slow was the lysosomal release of ATP (1310, 1413, 1947). Capacitance measurements further corroborated the slowness of astroglial exocytosis, finding it about two orders of magnitude slower than in neurons (901, 902).

This sluggish (when compared with neurons) exocytosis reflects several idiosyncrasies of astroglial excitation-secretion coupling. First, astrocytes contain fewer vesicles, and these vesicles are not clustered as it is typical for active zones of neuronal terminals (153, 169, 804). Second, the spatiotemporal evolution of Ca²⁺ signals in astroglia, which are often associated with Ca²⁺ release from the stores, is slower than in neuronal terminal where “trigger” Ca²⁺ signals originate solely from Ca²⁺ entry through voltage-gated channels. Third, astroglial SNARE complexes as well as their vesicular densities differ from their neuronal counterparts: in neurons exocytosis is supported by the ternary fusion complex formed by VAMP2, SNAP25, and syntaxin, whereas in astrocytes the ternary fusion complex assembles from VAMP2/3 or VAMP7, SNAP23, and syntaxin (1811). Substitution of SNAP25 with SNAP23 in the ternary complex halves its stability, arguably impeding the tethering/docking/fusion process. Further delays in exocytotic kinetics are associated with the densities of R-SNAREs associated with a single vesicle: there are ~25 VAMP2 molecules linked to a single vesicle in astrocyte versus ~70 molecules per vesicle in neurons (1625, 1716).

3. Nonexocytotic functions of vesicles in astrocytes

Intracellular vesicles contribute to the delivery of various molecular components to the plasmalemma. In particular, the VAMP2 labeled vesicles in astrocytes in vitro were immunopositive for the AMPA receptor subunits GluA1, 2, 3, which can be incorporated into the plasma membrane upon vesicular fusion (368). Similar mechanisms may regulate the density of the astroglial glutamate transporter EAAT2 (978, 1670), and disruption of this delivery pathway compromises astroglial glutamate uptake in pathology (1505). Vesicles in astrocytes in vitro also have been reported to carry CB₁ cannabinoid receptors (1284) as well as AQP4 water channels (1403). Vesicles also deliver membranes, and hence exocytosis is involved in membrane repair (43, 1659) and possibly in astroglial morphological plasticity (1796).

C. Diffusion Through Membrane Channels

Secretion by diffusion through plasmalemmal pores is driven by transmembrane concentration gradient, which for many biologically active molecules could be rather high. For example, cytosolic ATP concentration (~5 mM) exceeds ambient extracellular concentration (low nM) by ~10⁷ times, whereas glutamate ([Glu]_i ~0.3 mM, [Glu]_o ~25 nM) exceeds by 10⁴ times. Any conductive pore large enough to accommodate the secreting molecule can therefore act as a conduit, and nonvesicular release of various signaling molecules have been identified in different tissues [for instance in taste buds (871) or erythrocytes (1415)]. Connexin hemichannels (see sect. VIIH) were probably the first (711) and are certainly the most studied (572) diffusion secretion pathway in astrocytes. Activation of connexin hemichannels occurred in response to lowering extracellular Ca²⁺ (711) or metabolic inhibition (358), or treatment of cell cultures with proinflammatory cytokines (1457). Opening of these hemichannels was associated with the release of glutamate or ATP (143, 1681, 1919), and hence, the role of unpaired connexons in release of astroglial signaling molecules was considered (61, 365, 1553). In particular, connexon-mediated release of ATP has been recognized as a paracrine signal critical for the generation of intercellular Ca²⁺ waves in cultured astroglial cells (365, 1681). Pannexons also have been implicated in glutamate and ATP release from astroglia (572) and were shown to mediate release of d-serine (1306).

The repertoire of hemichannel-mediated secretion could be much wider. There are indications that numerous signaling molecules such as NAD⁺, adenosine, InsP₃, cyclic ADP-ribose, nicotinic acid adenine dinucleotide phosphate, lactate, and NO (249, 509, 596, 833, 1147, 1643) all can pass through Cx hemichannels in either direction (i.e., serving for secretion or uptake depending on concentration gradients). In addition, hemichannels could support uptake or release of several metabolites such as glucose, ascorbate, or glutathione (572, 1431). Both connexons and pannexons are also permeable to several synthetic molecules including Lucifer yellow, ethidium, calcein, propidium, 5 (6)-carboxyfluorescein and 2-(4-amidinophenyl)-1H-indole-6-carboxamide, and DAPI, which are often used for monitoring the activity of these channels (1525). Connexons are densely expressed at perivascular astroglial endfeet and may therefore contribute to astroglia-vasculature signaling. Release of ATP (with subsequent degradation to adenosine) from glia limitans endfeet through Cx43 hemichannels was claimed to induce dilatation of pial arterioles (1904). Release of ATP from astrocytes through Cx43 gap junctions was also shown to affect synaptic activity in hippocampal slices; Cx43 were activated by local drop in [Ca²⁺] following photoactivation of exogenous extracellular Ca²⁺ buffer (1754). The Cx43 hemichannels seem to be active in resting conditions when they provide constitutive ATP secretion to regulate basal excitatory transmission (324). Astroglial connexons (and possible pannexons) have been also implicated in release of lactate in acute slices from the brain stem, hippocampus, hypothalamus, and cortex (833). Inhibition of astroglial Cx43 channels in vivo by injection of selective inhibitor TAT-L2 peptide into basolateral amygdala impaired memory consolidation and resulted in amnesia for auditory fear conditioning. This can be rescued by coinjection of a somewhat eclectic mixture of neuroactive molecules (glutamate, glutamine, lactate, d-serine, glycine, and ATP) (1663). Opening of astroglial hemichannels is caused and regulated by many factors, which include membrane depolarization, mechanical stimulation, decrease in extracellular or increase in cytosolic [Ca²⁺], fluctuations in pH, reactive oxygen species, or intracellular phosphorylation of connexins (572). Activity of connexons is also regulated by growth factors, proinflammatory cytokines, or CB₁ cannabinoid receptors (535, 1275), whereas in tanycytes Cx43 channels are activated by extracellular glucose (1273).

A second route for secretion by diffusion is associated with plasmalemmal ion channels. The most characterized of these channels are P2X₇ purinoceptors operating in dilated (pore-forming) mode (see sect. VIIIA). The P2X₇ receptors have been implicated in the release of ATP and excitatory amino acids (457, 503, 1690). There is also evidence for glutamate release through TREK-1 channels following activation of metabotropic receptors. Apparently, G protein βγ-subunits directly activate TREK-1 channels by interaction with the NH₂ terminus of the latter (1893). Glutamate and GABA may also diffuse through anion channels, including volume-regulated anion channels and Best-1 chloride channels (1893). Glutamate released from astrocytes through these channels was claimed to activate synaptic NMDA receptors in CA1 pyramidal neurons (636), thus modulating synaptic plasticity (1323). Evidence about Best-1-mediated secretory pathway, however inspiring, is yet in need of independent confirmation.

D. Transporter-Mediated Release

Plasmalemmal transporters widely contribute to astroglial secretion; they are involved in release of many of the neuroactive molecules originating from astroglia. Specific roles of plasmalemmal transporters are summarized in TABLE 5 and discussed below.

E. Signaling Molecules Secreted by Astrocytes

1. Neurotransmitters

Experimental evidence accumulated over the last decades demonstrated that astrocytes are capable of releasing major neurotransmitters:⁸ glutamate, ATP, GABA, and glycine (see Refs. 55, 1383, 1526, 1922 for extensive reference lists). Astroglial secretion of ATP is probably the best documented, and it has been observed in cell cultures, in slices and, to some extent, in vivo and may occur through both exocytosis (933, 934, 1183, 1971) and diffusion through membrane channels (324, 1663). ATP released from astrocytes serves as a source of adenosine, which, in turn, provides for brain-wide inhibition and contributes to sleep homeostat (628, see also sect. XII).

Astroglial release of glutamate also received much attention starting from the first observation on cultured astroglia that demonstrated Ca²⁺-dependent glutamate release from astrocytes stimulated with metabotropic agonist bradykinin (1328). This initial finding was followed by numerous confirming experiments in vitro and some in situ, which further strengthened the concept of exocytotic release of glutamate from astroglial cells. Data for glutamate diffusion through plasmalemmal channels were also mounted (see Refs. 55, 1144, 1330, 1971 for references and also refer to the previous section).

Proving exocytotic astroglial release of neurotransmitters in vivo turned out to be a challenging task. The majority of the in vivo data were derived from a transgenic mouse model expressing a dominant negative (dn)SNARE (which is the cytosolic tail of VAMP2) in astrocytes (628, 1334). The cytosolic tail of VAMP2 competes with VAMP2 in the ternary complexes, thus suppressing exocytosis. The (dn)SNARE mice displayed changes in behavior and alterations in synaptic transmission which were thought to reflect occluded astroglial exocytosis (700, 933, 1183, 1765). At the same time there is some evidence for (dn)SNARE penetrating to neurons, thus raising the possibility that the impairment of neuronal, rather than astroglial, exocytosis may account for the phenotype observed (540, 1633). Glutamate may also be released from astrocytes via connexons and pannexons as well as through anion channels as has been discussed in the previous section. Recently a [H⁺]_i-dependent glutamate release pathway was detected in Bergmann glia, although the molecular nature of this pathway remains to be clarified (146).

Astrocytes in the neonatal optic nerve (1250), in the brain stem (198), and in the cerebellum (133, 1072) were claimed to contain GABA and express GABA synthesizing enzyme glutamic acid decarboxylase or GAD67. Immunoreactivity for GABA was also observed in ~80% of hippocampal astrocytes (955), although another report (1921) found only ~20% of GABA-positive astrocytes in hippocampus. In addition to GAD67-mediated synthesis, GABA may also be produced from putrescine degradation by monoaminoxidase B (MAOB) which is also predominantly an astroglial enzyme; this pathway was documented for Bergmann glia (1923).

Astrocytes do not express vesicular GABA transporter and hence are incapable of exocytotic GABA release. Nevertheless, Ca²⁺-independent GABA release was detected in cultured rodent (544, 545) and human (961) astroglia and confirmed in observations in situ on cerebellar Bergmann glia (98) and on astrocytes from olfactory bulb (900) as well as from hippocampus and cortex (670, 876, 955).

Astroglial release of GABA can be mediated either by the reverse mode of the GABA transporter (48, 544, 670, 876) or by diffusion through plasmalemmal anion channels, such as volume-regulated anion channels (900) or Best-1 anion channels. The latter pathway received much promotion recently (964, 1922), although the role for Best-1 had been already met with criticism (429). Whether astrocytes can indeed release appreciable quantities of GABA in physiological context remains doubtful. Uptake provides the only realistic source for GABA in astrocytes. Accumulated GABA is readily metabolized by being fed into tricarboxylic acid cycle by GABA transaminase (1575), and hence, cytosolic GABA concentration must be rather low. Cytosolic concentration of GABA, in turn, is critical for both transporter-mediated and for diffusional release. Surprisingly, information on intracellular GABA concentration in astrocytes in vivo is unavailable, and therefore, the probability and role of physiological astroglial GABA release is yet to be judged. Astrocytes are also reported to release glycine, either by reversed glycine transport (492) or through VRAC (327). In the brain stem and in the spinal cord, glycine is, arguably, released from astrocytes by the Na⁺-independent alanine-serine-cysteine 1 (Asc-1) transporter, and this release is critical for maintaining the glycine content in neuronal presynaptic terminal and hence for tonic inhibition of motor neurons (470).

2. Neuromodulators

a) l- and d-serine.

The concept of d-serine as a specific astroglial neuromodulator received much popularity in recent years following initial observations claiming the exclusive astrocytic presence of d-serine and its synthesizing enzyme serine racemase (1558, 1559, 1892). Physiological experiments were to follow; these demonstrated exocytosis of d-serine-containing vesicles from astrocytes in vitro (1071) and found that manipulating with astroglial Ca²⁺ signaling or with astroglial metabolism affects NMDA-mediated synaptic transmission in brain slices arguably through the lack of astroglia-derived d-serine (515, 675, 1161). Furthermore, inhibition of resting Ca²⁺ entry through genetic deletion of TRPA1 channels was found to decrease constitutive secretion of d-serine and reduce long-term potentiation (1608). Subsequent analysis, however, questioned all these results, when astroglial immunoreactivity for d-serine and serine racemase appeared to be largely an artifact (see Ref. 1891 for comprehensive account). More selective antibodies revealed predominantly neuronal expression of both d-serine and serine racemase [forebrain neurons (838, 1131), Purkinje neurons (1131) and magnocellular neurons (1859)]. Selective transgenic suppression of serine racemase in neurons caused its marked (~65%) decrease in cortical and hippocampal tissues, whereas selective interference with astrocytes produced only marginal effects (144). Release of d-serine from cortical slices was found to be Ca²⁺ independent, with the bulk of this release mediated by neuronal alanine-serine-cysteine transporter 1 (Asc-1/SLC7A10) (1502). Pharmacological inhibition or genetic deletion of Asc-1 reduced LTP at the hippocampal Schaffer collateral-CA1 synapse because of decreased d-serine release (1543). Finally, a conditional knockout of serine racemase from neurons but not from astrocytes impaired LTP at the same synapses (144); neuronal deletion of the enzyme also reduced dendritic remodelling and plasticity (94).

The astrocytic role in d-serine neuromodulation is, most likely, confined to the supply of the l-serine, the substrate for serine racemase. In the absence of l-serine, neuronal d-serine production cannot proceed. The l-serine is synthesized almost exclusively in astrocytes endowed with specific enzymes, such as for example 3-phosphoglycerate dehydrogenase, or Phgdh (1910, 1913). Genetic deletion of the latter specifically from astrocytes severely depressed (by ~80%) neuronal synthesis of d-serine (1913). Mutations of the Phgdh cause profound neurodevelopmental deficits, with patients showing low levels of d- and l-serine in the cerebrospinal fluid (889). Hence, astrocytes are indispensable for d-serine neuromodulation through production of l-serine and its shuttling to neurons.

b) kynurenic acid.

Kynurenic acid (or KYNA) is an endogenous inhibitor of NMDA receptors (by blocking the glycine binding site) and even more potent noncompetitive and allosteric inhibitor of α7 nicotinic receptors (699). KYNA is a product of tryptophan metabolism; de novo synthesis of KYNA occurs through transmutation of bioprecursor kynurenine catalyzed by kynurenine aminotransferases (KAT), of which KATII is the predominant brain enzyme present almost exclusively in astrocytes (614, 615). Astroglial secretion of KYNA contributes to regulation of glutamatergic (291, 1396), GABAergic (122), cholinergic (1969), and dopaminergic (1441) neurotransmission and hence is involved in regulation of various cognitive processes (1895). Impairments of astroglial KYNA production and secretion are connected to various pathologies, of which the most notable is the link to pathophysiology of schizophrenia (1583).

c) lactate.

In addition to being an energy substrate, lactate, produced in astrocytes through aerobic glycolysis and released by MCT1/4 transporters and/or hemichannels (833), may act as a signaling molecule mediating astroglial-neuronal communication (152, 433). Lactate directly modulates neuronal excitability by suppressing firing of hippocampal neurons in vivo (576) and of cerebral neurons in culture (224). In locus coeruleus, lactate release from optogenetically activated astrocytes excites neighboring neurons and triggers norepinephrine release both in slices and in vivo (1729). Molecular substrates of lactate signaling are yet to be fully identified; the hydroxycarboxylic acid (HCA)/GRP81 G protein-coupled receptor linked to the cAMP signaling cascade is one candidate; these receptors are expressed in many neurons and also in endfeet of astrocytes (951).

3. Hormones

a) thyroid hormones.

Astrocytes are an indispensable element of the thyroid regulation of the CNS. The l-thyroxine (or T₄) secreted by the thyroid gland represents a prohormone that needs to be converted (through de-iodination) into the biologically active triiodothyronine (or T₃). The T₄ crosses the blood-brain barrier by several dedicated transporters (1887), and the conversion into T₃ occurs in astrocytes, which specifically express the converting enzyme type 2 deiodinase (D2) (612). Uptake of T₄ into astrocytes is mediated by organic anion-transporting polypeptides 1c (OATP14/SLCO1C1) (421), and T₃ is released into extracellular space (for subsequent uptake into neurons and other glial cells) by L-type amino acid transporter 2 (LAT2/SLC7A8) (859).

b) neurosteroids.

Although all steroid hormones readily penetrate the blood-brain barrier, they are additionally synthesized within the CNS, which is considered to be steroidogenic tissue. Neurosteroids are produced in both neurons and neuroglia and are involved in local autocrine or paracrine signaling (6, 437). Neurosteroids, being lipophilic, do not require specific transmembrane translocation pathways. In hypothalamus, astrocytes synthesize progesterone, the synthesis of which is regulated by estradiol acting through membrane endoplasmic reticulum α receptors, transactivation of mGluR1, and Ca²⁺ release from the endoplasmic reticulum with subsequent activation of protein kinase A that stimulates progesterone synthesis from cholesterol (315, 1112). Estradiol (which derives through aromatase-induced conversion of testosterone) was not found to be produced in healthy astrocytes, which lack aromatase; however, the latter is upregulated following injury (294).

4. Growth factors

Astrocytes synthesize and release several growth factors, including fibroblast growth factor type 2 (FGF) (1756), epidermal growth factor (EGF) (1073), insulin-like growth factor (IGF) (320), and neurotrophins such as nerve growth factor (NGF), neurotropin-23 (NT-3), and brain-derived growth factor (BDNF) (453). Mechanisms of growth factor secretion remain largely unknown. In hippocampus, astrocytes provide for maturation of pro-BDNF secreted by neurons. Pro-BDNF is taken up into astroglial cells by endocytosis, and after conversion into the mature form, it is released by VAMP2-mediated exocytosis (147).

5. Neuropeptides

Astrocytes synthesize several neuropeptides in a region specific manner (1769). The main neuropeptides produced in astroglia include angiotensinogen (766, 1679), substance P (1113), somatostatin (1616), proenkephalin and enkephalin (1101, 1616), neuropeptide Y (NPY) (102), nociceptin (272), and atrial natriuretic peptide (ANP) (1095). Secretion of NPY is Ca²⁺ dependent and involves exocytosis of DCVs (1425). Similarly, astroglial ANP is also stored in vesicles and secreted by regulated exocytosis (911).

6. Polyamines

Polyamines, spermine and spermidine, influence various cellular processes, from gene expression and protein synthesis to protection (1349). In particular, polyamines act as endogenous modifiers of AMPA (447) and NMDA (1254) receptors, K_ir4.1 channels (1627), and connexons (134). In the CNS spermine and spermidine are predominantly found in astrocytes (950), whereas polyamine synthesizing enzymes mainly show neuronal localization. The pathway for polyamines shuttling to astroglia is unknown. Depolarization induces Ca²⁺-dependent release of polyamines from brain slices which may reflect exocytosis from astrocytes that express polyamine vesicular transporter VPAT/SLC18B1 (693).

F. Recapitulation

Astrocytes are dedicated secretory cells in the CNS and release ~200 bioactive molecules ranging from neurotransmitters and neuromodulators to growth factors and hormones. These molecules are released by several mechanisms, and quite often the same molecule can be secreted by different pathways. This, arguably, increases the versatility and flexibility of astroglial secretion. The detailed description of astroglial vesicular release (and in particular the molecular machinery responsible) in vivo remains to be accomplished. In contrast to neurons, where secretion is restricted (with some exceptions) to presynaptic terminals, astrocytes provide for humoral signaling by releasing bioactive agents into extracellular space, where they approach multiple targets. Gliosecretion regulates multiple brain functions from synaptic transmission to morphological plasticity and metabolism.

XII. PHYSIOLOGICAL FUNCTIONS OF ASTROGLIA

A. Ion Homeostasis in the CNS, or Ionostasis

Ionic homeostasis of the CNS is of paramount importance for nervous function, because it defines overall excitability and major signaling processes. Ion concentrations in the nervous tissue, however, are not static, and ion fluctuations regulate major systemic processes such as memory (683) or sleep (432). Astrocytes through numerous ion transporters are fundamental for ionostasis.

1. Potassium homeostasis

Control of interstitial K⁺ concentration is the most canonical function of astroglia. The critical role for astrocytes in maintaining extracellular K⁺ homeostasis in the CNS was proposed in the mid-1960s by Leif Hertz, Steven Kuffler, and Richard Orkand. Both energy-dependent NKA-mediated pathway (682) and diffusion through K⁺ channels (1276) have been suggested as underlying mechanisms. Rises of [K⁺]_o during neuronal activity are associated with repolarizing K⁺ efflux (710), with a single action potential increasing local [K⁺]_o by ~1 mM (1433). In addition, K⁺ exit is linked to an activation of postsynaptic ionotropic glutamate receptors (1465), with particularly large K⁺ efflux through NMDA receptors (1612) or with K⁺ release through K⁺-Cl⁻ cotransporter activated during GABAergic transmission (1822). Extracellular K⁺ may also rise following K⁺ efflux from astrocytes associated with operation of EAAT1/2 glutamate transporters. Arguably, most of K⁺ released in synaptic compartments is associated with dendritic arborizations, while axonal action potentials account for a smaller fraction of K⁺ entering the interstitial space (679, 741). Physiological increases in [K⁺]_o in the nervous tissue occurring in relation to neuronal activity are rather moderate; for example, mechanical stimulation of skin results in 0.4 mM [K⁺]_o increase in the spinal cord (669), while [K⁺]_o rises by 0.5 mM in occipital cortex in response to visual stimulation (1624). Maximal increase of 3 mM was recorded in the spinal cord in response to noxious cutaneous stimulation (1708). Direct repetitive electrical stimulation of the cortical surface or thalamic nuclei may increase [K⁺]_o in the somatosensory cortex by 8–12 mM (667), and even higher levels can be observed in pathology (683). It needs to be remembered, of course, that experimental techniques for ion concentration measurements have serious spatial limitations, and K⁺ concentration could be higher in tiny extracellular compartments. Astroglial regulation of extracellular K⁺ is mediated through several overlapping mechanisms, which can be generally classified into 1) diffusion through K⁺ channels, 2) active transport by NKA, and 3) transport by SLC transporters. All three mechanisms have been studied, and numerous experiments in vitro and in situ found evidence corroborating their contribution (683, 892, 1267, 1642).

The initial concept (892, 1276, 1843) of astroglial spatial K⁺ buffering postulated that K⁺ enters glia through plasmalemmal K⁺ channels; subsequently, K⁺ ions are redistributed through glial syncytium via gap junctions and released distantly. This concept has been found to operate also at a single-cell level in Müller glia. Here, K⁺ enters these cells though K_ir4.1 channels densely populating perisynaptic processes in the inner plexiform layer of the retina. Subsequently K⁺ is equilibrated through the cell and is released (again through K_ir4.1 channels) from the endfoot or from perivascular processes. This type of K⁺ buffering was defined as K⁺ siphoning (892, 1211). The concept of spatial K⁺ buffering rests on the ability of K_ir4.1 channels to accumulate K⁺ locally and on gap junctions, which provide for K⁺ redistribution. Experiments in vitro have demonstrated that K_ir4.1 channels indeed are mainly responsible for resting K⁺ permeability of astrocytes and could be considered as main players in K⁺ buffering (90, 436, 917, 1209). Nonetheless, analysis of conditional astroglia-specific K_ir4.1 knockout mice revealed neither the expected phenotype (neuronal hyperexcitability) nor significant alterations in the kinetics of [K⁺]_o following neuronal stimulation. Changes in [K⁺]_o dynamics were limited to moderate decrease in its recovery kinetics and appearance of pronounced undershoot following [K⁺]_o recovery (323). At the same time, genetic deletion of K_ir4.1 channels caused a significant (~20 mV) depolarizing shift in the resting membrane potential and decreased the resting K⁺ permeability. These data therefore argue against a dominant role of K_ir4.1 channels in K⁺ buffering. The role for K_ir4.1 channels in K⁺ clearance, however, remains debatable (943, 1244). Densities of K_ir4.1 channels markedly increase in postnatal development, and hence, experiments on younger animals could be somewhat misleading. Moreover, expression of K_ir4.1 channels differs between different brain regions, and therefore, the relative contribution of these channels to K⁺ buffering can have regional variability. The K_ir4.1 channels may be specifically important for K⁺ uptake from narrow spaces such as synaptic clefts, this role being in agreement with K_ir4.1 densely populating perisynaptic processes (1244).

The energy-dependent K⁺ buffering mediated through NKA has been discovered in experiments in cultured astrocytes (681). Numerous subsequent experiments performed in the in situ preparations revealed the leading role of NKA in extracellular K⁺ clearance associated with physiological neuronal activity (371, 942, 1433, 1902). In this scenario, K⁺ accumulated by astrocytes is subsequently released into the extracellular space to be taken up by neurons, thus fully restoring ionic gradients. Astrocytes therefore rapidly buffer excess K⁺ released during neuronal activity; when neurons stop firing, K⁺ is shuttled to the neuronal compartment (683, 944). Release of K⁺ is mediated through K_ir4.1 channels [deficient release may explain the increased undershoot in K_ir4.1 astroglial knockout mice (323)] and possibly through other channels or SLC transporters. The Na⁺/K⁺/Cl⁻ transporter NKCC1 was also suggested to participate in K⁺ buffering, and indeed, it can do so in astroglial cultures (942), although its role in situ and in vivo was not confirmed (942). The NKCC1 is activated by K⁺ increases above 10 mM (1845) or by extracellular hypertonicity (1418) and may therefore contribute to K⁺ clearance in pathological context. All in all, several astroglial systems are working in concert to ensure reliability and versatility of K⁺ homeostasis.

2. Chloride homeostasis

Strong stimulation of neuronal GABA receptors (induced for example by intense firing of activity of GABAergic interneurons) may deplete Cl⁻ from the synaptic cleft. Astrocytes, however, respond to GABA by Cl⁻ efflux, and hence, astroglial Cl⁻ secretion can therefore counteract Cl⁻ entry into neurons thus maintaining [Cl⁻]_o and preserving the driving force for Cl⁻ to sustain inhibitory neurotransmission (854). This hypothesis has been directly corroborated by recent experiments, which demonstrated an astroglial role in maintaining GABAergic neurotransmission. Moreover, these experiments revealed the role of gap junctions and astroglial syncytium for Cl⁻ homeostasis by showing that inhibition of connexons during intense stimulation of GABAergic transmission to CA1 neurons resulted in a collapse of the Cl⁻ gradient in these neurons (468).

3. Regulation of extracellular Ca²⁺

Astrocytes may also contribute to regulation of extracellular Ca²⁺ concentration. In the course of neuronal activity, Ca²⁺ concentration in narrow extracellular compartments and particularly in the perisynaptic cleft undergoes substantial fluctuations due to massive Ca²⁺ influx into neurons upon activation of Ca²⁺ channels. The extracellular Ca²⁺ concentration may decrease below 1 mM, which in turn can affect Ca²⁺ signaling both in the presynaptic terminal and in the postsynaptic compartment with obvious consequences for synaptic transmission (1519). In addition, lowering of [Ca²⁺]_o causes astroglial release of ATP, which excites hippocampal interneurons (1754). Lowering of extracellular Ca²⁺ concentration to ~0.5 mM was shown to trigger InsP₃-induced Ca²⁺ release from the endoplasmic reticulum in astrocytes (probably mediated by autocrine release of ATP) (1935). This may, at least hypothetically, aid in restoring [Ca²⁺]_o because Ca²⁺ can leave the astrocyte through either plasmalemmal Ca²⁺ pumps or Na⁺/Ca²⁺ exchangers.

4. Regulation of pH

Homeostatic control over cytosolic and extracellular pH is of a paramount importance for brain function, because even mild acidification or alkalinization significantly affects neuronal excitability and synaptic transmission. In resting conditions, the electrochemical gradient for H⁺ is aimed into the cytosol. Neuronal metabolism is the main source of extracellular protons, which are extruded by the Na⁺/H⁺ (NHE) transporter. The second major source of protons is associated with exocytosis of neurotransmitters because the lumen of vesicles is highly acidic (pH ~5.2–5.7; Ref. 1114). Protons may also be released from astrocytes together with lactate. Astrocytes regulate extracellular pH by 1) removing H⁺ in the course of operation of EAAT1/2-mediated glutamate uptake with a single H⁺ cotransported together with a glutamate molecule, and 2) by supplying extracellular space with HCO₃⁻ through the Na⁺/HCO₃⁻ (NBC) transporter, which can operate in both forward and reverse modes depending on membrane potential and [Na⁺]_i (401, 1497). The NBC is very sensitive to bicarbonate levels being active at [HCO₃⁻]_o as low as 0.3 mM (406). Astrocytes also release H⁺ by NHE and possibly by plasmalemmal V-type H⁺ pumps, which apparently play a dominant role in astroglial cells in the in situ rat optic nerve compared with the same cells in culture (640). This coincides with a substantially greater astroglial pH buffering capacity in situ versus in vitro (167, 640).

B. Water Homeostasis and Regulation of the Extracellular Space Volume

Astrocytes control water movements through several pathways, of which most notable are aquaporins and membrane transporters. In astroglial endfeet, aquaporin 4 (AQP4) channels are colocalized with K_ir4.1, with both channels being anchored to the endfeet membrane via dystrophin/α-syntrophin complexes containing a PDZ domain. This close colocalization and possible cooperation of AQP and K_ir4.1 channels links K⁺ and water movements (35). Functional coupling between K_ir4.1 and AQP4, however, remains debatable, with some evidence showing independent operation of these two channels (1940).

Astroglial water fluxes also contribute to regulation of extracellular volume (1184). Genetic deletion of AQP4 led to ~25% increase in the extracellular volume (1917), while animals lacking AQP4 in astrocytes had an increased brain water content (624). Astrocytes also contribute to dynamic changes in the extracellular volume that are associated with neuronal activity. Synaptic transmission evokes rapid and considerable (5–30%) reduction in the extracellular volume (430, 1710). This reduction is generally believed to reflect swelling of astroglial perisynaptic processes resulting from cotransport of water with K⁺, glutamate, bicarbonate, and other molecules (623, 1036). In these settings an important pathway for water transport is represented by plasmalemmal SLCs, such as EAAT1, GAT-1, MCT1, GLUT1, etc. with 40–500 molecules of water being cotransported along with the substrate against osmotic gradient. Some of these transporters (for example, EAAT1 or GAT-1) also possess a pore for passive water transport (1036). The AQP4 provides a conduit for water exit from astrocytes, and deletion of AQP4 enhances shrinkage of extracellular space (623).

C. Astrocytes and Homeostasis of Reactive Oxygen Species

Astrocytes are the key element of the CNS antioxidative system. Neuronal energy metabolism, which relies entirely on oxidative phosphorylation (127), constantly generates a high amount of reactive oxygen species (ROS), which need to be scavenged to avoid cellular damage. The antioxidant system of the CNS mainly relies on glutathione and ascorbic acid (1053). Glutathione scavenges ROS in a nonenzymatic direct reaction or else it acts as an electron donor for glutathione peroxidase. Astrocytes as a rule contain more glutathione compared with neurons (451). Neuronal synthesis of glutathione requires either cysteine or glutamylcysteine as obligatory precursors, both of which are provided by astroglia. Astrocytes accumulate cystine through the Sxc⁻ glutamate/cystine exchanger (232); subsequently, cystine is reduced to cysteine, which is further converted to glutamylcysteine (CysGlu). Astrocytes release cysteine and glutathione, the latter being converted into CysGlu by ectoenzyme γ-glutamyl transpeptidase (γGT). Both cysteine and CysGlu are accumulated into neurons for glutathione synthesis, with cysteine being transported by neuronal EAAT2/3 transporters (318). Neurons by themselves are unable to accumulate cystine for further conversion because they express very little of the Sxc⁻ transporter. In the presence of astrocytes in vitro, neurons sustain high levels of glutathione. Removal of astroglia from culture or inhibition of γGT prevents neuronal synthesis of glutathione and facilitates ROS toxicity (452). In coculture, a single astrocyte is able to protect up to 20 neurons (420).

Astroglial antioxidant capacity is also mediated by ascorbic acid, the reduced form of vitamin C. Astrocytes represent the reservoir for ascorbic acid (367, 1053); moreover, they are able to accumulate the dehydroascorbic acid derived from ROS oxidation and released by neurons and reduce it (using glutathione) to ascorbic acid (1464, 1886). Neuronal activity (probably through release of glutamate) stimulates astroglial release of ascorbic acid (1884). This release can be mediated through anion channels such as VRAC (1626) or even through hemichannels (363), although this latter mechanism has not yet been identified in astrocytes.

D. Neurotransmitters Homeostasis

Astroglia are fundamental for neurotransmitter turnover in the brain. Astrocytes remove and inactivate, through accumulation and metabolic conversion, glutamate, GABA, adenosine, and norepinephrine (FIGURE 29). In addition, astrocytes produce glutamine, an obligatory precursor for neuronal glutamate and GABA, which is indispensible for both excitatory and inhibitory neurotransmission.

FIGURE 29.

Astrocytes and neurotransmitter homeostasis. Astrocytes take up glutamate, GABA, adenosine, and monoamines. Glutamate is converted to glutamine (by glutamine synthetase, GS), which is shuttled to neurons for subsequent conversion into glutamate and GABA. GABA accumulated by astrocytes is mainly consumed by tricarboxylic acid cycle; adenosine is converted to AMP by adenosine kinase (ADK) while monoamines are degraded by astroglial monoamine oxidase B (MAO-B).

1. Glutamate and GABA

Glutamate, or glutamic acid, is ubiquitously present in all types of cells, and it acts as the major excitatory neurotransmitter in the CNS. During elementary acts of synaptic transmission, glutamate, released from neuronal terminals, can reach millimolar levels in the synaptic cleft (1092). The glutamate homeostatic system is responsible for 1) rapid removal of glutamate from the synaptic cleft, 2) control of glutamate spillover to neighboring synapses, and 3) rapid replenishment of the glutamate-releasable pool in the neuronal terminals. In addition, this system needs to keep ambient glutamate concentration rather low to prevent excitotoxicity. These multiple functions are mainly performed by astroglia.

First and foremost, astrocytes are the sole de novo synthesizers of glutamate from glucose in the CNS (685). This synthesis requires tricarboxylic acid cycle intermediate α-ketoglutarate (1574), which is produced by the enzyme pyruvate carboxylase exclusively expressed in astrocytes (1594, 1927). Astroglial cells express another enzyme fundamental for glutamate turnover, the glutamine synthetase. The exclusive astroglial expression of glutamine synthetase was discovered by Michael Norenberg and Antonio Martinez-Hernandes (1237). Glutamine synthetase catalyzes conversion of glutamate to glutamine. Glutamine in turn is a direct precursor for glutamate; neurons express phosphate-activated glutaminase that deaminates glutamine to glutamate. Glutamine synthetase is also a central cytosolic enzyme for detoxifying ammonium (NH₄⁺), which is converted into glutamine (359). In physiological settings, NH₄⁺ is released by active neurons and accumulated by astrocytes via diffusion through channels as well as by dedicated transporter (1064).

Second, astroglia are the main sink for glutamate released during neurotransmission. About 80% of extracellular glutamate in the CNS is taken up by astrocytes (380) using EAAT1/2 transporters (see sect. IXB1). Astroglial glutamate transport shapes glutamate dynamics in the synaptic cleft and prevents (or allows) spillover, thus contributing to the time course, efficacy, and outcome of synaptic transmission (1062, 1768). Knockout of astroglial glutamate transporters results in excitotoxicity and paralysis associated with increased glutamate level in the interstitium (1508). Third, astrocytes provide for the replenishment of glutamate in neuronal terminals. Glutamate entering into astrocytes undergoes conversion to glutamine (an energy-dependent reaction requiring one molecule ATP per one molecule of glutamate), which is subsequently shuttled to neurons through coordinated A and N systems of glutamine transporters (see sect. IXB2). Sodium entry, associated with EAAT1/2-mediated uptake of gltutamate, stimulates glutamine efflux thus coordinating glutamate/glutamine fluxes: increase in glutamate uptake increases glutamine release (1752, 1780). After entering neuronal terminals, glutamine is converted into glutamate in excitatory neurons; in the inhibitory neurons, glutamate is further converted into GABA. This sequence of transporting and biochemical events represents the glutamine-glutamate(GABA) shuttle (FIGURE 27), which is indispensible for maintenance of excitatory and inhibitory neurotransmission (242, 680). Part of the glutamate accumulated into astrocytes is metabolized to α-ketoglutarate by glutamate-dehydrogenase (GDH) or aspartate-glutamate transferase (AAT) and used for energy production; it is estimated that ~85% of glutamate is converted into glutamine and returned to neurons, whereas ~15% is oxidized (1507), thus contributing to cover energy costs of glutamate handling (1094). Astroglial glutamate transporters have been reported to colocalize with glycolytic enzymes (glutamate dehydrogenase and hexokinase) and with mitochondria (564), suggesting the existence of an “energy hub” supporting glutamine-glutamate (GABA) shuttle.

Postnatal development of the brain is associated with an increase in astroglial glutamate uptake (427) because of an increase in the expression of EAAT1/2 (1775). In neonatal neocortex, glutamate uptake is substantially lower than in hippocampus, which probably allows for glutamate spillover and activation of extrasynaptic receptors. Cortical maturation is accompanied by more stringent control of extracellular glutamate (641). In hippocampus there is also a developmental switch in transporters expression; in the neonatal tissue, the EAAT1 and (neuronal) EAAT3 predominate, whereas in adulthood the EAAT2 becomes the major form (97). Impairment of the glutamine-glutamate (GABA) shuttle affects neurotransmission. In the retina inhibition of glutamine synthetase rapidly (in 2 min) reduced the b-wave indicative of glutamatergic light response (103). Inhibition of astroglial Krebs cycle and hence inhibition of glutamate synthesis and thus conversion to glutamine suppresses glutamate-dependent memory consolidation in chicks; this can be rescued by injecting exogenous glutamine (574). Likewise, inhibition of glutamine synthetase rapidly depletes synaptic GABA content and reduces inhibitory neurotransmission in mouse hippocampus (1283). Astrocytes also remove extracellular GABA, which is catabolized through tricarboxylic acid cycle.

Astrocytes may also dynamically regulate extrasynaptic concentration of glutamate, not through its uptake but rather through its release via the cystine/glutamate antiporter Sxc⁻(1162). Increases in the extrasynaptic glutamate may affect neurons through both NMDA and metabotropic glutamate receptors. Genetic deletion of the catalytic subunit of the Sxc⁻ led to ∼50% decrease in extracellular glutamate measured by microdialysis (1080); a similar decrease was observed following chronic administration of cocaine or nicotine known to reduce expression of Sxc⁻ (87, 1162).

2. Adenosine homeostasis

Adenosine is ubiquitously present throughout the CNS, regulating numerous functions, of which the most notable is an inhibitory neuromodulation mediated through presynaptic A₁ receptors, stimulation of which suppresses release of major excitatory neurotransmitters, including glutamate, acetylcholine, dopamine, and serotonin (526). The sources of adenosine in the CNS are 1) ATP released from neural cells and degraded by ectonucleotidases (263) and 2) adenosine released from cells by equilibrative transporters; such release often accompanies cellular or metabolic stress. Adenosine, released by neurons, mediates synaptic depression and represents an autonomic feedback mechanism by which neurons downregulate excitatory transmission during episodes of prolonged activity (1011).

Adenosine is accumulated into astrocytes by equilibrating and concentrating transporters (see sect. IXB5), after which it is rapidly phosphorylated to AMP by adenosine kinase (ADK), which represents the key enzyme of adenosine metabolism (204). In the adult CNS, astrocytes express high levels of ADK (1686), which makes them important element of purines homeostasis and metabolism. Manipulations with adenosine kinase immediately affect presynaptic inhibition: overexpression of ADK results in reduction of adenosine level and promotes seizures (980); pharmacological inhibition of ADK increases neuronal firing (1298), whereas genetic deletion of ADK is incompatible with life (205).

3. Monoamines

Monoamines (norepinephrine, dopamine, and serotonin) in the CNS are catabolized through deamination or methylation, the former process being catalyzed by monoamine oxidase A and B (MAO A/B) and the latter by catecholamine O-methyl transferase (COMT). The newborn brain contains both MAO A and MAO B; however, in the postnatal development MAO B activity increases ~15 times, whereas MAO A only doubles (1928). Astrocytes in vitro were reported to express both MAO A and MAO B (684), whereas astroglia in situ express MAO B; moreover, astrocytes are the main containers of this enzyme in the CNS (973, 1468, 1548). The MAO B is inhibited by l-deprenyl, also known as selegiline (1468). The ¹¹C-deuterium-l-deprenyl is a popular (and the only) astroglial tracer for PET brain imaging (1481). The second monoamine catabolizing enzyme, the COMT, has been detected both in astrocytes and in processes of neurons (642, 835).

E. Lactate Production and Turnover

The concept of astroglia-mediated supply of neurons with the energy substrate rests, conceptually, on a disproportionally high uptake of glucose by astrocytes. It has been postulated that neurons utilize ~90–95% (75) of energy consumed by the brain, although later estimates found neuronal share exaggerated (21). In the resting brain, glucose, however, is shared roughly equally between neurons and astrocytes (336, 1206). Glucose metabolism, consequently, differs between neurons and astrocytes; the former rely preferentially on oxidative metabolism with high ATP production (218), whereas the latter utilize glucose mainly through aerobic glycolysis which yields relatively low amounts of ATP but produces substantial amount of lactate (109, 127, 183).

The astrocyte-neuron lactate shuttle hypothesis, ANLSH (1357, 1358), assumes that neuronal metabolism relies, during rest and in particular during activity, on lactate secreted by astrocytes. Astrocytic lactate production is linked to synaptic glutamate release, with subsequent Na⁺-dependent glutamate uptake into astrocytes. An increase in astrocytic [Na⁺]_i activates the NKA and increases aerobic glycolysis, thus producing lactate, which is subsequently transported to neurons to support their metabolism. Astrocytes are equipped with export lactate MCT1 and MCT4 transporters (see sect. IXB14) and hence secrete lactate produced by aerobic glycolysis. Astrocytes (in culture, in slices, and in vivo) were also reported to express lactate-permeable channels, which may rapidly release large quantities of lactate (1656). Lactate (which is present in the brain interstitium at ~1–1.5 mM concentration) is taken up and utilized by neurons for ATP production; moreover, neurons even seem to prefer lactate over glucose (127, 218, 1417). Neuronal mitochondria contain a dedicated lactate oxidation complex that facilitates lactate entry into the mitochondrial oxidative cycle (657). Finally, in vivo brain activity leads to a rapid increase in lactate levels; when activity terminates, lactate returns to the basal levels (274, 1058).

Several experimental arguments corroborated the ANLSH. For example, electrical activity can be sustained in cultures of neurons, in slice preparations, and in the isolated brain based on oxidation of lactate (775, 1582). Astrocytes in culture have been found to release higher quantities of lactate than cultured neurons and hence can be the lactate source. It has been argued, however, that this observation reflects predominant neuronal expression of lactate dehydrogenase 1 that is strongly inhibited by pyruvate and that the increase in pyruvate concentration that occurs during neural activity directs pyruvate towards the tricarboxylic acid cycle rather than towards conversion to lactate (325). This argument is further supported by the observation that cultured neurons release as much lactate as astrocytes when oxidative metabolism is blocked (1849). In the cerebellum, Bergmann glial cells take up several times more fluorescent glucose than Purkinje neurons (109). Neuronal activity in barrel cortex in vivo stimulated uptake of fluorescent glucose preferentially into astrocytes rather than neurons (336). However, this analysis was based on experiments using 6-deoxy-N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)-aminoglucose (6-NBDG), which cannot be phosphorylated in the cytosol by the hexokinase. In contrast, in vivo studies based on analysis of a near-infrared 2-deoxyglucose analog showed that neuronal uptake of the tracer is approximately two- to threefold higher than neighboring astrocytes across all regions studied (1020). Expression of hexokinase, which correlates directly with the rate of glucose metabolism in different regions, is consistently higher in neurons than astrocytes in mouse and human brain (1020). Moreover, an analysis of phosphorylated 2-fluoro-2-deoxy-d-glucose (FDG6P) in vivo found high concentrations of FDG6P in nerve terminals, suggesting that neuronal glucose-derived pyruvate (and not lactate derived from astrocytes) is the major oxidative fuel for active neurons, (1338). On the other hand, monitoring NADH levels in hippocampal slices indicated a clear separation between an increase in oxidative neuronal and glycolytic astroglial glucose metabolism (841), yet when these experiments were repeated in vivo, no clear separation was observed (840). In summary, although it is well documented that high neuronal activity is coupled with lactate production, definitive in vivo studies defining which, neuronal or astrocytic, lactate production predominates are yet to be done.

At the same time, the overall picture could become even more complicated, after it was found that up to 60% of lactate increase in the active brain derives from the blood (219). At the same time, it turned out that astrocytes are able not only to produce and release, but also to take-up and buffer extracellular lactate loads (546). Thus astroglia may have a dual role in metabolic support of neurons; it may produce and release lactate or else take lactate up from the blood and commute it to neurons (508).

In addition to producing and redistributing lactate, astrocytes are the sole possessors of brain glycogen. The glycogen granules in the brain were discovered in 1926 (716). In the mature brain, glycogen is present mostly in astrocytes, although it has been also detected in small sub-population of neurons in the brain stem and in some ependymal and choroid plexus cells (245, 295, 1048). Incidentally neurons do express glycogen synthase, which however is kept inactive by proteasomal-dependent mechanisms; when artificially activated it causes neuronal apoptosis (1823). Physiological activity affects glycogen levels in astrocytes; it is controlled by norepinephrine, serotonin, and VIP. Glycogen increases in sleep and anesthesia and decreases in awake and active brain, while glycogen mobilization helps to sustain neuronal activity (127, 246, 1047). Glycogenolysis is also critically important for maintenance of LTP (1706) and for memory consolidation (573). Are these physiological effects associated only with the glycogen function as an energy reserve? Total amount of glycogen in the brain is very low, being in a range of 0.5–1.5 g, which is 100 times less than in the liver (245). This amount of glycogen cannot sustain the brain activity for any reasonable time, and it might exert its physiological effects through yet unknown mechanisms, for example, through local signaling of undefined nature.

The metabolic role of astroglia and distribution of energy consumption between neurons and glia are far form being resolved. Futhermore, the metabolic plasticity of the brain may dynamically affect neuronal-glial energetic balance and lactate turnover.

F. Regulation of Synaptic Connectivity and Synaptic Transmission

1. Astroglial synaptic coverage

At least half of all synaptic contacts within the CNS are covered with perisynaptic astroglial membranous sheaths, which represent terminal extensions of peripheral astroglial processes (known as PAPs). The PAPs and perisynaptic processes are immunopositive for glutamate synthetase and astroglial glutamate transporters, and they do not express detectable levels of GFAP (415, 416, 1451). Perisynaptic glial compartments are generally devoid of organelles (1375, 1451), including endoplasmic reticulum (1340), although they may contain very small (0.2–0.4 μm) spherical mitochondria (419). The perisynaptic processes specifically express ezrin and radixin (415, 418, 952), which associate with actin and may contribute to morphological plasticity of astroglial processes. Rapid filopodial movements of astroglial processes have been characterized in vitro and in organotypic cultures (702, 952). This form of morphological remodeling may participate in synaptic plasticity (672), although it has yet to be confirmed in the in vivo brain.

The degree of astroglial coverage differs between brain regions and seems to reflect the nature of the synapse. In the neocortex 29–56% of excitatory synapses are enwrapped by glial processes; in layer IV of the somatosensory cortex, ~90% of synapses are covered with glial membrane (157). In the hippocampus the number of enwrapped synapses varies between 60 and 90% (1802, 1888). In the cerebellum, complex appendages emanating from the processes of Bergmann glial cells (FIGURE 30) cover almost 90% of synapses formed by climbing fibers and ~65% of synapses formed by parallel fibers on the Purkinje neuron (609, 1903). Degree of coverage also varies between different types of synapses. In the hippocampus, astrocytes cover ~50% of small macular synapses, while ~90% of large mushroom spines and perforated synapses are enwrapped with astrocytic processes (1888). In organotypic slices, 97% of complex synapses and 78% of simple synapses have astroglial coverage (1024). At the synapses formed by mossy fibers on granule neurons, coverage of individual synapses is ~15% (1128, 1711); however, astrocytes also cover whole glomeruli thus isolating them (1451).

FIGURE 30.

Microdomain organization of Bergmann glia and coverage of parallel fibers to Purkinje neuron synapses. A: reconstruction of an appendage based on electron microscopic data. a: Fluorescence light micrograph of a dye-injected Bergmann glial cell is shown; the red square (20 × 20 mm) corresponds to the portion that was reconstructed from consecutive ultrathin sections. b: One of the lateral appendages, arising directly from fiber with all the other side branches omitted for clarity. c: The same appendage as shown in b, but with one of the appendages marked by blue. This labeled structure is shown in isolation and at higher magnification in B. B: fine structure of appendages and relation to synapses. a: A small lateral appendage, arising from the reconstructed part of the glial fiber (blue in Ac), is shown as a slightly turned, isolated three-dimensional reconstruction (left). Electron micrographs of four sections contributing to the reconstruction (designated 1–4) are shown on the right; glial compartments appear black after conversion of the injected dye. The location of these sections in the reconstruction is indicated by the labeled arrows: 1, region directly contacting synapses; 2, glial compartments without direct synaptic contacts; 3, bulging glial structure containing a mitochondrion; 4, the stalk of the appendage. C: another example of a synapse contacted by glial compartments, appearing as black structures (white arrows); in this case, the postsynaptic element can be traced back along the spine (black arrows) to the Purkinje cell dendrite (reddish overlay; the presynaptic terminal is labeled by a green overlay). [From Grosche et al. (609).]

The perisynaptic astroglial processes covering synapses are exceedingly thin, their profiles being on average <200 nm (and often <100 nm) in diameter, although they have an exceedingly large surface (1451). The perisynaptic astroglial membrane represents the major part of the cell surface area, accounting for ~80% of the total astroglial plasmalemma, while contributing only a minor fraction (~4–10%) to the cellular volume; as a result, astroglial perisynaptic processes have an extremely high surface to volume ratio (~25 μm⁻¹; Ref. 608). Perisynaptic astroglial membrane is densely packed with receptors, ion channels, and multiple transporters; ion fluxes are particularly important for generation of local Ca²⁺ and Na⁺ signals, which couple homeostatic astroglial cascades with neuronal activity (1813). The perisynaptic processes are plastic structures; for example, they rapidly ensheath dendrites and synapses in neurons newly generated in dentate gyrus and integrated into adult hippocampus (912).

2. Tripartite synapse and synaptic cradle

Intimate morphological relations between astroglial membranes and synaptic structures, as well as wide expression of neurotransmitter receptors in astrocytes, led to a concept of a synaptic triad (855), which subsequently developed into the tripartite synapse model (58, 626). The tripartite synapse (FIGURE 31) highlights glial contribution to the regulation of synaptic transmission; astroglial processes become a legitimate part of synaptology together with pre- and postsynaptic neuronal compartments. The tripartite model further developed into a multipartite synapse, composed of the following components: 1) the presynaptic terminal; 2) the postsynaptic compartment, represented, for example, by the dendritic spine; 3) the perisynaptic process of the astrocyte; 4) the process of neighboring microglial cell that periodically contacts the synaptic structure; and 5) the extracellular matrix (ECM), which is present in the synaptic cleft and also extends extrasynaptically (390, 434, 494, 1205, 1812).

FIGURE 31.

Tripartite synapse and synaptic cradle. A: the original concept of tripartite synapse. “During synaptic neurotransmission, neurons release neurotransmitters from nerve terminals into the synaptic cleft to communicate with other neurons (1). The neurotransmitter released from the synapse (or other co-released neurotransmitter) can, under certain circumstances, “spill over” from the synaptic cleft and reach neurotransmitter receptors in adjacent astrocytes, eliciting intracellular increases in Ca²⁺ concentrations in the glial cells (2). The increase in the glial-cell Ca²⁺ concentration causes it to release a chemical neurotransmitter from the glial cell, which in the case of astrocytes is glutamate (3), that feeds back to the presynaptic nerve terminal to modulate synaptic neurotransmission.” [From Araque et al. (58). The figure from original submission (although left unpublished) was kindly provided by Prof. V. Parpura (Birmingham University, Birmingham, AL).] B: astroglial cradle embraces and fosters multi-partite synapse in the CNS. The majority of synapses in the brain and in the spinal cord are composed of several components that include the presynaptic terminal, the postsynaptic compartment, the perisynaptic process of the astrocyte, the process of neighboring microglial cell that periodically contacts the synaptic structure, and the extracellular matrix (ECM) present in the synaptic cleft and also extended extrasynaptically. Astroglial perisynaptic sheath enwraps synaptic structures and regulate, influence, and assist synaptogenesis, synaptic maturation, synaptic maintenance, and synaptic extinction. [From Verkhratsky and Nedergaard (1812).]

The tripartite synapse model principally focuses on a bidirectional rapid neuronal-glial communications; neurotransmitter-induced glial Ca²⁺ signaling with subsequent exocytotic release of neurotransmitters from the astroglia (the act of “gliotransmission”) lies at its core. The role of astroglia in regulation of synaptic connectivity is, however, immensely wider: astrocytes control emergence and shaping of synaptic networks, they regulate ionic homeostasis of the synaptic cleft, control neurotransmitters dynamics, prevent or allow neurotransmitter spillover, and contribute to synaptic extinction. These multiple roles of astroglia in synaptic physiology were synthesized in the concept of the astroglial cradle (FIGURE 31, Refs. 1205, 1812). This model regards astroglial coverage as a multipurpose tool, responsible for multiple functions mandatory for physiological neurotransmission.

3. Synaptogenesis, synaptic maturation, and elimination

Throughout life, synapses emerge, remodel, and disappear, which is an important part of neuroplasticity, learning and memory. Synaptic connections progress through several stages that include 1) formation of an initial contact between the terminal and postsynaptic neuron, 2) maturation, 3) stabilization and maintenance, and 4) elimination. Embryonic synaptogenesis proceeds mainly without glia. The main wave of genesis of excitatory glutamatergic synapses in mammals occurs in the first postnatal weeks; this massive emergence of synapses immediately follows the wave of astrogliogenesis (1115). The critical role for astrocytes in synaptogenesis was discovered in vitro: addition of astrocytes to pure neuronal cultures increased the number of synapses sevenfold (1386). This link between astroglia and synaptogenesis has been subsequently confirmed (487). Astrocytes support synaptogenesis through secreting multiple factors of which the most important are cholesterol (1090) and thrombospondins (486), although other factors [such as estradiol, protocadherins, or integrins (1385)] are also involved. Cholesterol, in particular, provides building material for new membranes, which appear during synaptogenesis, and cholesterol may also be locally converted into steroid hormones, which in turn can act as synaptogenic signals (486, 1385). Astrocytes also secrete hevin, which potentiate genesis of excitatory synapses and secreted protein acidic and rich in cysteine (SPARC), which inhibits hevin-induced synaptogenesis (918). Astrocytes remain indispensable for synapse formation throughout life; astroglial deficiency affects synaptogenesis in adult CNS and prevents regeneration after injury (1762). Astroglial factors are similarly important for maturation of synapses. These factors include activity-dependent neurotrophic factor and TNF-α, which regulate the trafficking of glutamate receptors into postsynaptic membranes, whereas cholesterol may enhance neurotransmitter release from presynaptic terminals (487). Astroglia-derived glypicans 4 and 6 facilitate synapse maturation by increasing the number of glutamatergic α-amino-3-hydroxy-5-methyl-isoxazole propionate (AMPA) receptors at postsynaptic sites (24).

Astroglial synaptic coverage may also exert dynamic control on the synaptic density. In hippocampus, for example, in the absence of astroglial perisynaptic processes, or in the conditions of disrupted ezrin signaling affecting their motility, dendritic spines disappeared faster (1226). Synapse stabilization in Purkinje neurons similarly requires astroglial coverage, which also regulates synaptic density (988). Astrocytes may regulate synaptic density through dynamic remodeling of coverage of postsynaptic membranes. In the lateral amygdala, for example, retraction of astroglial processes from synapses is a prerequisite for synaptic remodeling associated with implicit memory consolidation of Pavlovian threat conditioning (1285).

Astroglia contribute to synaptic elimination and regulate synaptic morphology. For example, astrocytes in vitro were reported to label synaptic terminals with complement factor C1q. This tag is subsequently recognized by microglia, which eliminate these tagged synapses by selective phagocytosis (1555); furthermore, astrocytes by themselves can engulf and eliminate synapses in the developing brain (334).

4. Astroglial release of neurotransmitters

Emergence of the concept of the tripartite synapse had stimulated great interest in astroglial secretion of neurotransmitters. Numerous studies, in vitro (in astroglial-neuronal cocultures), in situ, and in vivo have addressed this matter; many sets of data have been generated, and heated debates ensued (55, 632, 1205, 1526). As has been discussed above (see sect. XIE), astrocytes contain neurotransmitters and do possess several pathways for releasing glutamate, ATP, and GABA. Conceptually, however, these neurotransmitters are released from astrocytes slower than from neuronal terminals, and astroglial release seems to be more diffused in space, as astrocytes lack any kind of presynaptic active zones, where vesicles concentrate.

How does astroglial neurotransmitter release affect neurons? The spectrum of neuronal responses is quite wide. Astroglial release of glutamate, for example, has been shown to result in 1) inhibition of evoked and spontaneous excitatory or inhibitory postsynaptic currents (EPSCs or IPSCs) in hippocampus (57, 992); 2) potentiation of spontaneous excitatory postsynaptic currents (EPSCs) or inhibitory postsynaptic currents (IPSCs) in hippocampus (804, 822, 992, 1537,1538); 3) synaptic potentiation (1198, 1199, 1366); 4) increase in neuronal excitability (168); 5) generation of slow inward currents, or SICs, in hippocampus, cortex, brain stem, and spinal cord, with this being probably the most frequent observation (47, 100, 316, 502, 591, 826, 1539, 1542, 1607, 1906); 6) modulation of long-term potentiation or long-term depression (634, 1119, 1199); and 7) heterosynaptic depression (41). Astroglia-derived ATP and/or adenosine caused 1) suppression of EPSCs (238, 1069, 1334, 1592), 2) potentiation of EPSCs (594, 595), 3) inhibition of NMDA receptors through P2X-PCD95 signaling cascade (934), 4) generation of SICs (373), 5) increase in neuronal excitability (959), and 6) modulation of LTP (959, 1334). There is also some limited information about modulation of neuronal function with astroglia-released GABA (900, 964) or TNF-α (1668); the latter was even reported to induce “gliogenic” LTP in the spinal cord (910).

These multiple effects, which can be exerted presynaptically, postsynaptically, or extrasynaptically, modulate distinct neuronal mechanisms that most likely differ between brain regions, reflecting complexity of astroglial-neuronal communications. Astrocytes can release neurotransmitters by several mechanisms, which can prevail in different contexts. There are several arguments against Ca²⁺-dependent vesicular neurotransmitter release from astrocytes. First, artificial obliteration or enhancement of astroglial Ca²⁺ signaling does not produce any apparent effects on neuronal activity (14, 507, 1382). Second, extracellular levels of glutamate are not affected in astroglia-expressing dnSNARE (suppressor of exocytosis) mice (340). Third, the astroglial origin of glutamate that induces SICs remains to be unequivocally confirmed; similar effects could reflect, for example, somatodendritic release from neurons (see Ref. 1205 for details). Finally, stimulation protocols used to activate astrocytes in situ may not necessarily reflect physiological situation in vivo. Overall, astrocytes probably do release neurotransmitters, which, however, act in a slow and long-lasting manner; they target extrasynaptic low-affinity receptors and modulate multiple synapses, rather than affecting individual synaptic events.

5. Control of neurotransmitter dynamic in the synaptic cleft

Astrocytes, through their dedicated transporters, define dynamics of neurotransmitter concentration in the synaptic cleft and hence shape postsynaptic responses (1062, 1768). This is particularly prominent for glutamatergic transmission. Glutamate released from the presynaptic terminal, while diffusing to the postsynaptic membrane, faces rapid buffering by astroglial glutamate transporters densely populating the perisynaptic astroglial membranes. It has been estimated (1062) that glutamate, exocytosed from presynaptic terminal, encounters, within the diffusion distance of 0.5 μm, ~7,000 glutamate transporters, which substantially exceeds the number of AMPA receptors (~25) available for postsynaptic activation. Astroglial glutamate buffering reduces the amplitude and shortens NMDA receptor-mediated EPSCs in hippocampal (72, 428, 1106, 1764) and cerebellar (1289) synapses. Inhibition of glutamate uptake more than doubled the half time of the decay of the NMDA receptor-mediated EPSCs in CA1 hippocampal synapses (66). The AMPA receptor-mediated EPSCs, because of their fast desensitization, are generally insensitive to astroglial glutamate uptake (692, 1540), although in the auditory nerve synapse in the nucleus magnocellularis as well as in cerebellar parallel and climbing fiber synapses astroglial glutamate transport accelerates AMPA receptor-mediated EPSCs (1287, 1289, 1715). Removal of AMPA receptor desensitization with cyclothiazide renders AMPA receptor-mediated EPSCs sensitive to glutamate buffering (1764), again indicating a significant role of the latter in defining glutamate intracleft dynamics. Glutamate transporters similarly limit activation of neuronal metabotropic glutamate receptors: pharmacological inhibition (230, 1446) or genetic deletion (1062) of transporters resulted in severalfold potentiation of mGluR-mediated synaptic currents. Astroglial glutamate uptake seems to be regulated by neuronal activity. In mouse cortex, bursts of action potentials led to a rapid and short-lived (~50 ms) threefold suppression of glutamate uptake confined only to active synapses (64), which may reflect another mechanism of glia-dependent modulation of synaptic transmission.

Astroglial GABA transporters also have a role in controlling concentration of GABA in the synaptic cleft. The astrocytic GAT-3 for example was reported to regulate extracellular GABA and hence tonic GABAergic transmission in hippocampus (852). In thalamus, astrocytes express both GAT-1 and GAT-3 (387, 1828), activity of which affects kinetics of GABA_B receptor-mediated postsynaptic currents, with GAT-1 mainly affecting peak of IPSCs, while GAT-3 prevents neurotransmitter spillover (121).

6. Synaptic isolation

Astroglia, through perisynaptic processes and glutamate transporters, are instrumental for preserving spatial isolation of synaptic transmission, thus maintaining signaling specificity. Glutamate buffering prevents diffusion of neurotransmitters between adjacent synapses. Pharmacological inhibition of glutamate uptake with dl-threo-β-benzyloxyaspartic acid (dl-TBOA) significantly prolonged EPSCs in response to stimulation of multiple inputs in organotypic acute hippocampal slices, this being indicative of glutamate spillover (66). Similar results were obtained in cerebellum in parallel fibers to Purkinje neuron synapses of transgenic mice lacking EAAT1 transporters (1063). The degree of synaptic isolation is heterogeneous between CNS regions, and it changes during development. As a result, glutamate diffusion and synaptic interconnectivity may also change at different developmental stages and in different physiological context. Moreover, morphological plasticity of astroglial processes can rapidly affect synapse to synapse crosstalk.

7. Morphological plasticity of perisynaptic glial coverage

Dynamic changes in astroglial synaptic coverage represent another mechanism of regulation of synaptic transmission. Morphological plasticity of perisynaptic processes potentially may affect synaptic isolation, volume of the synaptic cleft, and availability of neurotransmitter transporters. Astroglial morphological plasticity is well known from in vitro studies in which various stimuli (such as norepinephrine or dibutyryl cAMP) result is rapid stellation (1479, 1794). Morphological remodeling of astroglial processes that affects neurotransmission has been subsequently described in the supraoptic nucleus in situ: in lactating rats, astroglial processes retracted thus permitting diffusion of glutamate and increasing presynaptic inhibition mediated by mGluRs (1260). In transgenic mice expressing Ca²⁺-impermeable AMPA receptors, Bergmann glial cells undergo retraction of perisynaptic processes with consequent potentiation of excitatory transmission (759). Astroglial morphological plasticity is activity dependent, and 24 h of whisker stimulation increases astroglial coverage (paralleled with an increase in EAAT expression) of synapses in the barrel cortex (566). Induction of LTP in hippocampus enhances motility of perisynaptic astroglial processes, which depends on glial Ca²⁺ signaling (1372). Astroglial synaptic coverage is highly dynamic and depends on the brain status. In sleep, for example, synaptic coverage decreases, while in wakefulness and sleep deprivation perisynaptic processes come closer to the dendrites and increase synaptic coverage (130).

G. Astrocytes Regulate Brain Microcirculation: The Concept of Neurovascular Unit

Activity of the brain is linked to local circulation, and increase in neuronal firing rapidly trigger vasodilatation of small vessels localized within ~200–250 μm from the site of increased neuronal activity. This is known as functional hyperemia that was discovered by Angelo Mosso (1159) and by Charles Roy and Charles Sherrington (1512), who postulated that “. . . the brain possesses an intrinsic mechanism by which its vascular supply can be varied locally in correspondence with local variations of functional activity” (1512).

Coupling of neuronal activity with local circulation is controlled by multiple mechanisms; neurons, for example, are known to release vasoactive agents such as NO or prostaglandins, and brain vasculature is richly innervated by neuronal projections that exocytose vasoactive neurotransmitters including acetylcholine, norepinephrine, serotonin, or dopamine (74, 300, 738). Compelling evidence indicates a specific role of astrocytes in neurovascular coupling and regulation of local blood flow. First, astrocytes were proposed to release arachidonic acid in response to stimulation with glutamate; subsequently, this arachidonic acid was suggested to be converted to a potent vasodilator epoxyeicosatrienoic acids, or EETs, by cytochrome P-450 (645).

This initial hypothesis was soon tested in experiments in brain slices, which revealed even more complex astroglial roles: glutamate stimulation of astrocytes and astroglial Ca²⁺ signals may evoke both vasodilatation and vasoconstriction. The vasodilatation was attributed to astroglia-derived prostaglandin E₂ (PGE₂) (1970), whereas vasoconstriction was found to be mediated by 20-hydroxyeicosatetraenoic acid (20-HETE) (1171). In the retina, both effects could be observed with vasodilatation being mediated by PGE₂ and EETs and vasoconstriction by 20-HETE (1110, 1127). Astroglia-dependent functional hyperemia mediated by prostaglandins (1718) as well as by EETs (995) was subsequently detected in vivo in the cortex. Astrocytes were also found to regulate resting vascular tone in similar bidirectional manner. In the in situ and in vivo cortex, the pressure-evoked vasomotor tone was controlled by astroglia-derived ATP; the latter was released following TRPV4-mediated [Ca²⁺]_i elevation (862). Similarly ATP released from astrocytes constricted arterioles in the retina (927). The tonic dilatation was also regulated by astrocytes, but in a COX1-dependent way (1503). The Ca²⁺ dependence of astroglia-vascular coupling has been questioned recently, when it appeared that type 2 InsP₃R^−/− mice (that are unable to generate global Ca²⁺ signals) functional hyperemia was preserved (210, 791, 1721).

An alternative mechanism associates vasoconstriction with the release of K⁺ from astroglial endfeet; increase in extracellular K⁺ supposedly hyperpolarizes vascular smooth muscle, hence causing vasodilatation. Initially the K_ir4.1 channels were suggested to mediate this K⁺ efflux (1342); however, this was negated by experiments on K_ir4.1^−/− mice which have fully preserved functional hyperemia (1109). The hypothesis, however, has resurfaced recently, and the Ca²⁺-dependent K⁺ channels were found to be responsible for K⁺ efflux from astroglial endfeet that subsequently hyperpolarized smooth muscle cells (511).

Accumulation of experimental evidence for astroglia-controlled neurovascular coupling strengthen the concept of neurovascular (or neuro-glio-vascular) unit as an assembly of neurons, interneurons, astrocytes, microglia, basal lamina, smooth muscle cells, pericytes, endothelial cells, and extracellular matrix, which by coordinated and reciprocal signaling, provides for local control of cerebral blood flow (646, 1173, 1870). Astroglia represent the central element of the neurovascular unit through integrating all other components that reside in its structural domain; moreover, the astroglial endfoot appears as a key element for neurovascular communication.

H. The Glymphatic System and the Unique Perivascular Space of the CNS

Any type of biological activity is linked to production of potential toxic waste products. In peripheral tissues, the lymphatic system plays an essential role in the removal of excess fluid and metabolic waste. The density of lymphatic capillaries matches the local rate of metabolism across multiple tissues and organs (1567). However, in spite of neural tissue being highly metabolically active, no conventional lymphatic system is present in the eye, in the brain, or in the spinal cord (9). Recent work has documented the existence of lymphatic vessels in close proximity to, but outside of, the brain in the dural and meningeal membrane (71, 1009). A characteristic feature of the CNS is the existence of a unique perivascular space created by astrocytic endfeet, which encase ~99% of the entire vasculature (1082). Although this observation has been questioned by the finding of a less complete coverage using cryofixation (897), it is clear that astrocytic endfeet form a donut-shaped tunnel system that surrounds arteries, capillaries, and veins. The separation of the vasculature from remaining tissue by vascular endfeet is unique to the CNS, as in peripheral organs blood vessels are either directly embedded into the tissue or are surrounded by loose fibrous structures. A series of recent studies have identified an organized pathway for interstitial solute clearance from the brain that utilizes the perivascular space as a highway for fast fluid transport. It was designated the “glymphatic system” (FIGURE 32) because it operates similarly to the peripheral lymphatic system and depends on astroglial AQP4 water channels (761). This system consists of a periarterial CSF influx path and a perivenous interstitial fluid (ISF) clearance route, which are coupled through convective interstitial bulk flow supported by astrocytic AQP4 channels (1204).

FIGURE 32.

Schematic outline of the glymphatic system. Convective glymphatic fluxes of cerebrospinal fluid (CSF) and interstitial fluid (ISF) propel the waste products of neuron metabolism into the paravenous space, from which they are directed into lymphatic vessels and ultimately return to the general circulation for clearance by the kidney and liver. [From Nedergaard (1204).]

The brain-wide glymphatic pathway serves as a waste disposal system for fluid, as well as for larger solutes and proteins, such as β-amyloid and tau, from the CNS. Quantitative analysis shows that 65% of β-amyloid is cleared by the glymphatic system (761). Glymphatic transport declines significantly as a function of aging and more rapidly in rodent models of Alzheimer's disease, stroke, and diabetes (542, 793, 904, 1365). With the use of magnetic resonance imaging (MRI) in rats (760), it was shown that clinically relevant lumbar intrathecal administration of CSF tracers permits the capture of all the key components necessary to assess glymphatic function (1914). It has been also demonstrated that the lateral and supine positions, which are the most popular sleep positions among humans and most animal species, are superior for β-amyloid clearance compared with the prone position (958). Most surprisingly, the sleep-wake cycle regulates glymphatic activity, which is primarily active during sleep or anesthesia (1899). The glymphatic pathway is also involved in brain metabolism, delivers both lipids and glucose, and plays a key role in the export of lactate (1020, 1021, 1432). Additionally, very recent work has documented the existence of glymphatic fluid transport in the human brain (472, 887). Overall, brain-wide astrocytic fluid transport may therefore serve as a critical pathway for removal of neurotoxic proteins that accumulate in neurodegenerative diseases.

I. Astroglia in Central Chemoreception of Oxygen, pH, and CO₂ and in Regulation of Respiration

Astrocytes throughout the brain possess an oxygen sensor linked to Ca²⁺ signaling. Decrease in Po₂ triggered [Ca²⁺]_i increase in cortical astrocytes in vivo, in astrocytes from brain slices, and in astrocytes in primary culture (isolated from cortex, hippocampus, midbrain, and brain stem) (45). The threshold for triggering astroglial [Ca²⁺]_i responses was determined at ~17 mmHg, being thus substantially lower than Po₂ threshold (~37 mmHg) in glomus cells of the carotid body (45). The actual oxygen sensor is associated with mitochondria; apparently hypoxia-induced reduction in respiration triggered production of ROS with subsequent activation of PLC and InsP₃-induced Ca²⁺ release from the endoplasmic reticulum. Hypoxia-induced Ca²⁺ signaling was claimed to initiate tetanus-toxin-sensitive (i.e., vesicular) ATP release from astrocytes; in the brain stem, this ATP in turn activated neuronal circuits responsible for respiration. Inhibition of astroglial ATP secretion suppressed hypoxia-induced increases in respiratory rate (45).

Astrocytes in the brain stem (in retrotrapezoid nucleus) act as central chemosensors for CO₂ and pH. Decrease in pH by 0.2–0.4 units triggers Ca²⁺ signals in brain stem astroglia with subsequent ATP release (598). This pH sensitivity is seemingly specific for brain stem astrocytes; cortical astroglia appear insensitive to pH fluctuations (843). In the brain stem, astroglia-derived ATP signals to respiratory neuronal network, thus initiating ventilatory response (598, 1875). The mechanism of pH-induced Ca²⁺ signaling involves Na⁺ entry through NBC1 with subsequent reversal of NCX and Ca²⁺ influx (1766). The ATP release pathway is seemingly Ca²⁺ dependent (implying a role for exocytosis), although evidence exists suggesting connexin hemichannels as a conduit (751).

J. Astrocytes in Regulation of Systemic Sodium Homeostasis

Brain chemosensors for the body sodium homeostatic system are localized in the circumventricular organs, which surround the ventricles. Changes in Na⁺ concentration in blood plasma or in cerebrospinal fluid modulate circumventricular neuronal activity and result in physiological (changes in kidney Na⁺ excretion) or behavioral (avoidance of dietary NaCl) responses (1630). Astrocytes in circumventricular organs and in particular in the subfornical organ act as sensors of systemic [Na⁺]. This is mediated by [Na⁺]_o-sensitive Na_x channels (see sect. VIIB2) specifically expressed by astrocytes in this part of the brain. Increase in plasma [Na⁺] above 140 mM activates Na_x, which results in an increase in [Na⁺]_i, activation of the Na⁺/K⁺ pump, and stimulation of aerobic glycolysis. Aerobic glycolysis produces lactate that is transferred to neighboring GABAergic neurons; consequently, neurons increase ATP production with subsequent closure of ATP-sensitive K⁺ channels, which causes neuronal depolarization and initiation of a systemic Na⁺-homeostatic response (1234, 1235, 1613). The role for astrocytes was further corroborated by the observation that genetic deletion of Na_x from astrocytes altered Na⁺-aversive behaviors following Na⁺ overload (1613).

K. Astroglia and Circadian Rhythms

The circadian clock is localized in the suprachiasmatic nuclei (SCN) within the anterior hypothalamus; this structure generates rhythmical activity is driven by transcription-translation feedback loops of ubiquitous canonical clock genes (1617). Neurons from SCN are capable to oscillate in isolation; in vivo however circadian rhythms are coordinated and synchronized by intercellular communications. Astroglial cells contain circadian genes that undergo rhythmic expression (1412, 1908). Astrocytes isolated from SCN are similarly capable of circadian activity, which results in circadian ATP release triggered by InsP₃-induced Ca²⁺ signaling (1067). This circadian ATP signaling is also controlled by mitochondria (261). Astroglia in the SCN display rhythmic morphological changes from stellate to more protoplasmic shape during the day/night cycle. This morphological remodeling is accompanied by changes in expression of GFAP and glutamate transporters, as well as with day/night changes in glial synaptic coverage which is higher at night (119, 780). Astrocytes have been also suggested to modulate SCN neurons by releasing TNF-α (461).

L. Astroglia and Sleep

Conceptually, sleep is regulated by the circadian clock (discussed above) and by the sleep homeostat, which controls the pressure to sleep (521). The sleep/wake cycle is associated with complex changes in astroglia (203). The astroglial transcriptome undergoes significant remodeling, with 1.4% of all transcripts showing changes in expression in sleep versus wakefulness (130). Astrocytes express 396 unique genes associated with wakefulness and 55 with sleep (130); at the same time, sleep does not affect the oligodendroglial transcriptome (131). Changes in astroglial transcriptome parallel morphological plasticity: sleep deprivation increases astroglial synaptic coverage in prefrontal cortex (130). To the contrary during normal sleep, the extracellular volume increases, which is likely to be mediated by changes in astroglial morphology (1899). Further evidence links astrocytes to fundamental elements of the sleep homeostat associated with an elevation of brain adenosine content in the wake state (1398, 1740). Conditional expression of dnSNARE in astrocytes (which apparently suppresses vesicular release of ATP and neurotransmitters; see sect. XIB) resulted in a reduced sleep pressure as well as prevented compensatory increases in sleep time following sleep deprivation in healthy wild-type animals (628). Very similar changes in sleep/wake behavior were observed in animals subjected to the intracerebroventricular infusion of adenosine A₁ receptor antagonist, 8-cyclopentyl-1,3-dimethylxanthine (CPT) (628). These observations have not been universally confirmed because the interstitial level of adenosine has been found unchanged in dnSNARE mice (540). The adenosine A₁ receptor pathways were also found to underlie increased sleep pressure under acute (induced by peripheral administration of lipopolysaccharide, LPS) inflammatory conditions (1183). It was further found, by using biosensor-based monitoring, that adenosine concentration indeed increases during wakefulness and sleep deprivation (1571). The astrocyte-sleep connection was further corroborated in experiments using astroglia-expressed channelrhodopsin-2: optical stimulation of hypothalamic astrocytes was found to induce sleep (1359). Finally, astrocytes may regulate sleep-wake cycle through a completely different mechanism, associated with dynamic ionostatic control of extracellular ionic composition (432).

M. Systemic Energy Homeostasis

A complex cellular network localized in the hypothalamic nuclei represented by the arcuate nucleus, the ventromedial hypothalamic nucleus, and the paraventricular nucleus (PVN) regulates body energy balance. These hypothalamic structures integrate numerous signals including major metabolites (glucose, free fatty acids, and amino acids) as well as dedicated hormones (leptin, ghrelin and insulin) and regulate appetite and food intake, deposition of fat, and energy expenditure (1157). Specialized hypothalamic astrocytes contribute to systemic energy homeostasis by glucose sensing, by providing the conduit for hormones to the hypothalamic neurons, and by directly affecting activity of these neurons.

1. Glucose sensing

The link between glucose concentration in the blood and stimulation of appetite (at low concentrations) or feeling of satiation has been proposed almost a century ago (287). Hypothalamus contains several population of “glucose-sensing” neurons, which respond to elevated glucose concentration with either increase (glucose-excited cells) or decrease (glucose-inhibited cells) in their firing rate (37, 1272). The sensitivity of these neurons to physiological glucose fluctuations remains debatable, although a subpopulation of glucose-inhibited orexin-producing neurons readily respond to physiologically relevant increases in glucose concentration (259). Tanycytes represent the second glucose-sensing cellular population of the hypothalamus. In the medial eminence, β2 tanycytes constitute the blood-brain barrier sealed by tight junctions between the somata of these glial cells; this barrier prevents diffusion of blood-originated molecules to the brain parenchyma. At the same time processes of β2 tanycytes establish direct contact with the fenestrated vessels of the pituitary portal blood system, thus being able to inspect the molecular composition of the blood, including glucose and hormones (1170, 1482, 1554).

These hypothalamic tanycytes have complex signaling: they express P2X and P2Y₁ purinoceptors, ACh receptors, and histamine receptors all linked to the generation of intracellular Ca²⁺ signals (376, 524). Furthermore, β2 tanycytes are functionally connected to propagating Ca²⁺ waves and are capable of releasing ATP (376, 524). Application of glucose or its nonmetabolizable analogs to the somata of tanycytes in acute slices triggered [Ca²⁺]_i elevation and Ca²⁺ waves; this signaling was mediated by ATP acting through P2Y₁ receptors to evoke Ca²⁺ release from the endoplasmic reticulum (206, 524). This glucose-induced Ca²⁺ signaling was cell autonomous and was replicated in the in vitro experiments: increases in extracellular glucose from 2 to 10 mM triggered Ca²⁺ signaling in cultured tanycytes (1273). These Ca²⁺ responses were similarly ATP dependent and required functional Cx43 hemichannels through which ATP was presumably released (1273). Mechanisms of glucose-induced Ca²⁺ signaling remain unclear; they may involve [Na⁺]_i rise (generated by glucose-Na⁺ cotransporter SGLT1) with subsequent reversal of NCX or operation of yet unknown glucose-sensitive metabotropic receptor coupled to PLC/InsP₃/Ca²⁺ release cascade (206). This astroglial signaling potentially may influence hypothalamic neurons; some initial data indicate that optogenetic stimulation of tanycytes through Ca²⁺-permeable channelrhodopsin2 resulted in excitation of neurons in the arcuate nucleus (207). Hypothalamic parenchymal astrocytes may also employ the GLUT2-glucokinase2 pathway for glucose-sensing via K-ATP channels (967), although this pathway requires physiological confirmation. Finally, conditional genetic deletion of astroglial insulin receptors affects glucose-sensing and systemic glucose metabolism (549).

2. Tanycytes provide a conduit for peripheral hormones to the brain

As mentioned above, tanycytes are capable of sensing circulating hormones, and they are the first to respond to intraperitoneal injection of leptin by leptin receptor (LepR)-mediated stimulation of STAT3 phosphorylation that occurs (as visualized by phosphorylated STAT3 immunoreactivity) in the processes contacting fenestrated capillaries (89). Moreover, leptin, which freely diffuses through these fenestra, is taken up by tanycytes (89, 1554), the process of which is somehow mediated by LepRs (89). It is hypothesized that captured leptin is transported through tanycytes and released into the parenchyma of hypothalamus from which it may further disseminate through the brain; this hypothesis has been further extended to ghrelin and possibly other peptide hormones (273, 345).

3. Astrocytes may regulate food intake

Food intake is controlled by two populations of hypothalamic neurons: agouti-related protein (AGRP) expressing neurons that stimulate food intake and pro-opiomelanocortin (POMC) neurons that inhibit food-seeking behavior (73). The role of astrocytes in regulation of food intake was first suggested based on the analysis of transgenic animals in which LepRs were specifically deleted from astrocytes. This manipulation reduced the number and complexity of hypothalamic (but not hippocampal) astroglia processes and decreased astroglial synaptic coverage on both AGRP and POMC neurons, which arguably led to an aberrant synaptic transmission and aberrant feeding behavior (861). Incidentally, astroglial deletion of LepRs prevented the development of morbid obesity in the modified mice (790). Stimulation of medial basal hypothalamic astrocytes carrying muscarinic receptor variant hM3Dq (using the Designer Receptor Exclusively Activated by Designer Drug,or DREADD technology) inhibited both basal feeding and feeding evoked by ghrelin. These effects were arguably mediated through stimulation of A₁ adenosine receptors located on AGRP neurons (1915). Such an artificial stimulation, however, may not necessarily reflect physiological context, and hence more research is needed.

N. Astroglia and Control of Reproduction

The gonadotropin-releasing hormone (GnRH), secreted by specific gonadotropin-releasing hormone neurons in the hypothalamus, controls coordinated release of luteinizing hormone and follicle-stimulating hormone, which both regulate reproductive function. There is some evidence for astrocytes influencing these events. Astroglia show hormone-dependent phasic morphological plasticity in the estrus cycle, which may regulate the density of synapses (e.g., GABAergic; Ref. 1319). Increase in astroglial membrane coverage at high estrogen levels prevents formation of inhibitory synaptic inputs in arcuate neurons (552) Steroid-induced remodeling of astroglia was also reported to reduce the number of dendritic spines in rat hypothalamus (1142). Thus astroglial morphological plasticity may represent a mechanistic link between plasma levels of 17β-estradiol and pulsatile secretion of gonadotropin hormones (422). Astrocytes also secrete several factors, which simulate GnRH release; these factors include transforming growth factor-β1, astrocyte-derived transforming growth factor-α, or neuroactive steroid metabolites (1051). It was also reported that oxytocin-induced PGE₂ secretion from astrocytes regulates pulsatile release of GnRH from neurons (1321).

O. Müller Glial Cells as Light Guides in Retina

The most exotic astroglial function belongs to Müller retinal glial cells. Müller cells perform all archetypal astroglial functions such as K⁺ homeostasis, glutamate uptake, regulation of extracellular pH, control of synaptogenesis, etc. (1450). In addition, Müller cells, which span the whole thickness of retina and are oriented along the direction of the light path, act as single-cell optical fibers that convey light directly to photoreceptors with minimal distortion or loss of photons (FIGURE 33). In the human retina, every Müller cell is coupled to one cone photoreceptor cell (responsible for sharp sight under daylight conditions, i.e., photopic vision) and several (~10) rod photoreceptor cells. The endfoot of the Müller cell is shaped as a funnel, and apparently acts as a light collector at the vitreal surface of the retina (522).

FIGURE 33.

Müller cells as light guides. Shown as an artistic impression, the light penetrates the retina using Müller glial cells as guides. (Artwork courtesy of Dr. Jens Grosche, Leipzig University.)

P. Recapitulation

The main and most fundamental function of astroglia is sustaining homeostasis at all levels of CNS organization. Employing a variety of mechanisms, astrocytes provide for molecular, cellular, structural, organ, and systemic homeostatic control. Through a synaptic cradle that embraces CNS synapses, astrocytes control the birth, maturation, and extinction of synapses; through releasing numerous neuroactive substances and regulating ionostasis, astrocytes affect neuronal excitability and synaptic connectivity. Astrocytes orchestrate homeostasis of the brain as an organ, particularly through establishing the glymphatic system that removes accumulated waste. Astrocytes are chemosensing elements of the brain contributing to systemic homeostasis of ions, metabolites, and energy.

Surprisingly limited efforts have been made into understanding how astrocytes participate in brain function on a macroscopic scale. The majority of functional studies of astrocytes have focused on local signaling. Even the archetypal supportive functions, such as astrocytic K⁺ buffering or glutamate uptake, have been generally studied at a cellular level. However, the brain is an organ assembled, on a macroscopic scale, to support the basic needs for energy metabolites and export of waste products. Since astrocytes play pivotal roles in both of these processes, a higher order of organization of astrocytes must exist to fulfil brain-wide tasks, in addition to the cellular specialization needed to meet local demands. The best example of a macroscopic function of astrocytes is the glymphatic system. Astrocytes create, by their vascular endfeet, the perivascular space which provides a low-resistance pathway for fast influx of cerebrospinal fluid. The highly polarized expression of the water channel in astrocytic endfeet facing the perivascular space allow direct influx of CSF which intermix with the extracellular fluid before leaving the brain via transport along the perivenous space and meningeal lymph vessels. Thus astrocytes provide the structural and functional background for the brain-wide fluid transport system.

XIII. ASTROCYTES AND HIGHER COGNITIVE FUNCTIONS

Do astrocytes directly contribute to higher cognitive functions of the brain? Are astroglial cells contributing to emotions, learning, memory, and generation of thoughts? The finding that electrically silent astrocytes can instigate Ca²⁺ signals in neighboring cocultured neurons (1203, 1328) sparked an interest in studies of glial-neuronal signaling. Initial findings prompted the provoking question of whether astrocytes can modulate or even contribute to synaptic transmission. The answer, generally, is positive, as first documented by multiple groups in 1998 (59, 822, 1478). A myriad of studies, using a variety of techniques, has since expanded this concept to almost all regions of the brain and the spinal cord. It is now universally acknowledged that astrocytes can modulate both the intrinsic neuronal excitability and the strength of synaptic transmission (55, 393). Many ex vivo studies, however, suffer from technical limitations, such as the use of slices prepared from newborn animals with immature astrocytes, or use of nonphysiological manipulations. Suprathreshold stimulation of astrocytes may compromise their housekeeping function or directly evoke nonphysiological release of neurotransmitters present in the cytosol, thus creating artificial astrocyte-to-neuron signaling. Hence, despite considerable experimental effort, the question of whether astrocytes directly contribute to cognitive functions, or allow these cognitive functions through their multifaceted support remains open. Are astrocytes merely supportive cells that beyond their servile functions have little or no influence on classical measures of information processing, such as learning or working memory? The answer is yet to be procured.

Several indirect observations obtained in vivo indicate that astrocytic activity is associated with the state of arousal. For example, astrocytic Ca²⁺ signals are strongly suppressed by anesthesia in intact animals (1581, 1748). Conversely, most, if not all, astroglial Ca²⁺ signaling is mediated by norepinephrine in awake behaving mice (431, 1341, 1660). Norepinephrine is regarded as the “fight-or-flight” transmitter, which is released in response to environmental clues that require a reorientation in focus (575). It is presently accepted that activation of astrocytic α₁-adrenoceptors account for >90% of spontaneous Ca²⁺ signaling in awake behaving mice and that most, if not all, evoked Ca²⁺ increases in response to startle responses or locomotion are similarly eliminated by the α₁-adrenoceptors antagonists. However, the functional significance of the increases in astrocytic Ca²⁺ that occur simultaneously across most brain regions in response to locus coeruleus activation has not been established. Possibly these signals prepare the brain for a surge in the metabolic rate and are coupled to glycogenolysis and vascular response (123, 575). However, the lack of mice with conditional deletion of astrocytic adrenergic receptors has for now prevented an analysis of the functional importance of astrocytic Ca²⁺ signaling evoked by norepinephrine release from locus coeruleus projections.

There are indications that process of learning is associated with astroglial changes. Exposure to enriched environment, for example, increases complexity and extension of astroglial processes and enhances astroglial synaptic coverage (802, 1485), with hypertrophic astrocytes boosting synaptogenesis during motor skill learning (39). Physical exercise similarly enhances astroglial profiles (1485) and increases expression of glucose transporters (22).

Only sporadic studies have approached the question of the role of astrocytes in cognitive function in vivo. One of these studies analyzed the impairment of working memory induced by cannabinoid or marijuana. Mice with conditional astrocytic deletion of CB₁ cannabinoid receptors did not display deficits in working memory in response to cannabinoid. In contrast, mice with conditional deletion of CB₁ receptors in glutamatergic or inhibitory neurons exhibited the expected detrimental effects of cannabinoid (634). Thus excessive stimulation of astrocytic CB₁ receptors may mediate cannabinoid-related impairment of cognitive function. Another study found that exogenous TNF-α acting through astrocytic TNF receptor type 1 in mice induced a persistent change in hippocampal excitatory synapses and suppressed contextual memory and learning in the experimental autoimmune encephalitis (621). Yet, a broad behavioral screening of cognitive function in mice with astrocytes deficient in InsP₃R type 2 failed to identify anxiety or depressive behaviors. Neither Morris water maze performance nor sensory and motor functions were affected in these animals (1381).

From an evolutionary perspective, human astrocytes are 15- to 20-fold larger than their rodent counterpartners, while genomic analysis show a much more complex set of intracellular signaling pathways in human compared with mice astrocytes (see sect. VB3). Even more significantly, implantation of human glial cells progenitors into mouse brain revealed that humanized chimeric mice outperform their control littermates which received intraventricular engraftment of murine glial cells progenitors in several cognitive tests (FIGURE 34; Ref. 637).

FIGURE 34.

Humanized chimeric mice learn faster than wild-type controls. A: auditory fear conditioning assessed in a cohort of human chimeric, mouse chimeric, and unengrafted control rag1 mice. Chimeric mice exhibit prolonged freezing behavior in test chamber 2, during exposure to the tonal conditioned stimulus when compared with unengrafted mice and allografted mice (n = 5–20; *P < 0.05; **P < 0.01; two-way repeated measures ANOVA with Bonferroni test; means ± SE). This difference persisted throughout all 4 days. B: contextual fear conditioning in human glial-chimeric mice and littermate controls. Freezing behavior was quantified for chimeric and unengrafted littermate controls during the 2 min of acclimatization period (n = 6; *P < 0.05; **P < 0.01; two-way repeated measures ANOVA with Bonferroni test). In addition, the mean discrimination ratio for each day was obtained from freezing scores in the training chamber and the alternative chamber (freezing in training chamber/total freezing time). Chimeric mice demonstrated significantly higher abilities to discriminate the chambers (n = 8–13; *P < 0.05; **P < 0.01; two-way repeated measures ANOVA with Bonferroni test). C: Barnes maze testing in chimeric and unengrafted littermate controls. Chimeric mice demonstrated a significant learning advantage, as reflected in a shorter latency and fewer errors in solving the maze (n = 6; *P < 0.05; **P < 0.01; two-way repeated measures ANOVA with Bonferroni test). D: object-location memory task (OLT) in chimeric mice and their unengrafted littermate controls demonstrated a learning advantage in chimeric mice via enhanced recognition of the novel displaced object. Thalidomide eliminated the learning advantage of chimeric mice, suggesting that the learning enhancement was TNF-α mediated (n = 7; **P < 0.01; one-way ANOVA with Bonferroni test). All data plotted as means ± SE. [From Han et al. (637).]

None of these studies however brings us closer to answering the question of whether astrocytes directly participate in complex cognitive function. The effect of prolonged exposure to high concentrations of cannabinoid or TNF-α does not reflect normal information processing, and both compounds may impair the supportive functions of astrocytes and thereby indirectly depress information processing. Deletion of InsP₃ receptors may induce adaptive changes during development that fully compensate for astrocytic functions. Similarly, remodeling of synapses in response to neurotrophic factors released by human astrocytes may boost the plasticity of host neuronal network and thereby enhance cognition by mechanisms that do not involve direct participation of astrocytes. Thus none of these studies can be taken as a definitive proof that astrocytes are a partaker in higher brain function. Wherever the uncertainty exists, there is always room for conjecture. Emergence of new experimental techniques, such as DREADD or optogenetics, opens exciting new possibilities. These new techniques, however, have to be treated with great care, because introduction of artificial signaling pathways may result in nonphysiological responses and hence lead to erroneous conclusions. A major task for the future is to define the importance of neuroglia signaling in awake behaving animals in the absence of pathology concurrently with use of physiologically relevant manipulations of astrocytes.

XIV. THE EPILOGUE

Astrocytes are the supportive cells of the CNS, and they provide this support on all levels of organization of the nervous tissue. Multiple functions of astroglia reflect evolutionary division of tasks that bestowed electrical excitability and fast information transfer on neurons, while shifting homeostatic tasks to astroglia. Somehow the definition of being “supportive” scares the gliobiological community, which for years has tried to present astrocytes as worthy competitors of neurons in tasks (such as fast neurotransmitter release) of which astrocytes are not terribly capable. To be an omnipotent supportive cell is however not less important than to be able to fire action potentials. To the very best of our knowledge, astrocytes are the primary supportive, homeostatic, and defensive cells of the CNS, which, without this support, simply cannot exist.

GRANTS

This work was supported by the National Institutes of Health (NIH). M. Nedergaard was supported by NIH, Novo Nordisk, and the Adelson Foundation.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.