Astrocytes are housekeepers of the central nervous system (CNS) and are important for CNS development, homeostasis and defence. They communicate with neurones and other glial cells through the release of signalling molecules. Astrocytes secrete a wide array of classic neurotransmitters, neuromodulators and hormones, as well as metabolic, trophic and plastic factors, all of which contribute to the gliocrine system. The release of neuroactive substances from astrocytes occurs through several distinct pathways that include diffusion through plasmalemmal channels, translocation by multiple transporters and regulated exocytosis. As in other eukaryotic cells, exocytotic secretion from astrocytes involves divergent secretory organelles (synaptic‐like microvesicles, dense‐core vesicles, lysosomes, exosomes and ectosomes), which differ in size, origin, cargo, membrane composition, dynamics and functions. In this review, we summarize the features and functions of secretory organelles in astrocytes. We focus on the biogenesis and trafficking of secretory organelles and on the regulation of the exocytotic secretory system in the context of healthy and diseased astrocytes.
Astrocytes, secretory cells of the CNS
The concept of astrocytes as secretory cells is almost as old as the discovery of these glial cells. The secretory potential of astrocytes became known only 15 years after Michael von Lenhossék coined the term “astrocyte” (von Lenhossék, 1895). In 1909, Hans Held observed, using the molybdenum haematoxylin stain, granular inclusions in neuroglial processes, which he interpreted as a sign of active secretion (Held, 1909). A year later, Jean Nageotte reported secretory granules in glial cells of the grey matter (i.e. astrocytes) using the Altmann method of fucsin labelling. Nageotte concluded that he was “able to present evidence of a robust and active secretion phenomenon in the protoplasm of these cells” (Nageotte, 1910). These granules, later called gliosomes by Alois Alzheimer (see (Glees, 1955) for historic narration), were often observed, and the hypothesis of astroglial secretion was also entertained by Wilder Penfield (Penfield, 1932). Of note, this early 20th century term should not be confused with the recent use of the name gliosomes for describing glial sub‐cellular re‐sealed particles (Nakamura et al, 1993) containing transmitter‐laden vesicles (Stigliani et al, 2006). Be this as it may, both Nageotte and Penfield regarded astrocytes as true endocrine elements that release their secretions into the blood from their endfeet tightly associated with the brain vasculature. This endocrine role of astroglia has not been experimentally confirmed. However, research carried out in recent years has provided a remarkable body of evidence indicating that astrocytes secrete diverse substances that contribute to the regulation of CNS development and homeostasis, synaptogenesis and cognitive function. In that, astrocytes act as a part of a neuroglial secretory network, which, by analogy with the endocrine system, can be defined as the gliocrine system of the CNS (Vardjan & Zorec, 2015). Other known cellular components of the gliocrine system are microglia and oligodendroglia, which all secrete numerous factors important for trophic support, homeostatic control and defence of the nervous tissue. It is highly likely that NG2 cells could be annexed to the gliocrine system, albeit experimental evidence on this account is lacking at present. Astroglia‐derived secretory substances include (Table 1): (i) classical neurotransmitters, (ii) neurotransmitter precursors, (iii) neuromodulators, (iv) hormones and peptides, (v) eicosanoids, (vi) metabolic substrates, (vii) scavengers of ROS, (viii) growth factors, (ix) various factors that can be defined as “plastic” (e.g. factors that regulate synaptogenesis and synaptic connectivity) and, finally, (x) pathologically relevant molecules such as inflammatory factors. These different molecules are released by astrocytes through several pathways (Fig 1 and see Malarkey & Parpura, 2008 for details) represented by: (i) vesicle‐based exocytosis (e.g. that of d‐serine (Martineau et al, 2013) or glutamate (Montana et al, 2004); (ii) diffusion through plasmalemmal pores/channels (e.g. release of ATP and/or glutamate through anion channels, connexin hemichannels or dilated P2X7 receptors, Cotrina et al, 1998; Suadicani et al, 2006) and (iii) extrusion through plasmalemmal transporters (e.g. the release of GABA via the reversed operation of GAT‐3 transporters, Unichenko et al, 2012). Often, the same molecule can be released through different pathways, which affects the complexity/specificity of its action. The release of these molecules to the extracellular space, along with their subsequent transport by the convective glymphatic system (Thrane et al, 2014), occurs within various brain regions in different time spans and with multiple functional consequences. In this review, we primarily focus on the exocytotic secretory pathway.
Exocytosis: multiple mechanisms
Exocytotic release, engaging various types of membrane‐bound organelles laden with heterogeneous cargo, emerged early in evolution (Vardjan et al, 2010; Spang et al, 2015) and is present in the majority of eukaryotic cells. Fusion of organelles with the plasma membrane is key for intercellular signalling and for targeting various molecules (e.g. receptors or transporters) to the plasmalemma. Exocytosis is regulated by cytosolic free calcium ions and can occur either without stimulation (constitutive secretion) or in response to exogenous stimulation (regulated secretion, Kasai et al, 2012). In the brain, neurones are an exemplary model to study the exocytotic signalling pathway due to the spatially and temporally precise release of neurotransmitters at chemical synapses. Astrocytes are similarly capable of exocytosis, but this process is different in terms of spatial arrangements, kinetics and molecular mechanisms.
Vesicular release is supported by the evolutionary conserved family of SNARE proteins (Sollner et al, 1993). They are further divided into two categories, R‐SNAREs and Q‐SNAREs (Fasshauer et al, 1998; Jahn & Scheller, 2006). The former are associated with the vesicular membrane (also referred to as VAMPs), while the later are either integral plasma membrane proteins (e.g. syntaxins) or proteins associated with the plasma membrane (e.g. SNAP25 in neurones or SNAP23 in astrocytes). In the presence of supra‐threshold cytosolic Ca2+ concentrations, R‐SNARE and Q‐SNARE proteins form the ternary SNARE complex by contributing their SNARE domains (one from each VAMP2 and syntaxin and two from SNAP23/25) to form a 4 α‐helical bundle (SNAREpin). This bundle facilitates the fusion of vesicular and plasma membranes (Sutton et al, 1998; Weber et al, 1998). Kinetics of exocytosis is highly heterogeneous (Table 2). Fusion develops in < 1 ms in fast CNS synapses, whereas in endocrine or in kidney cells exocytosis proceeds over many hundreds of milliseconds or even seconds (Coorssen & Zorec, 2012; Kasai et al, 2012; Neher, 2012). Time course of exocytotic release is determined by several factors. First, it is the sensitivity of secretory apparatus to [Ca2+]i, which is heterogeneous in different cell types. Second, the spatiotemporal progression of local [Ca2+]i signals differs markedly between cells. For instance, in synaptic terminals excitation–secretion coupling is exceedingly fast due to the organisation of Ca2+ nanodomains that reflect a close proximity of the Ca2+ source and exocytotic machinery (Eggermann et al, 2012). Finally, slow regulated exocytosis may also evince a distinct vesicle nanoarchitecture (e.g. arrangement and density of R‐SNAREs, see Fig 2) and the heterogeneity of Q‐SNAREs (Takamori et al, 2006; Singh et al, 2014). Multiple mechanisms controlling exocytosis may coexist within the confinement of a single cell resulting in complex kinetics of secretion (Rupnik et al, 2000).
Diversity of astroglial secretory organelles
Eukaryotic cells produce different types of membranous secretory organelles that are classified as intracellular or extracellular. Intracellular vesicles are represented by transport vesicles, lysosomes and various types of secretory vesicles, whereas extracellular vesicles are ectosomes, exosomes, microvesicles (microparticles), membrane particles and apoptotic vesicles (van der Pol et al, 2012; Cocucci & Meldolesi, 2015). Intracellular vesicles are cellular organelles that may completely fuse with cellular membranes, whereas extracellular vesicles are membranous compartments released into the surrounding environment. Generally, vesicles undergoing constitutive or regulated exocytosis derive either from the trans‐Golgi network or from early or recycling endosomes, although multivesicular bodies and lysosomes have been reported to undergo exocytosis under certain conditions.
Several secretory organelles undergo regulated exocytosis in astrocytes (Fig 3). These include clear electron lucent SLMVs that morphologically resemble synaptic vesicles (Bezzi et al, 2004; Crippa et al, 2006; Jourdain et al, 2007; Bergersen & Gundersen, 2009; Martineau et al, 2013), DCVs (Calegari et al, 1999; Parpura & Zorec, 2010) and secretory lysosomes (Zhang et al, 2007; Li et al, 2008; Verderio et al, 2012). All these organelles can store and release low (amino acids) and/or high (peptides and proteins) molecular weight chemical transmitters (Parpura & Zorec, 2010; Gucek et al, 2012; Vardjan & Zorec, 2015). Secretory vesicles can also act as recycling vesicles that take up extracellular molecules (e.g. by endocytosis) and promote their subsequent release (Vardjan et al, 2014b). This function may be essential for defining the composition of the cerebrospinal fluid and for the function of the glymphatic system (Thrane et al, 2014).
Synaptic‐like microvesicles carry amino acids
Astroglial SLMVs typically have a diameter of 30–100 nm and appear in pairs/groups of 2–15 vesicles (Bezzi et al, 2004; Jourdain et al, 2007; Bergersen et al, 2012; Martineau et al, 2013). They are much less numerous compared to synaptic vesicles in nerve terminals where these organelles exist in groups of hundreds to thousands. Larger SLMVs (diameter of 1–3 μm) have also been identified in astrocytes in hippocampal slices. These vesicles may be generated by intracellular fusion of smaller vesicles and/or other organelles in response to a sustained increase in [Ca2+]i or mechanical stimulation (Kang et al, 2013), but it is not clear whether they contribute to physiological secretion.
Concentrating neurotransmitters into vesicles is accomplished by vesicular neurotransmitter transporters or VNTs, which differ from the transporters at the plasma membrane with respect to energy coupling, substrate specificity and affinity. Six types of VNTs have been identified so far, including transporters for glutamate (VGLUT1‐3), acetylcholine (vAChT), monoamines (VMAT1‐2), GABA/glycine (VIAAT, also named VGAT), and more recently transporters for ATP (VNUT) and, possibly, for aspartate (sialin/VEAT) (Chaudhry et al, 2008; Sawada et al, 2008; Blakely & Edwards, 2012). Accumulation of d‐serine in SLMVs is mediated by vesicular d‐serine transporter, VSerT (Martineau et al, 2013), although its molecular identity remains elusive. The VNTs are essential molecular components of chemical transmission and the fingerprint of regulated exocytosis. Some VNTs such as VGLUT1‐3 have been identified in cultured astrocytes (Fremeau et al, 2002; Bezzi et al, 2004; Kreft et al, 2004; Crippa et al, 2006; Montana et al, 2006). Analyses of astrocytes in situ using gene chip microarray, single‐cell RT–PCR and immunostainings (Bezzi et al, 2004; Li et al, 2013; Sahlender et al, 2014) have produced variable results and in some cases have challenged the presence of VNTs and thus the concept of astroglial exocytosis. Nonetheless, immunogold electron microscopy, confocal microscopy and single‐cell RT–PCR have shown that sub‐populations of astrocytes in the brain express VGLUT1 (Bezzi et al, 2004; Bergersen et al, 2012; Ormel et al, 2012), VGLUT2 (Bezzi et al, 2004) and VGLUT3 (Ormel et al, 2012).
In astrocytes, SLMVs primarily store glutamate and d‐serine, an agonist of glycine regulatory site of NMDA receptor (Martineau et al, 2008, 2013; Bergersen et al, 2012). In cultured astrocytes, SLMVs co‐localise with d‐serine (Mothet et al, 2005; Martineau et al, 2013) and VGLUTs, suggesting that glutamate and d‐serine may reside in the same secretory organelle (Bezzi et al, 2004; Ormel et al, 2012). This contrasts the in situ evidence showing that glutamate and d‐serine are stored in distinct SLMVs within the same astrocyte (Bergersen et al, 2012). Direct comparison of astroglial SLMVs (Crippa et al, 2006; Martineau et al, 2013) and neuronal synaptic vesicles shows that astrocytic vesicles in bulk contain d‐serine and glutamate, whereas neuronal synaptic vesicles in bulk contain glutamate, glycine and GABA but are devoid of d‐serine (Martineau et al, 2013; Sild & Van Horn, 2013). Astrocytes from various brain regions, including the hippocampus and cortex, and Bergmann glial cells in the cerebellum contain SLMVs (Bergersen et al, 2012; Ormel et al, 2012). These vesicles are present in perisynaptic processes as well as in somata (Bezzi et al, 2004; Montana et al, 2004; Ormel et al, 2012). The release of both glutamate and d‐serine from astrocytes is Ca2+‐dependent and is blocked by tetanus toxin that cleaves astrocytic R‐SNAREs VAMP2 and VAMP3 (Bezzi et al, 2004; Mothet et al, 2005; Martineau et al, 2008, 2014; Henneberger et al, 2010; Parpura & Zorec, 2010; Kang et al, 2013; Shigetomi et al, 2013).
Dense‐core vesicles carry peptides
The DCVs are the main component for the storage and release of neuropeptides and hormones from neuroendocrine cells (Burgoyne & Morgan, 2003) and neurones (Klyachko & Jackson, 2002). These vesicles also contain ATP, which is likely accumulated into DCVs via VNUTs, albeit the presence of this transporter on these organelles has not yet been reported. The ultrastructure characteristics of astroglial DCVs are similar to those of neuroendocrine cells and neurones, although their core seems not as dense as in neuroendocrine cells (Potokar et al, 2008). The actual fraction of DCVs in astrocytes is quite small; for example, VAMP2‐positive DCVs represent only 2% of the total number of vesicles examined (i.e. clear and dense‐core vesicles, Crippa et al, 2006). Astroglial DCVs are generally larger than SLMVs, being ~100–600 nm in diameter (Calegari et al, 1999; Hur et al, 2010; Prada et al, 2011), albeit ANP‐storing vesicles can have diameters as small as 50 nm (Potokar et al, 2008). DCVs from cultured astrocytes contain the secretory proteins secretogranins II (Calegari et al, 1999; Paco et al, 2009; Prada et al, 2011) and III (Paco et al, 2010), chromogranins (Hur et al, 2010), ANP (Kreft et al, 2004; Paco et al, 2009), neuropeptide Y (Ramamoorthy & Whim, 2008; Prada et al, 2011) and ATP (Coco et al, 2003; Pangrsic et al, 2007). The DCVs containing secretogranins were also identified in astrocytes from human brain samples (Hur et al, 2010), confirming the existence of DCVs in situ.
Secretory lysosomes in astrocytes
In cultured astrocytes, secretory lysosomes contribute to the storage and Ca2+‐dependent exocytotic release of ATP (Jaiswal et al, 2007; Zhang et al, 2007; Li et al, 2008). Diameters of secretory lysosomes are between 300 and 500 nm, and they coexist with SLMVs within the same astrocyte (Liu et al, 2011) and can be labelled with dextrans (Jaiswal et al, 2002; Vardjan et al, 2012), FM dyes and MANT‐ATP (Zhang et al, 2007). These organelles are seemingly devoid of VGLUTs and VAMP2 (Zhang et al, 2007; Liu et al, 2011), while expressing lysosomal‐specific markers such as cathepsin D, LAMP1 (Zhang et al, 2007; Martineau et al, 2008), ras‐related protein Rab7, and VAMP7 (Chaineau et al, 2009). Secretory lysosomes also express VNUT (Sawada et al, 2008) that is needed for the accumulation of ATP (Oya et al, 2013). Exocytosis of lysosomes in astrocytes relies mainly on tetanus toxin‐insensitive VAMP7, allowing for the release of both ATP and cathepsin B. Downregulation of VAMP7 expression inhibits the fusion of ATP‐storing vesicles and ATP‐mediated intercellular Ca2+ wave propagation (Verderio et al, 2012), a form of long‐range communication in the astroglial network (Cornell‐Bell et al, 1990). Fusion of secretory lysosomes is triggered by slow and locally restricted Ca2+ elevations (Li et al, 2008), which are distinct from Ca2+ spikes that are linked to SLMV fusion (Verderio et al, 2012). Similarly to other cells, secretory lysosomes in astrocytes are likely to play a role in membrane repair (Andrews & Chakrabarti, 2005).
Extracellular vesicles are broadly divided into exosomes and ectosomes. ECVs are typically loaded with a wide spectrum of bioactive substances including cytokines, signalling proteins, mRNA and microRNA (Mause & Weber, 2010). Exosomes are vesicles of 40–100 nm in diameter, produced through the formation of MVBs and their subsequent fusion with the plasma membrane (Mathivanan et al, 2010). Ectosomes, on the other hand, range from 100 to more than 1,000 nm in diameter and are formed and released by shedding off the plasma membrane.
The formation of exosomes follows the typical endocytic route, where transmembrane proteins are endocytosed and trafficked to early endosomes and subsequently to late endosomes. Intraluminal vesicles are generated by neutral sphingomyelinase 2 and ceramide‐dependent process (Trajkovic et al, 2008) that also requires the ESCRT to generate MVBs (van Niel et al, 2006). Fusion of MVBs and release of exosomes involve Rab11, Rab27 and Rab35 (Vanlandingham & Ceresa, 2009; Hsu et al, 2010; Ostrowski et al, 2010; Baietti et al, 2012). During differentiation, MVBs become enriched in lipids such as cholesterol, lysobisphosphatidic acid and sphingomyelins containing ceramide (Kobayashi et al, 1998; Chevallier et al, 2008).
Ectosomes form by direct budding off the plasma membrane (Thery et al, 2009). Similarly to exosomes, ceramide is required for ectosome release (Bianco et al, 2009); ceramide together with ESCRT subunits participate in ectosome assembly and budding. During ectosome shedding from the plasma membrane, ARRDC1 interacts with ESCRT component TSG101 (Nabhan et al, 2012).
Astrocytes release both types of ECVs. Ectosome shedding from astrocytes occurs upon the activation of P2X7 purinoceptors and involves the rapid activation of acid sphingomyelinase that moves to the plasma membrane outer leaflet. Sphingomyelinase alters membrane structure/fluidity leading to vesicle blebbing and shedding (Bianco et al, 2009). Diameters of ectosomes shed by cultured astrocytes from the 2‐day‐old rat cortex vary between 100 and 1,000 nm (Proia et al, 2008; Bianco et al, 2009). Some ECVs are even larger. For example, in cultured human foetal astrocytes spontaneous shedding of large (~4 μm diameter on average) ECVs has been detected. These large ECVs can contain mitochondria and lipid droplets and are decorated with β‐1 integrin, a shedding marker (Falchi et al, 2013). The physiological relevance of this type of ECVs remains to be resolved; it cannot be excluded that they represent apoptotic bodies. Astrocyte‐derived ectosomes carry numerous factors that regulate the activity of neighbouring cells including fibroblast growth factor 2 and vascular endothelial growth factor (Proia et al, 2008), IL‐1β (Bianco et al, 2009), nucleoside triphosphate diphosphohydrolases (Ceruti et al, 2011) and matrix metalloproteinases (Sbai et al, 2010). Ectosomes also contain acid sphingomyelinase and high levels of phosphatidylserine on their membrane outer leaflet (Bianco et al, 2009). Finally, both exosomes and ectosomes contain nucleic acids, mainly microRNAs, small RNA regulators that have essential roles in different biological processes. Exosomes containing microRNAs can be utilized in communication between astrocytes and neurones. For instance, astrocytes treated with morphine and HIV Tat increase the expression of miR‐29b that is released by exosomes; miR29b in turn is taken up by neurones where it downregulates the expression of PDGFB receptors (Hu et al, 2012). Of note, astrocyte‐derived exosomes have been reported to contain mitochondrial DNA (Guescini et al, 2010). Whether ECVs also carry neurotransmitters is yet to be elucidated.
Exosomes are released from astrocytes in response to oxidative and heat stress (Taylor et al, 2007) and also in pathological conditions. Secretion of exosomes containing PAR4 and ceramide is increased in astrocytes surrounding amyloid plaques in a mouse model of familial Alzheimer's disease. These PAR4‐ and ceramide‐enriched exosomes are subsequently taken up by astrocytes and induce apoptosis even in the absence of β‐amyloid (Wang et al, 2012). Given the role of PAR4‐ and ceramide‐containing exosomes in apoptotic processes, they are defined as “apoxosomes”. Whether exosomes are discharged by astrocytes through a physiologically regulated process and whether exosomal release in vivo has a physiological function remains unclear. Owing to the lack of methods to specifically block exosome secretion without affecting secretion of other membrane vesicles, the resolution to these issues cannot be reached at the time being.
Molecular machinery of astroglial exocytosis
Twenty years ago, it was demonstrated that SNAREs are present in cultured astrocytes (Parpura et al, 1995). Subsequent studies have revealed that astrocytes express proteins characteristic for neuronal synaptic vesicles such as VAMP2 and proteins that are found in exocytotic trafficking vesicles of non‐neuronal cells such as SCAMP and VAMP3 (Parpura et al, 1995; Maienschein et al, 1999; Wilhelm et al, 2004; Mothet et al, 2005; Crippa et al, 2006; Montana et al, 2006; Martineau et al, 2008). The lysosome‐associated TI‐VAMP/VAMP7 is expressed in astrocytes (Zhang et al, 2007; Martineau et al, 2008; Verderio et al, 2012) along with components of secretory machinery, Q‐SNARE proteins SNAP23, syntaxins 1, 2, 3 and 4 (Hepp et al, 1999; Zhang et al, 2004b; Paco et al, 2009), SNARE‐associated proteins such as synaptotagmin 4 (Zhang et al, 2004a) and isoforms of Munc18 (Paco et al, 2009). Cleavage of SNARE proteins with tetanus or botulinum neurotoxins (Verderio et al, 1999) reduce glutamate and d‐serine and, to a lesser extent, ATP release in cultured astrocytes (Coco et al, 2003). Similarly, the treatment with tetanus toxin suppressed exocytosis measured by monitoring amperometric spikes (Chen et al, 2005) or by recording membrane capacitance (Kreft et al, 2004). The residual, toxin‐insensitive component of Ca2+‐evoked exocytosis could be due to the contribution by other secretory organelles, such as lysosomes carrying toxin‐insensitive VAMP7 (Verderio et al, 2012).
Several SNARE proteins, including VAMP2 (Wilhelm et al, 2004), VAMP3 (Bezzi et al, 2004; Zhang et al, 2004a; Jourdain et al, 2007; Bergersen & Gundersen, 2009; Schubert et al, 2011), VAMP7 (Verderio et al, 2012), SNAP23 and syntaxin 1 (Schubert et al, 2011), have been detected in astrocytes in situ. VAMP2 and 3 co‐localise with VGLUT1 and 2 on SLMVs that store glutamate (Bezzi et al, 2004; Zhang et al, 2004a; Jourdain et al, 2007; Bergersen & Gundersen, 2009) and likely d‐serine (Martineau et al, 2013). Several synaptotagmin isoforms including synaptotagmins 4, 5, 7 and 11 are also present in astrocytes (Zhang et al, 2004a; Mittelsteadt et al, 2009). However, typical neuronal SNARE‐associated proteins, such as synaptotagmins 1 and 2, and synaptophysin (Wilhelm et al, 2004) have not been observed in astrocytes in situ.
Inactivation of VAMP2 and/or VAMP3 by tetanus neurotoxin abolished the release of glutamate (Jourdain et al, 2007; Perea & Araque, 2007) and, likely, d‐serine in astrocytes in brain slices (Henneberger et al, 2010). A transgenic mouse model expressing a dominant negative (dn) SNARE (i.e. the cytosolic tail of VAMP2) in astrocytes (Pascual et al, 2005; Halassa et al, 2009) showed changes in behaviour, synaptic transmission and maturation of neurones (Pascual et al, 2005; Hines & Haydon, 2013; Nadjar et al, 2013; Turner et al, 2013; Lalo et al, 2014; Sultan et al, 2015), suggesting a role for astrocytic VAMP2‐dependent exocytosis in vivo. Of note, VAMP2 cytosolic tail is supposed to compete with VAMP2 for binding to other components forming the ternary complexes, leading to the reduced number of complexes formed and hence inhibiting regulated exocytosis. Although these experiments provide strong support for a function of SNARE proteins in astroglial regulated exocytosis, there are indications that neurones might also express dnSNARE in the transgenic mice, thus raising the possibility that the impairment of neuronal, rather than astroglial, exocytosis may account for the phenotype observed (Fujita et al, 2014). The debate that ensued (Sloan & Barres, 2014) highlights technical matters, and particular aspects of astroglial glutamate secretion in the context of synaptic transmission, without questioning the general concept of exocytosis‐mediated astroglial secretion. These technical dissensions nonetheless emphasize the need for refining the existing experimental strategies and developing new approaches directly attacking the various facets of astroglial secretion in physiological and pathophysiological contexts (Jahn et al, 2015).
Astroglial exocytosis is slow
Visualisation of fluorescently labelled VGLUT1/2‐containing vesicles revealed that fusion events in isolated astrocytes occur within hundreds of milliseconds after the increase in cytosolic Ca2+ (Bezzi et al, 2004; Cali et al, 2008; Marchaland et al, 2008; Santello et al, 2011). Even slower kinetics of vesicular fusions has been reported by using synapto‐pHluorin (spH), a fluorescently tagged‐VAMP2, (Bowser & Khakh, 2007). Treatment of astrocytes with the Ca2+ ionophore ionomycin triggered exocytotic fusion of spH‐labelled SLMVs within seconds (Liu et al, 2011). Similarly, the TIRF microscopy (Malarkey & Parpura, 2011) showed slow exocytotic bursts occurring within seconds after mechanical stimulation of astrocytes. Secretion of NPY from peptidergic vesicles occurred with a > 1‐min delay after stimulation (Ramamoorthy & Whim, 2008; Prada et al, 2011). Exocytotic release from peptidergic vesicles in 8‐Br‐cAMP‐matured astrocytes also began minutes after the stimulation (Paco et al, 2009). Similar observations have been made for secretory lysosomes, which labelled with FM dyes fused with the plasma membrane with an ~1‐min delay after exposure of astrocytes to Ca2+ ionophores or ATP (Zhang et al, 2007; Li et al, 2008). Exocytotic fusion of quinacrine‐loaded vesicles that express lysosomal VAMP7 occurred with a > 2‐min delay after exposure to various stimuli including ionomycin, glutamate, ATP or UV‐induced Ca2+ uncaging (Kreft et al, 2004; Pangrsic et al, 2007; Pryazhnikov & Khiroug, 2008). Likewise, EGFP‐LAMP1‐ and FITC‐dextran‐labelled lysosomes underwent exocytotic fusion with a > 40‐s delay after administration of ionomycin (Liu et al, 2011) or the group I mGluR agonist DHPG (Jaiswal et al, 2007).
Taken together, these imaging data indicate that in contrast to neurones, where the fusion occurs within < 0.5 ms after the Ca2+ entry into the cytosol (Neher, 2012; Sudhof, 2012), exocytotic release of various molecules from astrocytes is a much slower process, occurring with a substantial post‐stimulus delay (Vardjan et al, 2015). Indeed, capacitance measurements on isolated astrocytes confirm that the kinetics of vesicle fusion is at least 2 orders of magnitude slower than in neurones (Fig 4 and Table 2; Kreft et al, 2003, 2004). Incidentally, inhibition of astroglial exocytosis (using astroglia targeted expression of dnSNARE or pharmacological tools) affects only slow electrical oscillations in the cortex (Fellin et al, 2009), while fast neuronal electrical activity seems to be unaffected by corrupted (using a mouse model rendering a reduction of VAMP1‐3 expression in Müller cells, a specialized astroglia of the retina) gliotransmission (Slezak et al, 2012).
The somewhat lethargic kinetics of astroglial vesicular release likely reflects distinct organisation of the exocytotic machinery. First, electron microscopy studies (Bezzi et al, 2004; Jourdain et al, 2007; Bergersen et al, 2012) have shown that astrocytes lack structurally organised vesicle clusters typical of the active zone present in presynaptic terminals, which may make the stimulus–secretion coupling looser. In neurones, SNARE proteins are associated with vesicles clustered at active zones that are essentially release sites. This spatial localisation arguably is linked to the minimisation of the delay between the stimulus and the secretory output (Kasai et al, 2012). Second, the SNARE components and SNARE‐associated proteins of the exocytotic apparatus are not identical in astrocytes and neurones, neither is the stability of SNARE complexes, nor are the numbers of SNARE molecules associated with a single vesicle.
VAMP isoforms with similar structural properties can participate in the formation of several different SNARE complexes (Wilhelm et al, 2004; Montana et al, 2009), which may affect the mechanism of vesicle fusion with the plasma membrane. In neuronal terminals, the ternary fusion complex forms between VAMP2, SNAP25 and syntaxin, whereas in astrocytes the ternary SNARE fusion complex assembles from VAMP2/3 or TI‐VAMP, SNAP23 and syntaxin (Montana et al, 2009; Hamilton & Attwell, 2010). At a single molecule level, the presence of SNAP23A (as opposed to SNAP25B) in the ternary complex decreases the complex stability by half, arguably retarding the tethering/docking/fusion process (Fig 4B). Moreover, the density of R‐SNAREs associated with a single vesicle in astrocytes is lesser than in neurones; in the latter, a single synaptic vesicle contains ~70 VAMP2 molecules (Takamori et al, 2006) vs. ~25 in a single astroglial vesicle (Singh et al, 2014). This paucity of VAMP2 would lead to the reduced density of ternary SNARE complexes, which would contribute to further retardation of docking and fusion process in astrocytes.
Trafficking of astroglial vesicles
Quantitative measurements of secretory vesicles mobility in astrocytes (Potokar et al, 2005, 2013b) revealed two types of vesicle mobility: (i) directional, when vesicles travel along tracks, such as cytoskeletal elements, including intermediate filaments, or (ii) non‐directional, characterised by contorted vesicle trajectories, typical for the Brownian movement of particles. These experiments also unveiled a dichotomy of vesicular traffic: glutamatergic vesicles accelerate with an increase of cytosolic Ca2+ (Stenovec et al, 2007), whereas peptidergic vesicles and endolysosomes slow down (Potokar et al, 2008, 2010). Such stimulation‐dependent vesicle mobility regulation has not been observed in neurones and may represent an adaptive mechanism for astrocytes to redistribute vesicles to the correct location.
Astroglial exocytosis in physiology and pathophysiology
Endo/exocytosis of BDNF
BDNF is a powerful regulator of neuronal plasticity (Poo, 2001). Its synthesis, which occurs in both neurones and astrocytes (Lu et al, 2005; Juric et al, 2008), yields two distinct forms: pro‐BDNF (which binds to and acts through the pan‐neurotrophin receptor p75) and mature BDNF that stimulates TrkB receptor. Neurones often release pro‐BDNF, which undergoes maturation either extracellularly (by tissue plasminogen activator/plasmin) or in astroglia. The latter pathway has been demonstrated in hippocampal slices and involves endocytotic uptake of pro‐BDNF by astrocytes, in response to a strong electrical stimulation of neurones, conversion of pro‐BDNF into the mature form in astrocytes and subsequent VAMP2‐mediated exocytotic release of mature BDNF from these glial cells (Bergami et al, 2008).
Secretion of peptides
Astrocytes also synthesize and secrete NPY, a peptide widely distributed throughout the mammalian nervous system (Barnea et al, 1998, 2001), where it acts as a neuroproliferative factor (Hansel et al, 2001; Geloso et al, 2015) and regulates the growth of vascular tissue (Zukowska‐Grojec et al, 1993). Release of NPY is activated by an mGluR‐linked increase in cytosolic Ca2+ and proceeds through exocytotic fusion of DCVs (Ramamoorthy & Whim, 2008). Several types of natriuretic peptides, including ANP, brain natriuretic peptide and C type natriuretic peptide (CNP), are present in the CNS (Potter et al, 2006). ANP in particular is present in neurones and astrocytes in various brain regions (McKenzie et al, 2001). Natriuretic peptides exert their actions by 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 (Lucas et al, 2000). NPR‐A and NPR‐B are plasmalemma‐bound guanylyl cyclase receptors, which mediate intracellular signalling by increasing intracellular cGMP. NPR‐C is a “clearance receptor” that removes peptides from the extracellular space, but does not itself possess guanylyl cyclase activity (Potter et al, 2006; Rose & Giles, 2008). In astrocytes, ANP is stored in vesicles and released into the extracellular space by regulated exocytosis (Krzan et al, 2003). The astroglial ANP content significantly increases after experimental brain infarction (Nogami et al, 2001), suggesting that this gliosignalling molecule may regulate the cerebral blood flow. ANP is also involved in the control of systemic salt intake as the loss of ANP receptors eliminates the inhibition of salt‐seeking behaviour caused by a NaCl load (Blackburn et al, 1995).
Delivery of receptors, channels and transporters to the plasma membrane
Ionotropic glutamate receptors.
The VAMP2‐positive vesicles of cultured astrocytes are immunopositive for the AMPA receptor subunits GluA2,3 and, to a lesser extent, for GluA1 (Crippa et al, 2006). This presence of GluA2,3 subunits on VAMP2‐positive vesicles suggests a vesicle‐mediated mode of AMPA receptor delivery to the astrocytic plasma membrane, as previously described in neurones (Passafaro et al, 2001).
Glutamate transporter EAAT2.
Astrocytes play a key role in the uptake of glutamate released during synaptic transmission (Danbolt, 2001). Glutamate clearance is a function of Na+‐dependent excitatory amino acid transporters EAAT1 and EAAT2, which are predominantly expressed in astroglia (Zhou & Danbolt, 2013). The efficacy of the clearance directly depends on the density of transporters in the plasma membrane (Robinson, 2002; Huang & Bergles, 2004). The density of EAAT2 in astrocyte plasmalemma is controlled by exo‐/endocytosis (Stenovec et al, 2008; Li et al, 2015). Aberrant trafficking of EAAT2‐containing vesicles to the plasma membrane may compromise glutamate uptake and contribute to neuronal excitotoxicity such as seen in amyotrophic lateral sclerosis (Rossi, 2015).
G protein‐coupled receptors (GPCRs).
Astrocytes express several types of GPCRs such as cannabinoid receptor 1, CBR1 (Navarrete & Araque, 2008), chemokine receptor CXCR4 (Bezzi et al, 2001), mGluR5 (Kirischuk et al, 1999) and P2Y1 purinoceptors (Domercq et al, 2006). The delivery of GPCRs to the plasmalemma may involve vesicular transport. CBR1 is mainly expressed in acidic intracellular organelles that co‐localise with endocytic compartments. While trafficking of CBR1 has been studied using CBR1 fluorescent proteins chimeras, the mechanism by which it reaches the surface of astrocytes, whether by a constitutive recycling pathway or by a Ca2+‐dependent mechanism such as exocytosis, remains to be determined (Osborne et al, 2009).
Astrocytes express aquaporin 4 (AQP4), a channel that is critical for brain water homeostasis (Nagelhus & Ottersen, 2013). Distribution of AQP4 in astrocytes is highly polarised being mainly confined to endfeet, and to a lesser extent, to perisynaptic processes (Nielsen et al, 1997; Nagelhus et al, 1998; Arcienega et al, 2010). Water transport mediated by AQP4 contributes to pathology and is important for astrocyte swelling and brain oedema formation/resolution in vitro (Yamamoto et al, 2001; Arima et al, 2003) and in vivo (Ke et al, 2001; Papadopoulos et al, 2004). Water transport through the cell membrane is regulated by the permeability properties of AQP4 (Gunnarson et al, 2008; Nicchia et al, 2011), the heterogeneity of AQP4 crystalline‐like orthogonal arrays of particles (Hirt et al, 2011) and, as recently suggested, by trafficking of AQP4‐containing vesicles to/from the plasma membrane (Potokar et al, 2013a). In unstimulated conditions, the mobility of vesicles containing AQP4 resembles the mobility of slow recycling and endosomal vesicles. This mobility of AQP4e isoform‐laden vesicles correlated with changes in the AQP4 presence at the plasma membrane. Hypoosmotic stimulation, which induces astrocyte swelling, triggered a transient reduction in AQP4e isoform vesicle mobility mirrored by the transient increase in the AQP4 plasma membrane expression. These data indicate that the regulation of vesicle mobility is an important mechanism to alter the delivery/retraction ratio of AQP4 vesicles to/from the astroglial plasma membrane.
Control of sleep homeostat
An increase of adenosine levels in the extracellular space promotes sleepiness, while; adenosine receptor antagonists promote wake‐fulness (Basheer et al, 2004). It turns out that astrocytes are key for controlling adenosine levels and they do so via SNARE‐dependent release of ATP that is converted to adenosine extracellularly by ecto‐nucleotidases (Pascual et al, 2005). Thus, astroglial sourced adenosine is essential for the regulation of sleep homeostat and for responses to sleep deprivation (Halassa et al, 2009).
Astrocytes also contribute to the extracellular level of glutamate in the nucleus accumbens core (NAcore) by a SNARE‐dependent process. At behavioural level, cue‐induced reinstatement of cocaine seeking in rats extinguished from cocaine was inhibited by glutamate release from astrocytes, which action was mediated via the group II mGluRs (Scofield et al, 2015). Of note, stimulation of inhibitory presynaptic mGluR2/3 receptors reduces synaptic glutamate release in the NAcore, preventing drug seeking. Cocaine addiction is also characterised by impaired NMDA receptor‐dependent synaptic plasticity in the NAcore. It has been shown that cocaine‐induced deficits in NMDAR‐dependent long‐term potentiation and depression result partially from reduced release of d‐serine from astrocytes (Curcio et al, 2013). Administration of d‐serine directly into the NAcore in vivo blocked behavioural sensitisation to cocaine. Accordingly, d‐serine and glutamate could team up to regulate the cocaine sensitisation state.
Glial exocytosis in neuroinflammation: secretion of complement proteins
The complement system represents one of the most fundamental immune regulating cascade, defining various aspects of tissue defence (Holers, 2014). Complement proteins C3a and C1q are present in the CNS, where they regulate neurogenesis, neuronal survival and synaptic elimination (Stevens et al, 2007; Shinjyo et al, 2009). The C3a complement protein is produced and secreted from astroglia; this secretion is disrupted by brefeldin A, which interferes with anterograde transport from the endoplasmic reticulum to the Golgi apparatus, thus indicating the specific role for the secretory pathway (Lafon‐Cazal et al, 2003). NF‐κB signalling promotes secretion of C3a and excessive NF‐κB activation may increase astroglial C3a release that in turn can contribute to neurodegeneration (Lian et al, 2015).
Glial exocytosis in neuroinflammation: antigen presentation
In neuropathology, astrocytes often become reactive, which leads to their morphological and biochemical remodelling; the reactivity is manifested by an increased expression of intermediate filaments (most notably GFAP and vimentin) (Burda & Sofroniew, 2014; Pekny et al, 2014; Sofroniew, 2015). Reactive reprogramming of astrocytes also affects vesicle delivery. Exposure of otherwise immunologically silent astrocytes to interferon‐γ, a proinflammatory cytokine, initiates expression of MHC‐II molecules and surface antigens causing astroglial cells to behave like nonprofessional antigen‐presenting cells (Vardjan et al, 2012). It has been suggested that IFN‐γ‐activated astrocytes participate in antigen presentation and activation of CD4 helper T cells in immune‐mediated disorders of the CNS including multiple sclerosis (Fontana et al, 1984; Soos et al, 1998) and experimental autoimmune encephalomyelitis (Shrikant & Benveniste, 1996).
The delivery of MHC‐II molecules to the cell surface of antigen‐presenting cells is mediated via a cytoskeletal network and requires the fusion of MHC‐II‐carrying late endolysosomes with the plasma membrane. Actin microfilaments (Barois et al, 1998), microtubules (Wubbolts et al, 1999; Vyas et al, 2007) and their motor proteins (Wubbolts et al, 1999; Vascotto et al, 2007) mediate trafficking of MHC‐II compartments in antigen‐presenting cells. Recently, the role of intermediate filaments (GFAP and vimentin) in MHC‐II trafficking was investigated in IFN‐γ‐activated astrocytes (Vardjan et al, 2012). In IFN‐γ‐activated astrocytes, upregulation of intermediate filaments allows for a faster and therefore more efficient delivery of MHC‐II molecules to the cell surface (Vardjan et al, 2012). Reduced mobility of late endolysosomes due to an increase in [Ca2+]i may increase their probability of docking and fusion to the plasmalemma (Potokar et al, 2010), which, in astrocytes acting as antigen‐presenting cells, may provide an additional regulatory mechanism that controls the delivery of MHC‐II molecules to the cell surface (Vardjan et al, 2012). Besides IFN‐γ, endogenous suppressors, including norepinephrine, regulate the expression of MHC‐II molecules in astrocytes (Frohman et al, 1988; De Keyser et al, 2004). The effects of norepinephrine are mediated through the activation of G protein‐coupled β‐adrenergic receptors on astrocytes and the activation of the cAMP signalling pathway (Vardjan et al, 2014b). However, it is unclear how this pathway controls the vesicular delivery of MHC‐II molecules to the plasma membrane. These regulatory mechanisms may enable antigen‐presenting reactive astrocytes to respond rapidly and in a controlled manner during CNS inflammation. Incidentally, cultured astrocytes expressing mutated (M164V) presenilin 1 have impaired vesicular trafficking, which may be related to compromised defensive capabilities of astrocytes in the neurodegeneration context (Stenovec et al, 2016).
Glial exocytosis in neuroinflammation: release of cytokines with ECVs
Human astrocytes express a large number of cytokines (Choi et al, 2014). The mechanisms by which astrocytes secrete these cytokines are still to be defined. However, the release of pro‐inflammatory cytokines, and in particular IL‐1β, has been extensively characterised in microglia. Microglia extracellular vesicles express IL‐1β, IL‐6, inducible nitric oxide synthase and cyclooxygenase‐2 (Bianco et al, 2009; Verderio et al, 2012). Microglial ectosomes contain the cytokine IL‐1β (Bianco et al, 2005, 2009). Pro‐IL‐1β is incorporated into ectosomes together with pro‐caspase‐1, the enzyme responsible for IL‐1β maturation, P2X7 receptor (Bianco et al, 2005), and likely with other inflammosome components, as described in monocytes (Qu et al, 2007; Sarkar et al, 2009). As a consequence of the assembly of this multiprotein complex, mature IL‐1 β (as well as IL‐18) is released from ectosomes upon ATP stimulation. It is possible that proinflammatory cytokines from astrocytes follows a similar route, employing extracellular vesicles.
Astrocytes express a complex exocytotic machinery that is associated with several types of secretory vesicles involved in the secretion of a wide variety of neurotransmitters, neurotransmitter precursors, hormones, trophic and plastic factors, etc. Astroglial secretion contributes to the intrinsic CNS gliocrine network that provides for the regulation of multiple physiological and pathophysiological processes. Likely owing to the difference in secretory machinery, astroglial exocytosis is much slower that the neuronal counterpart. This fundamental difference reflects distinct physiological specialisation of astroglia as a key homeostatic component of the neural network.
Conflict of interest
The authors declare that they have no conflict of interest.
Authors' research was supported by the Alzheimer's Research Trust (UK) to AV and by the grant (agreement from August 27 2013 no. 02.B.49.21.0003) between The Ministry of Education and Science of the Russian Federation and Lobachevsky State University of Nizhny Novgorod and by the grant of the Russian Scientific Foundation no. 14‐15‐00633; M.M was supported by Italian Ministry of Health GR‐2011‐02347377, Cariplo 2014‐0655, Progetto CNR Invecchiamento‐ CUP B44G13000080005; V.P.'s work is supported by the National Institutes of Health (HD078678). J.P.M.'s work is supported by the CNRS, the University Aix Marseille, and by the Fondation pour la Recherche Medicale. R.Z.'s work is supported by The Research Agency of Slovenia grants no. P3 310, J3 3236, J3 4051, J3‐4146, J3 6790 and J3 6789.
FundingAlzheimer's Research Trust (UK)02.B.49.21.0003
See the Glossary for abbreviations used in this article.
- Secretory vesicles
- synaptic‐like microvesicles
- dense‐core vesicles
- extracellular vesicles
- multivesicular bodies
- Proteins mediating exocytosis
- the soluble N‐ethyl maleimide‐sensitive fusion protein attachment protein receptor. SNAREs are further sub‐classified into R‐SNAREs and Q‐SNAREs. R‐SNAREs are proteins contributing arginine (R) to the ionic layer of the ternary SNARE complex, whereas Q‐SNAREs contribute glutamine (Q)
- vesicle‐associated membrane protein Astrocytes express VAMP2, also known as synaptobrevin 2, VAMP3, also called cellubrevin, and tetanus toxin‐insensitive VAMP (TI‐VAMP), molecularly defined as VAMP7
- synaptosome‐associated protein of 23 kDa
- synaptosome‐associated protein of 25 kDa
- secretory carrier membrane protein
- Vesicular neurotransmitter transporters
- vesicular neurotransmitter transporter
- vesicular glutamate transporters, which belong to the SLC17 solute carrier family. All three known types, VGLUT1 (SLC17A7), VGLUT2 (SLC17A6) and VGLUT3 (SLC17A8), are expressed in astrocytes
- VAChT (SLC18A3)
- vesicular acetylcholine transporter
- VMAT1 and 2 (SLC18A1 and SLC18A2, respectively)
- vesicular monoamine transporters 1 and 2. VMAT 1 is also known as chromaffin granule amine transporter (CGAT)
- VGAT (SLC32A1)
- vesicular GABA transporter, also known as vesicular inhibitory amino acid transporter (VIAAT)
- VNUT (SLC17A9)
- vesicular nucleotide transporter
- VEAT (SLC17A5)
- vesicular excitatory amino acid transporter, also known as sialin
- Other abbreviations
- 8‐bromo‐adenosine 3′,5′‐cyclic monophosphate, a membrane‐permeable form of cAMP
- α‐amino‐3‐hydroxy‐5‐methyl‐4‐isoxazolepropionic acid
- atrial natriuretic peptide
- arrestin domain‐containing protein 1, which interacts with the ESCRT
- aquaporin 4
- brain‐derived neurotrophic factor
- brefeldin A
- central nervous system
- (R/S)‐3,5‐dihydroxyphenylglycine, an antagonist of mGluRs
- excitatory amino acid transporter
- enhanced green fluorescent protein
- endosomal sorting complexes required for transport
- fluorescein isothiocyanate
- FM dyes
- lipophilic styryl compounds used for studying vesicular recycling at the plasma membrane. Initially, they were synthesised by Fei Mao, hence FM
- glial fibrillary acidic protein
- glutamate receptors, ionotropic AMPA type
- G protein‐coupled receptor
- HIV Tat
- human immunodeficiency virus trans‐activating proteins
- IL‐1β, IL‐6, IL‐18
- interleukin‐1β, interleukin‐6, interleukin‐18, respectively
- lysosome‐associated membrane glycoprotein 1, a lysosomal marker
- (2′‐(or‐3′)‐O‐(N‐methylanthraniloyl) adenosine 5′‐triphosphate, a fluorescent analogue of ATP
- major histocompatibility complex molecule class II
- metabotropic glutamate receptors
- nuclear factor kappa‐light‐chain‐enhancer of activated B cells
- neuropeptide Y
- natriuretic peptide receptor
- protease‐activated receptor 4
- platelet‐derived growth factor subunit B
- Rab proteins
- a family of proteins, which are numerically denoted (e.g. Rab7, Rab11, Rab27 and Rab35) as members of the Ras superfamily of monomeric G proteins
- reactive oxygen species
- solute carrier
- total internal reflection fluorescence
- TrkB receptor
- tropomyosin‐related kinase receptor
- tumour susceptibility gene 101, a component of ESCRT
- © 2016 The Authors