Synaptic plasticity has many manifestations. The typical types of synaptic plasticity include LTP and LTD. Studies have shown that astrocytes can regulate synaptic transmission and participate in the regulation of excitatory synaptic structure and function by releasing glial transmitters such as glutamate and D-ser in response to Ca²⁺ influx (Araque et al., 2014).

ECM molecules derived from astrocytes are also important factors in regulating synaptic function. In the central nervous system, the most prominent extracellular matrix structure is the peripheral neural network (PNN) (Lau et al., 2013). The PNN has the functions of producing polyanions that buffer the local environment, as well as capturing and locally concentrating growth factors and nutrients. Additionally, it can act as a diffusion barrier for extracellular signal molecules, including neurotransmitters in synapses, which helps prevent neurotransmitters from overflowing. Accordingly the PNN plays a role in stabilizing synapses, synaptic plasticity, and neuronal surface isolation (Suttkus et al., 2016; Sonntag et al., 2018).

Structurally, the PNN maintains the stability of chondroitin sulfate proteoglycans (CSPGs) binding to the hyaluronic acid (HA) skeleton by connecting proteins (Crtl1/Hapln1 and Bral2/Hapln4). In terms of composition, HA is the main component of the PNN, and CSPGs are its main proteoglycans. In the central nervous system, many CSGPs with different core proteins are secreted by astrocytes, such as brevican, versican and so on (Bandtlow and Zimmermann, 2000; John et al., 2006).

In EMC, the PNN is highly dynamic. After maturation, the PNN will establish a microenvironment between neurons and synapses, which will affect synaptic plasticity and determine where synapses form. The loosening of the PNN can destroy the stability of the inhibitory environment and release growth factors to promote synaptic plasticity (Dzyubenko et al., 2016).

When chondroitinase is injected into the central nervous system, it damages the PNN-like structure by destroying the chondroitin sulfate chain and hyaluronic acid (HA), which in turn impairs the transmission of excitatory synapses (Pyka et al., 2011). Therefore, PNN is a powerful factor in the effect of astrocytes on synaptic plasticity.

Synapses are also affected by the direct adhesion between neurons and astrocytes. One factor mediating astrocyte-neuron adhesion is γ-Pcdh, and this astrocyte protein can promote the development of excitatory and inhibitory synapses (Garrett and Weiner, 2009; Pyka et al., 2011).

Studies have shown that the activation of protease-activated receptor 1 (PAR1) in hippocampal CA1 astrocytes leads to the opening of the Ca²⁺-dependent glutamate-permeable anion channel Best1, which mediates Ca²⁺-dependent glutamate release from astrocytes, thereby increasing the concentration of glutamate in the synaptic space (Woo et al., 2012). Synaptic NMDAR is the main target of Best1-mediated glutamate release in astrocytes. Increased activation of synaptic NMDAR leads to the enhancement of NMDAR-dependent synaptic transmission. When synaptic glutamate increases glutamate secretion in astrocytes mediated by Best1, the threshold of NMDAR-dependent LTP decreases (Park et al., 2015).

Astrocytes can secrete a variety of cytokines and neurotransmitters that participate in the regulation of synaptic plasticity. Hevin is a synaptic protein secreted by astrocytes. In the developing cortex, Hevin is necessary for maintaining synaptogenesis and maturation of the dendritic structure in the thalamic cortex (Risher et al., 2014). The transsynaptic adhesion between presynaptic neurexins (NRX) and postsynaptic neuroligins (NL) is essential for the formation and maturation of excitatory and inhibitory synapses (Baudouin and Scheiffele, 2010). Hevin can bridge Neuresin-1α and NeuroLigin-1B to assemble glutamatergic synapses. Thus, astrocytes change the cross-synaptic interaction between NLS and NRXs via Hevin, thereby regulating the formation and plasticity of excitatory synapses (Singh et al., 2016). Other synaptic proteins such as thrombospondin (TSP) play an important role in the formation of excitatory synapses (Chung et al., 2015; Baldwin and Eroglu, 2017). During development, the nervous system undergoes a process of forming a large number of neural circuits, during which the expression of TSP-1 and TSP-2 in astrocytes is maintained at a high level. Conversely, a lack of TSP will lead to a decrease in the number of excitatory synapses (Christopher et al., 2018). Glypicans can also increase the formation of excitatory synapses. Glypican4 and Glypican6 secreted by astrocytes can recruit AMPA glutamate receptors to the surface of postsynaptic cells and induce the formation of active excitatory synapses by increasing the density of receptors on the surface of postsynaptic cells (Allen et al., 2012; Baldwin and Eroglu, 2017; Farhy-Tselnicker et al., 2017). Transforming growth factor β1 (TGF-β1) is also secreted by astrocytes. As a member of the TGF-β superfamily, TGF-β1 has been recognized as a neuroprotective and neurotrophic factor in previous studies. Recent studies on the role of TGF-β1 in synaptic transmission have found that it is expressed in hippocampal neurons in an activity-dependent manner and participates in synaptic regulation (Caraci et al., 2015, 2018). In vivo, the levels of AMPA and NMDA receptor subunits increase in the hippocampi of mice with TGF-β1 overexpression, and the induction of excitatory synapses by TGF-β1 is mediated by NMDA receptor activity and the NMDA coactivator D-serine (Bae et al., 2011).

In addition, TGF-β1 can also induce a switch from the early stage of LTP, which is independent of protein synthesis (E-LTP) into the late stage that is dependent on protein synthesis (L-LTP), which indicates that TGF-β1 can increase neuronal excitability by inducing long-term dissimilation of ganglia (Caraci et al., 2015).

Brain-derived neurotrophic factor (BDNF) also plays an important role in regulating synaptic plasticity. BDNF works through two receptor systems, TrkB (tropomyosin-related kinase B) and P75NTR (Baldwin and Eroglu, 2017; Holt et al., 2019). Interestingly, the interaction of BDNF through p75NTR and Trk receptors often has opposite effects. For example, Trk receptor attachment almost always promotes neuronal survival and differentiation (Patel et al., 2000), whereas p75NTR involvement often promotes apoptosis (Defreitas et al., 2001). Similarly, in the case of synaptic plasticity, BDNF-mediated activation of TrkB promotes hippocampal LTP, whereas the interaction of neurotrophic factor with p75NTR instead promotes hippocampal LTD. A major pathway involved in p75NTR activation by neurotrophic factors is the Jun kinase signaling cascade, which leads to the activation of p53 and apoptosis. In addition, the Jun kinase cascade induces the expression of Fas ligands in neuronal cells, which promotes apoptosis by activating Fas receptors. There are important intermediates in the Jun kinase cascade pathway, such as NRIF and the E3 ubiquitin ligase Traf6. When these intermediates are absent, Jun kinase signaling in sympathetic neurons will also be blocked (Yeiser, 2004; Linggi et al., 2005). In PC12 cells and sympathetic neurons, the activation of the Jun kinase cascade involves Cdc42, and apoptosis is strongly inhibited by dominant-negative Cdc42 mutations (Bazenet and Rubin, 1998). Binding of neurotrophic factor to p75NTR also promotes NF-κB activation, thereby promoting NF-κB-dependent neuronal survival (Middleton, 2000). The involvement of neurotrophic factors in p75NTR also controls Rho family GTPase activity. It was shown that p75NTR activates RhoA through direct interaction, thereby inhibiting neurite growth. The binding of neurotrophic factor to p75NTR eliminates p75NTR-dependent RhoA activation and stimulates neurite growth (Yamashita et al., 1999). All of these studies illustrate the self-evident importance of p75NTR as a death receptor in neural and synaptic development.

In terms of synaptic plasticity, the effect on LTP is mainly mediated by the TrkB receptor. Stable LTP in rodent DG requires a BDNF-TrkB signal period lasting more than 8 h after LTP induction (Panja et al., 2014). Unprocessed BDNF is also called proBDNF, and its synaptic effect is opposite to that of mature BDNF, which acts through TrkB. By contrast, proBDNF has a high affinity for the p75NTR (Chao and Bothwell, 2002; Lu, 2003). ProBDNF binds to P75NTR first, and their binding induces LTD (Rösch et al., 2005). In the normal physiological state, the expression of P75NTR in the nervous system is very limited, and it is upregulated in pathological conditions such as inflammation or seizures. Therefore, the influence of BDNF on LTD is minimal (Figure 2). Studies have also shown that cholecystokinin (CCK) and glutamate cooperate with the CCK receptor and mGluR5 in astrocytes to trigger the release of the glial transmitter ATP, which increases the release of GABA from adjacent inhibitory terminal synapses, and switches the state of GABAergic synapses in the dorsomedial hypothalamic nucleus of rats from LTD to LTP (Crosby et al., 2018).

GABA is the main inhibitory neurotransmitter, and astrocytes express the GABA transporters GAT1 and GAT3 (Scimemi, 2014). Changes in the expression and activity of GAT1 or GAT3 affect inhibitory synaptic transmission in inter-hippocampal neurons (Beenhakker and Huguenard, 2010). GATs control the excitability of neurons in the neural network by regulating the level of extracellular GABA (Muthukumar et al., 2014). Furthermore, there is a signaling pathway between astrocytes and the GABA system. Astrocytes can enhance the current mediated by GABA in neurons and increase the transmission of inhibitory synapses in hippocampal CA1 neurons through Ca²⁺ signals. GABA selectively opens the Cl-channel by activating the type A GABA receptor (GABAAR), which induces an inhibitory potential and mediates rapid synaptic inhibition. The activation of GABAb elevates the Ca2+ level, which at the same time enhances the Ca2+-dependent release of gliotransmitters from astrocytes (Kang et al., 1998). It is now generally accepted that synaptic inhibition is mediated by GABAA receptors and that potentiation involves astrocyte GABAB receptors, astrocyte glutamate release, and presynaptic metabotropic glutamate receptors. And interneuron activity regulates the release of transmitters from the same synapses according to the firing rate, which suggesting that mechanisms of synaptic inhibition can coexist at the same synapse as those responsible for synaptic potentiation and that the end result is regulated by the firing rate of the interneuron (Perea et al., 2016). In a mouse model of Alzheimer's disease (AD), the excess GABA released by reactive astrocytes through the Best1 channel in the hippocampus acts on presynaptic GABA receptors and leads to catatonic inhibition of the dentate gyrus granulosa cells in the hippocampus (Lee et al., 2010; Jo et al., 2014). In terms of memory impairment caused by AD, inhibition of GABA produced by reactive astrocytes or blocking of GATs by different means can restore synaptic plasticity, learning, and memory impairment caused by AD. The application of GABA receptor antagonists in AD mice can improve hippocampal LTP, alleviate the learning impairment, and protect memory to different degrees (Yuji et al., 2008). Therefore, it stands to reason that inhibition of the synthesis or release of GABA may be a powerful method for treating memory impairment caused by AD.

TGF-β1 not only promotes the formation of excitatory protrusions, but also promotes the formation of inhibitory synapses. It induces the formation of inhibitory synapses by increasing the phosphorylation of CaMKII downstream of NMDA receptors, as well as increasing the expression of NL2 and aggregation of Gephyrin/NL 2, which is an important component of inhibitory synapses (Diniz et al., 2015).

At present, it is believed that Chordin-like1 (Chrdl1) factor secreted by astrocytes can regulate the level of synaptic GluA2AMPAR, mainly by increasing its quantity. In the developing brain, an important sign of the stability and maturation of excitatory glutamatergic synapses is that AMPAR subtypes on excitatory synapses are replaced by Ca²⁺-opaque channels containing GluA2 from Ca²⁺ permeable channels (Brill and Huguenard, 2008). In other words, by reducing the entry of calcium into postsynaptic cells, excitatory glutamatergic synapses tend to mature, stabilize, and limit synaptic plasticity mediated by Ca²⁺-dependent pathways (Henley and Wilkinson, 2016). However, the mechanism of the iconic transformation of stable mature synaptic AMPAR is not clear. Studies show that synaptic GluA2 AMPAR decreases and synaptic remodeling increases in response to changing sensory inputs in Chrdl1KO mice. In other words, Chrdl1 expressed by astrocytes limits synaptic plasticity by promoting GluA2-dependent synaptic maturation (Henley and Wilkinson, 2016).

With regard to the effect of ATP on LTD, researchers found that selective stimulation of astrocytes through ChR2 could lead to LTD, while blocking P2Y receptors in neighboring hippocampal neurons could prevent hLTD. Thus, ATP secreted by astrocytes mediates hippocampal LTD by activating neuronal P2Y1 receptors (Chen et al., 2013).

Recent studies have shown that the number and transmission of inhibitory synapses in neurons co-cultured with astrocytes is increased compared with neurons cultured alone, and their plasticity is enhanced (Sorg et al., 2016). Interestingly, these effects disappeared after treatment with selective Krebs cycle inhibitors such as fluoroacetic acid, indicating that the regulation of GABAergic synaptic development depends on a series of metabolic pathways in astrocytes, including the Krebs cycle. In other words, key metabolic enzymes expressed by astrocytes, such as glutamine synthetase, contribute to the plasticity of GABAergic neuronal networks (Kaczor and Mozrzymas, 2017).

Synaptic phagocytic receptors MEGF10 and MerTK are expressed on the surface of astrocytes. Astrocytes can eliminate excitatory or inhibitory synapses by interacting with MEGF10 and MerTK, which can recognize phosphatidylserine in target fragments as an opsonic signal to initiate phagocytosis and drive synaptic remodeling (Figure 3) (Chung et al., 2013). This process of synaptic elimination is strongly dependent on neural activity. Recent studies have shown that astrocytes regulate synaptic elimination by activating purinergic signals in an ATP-dependent manner mediated by IP3R2 receptor release (Yang et al., 2016).

In addition, studies have shown that astrocytes can express Ephrin-B1, which is a membrane binding protein and the ligand of the EphB receptor. The cross-synaptic ephrin-B/EphB interaction between neurons is essential for the formation and maintenance of synapses in the mouse brain (Henkemeyer et al., 2003; Grunwald et al., 2004; Xu and Henkemeyer, 2012). Astrocytic ephrin-B1 can compete with neuronal ephrin-B1 and trigger astrocyte-mediated elimination of synapses containing EphB receptors through cross-synaptic endocytosis. The loss of the neuronal EphB receptor weakens the ability of astrocytes expressing functional Ephrin-B1 to phagocytize synaptosomes in vitro, while the overexpression of Ephrin-B1 in astrocytes impairs long-term contextual memory. Thus, astrocytic Ephin-B1 regulates long-term contextual memory by limiting the formation of new synapses in the adult hippocampus (Koeppen et al., 2018).

Astrocytes are the main source of apolipoprotein E (ApoE) in the central nervous system and transport cholesterol to neurons through apolipoprotein E receptors, a member of the (LDLR) family of low-density lipoprotein receptors. ApoE exists in three subtypes, ApoE2, ApoE3, and ApoE4, whereby the latter is related to the decrease of dendritic spine density in memory retention (Caselli et al., 2004). The formation of dendritic spines is the most important feature of synaptic plasticity of mature neurons. At the same time, AopE4 is also the strongest genetic risk factor for late-onset Alzheimer's disease (LOAD).

Astrocytes can also initiate the classical complement cascade by secreting TGF-β and inducing the expression of complement protein C1q via the TGFBR2 receptor. C1q is then recognized and taken up by microglia through C3R-mediated phagocytosis (Schafer et al., 2012; Bialas and Stevens, 2013). In this way, astrocyte-derived TGF-β regulates the expression of neuronal C1q to initiate complement activation and microglial-mediated synaptic pruning (Figure 2).

In addition to complement C1q, IL-33 expressed by astrocytes also drives microglial synaptic phagocytosis and limits the number of excitatory synapses (Nguyen et al., 2020). These studies have shown that astrocytes can mediate the indirect effects of microglia-dependent synaptic elimination through C1q or IL-33.

Autism spectrum disorders (ASD) are a heterogeneous group of neurological developmental conditions characterized by repetitive movements, impairments in social interaction, and altered vocal communication as the core symptoms (Lai et al., 2014). The exact pathophysiological basis of ASD is currently unknown, but a growing body of evidence supports dysregulated neuroinflammation and dynamic imbalances in synapse formation (Christopherson et al., 2005; Kucukdereli et al., 2011), pruning, elimination (Chung et al., 2013), and plasticity (Henneberger et al., 2010) as important factors contributing to ASD. Neuroinflammation is characterized by the persistent activity and proliferation of glial cells (e.g., astrocytes and microglia) following injury, infection, or disease. Astrocytes respond to neuroinflammation through morphological changes and the release of pro-inflammatory mediators (Hanisch and Kettenmann, 2007; Kierdorf and Prinz, 2013). Since astrocytes are closely associated with the connection and regulation of neural circuits, early disruption of astroglial homeostasis may lead to abnormal brain development and the ASD phenotype. Autopsy studies have found that individuals with ASD have increased numbers of neurons in the cerebral cortex, white matter, and especially the cerebellum (Morgan et al., 2010; Vargas et al., 2010). Western blot analysis indicated that GFAP protein expression was significantly increased in the cerebellum, frontal middle gyrus, and anterior cingulate gyrus of autistic patients, suggesting that astrocyte immunoreactivity was also increased (Vargas et al., 2010). Recent studies have shown that astrocytes control excitatory synaptogenesis through the secretion of platelet responsive protein (TSP), which acts through its neuronal receptor calcium channel subunit α2δ-1 (Eroglu et al., 2009). The α2δ-1 receptor is associated with neurological disorders such as epilepsy, neuropathic pain, intellectual disability and ASD (Newton et al., 2001; Iossifov et al., 2014). The small Rho GTPase Ras-Related C3 Botulinum toxin substrate 1 (RAC1) is a synaptic signaling protein downstream of the TSP-α2δ-1 axis. RAC1 is a key component of the cascade that mediates synaptic and spinal growth (Christopher et al., 2018). This provides a new perspective for intervention in diseases including ASD and epilepsy, which are strongly associated with abnormal RAC1 signaling (Zeidán-Chuliá et al., 2013; Tejada-Simon, 2015). Since astrocytes can control neuronal RAC1 signaling through TSP-α2δ-1, understanding the regulatory mechanism of the TSP-α2δ-1-RAC1 pathway in astrocytes may uncover new potential therapeutic targets for psychiatric disorders such as ASD in the future.

Rett syndrome (RTT) is a rare X-linked neurodevelopmental disorder resulting in a range of symptoms including autistic features, intellectual impairment, motor degeneration and autonomous nervous system abnormalities. Loss of function of the X-linked gene MECP2, which encodes methyl CpG-binding protein 2, is the main cause of RTT (Amir et al., 1999). MECP2 is both a transcriptional activator and a transcriptional repressor that regulates synaptic activity in an activity-dependent manner (Qiu et al., 2012). Although few early studies explored the role of glial cells in RTT, advances in technology have shown that MECP2 is expressed in both astrocytes and neurons (Grunseich et al., 2009). In fact, a recent study found that naïve dendrites of primary WT neurons co-cultured with MECP2-deficient primary astrocytes were poorly developed compared to neurons co-cultured with WT primary astrocytes (Grunseich et al., 2009; Maezawa et al., 2009). Moreover, dendritic and synaptic morphology could be restored in vivo after restoring MECP2 (Lioy et al., 2011). These findings that MECP2-deficient astrocytes are detrimental to neuronal and dendritic growth and development, and that astrocytic MECP2 deficiency is an important cause of RTT. This morphological abnormality of neurons and dendrites has been attributed to neurotoxins released by MECP2-deficient astrocytes. However, how the astrocytes of RTT patients exert neurotoxic effects remains unknown. Furthermore, the transcription of the neurotrophic factor BDNF secreted by astrocytes is also regulated by MECP2, and overexpression of BDNF in MECP2-deficient mice improved motor function and restored the normal electrophysiological activity of somatosensory pyramidal neurons (Chang et al., 2006). Based on this, we can ask the question whether MECP2 can also control other factors secreted by astrocytes (such as TSPs, Hevin, etc.).

Among the many forms of inherited intellectual disability, Fragile X syndrome (FXS) is one of the most common. The trinucleotide repeat amplification of the Fragile X mental retardation 1 (FMR1) gene can lead to the loss of Fragile X mental retardation 1 protein (FMRP), which plays a number of important roles in the human body, such as binding to proteins involved in the regulation of genomic stability, cell differentiation and other processes (Contractor et al., 2015). It also plays an important role in the translation, transport and targeting of neuronal mRNAs, which in part control synaptic plasticity, development and elimination (Pfeiffer and Huber, 2009). FXS is mainly characterized by moderate to severe mental retardation, autistic traits, susceptibility to seizures, and behavioral abnormalities (Penagarikano, 2007). FMR1 knockout (KO) mice have phenotypes such as learning and intellectual deficits (Comery et al., 1997; Wisniewski et al., 2010), as well as changes of social behavior (Mckinney et al., 2010) that are similar to those of human FXS patients.

In vivo imaging studies on global FMR1 KO mice indicated that the lack of FMRP leads to reduced synaptic stability (Feng et al., 2010; Padmashri et al., 2011). In addition, the density of immature dendritic spines in cortical vertebral neurons is greatly increased in both FXS patients and adult FMR1 KO mice (Dailey and Smith, 1996). The morphology and density of dendritic spines has important implications for proper synaptic connectivity and function (Sala and Segal, 2014). Therefore, the lack of FMRP will lead to abnormal development of dendritic spines and thus affect synaptic connections and functions. For example patients with fragile X syndrome, the spine is usually longer and thinner than the average person and there are more spines exist per unit length of the dendrite (Irwin et al., 2001), which lead to a series of abnormalities social behavior such as verbal language deficits, reduced eye contact, social and generalized anxiety, sensory hypersensitivity, and difficulty regulating attention and activity levels (Lewis et al., 2010). These abnormalities in social behavior will continue from the onset into adulthood. Also, Patients with fragile X syndrome have a learning disability, as both learning and memory are impaired in patients with fragile X syndrome. A recent study showed that selective deletion of the FMR1 gene in astrocytes results in reduced expression of the astroglial glutamate transporter protein GLT1 (Higashimori et al., 2016). The primary role of GLT1 is the regulation of the extracellular synaptic glutamate concentration, and GLT-1-mediated loss of glutamate uptake increases extracellular glutamate levels, thereby increasing excitability of layer V pyramidal neurons (Higashimori et al., 2016). The increased glutamate concentration may promote the growth of new dendritic spines (Kwon and Sabatini, 2011). Thus, the dysregulation of glutamate homeostasis caused by the absence of FMRP in astrocytes may partially explain the increased density of immature dendritic spines in FMR1 KO mice. Astrocyte-specific deletion of FMR1 leads to a significant increase in immature spines in the motor cortex of mice, which manifests as abnormal synaptic morphology and reduced synaptic density in hippocampal neurons. This is due to the overproduction of dendrites during neuronal development and abnormal or insufficient pruning (Hodges et al., 2016). Interestingly, normal astrocytes prevent such abnormal synapses and reduced density of hippocampal neurons. Therefore, astrocytes play an important role in the abnormal synaptic development of FMR1 KO mice.

The group I mGluRs (mGluR1/5)-mediated LTD is a specific protein synthesis-dependent synaptic plasticity, which can be increased in FMR1KO mice (Huber et al., 2002). This groundbreaking discovery led to a new mGluR-based theory of FXS, which states that dysregulated mGluR1/5-mediated protein synthesis-dependent forms of synaptic plasticity contribute to the pathology of FXS (Bear et al., 2004). The synthesis of this specific protein leads to the internalization of AMPAR, which is the key mechanism of mGluR-LTD. Unfortunately, this mechanism seems to be damaged in FXS. Some studies have found that the mGluR5-mediated internalization of AMPAR increases when FMRP levels decrease (Nakamoto et al., 2007). Therefore, we have reason to believe that the role of FMRP in regulating mGluR1/5-induced protein synthesis may be the basis of LTD, exemplified in FXS mice. In addition, the sensitivity of mGluR1/5 to glutamate released by synapses in the hippocampus is enhanced in FXS mice, and the resulting loss of FMRP also leads to prolonged epileptiform discharges (Chuang, 2005; Bianchi et al., 2009). Moreover, the synthesis and release of endogenous cannabinoids in the hippocampus cannabinoid (eCB) mediated by mGluR1/5 is in enhanced following the deletion of FMRP (Zhang and Alger, 2010). In fact, there are also defects of the mGluR1/5-dependent pathway in other areas of the FXS mouse brain, such as abnormal synaptic plasticity of mGluR1/5-dependent neurons in the amygdala of FMR1 KO mouse (Suvrathan et al., 2010). Similar to the excessive LTD in the hippocampus, there is also excessive mGluR1/5-dependent LTD in the cerebellum of FXS mice (Koekkoek et al., 2005). Since different areas of the brain correspond to different functions, defects of the mGluR1/5-dependent pathway in different areas of the FXS brain may not only lead to cognitive dysfunction, but also lead to epilepsy (prolonged epileptiform discharges in the hippocampus), anxiety (abnormal synaptic plasticity in the amygdala) and motor deficits deficit (cerebellum). Although most of these studies are based on FXS animal models that may not always completely reflect the human phenotype, the successful validation of mGluR theory in different animal models further confirmed that FMRP plays a vital role in stimulation-induced mGluR1/5 signal transduction, and the potential value of glutamate derived from astrocytes via to activate mGluR1/5 as a therapeutic target for the treatment of FXS.

Trisomy of chromosome 21 causes Down syndrome (DS), which is the most common cause of inherited mental retardation. Intellectual disability in DS includes deficits in cognitive function, interpersonal interaction and communication. Abnormalities in the structure and function of dendritic spines are one of the prominent features of DS. Altered dendritic spine morphology and reduced density may underlie altered neuronal and synaptic plasticity (Benavides-Piccione et al., 2004), ultimately leading to cognitive deficits. In the cortical and hippocampal structures of DS fetuses and neonates, the morphology of dendritic spines exhibits abnormal changes, such as unusually long spines, shorter spines, and a decrease in synaptic density (Marin-Padilla, 1972). It was found that this abnormal dendritic spine development as well as the reduced synaptic density and activity in DS patients is closely related to astrocytes, and that the astrocyte-secreted TSP-1 protein is a key factor that regulates the number and morphology of dendritic spines. TSP-1 levels are significantly reduced in the astrocytes of DS patients, and restoration of TSP-1 levels can rescue the aforementioned morphological abnormalities and synaptic defects in the dendritic structures of hippocampal and cortical neurons (Octavio et al., 2010). Such findings highlight the role of astrocytes in synaptic defects in DS, and although there are no effective methods to date to prevent or restore such synaptic defects, these results demonstrate the potential of astrocyte-secreted molecules such as TSP-1 as therapeutic targets for symptomatic relief in DS and other neurological disorders associated with synaptic defects.

Indeed, astrocytes may influence numerous neurological disorders in multiple ways, and it is well-known that targeting neurons will be far more difficult than targeting astrocytes, which speaks volumes about the potential of astrocytes as therapeutic targets for the prevention and treatment of related diseases.