Astrocytes are a group of cells with heterogeneity, which have not been classified according to their anatomical location, molecular composition, type of neurotransmitter used or their response to diverse brain traumas (Chaboub and Deneen, 2012). Astrocytes occur more than five times more frequently than neurons. Many types of astrocytes exist in the CNS, including fibrous and protoplasmic glial cells, and numerous anatomical differences have been found between them. Fibrous astrocytes (located in the brain's white matter), exhibit strong interfascicular processes, whereas, protoplasmic glial cells (found in grey matter) are more predominant, contain more organelles and exhibit highly branched tertiary processes. Both of them are morphologically complex with star-shaped appearances and radially project a multitude of repeatedly branching variable diameter cellular processes that exhibit terminal footplate protuberances. These end-feet support and maintain the blood-brain barrier (BBB) which itself is constituted by the endothelial cells of brain microvessels. Additionally, adjacent to the pia mater, these two types of astrocytes extrude processes that run perpendicular to the pia to form the Pia-glial membrane.

Glial fibrillary acidic protein (GFAP) is the major intermediate filament (IF) protein of astrocytes. It participates in the formation of the cytoskeleton and maintainenance of its tensile strength. The morphological heterogeneity of astrocytes can be observed by three-dimensional reconstruction with antibodies against GFAP where the branched, tubular processes are the main manifestation of this heterogeneity. Astrocytes are a general term for a group of cells whose morphology differs depending on their subtypes, molecular diversity, and anatomical location (Hu et al., 2016). For example, glial processes appear in the cortex less regularly than in the hippocampus. The protein connexin-43 (Cx43) is not expressed by neurons and is strictly excluded from oligodendroglia, microglia and NG2 cells (Theofilas et al., 2017). For this reason, a connexin-43-GFP reporter mouse has recently been adopted to highlight the heterogeneity of astroglial cell morphology. Research into mechanisms of astrocyte morphology has shown that the dynamic recombination of actin cytoskeleton is controlled by the small GTPase of the Rho family and this is now generally believed to be the cause of cell morphology change. Although, some cytokine and cell adhesion factors are known to participate in morphological changes of astrocytes as well. CD44 is an integral membrane glycoprote which functions as an adhesion molecule for epithelial cells (Kondo et al., 1998). The overexpression of CD44 results in a larger cell size and astrocytes appear to exhibit a flat shape, while heparin-binding epidermal growth factor (HBEGF) alters the shape of astrocytes to radial glial-like phenotypes and induces expression of IF Further, RT-qPCR analysis has indicated that HB-EGF affects the expression of the Notch signaling pathway gene to play a role in the morphological heterogeneity of astrocytes (Puschmann et al., 2014).

Astrocytes in healthy CNS are considered to be the main homeostatic cells. Studies have reported that these cells are responsible for the maintenance of internal homeostasis, including ion, neurotransmitter, metabolic, synaptic, and organ homeostasis. They also regulate blood flow and microcirculation. The cell membrane of brain endothelial cells (BECs) and perivascular end-feet of astrocytes ensure the integrity of the BBB (Yamamizu et al.,2017). Astrocytes play an important role in the homeostatic regulation of the cerebral microenvironment due to their secretion of chemical mediators capable of rapid regulation of endothelial cell permeability (Abbott, 2002; Verkhratsky et al., 2017).

Astrocytes fulfill various homeostatic functions involving potassium (K⁺), chloride (Cl^-), hydron (H⁺), sodium (Na⁺), and calcium (Ca²⁺). K⁺ efflux is associated with excessive neuronal activity, and K⁺ in extracellular space is consequently moved into astrocytes by active transport that mediates local uptake to buffer extracellular K⁺. This process is enhanced by the Kir4.1 channel in a VGLUT1-dependent manner (Cai et al., 2011), the Na⁺-K⁺-Cl^--cotransporter-1 (NKCC1) (Amadeo et al., 2018), the Na⁺-K⁺-ATPase pump (Du et al., 2016) and Cx43 (Lykke et al., 2017). The molecular mechanism of maintaining Cl^- ion homeostasis by astrocytes includes GABA receptor anion channels (Fraser et al., 1995) and ClC-2. Ca²⁺ ion homoeostasis relies on a Na⁺/Ca²⁺ exchanger 1-3 (NCX1-3) (Verkhratsky et al., 2012) and plasmalemmal Ca²⁺ pump (PMC) and is controlled by the coordination of IP3R and store-operated Ca²⁺ channels (Blaustein et al., 2002; Liu et al., 2017; Reyes et al., 2012; Sakuragi et al., 2017). Neurotransmitter homeostasis refers to glutamate, GABA, glycine, and adenosine.

The developmental progression of astrocytes can be identified and visualized by their morphological criteria and the expression of specific proteins as characterized markers such as GFAP, vimentin, protein S100, glutamate transporter GLT1, glutamine synthetase (GS), Kir4.1 channel, water channel aquaporin 4 (AQP4) and Cx43 (Morel et al., 2014). Astrocyte precursors express intermediate proteins of vimentin, nestin, and synemin, while mature and differentiated astrocytes express only GFAP and vimentin.

Antibodies against GFAP, the most widely used immunostain, and the expression of GFAP significantly increased at three, six and twentyfour hours post-TBI (Huang et al., 2015). Compared with GFAP, the immune response cells of N-myc downstream regulated gene2 (NDRG2) are more abundant and more evenly distributed in the brain. For example, there are cells strongly positive for NDRG2 in the caudate nucleus, where GFAP rarely occurs. Vimentin is one of the proteins in IF that provides elasticity and is responsible for maintaining cytoskeletal integrity.

S100B has long been recognized as an important marker of TBI and is produced by activated astrocytes, which can further promote the activation of microglia, increase the production of inflammatory factors such as TNF-α, IL-1β, and Il-6 and cause long-term neurological dysfunction. Animal models have shown a significant increase in S100B levels in the brain lasting up to five days after TBI (Kabadi et al., 2015). Interestingly, proliferating cell nuclear antigen was detected as S100A6+ in astrocytes indicating that S100A6 (calcyclin) is a novel marker for astrocyte precursors (Yamada and Jinno, 2014). The differentiation of astrocytes can be visualized by glycogen metabolism and glycogen is a hallmark of mature astrocytes (Brunet et al., 2010). The marker that identifies the different subpopulations of astrocytes is also useful because the mechanisms that underly astrocyte diversity remain unclear and elucidating their functional heterogeneity is a difficult task. Fortuitously, a recent study has reported that a CS56-specific antibody could be used as a novel marker to detect astrocyte subpopulations (Okuda, 2018). Additionally, GS, a 38 kD breakdown product of aldolase C (ALDOC), ubiquitin carboxyterminal hydrolase L1 (UCH-L1), brain lipid binding protein (BLBP) and astrocytic phosphoprotein (PEA15) have been tested as candidates for TBI evaluation (Guida et al., 2017).

However, GS was was found to be an unpromising marker because of its variability in many neurological diseases, such as cerebral hypoxia, epilepsy and chronic neurodegenerative processes. Additionally, Studues have shown that there is no significant correlation between the expression of GS and GFAP (Bartlett and Li, 1990). UCH-L1 level wase elevated in cerebrospinal fluid (CSF) 24 hours post-TBI while blood levels were unchanged. During the first week after TBI, ALDOC levels were significantly raised, but stable. Levels of BLBP and PEA15 can be associated with injury progression. ALDOC, BLBP and PEA15 show a sharp increase in both severe and mild TBI blood and CSF tests; 25 kDGFAP-BDP appeared overnight, but was rarely present in mild TBI (Agoston and Shutes-David, 2017). Moreover, astrocytic Ngb expression has been observed in sub-acute and chronic TBI (Chen et al., 2015).

In addition to biomarkers, a variety of advanced imaging tools are also used to study astrogliosis. Manganese-enhanced magnetic resonance imaging (mMRI) can be employed during the subacute phase of TBI (7--14 days) when astrogliosis is characterized by crescent-shaped T1-weighted signal enhancements surrounding the impact core. Experiments have confirmed that the hyperintensive area is glial scar tissue, while the void is an impacted necrotic area (Talley Watts et al., 2015). MMRI employs manganese is an analog of calcium, dysfunction of which is associated with secondary insult in TBI. It was also discovered that diffusion tensor imaging (DTI), an advanced MRI technology, was more sensitive to microstructural abnormalities of the CNS than standard MRI. GFAP expression in the cortical gray matter of a TBI model was significantly and positively correlated with fractional anisotropy (FA) in the acute (1--7 days) and subacute (7--14 days) periods after TBI due to the increased content and directional cohesiveness of astrocytes.

In white matter, decreased FA indicates neuronal demyelination and axonal degeneration, while increased FA is associated with glial scar formation but not axonal regeneration. In the subacute phase, the mean kurtosis (MK) of the ipsi- and contralateral damaged regions was significantly increased, and MK improvement was inversely proportional to distance from affected areas. Immunohistochemical analysis showed that MK increase was related to the proliferation of reactive astrocytes (Zhuo et al., 2012). Thus, FA and MK in DTI can be used to study and monitor reactive astrogliosis.

Regrettably, no causal relationship has been found between diffusion magnetic resonance imaging (dMRI) at all posttraumatic time points and reactive astrocytes, so dMRI is less valuable for the study of astrogliosis (Zhuo et al., 2012). In addition to various MRI imaging modalities, the major imaging innovations for investigation of astrogliosis are two-photon imaging in vivo, positron emission tomography (PET) and live tissue imaging. Intracellular Ca²⁺ acts as a second messenger and its transient changes can be used as an indicator of the physiological activity of astrocytes at different times after TBI. On this basis, astrocytes can be imaged in vivo with a two-photon microscope (Shandra and Robel, 2019). Although SPECT has a higher diagnostic value in mild TBI than CT and MRI (Raji et al., 2014), it is a poor discriminator of astrogliosis. Translocator protein 18 kDa (TSPO) PET imaging has demonstrated that in addition to reactive astrocytes, microglia cells can also overexpress selectively binding TSPO radioligands after TBI (Lavisse et al., 2012), thus with SPECT it is difficult to elucidate astrocyte changes. Confocal microscope live tissue imaging can be used to explore interactions between reactive astrocytes and immune cells during the period of inflammatory response that follows TBI (Escartin and Murai, 2014). Overall, advanced imaging techniques, particularly different types of MRI, have led to new insights into the function of astrocytes.

The morphological changes of astrocytes following TBI include cell hypertrophy, synaptic elongation and increase in number (See Table 2). Palisading processes grow about one week after injury and glial scars are formed in severe cases (Bitner et al., 1987). Additionally, these changes are accompanied by both proliferation and migration of reactive astrocytes and elevation of GFAP that are related to the severity of the injury (Fix et al., 1996). Acute and diffuse lesions without tissue damage cause slight and reversible hypertrophy of astrocytes and in this case, reactive astrogliosis is limited to local areas without proliferation. More persistent and more severe lesions result in more pronounced cell hypertrophy, increased growth and neurite extension toward the injured site, along with astrocyte proliferation and migration. In highly reactive astrogliosis, glial scars are predominant. Astrocytes appear to begin hypertrophy 2--3 days after injury and, accompanied by glial scar formation, remain hypertrophic up to seven days after injury. Reactive astrogliosis can be maintained for up to 60 days. The multiple processes of hypertrophic astrocytes project to an injured area, but as the distance from the injury site increases hypertrophy decreases. Finally, brain tissue without substantial damage up-regulates GFAP but shows no signs of hypertrophy. One factor that causes astrocyte hypertrophy is the endothelin B receptor (ETBR) (Rogers et al., 2003), while another major driver of hypertrophy and growth in the formation of scar-producing astrocytes is the cytoskeleton (microtubules, IF and actin filaments in microfilament).

Table 2 Morphological change, proliferation and migration of astrocytes in response to TBI

ETBR is a G-protein-coupled receptor that activates the phosphatidylinositol pathway. Its ligand ET boosts mitosis of cells, induces glutamate efflux and promotes dynamic change of actin filaments. ETBR is the major ET receptor type expressed on astrocytes in in vitro (Rogers et al., 1997). Experiments in vitro have shown that treatment by ET1 induces the hypertrophy and proliferation of normal astrocytes and the infusion of bosentan (a mixed ETAR and ETBR antagonist) causes a significant decrease of astrocyte hypertrophy in damaged nerve tissue. The up-regulation of ETBR is a potential cause of astrocyte hypertrophy after TBI, but its specific signaling pathway is still unclear (Rogers et al., 2003). In addition to inducing astrocyte hypertrophy, ETBR is also involved in astrocyte proliferation, and loss- and gain-of-function studies and promoter activity assays (Li et al., 2010) have shown that Jagged1/Notch1 signaling enhances phosphorylation signaling and transcription. The level of signal transducers and activators of transcription 3 (STAT3) indirectly increases ETBR expression and its ligand (ET1) promotes mitosis of astrocytes.

It has further been confirmed that the Notch1-STAT3-ETBR axis is one of the pathways that promotes reactive astrogliosis proliferation following brain injury (LeComte et al., 2015) and astrocyte proliferation instead of migration can be detected on day 5--7 after injury. GFAP-positive astrocytes were found to proliferate on day one, three and seven after injury, and the number of proliferating astrocytes reached their peak on the first day after injury. The DIX domain containing 1 (Dixdc1) and transforming control tumor protein (TCTP) have also been found to be involved in the regulation of astrocyte proliferation after TBI. Dixdc1 is a positive regulator of the Wnt signaling pathway and is upregulated in reactive astrogliosis (Lu et al., 2017). TCTP is a ubiquitous and highly conserved protein that plays a broad role in the regulation of cell proliferation, growth, apoptosis and cell cycles and regulates astrocyte proliferation as well as migration. Inhibition of TCTP also reduced migration of astrocytes, as the recombination of microtubules and F-actin is disrupted after siRNA transfection (Ren et al., 2015).

More severe TBI occurs in secondary cerebral edemas that account for most early deaths after TBI. Edema can be caused by the destruction of ionic homeostasis due to the incapacitation of astrocytes after TBI. Zhang et al. (2010) have shown that because extracellular glutamate levels increase after brain injury, microglia can release cytokines and directly contact astrocytes. Additionally, trauma significantly up-regulates the activity of neuronal NKCC and astrocyte AQP4 is significantly elevated after TBI, all of which can lead to cerebral edema. A study of the distribution of AQP4 in the CSF of patients who suffered TBI showed a significant increase of AQP4 in these patients (Lo Pizzo et al., 2013). So, some investigators (Kimelberg, 2005) believe that AQP4 is also the cause of astrocyte edema.

Tensile strength and cell polarity are important for morphological changes in astrocytes and a scratch damage model is commonly employed to trigger and estimate astrocyte polarization and migration towards the wound area (Magdalena et al., 2003b). A major tensile force originates from the dynamic reorganization of actin filaments and cell polarity is derived from microtubule-induced Golgi reorientation. Such directional extension towards the movement leading edge of cell movement allows astrocytes to migrate to a wound (Magdalena et al., 2003a). The key molecule for directed elongation of astrocytes is the small GTPase Cdc42, which regulates the arrangement of microtubule tissue centers (MTOCs) and Golgi through the mPar6-PKCzeta signaling pathway. Cdc42 polarizes transport of cytoskeletal regulators towards the leading edge to force microtubules to the tip of the protrusion; thereby, elongating the processes of reactive astrocytes (An et al., 2016; Robel et al., 2011). Significant up-regulation of IF in reactive astrogliosis performs the function of maintaining cell polarity and orientation of both motion and nuclear localization. IF plays a major role in the formation of reactive astrocytes and glial scars, so it is believed to be the most relevant cytoskeletal component. In short, the various components of the cytoskeleton are complexly involved in astrocyte alterations, an extension of processes and migration to the injured site (Schiweck et al., 2018).

Damaged cells release, among other molecules, heat shock protein, MMP-9 and ATP to many pattern recognition receptors (PRRs) that are expressed on the cell surface or interior of astrocytes, such as Toll-like receptors (TLR), that identify various types of TBI damage-associated molecular patterns and then activate astrocytes at the peak level of their inflammatory response. ATP induces the activation of a p2x7-dependent inflammasome via the pannexin one channel (Corps et al., 2015; Russo and McGavern, 2016). The release of MMP-9 increases astrocyte exposure to pro-inflammatory cytokines such as IL-1β by activating the ERK pathway (Ralay Ranaivo et al., 2011). Reactive astrocytes then produce an immune response that releases cytokines and chemokines and activates the migration of both microglia and astrocytes to the injury site to produce an anti-inflammatory response to TBI. Meanwhile, neutrophils and monocytes enter the brain through the meningeal vessels and choroid plexus and initiate a wider range of inflammatory effects (Makinde et al., 2017). Neutrophils resist infection while releasing ROS, which causes tissue damage. Importantly, neurons also express TLRs, which allow them to respond to TLR-driven inflammatory responses and become damaged (Lozano et al., 2015). Also, inflammation damages the absorption of glutamate by astrocytes, resulting in excitotoxic harm, with further aggravation of nerve damage (Cekanaviciute and Buckwalter, 2016). Astrocytic aggravation of inflammation is also manifested in biochemical and metabolic signals. In vitro, the activity of NF-κB, the expression of pro-inflammatory cytokine and the release of MMP9 activity were up-regulated in reactive astrocytes 24 hours after TBI (Pan et al., 2012). The antioxidant response element (ARE) depletes Nrf2 during oxidative stress, activating the DNA binding activity of NF-κB in astrocytes and downstream pro-inflammatory cytokines. This causes increased the DNA binding activity of CD24 and NF-κB in astrocytes and downstream pro-inflammatory cytokines. All of these processes induce more serious inflammation (Li et al., 2014, 2015).

Alternatively, reactive astrocytes avoid excessive inflammation through two mechanisms. Either the glial scar limits the migration of inflammatory cells into the CNS or chemokines regulate the activation of white blood cells. Indeed, CCL2 not only recruits blood-derived macrophages that lead to neuroinflammation and denaturation after trauma but also inhibit the stimulation of astrocytes by IL-1β. It is by these means reactive astrocytes reduce their production of pro-inflammatory cytokines, thus to some extent protecting nerve cells.

Glial scar tissue is a physical boundary formed by the composition of ECM and active astrocytes after proliferation and migration. The glial scar encloses the injured area and forms a physical barrier against the expansion of inflammation and injury. It distributes along the interface of all blood vessels and becomes located both at the junction between the brain parenchymal cells and the BBB's endothelial cells and at peripheral cells. Since WBC must enter the CNS from the bloodstream through the endothelial cells, when present, glial scars restrict the entry of excessive white blood cell (WBC) into the CNS (Cekanaviciute and Buckwalter, 2016).

It is generally believed that the glial scar forms an inhibiting barrier for axonal regeneration. Additionally, the blood supply is affected by glial scar hyperplasia. Glial scars also release axonal growth inhibitors, such as proteoglycans and tenascins and ephrinb1 in astrocytes competes with ephrin-b1 in neurons, all of which inhibit new synapse formation. Neurons that grow under the atypical astrocytic cell pattern are malnourished, but their survival is not affected (Liu et al., 2015; Nikolakopoulou et al., 2016). However, there is an alternative view that in the early period of TBI glial scars repair and support axon regeneration with nerve growth factor (NGF), glial nerve growth factor (GDNF) and glial cell maturation factor. Here, through gene-targeted functional deletions in adult mice, it has been found that preventing or removing astrocyte scars does not lead to spontaneous regrowth of sensory or serotonin axons. In contrast, RNA sequencing showed that after lesion clearance, reactive astrocytes expressed and constantly delivered axon-specific growth factors through the formation of glial scars, stimulating regeneration of robust sensory axon (Li et al., 2018).