In 1856, Rudolf Virchow described for the first time a type of cell with neuron-supportive functions (Virchow, 1856). Then, in 1895, MV Lenhossék proposed the name astrocyte (“Astrocyten”) for this type of neuron-supporting cell (Lenhossék, 1893). Cortical astrocytes originate from radial glia derived from the neuroepithelial cells, radial glial cells originate from the cortical ventricular zone and are characterized by a long basal process that extends from the cortical ventricular zone to the pial surface (Arellano et al., 2021). During embryonic development, radial glial cells generate intermediate glial progenitors via asymmetric division, and these progenitors then migrate, proliferate, and finally transform into astrocytes in nerve tissue. After birth, astrocytes are primarily generated through the direct transformation of radial glial cells in the ventricular zone, the migration and development of postnatal progenitors in the subventricular zone, and the symmetrical division of differentiated astrocytes (Levison and Goldman, 1993; Ge et al., 2012; Verkhratsky and Nedergaard, 2018; Abdeladim et al., 2019). NG2 glial cells comprise another possible source of astrocytes (Nishiyama et al., 2016). Here, astrocytes undergo limited migration along with radial glial processes (Jacobsen and Miller, 2003). Astrocytes of different origins are phenotypically diverse, which is a partial manifestation of the heterogeneity of astrocyte morphology and function (Magavi et al., 2012; Tsai et al., 2012; Molofsky and Deneen, 2015). A combination of heredity, development, and phenotype renders astrocytes a truly opportunistic cell with lifelong adaptive plasticity.

Under physiological conditions, astrocytes perform a variety of functions primarily associated with the maintenance of CNS homeostasis, including the formation and maintenance of the blood–brain barrier (BBB) and blood–spinal cord barrier (BSCB), signal transmission across synapses, the maintenance of neuronal function, and metabolic regulation (Molofsky and Deneen, 2015). In a pathological background, however, astrocytes can become activated. The lifelong adaptive plasticity of these cells and the complexity of the disease background determine the diversity of astrocyte subpopulations after injury (Verkhratsky and Nedergaard, 2018). Following CNS insult, activated astrocytes can sequentially display two different histological phenotypes over time, first becoming reactive astrocytes (RAs), and then scar-forming astrocytes (SAs; Hara et al., 2017). This sequential phenotypic change from the resting state to the activated state is referred to as reactive astrogliosis (Zamanian et al., 2012). However, this histological classification method fails to clearly define RAs and SAs as it is neither objective nor quantitative.

In 2017, Hara et al. (2017) were the first to define several RA- and SA-specific marker genes in the mouse. Plaur, Mmp2, Mmp13, Axin2, Nes, and Ctnnb1 were classified as RA marker genes, while SA markers included Cdh2, Sox9, and chondroitin sulfate proteoglycan (CSPG)-related genes, such as Xylt1, Csgalnact1, Chst11, Pcan, Acan, and Slit2. Nevertheless, RAs and SAs both display high expression levels of several proteins, including GFAP, nestin, β-catenin, N-cadherin, and SOX9. As the disease progresses, there is an overlap of RA subpopulations and RAs interact with Col1 and are converted into SAs via the integrin/N-cadherin pathway (Hara et al., 2017; Li X. et al., 2020). This research is of great significance to the understanding of SAs, but due to the lack of further research, the function of SA is not yet clear. Recently, Escartin et al. (2021) redefined RAs as ‘astrocytes that undergo molecular, morphological, and functional changes in response to pathological stimuli from surrounding tissue, such as CNS disease, injury, and deleterious experimental manipulation, among others. High GFAP expression levels and cell hypertrophy are considered the minimum criteria for defining RAs (Liddelow et al., 2017).

In addition to the above classification of astrocytes (RAs and SAs), RAs are also divided into different astrocyte subgroups. In 2012, Zamanian et al. undertook a genomic analysis using two mouse injury models (inflammation and cerebral ischemia models) to profile RA phenotypes. The authors found that the RA phenotype was dependent on the type of inducing injury, and identified high Lcn2 and Serpina3n expression levels as strong markers of RA phenotype (Zamanian et al., 2012). In 2017, Liddelow et al. found that neurotoxic RAs, which they named A1 astrocytes, were induced by cytokines (TNF-α, IL-1α, and complement component C1q) secreted by activated microglia, whereas neuroprotective RAs, termed A2 astrocytes, were induced under ischemic and hypoxic conditions. As shown in Figure 1. The neurotoxic effect of complement component 3 (C3), a strong marker of A1 astrocytes, has been confirmed in a variety of CNS diseases, especially the interaction between the C3 cleavage fragment, C3a, and its receptor, C3aR, on neurons (Guo et al., 2010; Lian et al., 2015; Li J. et al., 2020; Yadav et al., 2021). However, the A1 and A2 phenotypes were not proposed to be universal or all-encompassing, they were widely misinterpreted as evidence for a binary polarization of reactive astrocytes in either neurotoxic or neuroprotective states, which could be readily identified in any CNS disease, acute or chronic, like the once-popular, but now discarded, Th1–Th2 lymphocyte and M1–M2 microglia polarization theories. Any binary classification method cannot show the diversity of astrocytes across diseases. More importantly, in mouse models of CNS damage, a RA subset was usually a mixture of A1 and A2 or pan-reactive transcripts (Das et al., 2020). So, Escartin et al. (2021) recommend moving beyond the A1–A2 labels and the misuse of their marker genes. In fact, the latest works of the original authors who studied these subtypes no longer use A1/A2. Guttenplan et al. (2021) used the induction conditions of A1 astrocytes but called the induction results neurotoxic reactive astrocytes. Hasel et al. (2021) used the term neuroinflammatory astrocyte, and used the pattern of “Y-zone X-positive astrocytes showing Z phenomenon” to describe the neuroinflammatory astrocyte subgroups he discovered. Based on existing knowledge, this is an ideal way of naming. However, the A1/A2 classification of RAs is still widely used.

In our opinion, under certain conditions, neurotoxic reactive astrocytes, neuroinflammatory astrocytes, and A1 astrocytes are almost the same. In vitro, neurotoxic reactive astrocytes and A1 astrocytes are induced in the same way. In the brain of LPS-induced systemic inflammation mouse model, Liddelow et al proposed the concept of A1 astrocytes, and Hasel et al. proposed various neuroinflammatory astrocyte subtypes, A1 astrocytes can be regarded as a subgroup of neuroinflammatory astrocytes. Neurotoxic reactive astrocytes emphasized function, while neuroinflammatory astrocytes emphasized background, both concepts include A1 astrocytes. At present, users of the A1/A2 concept all regard A1 as the representative of neurotoxic astrocytes and A2 as the representative of neuroprotective astrocytes. However, considering their functional heterogeneity, it is likely that not all neurotoxic RAs are A1 astrocytes, and neither are A2 astrocytes. In the background that current knowledge does not allow objective classification of astrocytes, the use of a binary description of reactive astrocytes (A1/A2, neurotoxicity/neuroprotective), seems unavoidable. Recently, Escartin et al. (2021) reached a consensus that the field should move beyond binary descriptors and embrace objective classification based on their increasingly complex functional heterogeneity. And the work by Liddelow and Hasel supports this view (Hasel et al., 2021).

Astrocytes are key factors in secondary neuronal damage and repair inhibition largely due to their dual role in the regulation of neuroinflammation and glial scar formation after CNS injury (Liddelow and Barres, 2017; Adams and Gallo, 2018). This dual role requires the accurate classification of astrocyte subpopulations. In this review, we will focus on the heterogeneity of astrocytes and astrocyte targeted therapy strategies after CNS injuries (TBI and traumatic SCI) to help the development of targeted therapy strategies based on these precise classification of astrocytes.

Owing to the preventability of most CNS injuries and the complex and expensive medical care they require, TBI and SCI are increasingly recognized as global health priorities. In 2016, approximately 27.08 million new cases of TBI and 0.93 million new cases of SCI were diagnosed. The age-standardized incidence rate was reported to be 369 per 100,000 population for TBI and 13 per 100,000 for SCI (GBD 2016 Traumatic Brain Injury and Spinal Cord Injury Collaborators, 2019). TBI alone caused annual global economic losses of $US400 billion (Maas et al., 2017). From 1990 to 2016, the age-standardized prevalence of TBI increased by 8.4%, whereas that of SCI did not change significantly. However, given the increase in population density, population aging, and the increased use of motor vehicles, the number of people with SCI is expected to increase. TBI has a higher mortality rate (higher acute injury-related mortality), while TSCI is characterized by a higher standardized mortality rate (shorter long-term life expectancy for SCI survivors; Badhiwala et al., 2019). Public health initiatives to prevent injuries, such as the use of bicycle helmets, fall prevention, policy changes affecting the impact of sports, and other public safety measures, are very effective in reducing the morbidity and mortality associated with TBI and SCI (Taylor et al., 2017). The focus of clinical management involves reducing intracranial pressure, medullary cavity pressure, and cerebral edema, as well as systemic supportive treatment (Maas et al., 2021). In most cases, the effects of these interventions on patients are disappointing (Maas et al., 2017). The burden of disability due to CNS injury can also have a devastating effect on the families of patients because it prevents them from engaging in economic activities.

TBI is divided into focal tissue damage and diffuse tissue damage. Focal injuries are caused by direct impact and include scalp injuries, skull fractures, brain contusions, cerebral hemorrhage, and stroke, which form focal TBI lesions that can vary greatly in size (Gaetz, 2004). Diffuse injury is caused by acceleration–deceleration forces, including hypoxia–ischemic injury, meningitis, and vascular injury (Gaetz, 2004). However, tissue damage after TBI is rarely purely focal or diffuse, and a single case usually involves multiple focal and diffuse lesions (Skandsen et al., 2010). TBI-related tissue pathology and its functional consequences are heterogeneous and determined largely by: (a) the mechanical properties of the injury; (b) the degree of injury severity (mild, moderate, or severe); and (c) the anatomical location of the injury (Burda et al., 2016). The spinal cord has a unique anatomical structure and the impact of scars on the function of the spinal cord at later stages of SCI can be devastating. Consequently, greater attention is given to pathological changes occurring over time. Several key time points are worth noting, such as the 3rd day after injury when inflammation peaks.

Traumatic injury in the CNS is characterized by transient mechanical damage and subsequent delayed non-mechanical damage (Burda et al., 2016). Primary injury in the brain is caused by mechanical force, which immediately leads to contusion and bleeding in the affected area. In the spinal cord, injury usually relates to vertebral fracture or dislocation (Oyinbo, 2011). The secondary injury occurs hours, days, months, or even years after the initial injury, and is characterized by the expansion of tissue damage from the center of the disease. According to the research in the rodent model of TBI, secondary injury can be simply divided into two parts. The first is inflammation, which peaks on the 3rd day after injury (Susarla et al., 2014). Under the stimulation of a wide variety of pro-inflammatory factors produced as a result of the primary injury, microglia and astrocytes are activated, peripheral immune cells are recruited, and the inflammation cascade is initiated. These effects are accompanied by the destruction of the neurovascular unit, glutamate accumulation, oxidative stress, axonal damage, and neuronal death (Gyoneva and Ransohoff, 2015). The second part involves scar formation, in which glial scars begin to form on day 7 post-injury (Villapol et al., 2014). The glial scar surrounds the site of injury and limits the spread of a strong inflammatory response (Burda and Sofroniew, 2014); however, glial scars secrete a variety of cytokines and proteoglycans that promote neurotoxicity and inhibit axon regeneration, respectively (Silver and Miller, 2004). The outcome of glial scarring is the development of a fibrotic scar, which creates a physical and chemical barrier to axon regeneration and nerve function recovery after injury (O’Shea et al., 2017).

The role of an astrocyte is determined by its subgroup status and the surrounding environment. This diversity of astrocyte function directly affects the inflammatory response and glial scar formation after injury. After an injury, astrocytes interact with surrounding cells, such as neurons, microglia, and endothelial cells, that together constitute the post-injury microenvironment, which plays a pivotal role in disease development (Abbott et al., 2006; Valori et al., 2019).

Although primary CNS injuries cannot be treated, secondary injuries provide a therapeutic window for the treatment of the resulting diseases (Wang et al., 2014). Accordingly, to identify effective treatment strategies, research attention has increasingly focused on the role of astrocytes in the pathology of CNS damage.

In response to CNS damage, naïve astrocytes are activated and transform into RAs. This transformation involves changes in morphology, increased expression of the intermediate filament proteins GFAP and vimentin, as well as increased proliferation and secretion of inflammatory mediators and growth factors (Karve et al., 2016). After TBI in mouse, astrocytes react within 24 h and reach a peak of approximately 3–7 dpi, showing a continuous reactive state (Susarla et al., 2014). A recent study conducted using a mouse CCI (chronic constriction injury) model reported the occurrence of astrocyte hypertrophy in the lesion site and surrounding area at 3 days post-injury (dpi). At 7 dpi, the morphological changes became long-lasting, and glial scars began to form (Villapol et al., 2014). In this model, reactive gliosis persisted for up to 60 dpi, indicative of a continuous response of astrocytes to brain injury (Villapol et al., 2014). In another study, after sensorimotor cortex aspiration in adult rats, astrocyte activation lasted for 16 weeks (Basiri and Doucette, 2010).

RAs are astrocytes that undergo molecular, morphological, and functional changes in response to pathological stimuli from surrounding tissue, such as CNS disease, injury, and deleterious experimental manipulation, among others. As mentioned before, the lifelong adaptive plasticity of astrocytes and the complexity of the disease background determine the diversity of astrocyte subpopulations after injury. In animal models of TBI, P2Y (1)R stimulation was shown to reduce the severity of brain edema and cytotoxic swelling (Talley Watts et al., 2013). However, the results of another study suggested that microglia could convert astrocytes into neurons by mediating the downregulation of P2Y (1)R (Shinozaki et al., 2017). Early et al. (2020) proposed that astrocytes exhibited age-related progressive reactive astrocyte response by the models of TBI in mice of different ages. Recently, Hasel et al. (2021) successfully demonstrated the heterogeneity of RAs in the brain of LPS-induced mouse models. They used single-cell sequencing combined with spatial transcriptomics and in situ hybridization techniques to show that RAs were transcriptome and spatially heterogeneous under inflammatory conditions; and clarified the highly expressed genes and possible functions of RA subtypes in different anatomical locations (Hasel et al., 2021). Combined, the findings of all these studies have highlighted the high heterogeneity of RAs, which can lead to both neuroprotective and toxic effects after CNS injury (Miller, 2018). Differences in in vitro induction conditions; species used in animal models; injury type, degree, and location; and time passed after the injury have all contributed to the contrasting results obtained in different studies. All these make the precise typing of RAs more difficult.

As early as 2012, Zamanian et al. (2012) discovered a potentially harmful subgroup of RAs. Subsequently, Liddelow et al. (2017) proposed a neurotoxic RA with C3 as a molecular marker and named it A1 astrocytes. A1 astrocytes were induced by cytokines (TNF-α, IL-1α, and complement component C1q) secreted by activated microglia. Although the concept of A1 is not relevant in this field, many previous research results of A1 neurotoxic astrocytes can help subsequent research on the neurotoxic subpopulations of RAs. A1 astrocytes lose many basic functions and gain harmful ones when compared with normal astrocytes. Namely, A1 astrocytes have fewer synapses and a weaker ability for synapse induction; impaired myelin scavenging ability; they can inhibit oligodendrocyte maturation; exhibit stronger neurotoxicity; and kill CNS neurons that have severed axons (Liddelow et al., 2017; Li X. et al., 2020). A1 astrocytes have a significantly different morphology: long dendrites (Zou et al., 2019). This suggests that the morphology of RAs may be changeable. Adding morphological features to the subgroup division can make the typing more specific and accurate. A1 astrocytes are found in a variety of CNS injuries and neurodegenerative diseases but are also present during the normal aging process (Clarke et al., 2018; Yun et al., 2018; Zheng et al., 2021). Alawieh et al. showed that a significant increase in C3 levels after CNS injury triggers continuous microglia degeneration and astrocyte activation, reduces dendrite and synapse density, and ultimately leads to the loss of neurons (Alawieh et al., 2018; Clark et al., 2019). After SCI, mice with C3 deficiency have reduced inflammation and secondary damage and better nerve regeneration and functional recovery after injury compared with that for normal mice (Guo et al., 2010). However, mice with C3aR deficiency show abnormal neurodevelopment that persists into adulthood, and is characterized by locomotive hyperactivity and altered cognitive functions (Pozo-Rodrigálvarez et al., 2021). Wang et al. (2021) proposed a more radical possibility, namely, that A1 astrocytes could directly kill neurons by secreting neurotoxic C3. Several studies have reported that C3 is closely related to the onset of multiple neurodegenerative diseases (Lian et al., 2015; Litvinchuk et al., 2018). These observations suggest that the basic C3 level is necessary for the maintenance of a normal physiological environment in the CNS, whereas excessive C3 availability produces neurotoxic effects after injury. However, it must be acknowledged that the expression of a singular marker “C3” is not a definitive marker that identifies A1 astrocytes. The work of Boisvert et al. (2018) showed that C3 was upregulated on astrocytes in the condition of aging, and did not necessarily, or categorically, indicate A1 astrocytes. Therefore, it is neither accurate nor objective that C3 is used as a singular marker of A1 astrocytes in injury and diseases in humans and other models. Recently, Guttenplan et al. (2021) proposed that saturated lipids contained in APOE and APOJ lipid particles mediated the neurotoxicity of RAs. Astrocytes specifically knock out saturated lipid synthase ELOVL1 to eliminate the formation of long-chain saturated lipids, which reduced astrocyte-mediated toxicity.

In CNS injury, a variety of substances and intracellular signal pathways are involved in the induction and transformation of the functions of RAs (neurotoxicity and neuroprotection; Table 3). For instance, the activation of the NF-κB and Notch signal pathways promotes A1 transformation, while exposure to mesenchymal stem cell (MSC)-derived exosomes, which play anti-inflammatory and neuroprotective roles after SCI, suppresses A1 astrocyte numbers by inhibiting the NF-κB signaling pathway (Wang et al., 2018; Liu et al., 2019; Qian et al., 2019). Additionally, activating the FGF2/FGFR1 pathway can reverse the increase in C3 expression levels in astrocytes following ultrasound exposure (Zou et al., 2019). After SCI, the application of electrospun fiber was reported to promote the expression of A1-specific markers, but electrospun fiber-containing TGF elicited the opposite effect (Gottipati et al., 2020). In comparison, in an IL-1β-induced neonatal rat model of white matter injury, astrocytes showed A2 reactivity (Shiow et al., 2017). After TBI, neuron-derived prokineticin 2 and astrocyte-derived estrogen activated STAT3 signaling pathway in astrocytes, leading to the upregulation of A2 astrocytes (Neal et al., 2018; Ma et al., 2020; Wang J. et al., 2020). We have previously shown that miR-21, a regulator of the STAT3 pathway, can transform neurotoxic (A1) RAs into an A2 phenotype (Su et al., 2019). MFG-E8, MSC-derived extracellular vesicles (EVs), Wnt-3a, and Trkβ have also been shown to be involved in A1/A2 transformation (Xu et al., 2018; Zhang D. et al., 2019; Kaminski et al., 2020; Miyamoto et al., 2020). Interestingly, FGF2 can inhibit the TGF-β1-induced increase in GFAP expression in astrocytes (Tran et al., 2018). The antagonism between different molecules that induce the same phenotype further underlines the need for the development of a more precise method for typing RAs.

Following CNS injury, naive astrocytes transform into RAs, and then eventually SAs, leading to impaired axon regeneration and functional recovery. This continuous phenotypic change is a manifestation of astrocyte reactivity, which was once considered to be a unidirectional and irreversible process (Hara et al., 2017). Diseases and injuries of the CNS are usually accompanied by a certain degree of scar formation, although scar formation differs according to disease and injury (Smith et al., 2015). Glial scars are mainly involved in the repair process after CNS injury. After SCI, damage repair efficiency is low and the resulting pathological changes cannot be overcome. Consequently, here, we focus on astrocyte-mediated scar formation after SCI (Bradbury and Burnside, 2019). SCI lesions exhibit three compartments: a non-neural (stromal) lesion core, astrocyte scar borders, and spared but reactive neural tissue. SAs participate in the formation of astrocyte scar borders (Sofroniew, 2018). The scarring process begins on day 7 post-injury and involves the misalignment of activated astrocytes and the deposition of inhibitory CSPGs. SAs can be identified from 14 dpi (Hara et al., 2017).

Various mediators are involved in glial scar formation, including TGF-β1/2, IFN-γ, FGF, MMP-9, fibrinogen, and STAT3 (Moon and Fawcett, 2001; Herrmann et al., 2008; Hsu et al., 2008; Schachtrup et al., 2010). The glial scar represents a physical barrier that enwraps damaged tissues and restricts the migration of inflammatory cells from the non-neural lesion core to the CNS parenchyma (Voskuhl et al., 2009; Sofroniew, 2015). Glial scars fill the interstitial spaces and induce the formation of new capillaries (Rolls et al., 2009). RA ablation impairs glial scar formation, leading to extensive infiltration of inflammatory cells and loss of neurons (Gu et al., 2019). Importantly, however, RA ablation also exerts an unwelcome inhibitory effect on axon regeneration (Anderson et al., 2016). CSPGs deposited in glial scars inhibit oligodendrocyte precursor cell differentiation and remyelination, the two most important processes underlying axon regeneration. CSPG inhibition or inactivation effectively improves motor function (Bradbury et al., 2002; Silver and Miller, 2004; Siebert et al., 2011; Lang et al., 2015; Tran et al., 2018). Wallerian degeneration of damaged axon protrusions leads to continuous extracellular deposition of axons and myelin debris. Myelin-related molecules (MAG, Nogo, OMGP), in conjunction with CSPGs, inhibit neuronal regeneration and neural plasticity (Sofroniew, 2018). However, the deletion of CSPG-related genes or CSPG receptor blockade only enhances synaptic remodeling and cannot directly overcome the protective effects of the astrocyte scar and lesion cores of non-neural tissue to produce meaningful spontaneous axonal regeneration (Hossain-Ibrahim et al., 2007; García-Alías et al., 2009). A combination of TGF-β1/2 antibodies reduced CNS scar formation in an adult rat model of brain injury; however, this was not accompanied by an increase in axon regeneration (Moon and Fawcett, 2001). GFAP^−/−vim^−/− mice show normal scar formation after TBI or SCI, but the scar density is low and accompanied by bleeding (Pekny et al., 1999). Three genetically targeted loss-of-function interventions—preventing astrocyte scar formation, attenuating scar-forming astrocytes, and ablating chronic astrocytic scars—all failed to promote spontaneous axon regrowth. However, exogenous administration of axon-specific growth factors, coupled with growth-activating priming injuries, stimulated axon regeneration, which was reversed by glial scar ablation (Anderson et al., 2016).

Glial scars transform into fibrous scars 14 dpi, and SAs are produced at the same time. SAs are known to originate from the interaction between RAs and type I collagen via the integrin/N-cadherin pathway. Antibodies targeting collagen-binding integrin and N-cadherin neutralizing antibodies both inhibited this process (Hara et al., 2017). Immunofluorescence analysis identified the presence of SOX9-positive nuclei in astrocytes of a wild-type brain scar 30 days after the cortical puncture. In contrast, SOX9 expression was strictly limited to the cytoplasm in the DBN−/− brain. DBN may also participate in the transformation of RAs into SAs (Schiweck et al., 2021). Inhibiting the RA/SA conversion may represent an ideal treatment for CNS injury. For this, the restrictive effect of RAs on inflammation should not be affected, only the formation of the glial scar boundary should be inhibited so as to alleviate the inhibitory effect of the surrounding environment on axon regeneration.

In summary, the dual role of the glial scar in axon regeneration may result from the low inherent regeneration potential of neurons. The growth-activating effect of the glial scar cannot bridge the gap between the neuronal regeneration potential and the physical hindrance represented by glial scars; when a glial scar is ablated, neurons cannot regenerate axons on their own without the growth-activating effect of the glial scar. Han et al. (2020) proposed to increase the intrinsic regenerative power of neurons by restoring cellular energy, and successfully promoted the germination and regeneration of axons after SCI by enhancing mitochondrial transport and energy metabolism. Therefore, in the case of preserving glial scars, enhancing the regeneration potential of neurons may also be a feasible treatment option.

Based on the dual role of astrocytes in CNS injury, multiple attempts have been undertaken to enhance the beneficial effects of astrocytes or reduce their harmful effects. Here, we mainly review the existing attempts at astrocyte-targeted therapy (Table 4).

The importance of astrocytes in CNS disease and injury is widely recognized; however, our understanding of astrocyte functions is still in its infancy. The continuous development and breakthrough of instruments and technologies provide conditions for accurate typing of astrocytes. The combination of single-cell and spatial transcriptome sequencing shows promise as a means of determining astrocyte heterogeneity after injury. Through the sequencing of several key times after injury, the time and space distribution of each astrocyte subpopulation can be determined. For example, astrocyte subpopulation D appears on the 7th day after SCI, mainly distributed in the core of injury. Further investigations to determine the temporal and spatial specificity of different astrocyte subpopulations with their specific genetic markers, thereby revealing their respective roles in injury, will provide a more precise indication to allow the targeting of specific astrocyte subpopulations for the treatment of CNS injuries. Such as the study of Hasel et al. (2021), in the mouse inflammation model, they divided astrocyte subgroups according to the difference between transcriptome and anatomical location and found that Cluster 8 is widely present in inflamed brains, but few in normal brains. In subsequent studies, treatment attempts can be made against Cluster 8 to inhibit the production of Cluster 8, or convert Cluster 8 into a neuroprotective or even neutral RAs subgroup to reduce inflammation. Although they have been proposed to express unique marker genes, little is known regarding the process involved in the transformation between RAs and SAs given that research attention has primarily focused on inflammation and glial scar formation after injury. In the absence of theoretical support, there is no way to talk about the treatment of targeted SA. As detailed in this review, clarifying how SAs are generated may provide ideal treatment and management options for CNS injuries. Based on the precise type of astrocytes, targeting harmful RA subgroups in the early stage of injury to reduce neuronal death and tissue destruction, and changing the extracellular matrix and reducing scar formation through the regulation of SA in the later stage to weaken the external inhibitory factors of nerve regeneration. This kind of treatment is worth looking forward to.

BN designed the research and revised the manuscript. YZ found some articles. GY wrote the article. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.