Astrocyte polarization in glaucoma: a new opportunity.

Astrocyte polarization is a new concept which is similar to microglia polarization and in which astrocytes are classified as A1 (neurotoxic) and A2 (neuroprotective). Several studies on astrocyte polarization have focused mainly on neurodegenerative diseases, trauma, and infections. However, the role of astrocyte polarization in glaucoma, a neurodegenerative disease, has not been fully explored. In this review, we first describe the characteristics of astrocyte astrogliosis in glaucoma, including morphological, molecular, proliferative and functional changes. We then summarize understanding of astrocyte polarization in other diseases, and show that A1 astrocytes are involved in the death of retinal ganglion cells in glaucoma, and that their neurotoxins kill only damaged retinal ganglion cells. Based on this, we propose new interesting conjecture on astrocyte polarization in glaucoma: (1) That the neurotoxin from A1 astrocytes is a product of the complement system (membrane-attacking complex), since this system is known to mediate synaptic elimination and the C3 expression is clearly increased in A1 astrocytes; (2) that reactive scar-forming astrocytes in the optic nerve head may be classified as A2 astrocytes since their ablation leads to a worse prognosis in glaucoma. Finally, current therapeutic research progress on astrocyte polarization in other diseases is also addressed. Regulation of astrocyte polarization can be achieved by extracellular microglia-related and intracellular pathways. Reduced A1 or increased A2 astrocytes can rescue the nerve. For example, glucagon-like peptide-1 receptor agonist rescues retinal ganglion cells by reducing A1 astrocytes via the extracellular microglia-related pathway in an ocular hypertension model, suggesting that regulation of astrocyte polarization as a therapeutic target in glaucoma is feasible.

Introduction

Glaucoma is a neurodegenerative disease characterized by retinal ganglion cell (RGC) death and axonal degeneration. The predominant risk factors for this disease include age and intraocular pressure (IOP) (Quigley, 1993). Currently, glaucoma is mainly treated by reducing IOP, however, the presence of normal-tension glaucoma indicates that there are other pathogenic mechanisms in addition to pathological ocular hypertension (OHT). Neuroinflammation is the main cause of persistent secondary nerve damage after alleviation of high IOP, and has recently attracted significant research attention.

However, neuroinflammation is not conventionally considered part of the inflammatory response, which involves only immune cells. Glial cells (microglia and astrocytes) are resident immune monitoring cells in neuroinflammation (Soto and Howell, 2014; Williams et al., 2017). Microglia have been more widely investigated than astrocytes, as they are considered to originate from tissue macrophages, thus exhibiting more immune effects (Ginhoux et al., 2010).

Astrocytes are mainly located in the retinal nerve fiber and ganglion cell layers and the optic nerve (Reichenbach and Bringmann, 2020). Notably, the optic nerve head (ONH) structures and astrocyte components differ between primates and rodents. In primates, the lamina cribrosa is composed mainly of collagen, with abundant astrocytes on the surface and lamina cribrosa cells inside the cribriform plates (Anderson, 1969; Hernandez, 2000; Hernandez et al., 2008). However, in rodents, the structure is known as the glial lamina and is almost entirely composed of astrocytes, without collagen (Johansson, 1987; May and Lütjen-Drecoll, 2002; Howell et al., 2007). Both laminal forms are consistent with the site of neuroinflammation in glaucoma, thus the role of astrocytes requires further exploration (Williams, et al., 2017). Astrocytes are also located in the distal visual pathways, including the lateral geniculate nucleus and visual cortex and are activated under conditions of high IOP, indicating that glaucoma is highly correlated with the central nervous system (Lam et al., 2009; Shimazawa et al., 2012; Fujishiro et al., 2020).

Current reviews on neuroinflammation in glaucoma are mainly focused on microglia, and the role of astrocyte polarization in glaucoma has not yet been reported. The purpose of this review is to explore the relationship between astrocyte polarization and glaucoma and explain the feasibility of astrocyte polarization as a therapeutic target for neuroprotection and neural regeneration.

Retrieval Strategy

Articles published from January 1980 to November 2021 were retrieved from PubMed in this narrative review. We used the following search strategy: (A1 astrocyte OR A2 astrocytes OR Astrocytes OR Neuroinflammation) AND (Glaucoma OR Neurodegenerative Disease OR Optic Nerve OR Retina). The final retrieval date was November 29, 2021. The results were screened based on titles and abstracts, and only studies conducted on rodents and primates and published in English were included.

Reactive Astrogliosis

Previous studies reported that resting astrocytes are activated into reactive astrocytes in response to injury through a process known as “reactive astrogliosis”. The process comprises four major changes () (Sofroniew, 2009), which are described below.

Reactive astrogliosisthe in glaucoma and regulation of astrocyte polarization.

Reactive astrogliosis brings morphological, proliferative, molecular, and functional changes in glaucoma. Astrocyte functional polarization is the main functional change. Regulation of astrocyte polarization is a new opportunity for glaucoma therapy through the extracellular microglia-related pathway (a) and intracellular pathway (b). C1q: Complement component 1q; C3: complement component 3; IL: interleukin; RGCs: retinal ganglion cells; S100A10: S100 calcium binding A10; TGF-β: transforming growth factor-β; TNF-α: tumor necrosis factor-α.

Morphological changes

Astrocytes are grouped into two main types based on their morphology including protoplasmic astrocytes in the retina and fibrous astrocytes in the ONH (Büssow, 1980). The two types are in contact with blood vessels and neurons. However, protoplasmic astrocytes are closer to neuronal cell bodies and synapses, whereas fibrous astrocytes are in contact with the nodes of Ranvier on the neuronal axons (Allen and Barres, 2009).

Reactive protoplasmic astrocytes in the retina constrict themselves but the structure of the spatial domain is maintained (). This situation may be implicated in disrupting neuronal homeostasis and the blood-eye barrier.

Astrocyte in the stretched preparation of retina labelled by anti-GFAP antibody (green).

This figure is the author’s unpublished data. An OHT mouse model was induced by injecting microbeads into the anterior chamber. Completely untreated mice were included in the control group. Retina was prepared by immunofluorescence. Figures were taken with confocal microscope, each magnified 20×. The expression of GFAP is stronger in eyes with OHT. Astrocytes in eyes with OHT become thinner and their secondary processes are decreased, resulting in reduced synaptic and vascular. GFAP: Glial fibrillary acidic protein; OHT: ocular hypertension.

Reactive fibrous astrocytes in the ONH undergo biphasic morphological changes, based on axonal changes. Astrocytes retract their primary processes and redistribute to the edge of the nerve to reduce their spatial coverage during early axon expansion (Cooper et al., 2016). In addition, their fine branches are reoriented (Tehrani et al., 2014). Once axons are lost, astrocytes partially re-extend long processes and restore their uniform distribution from the center to the edge to resume near-normal morphology and an extensive spatial overlap (Sun et al., 2010; Bosco et al., 2016; Cooper et al., 2018). Astrocytes close to the myelination transition zone extend characteristically glaucomatous longitudinal processes (Wang et al., 2017).

Molecular changes

Reactive astrocytes upregulate “Pan-reactive” molecules including glial fibrillary acidic protein (GFAP), vimentin and lipocalin-2, representing the common overlap molecule under neuroinflammation and ischemia (Zamanian et al., 2012). Structurally, GFAP is a cytoskeleton molecule (Middeldorp and Hol EM, 2011) which exhibits increased expression in reactive astrocytes of most species, thus it is often used to label astrocytes to evaluate their morphological changes (Wang et al., 2000, 2002; Ramírez et al., 2010; Gallego et al., 2012). Notably, Müller cells also express GFAP (Ramírez et al., 2010; Gallego et al., 2012). In addition, actin can indicate finer branches that GFAP cannot indicate (Tehrani et al., 2014). Structural molecular changes of ONH astrocytes in response to high IOP are mediated by integrin signaling (Tehrani et al., 2016).

Moreover, metabolic molecules in astrocytes undergo redistribution. For example, the transfer of glucose is driven by a concentration gradient, from the non-OHT contralateral eye with a high concentration to the OHT eye with a high demand to support nerve function (Cooper et al., 2020). Astrocytes are highly interconnected by gap junctions which are composed of the protein connexin-43 (Cx43), thus functioning as a broader network (Boal et al., 2021). In OHT models, Cx43-mediated astrocyte network transfers metabolic resources from a non-OHT eye to an OHT eye. Although this may improve axonal function and vision in the OHT eye, it renders the non-OHT eye vulnerable to biological energy stress (Cooper et al., 2020).

Proliferation

Reactive astrogliosis is a cascade based on the the severity of pathophysiology. Proliferation of astrocytes is initiated by severe irritations, such as chronic neurodegenerative injury, diffuse trauma, diffuse ischemia, or some types of infection (Sofroniew, 2009, 2014). One study using a chronic glaucoma model reported that reactive non-proliferative astrogliosis is prevalent in the retina (Inman and Horner, 2007). Astrocytes proliferate and participate in the remodeling of ONH structure (Hernandez, 2000; Lozano et al., 2019). The explanation may be that the retina, as a whole, distributes the OHT evenly, whereas the ONH as a structurally different region of the retina may be subjected to more pronounced pressure. Significant biomechanical deformation appears in the ONH. Therefore, structural remodeling and astrocytes proliferation can cope with this deformation.

Functional changes

Functions of resting astrocytes include (1) regulation of the concentrations of ions and neurotransmitters to ensure internal environment balance, (2) nerve nutritional support, (3) synaptic formation, and (4) as a member of the neurovascular unit to form the second line of the blood-retinal barrier with the parenchymal basement membrane (Liddelow and Barres, 2015; Lefevere et al., 2020). However, reactive astrocytes are functionally polarized in neuroinflammation. Reactive astrogliosis in lipopolysaccharide- and stroke-induced neuroinflammation models exhibit opposite pro- and anti-inflammatory effects, respectively. Moreover, genomic analysis based on transcriptome databases shows two subtypes of reactive astrocytes named A1 and A2 phenotypes (Zamanian et al., 2012; Liddelow et al., 2017).

Astrocyte Polarization

A1 astrocytes

A1 astrocytes are mainly induced by interleukin-1α (IL-1α), tumor necrosis factor-α (TNF-α), and complement component 1q (C1q) secreted by microglia in the neuroinflammatory model, all of which are indispensable (Liddelow et al., 2017). Mitochondrial fragments released by microglia (Joshi et al., 2019) or pre-treated with interleukin-18 (IL-18) in vitro (Hou et al., 2020) can trigger astrocytes to be polarized into the A1 phenotype which is harmful to neurons.

A1 astrocytes lose the normal functional characteristics including phagocytosis, glutamate absorption and cultivate synaptic formation (Liddelow et al., 2017; Barbar et al., 2020; Li et al., 2020). Liddelow et al. (2017) reported that most or all RGCs died after culture with A1 astrocytes. This and other research indicates that A1 astrocytes may secrete unknown neurotoxins, with similar toxic effects on cortical neurons, embryonic spinal motor neurons and oligodendrocytes (Liddelow et al., 2017; Barbar et al., 2020) mainly via induction of neuronal apoptosis (Liddelow et al., 2017).

In addition, A1 astrocytes upregulate several genes previously shown to be synapse-destructive (Stevens et al., 2007) such as complement component 3 (C3), a specific marker for A1 astrocytes (Liddelow et al., 2017; Conley et al., 2019). C3 of astrocytes is a downstream protein of C1q which is one of the three main factors secreted by microglia, participates in the complement cascade and mediates synaptic elimination during development of the central nervous system (Stevens et al., 2007). Therefore, complement cascades are abnormally activated to eliminate normal synapses when reactive astrocytes are polarized to A1 astrocytes. This may explain why A1 reactive astrocytes exhibit decreased synaptic function (Liddelow et al., 2017).

The neurotoxins secreted by astrocytes have not been fully elucidated to date. Membrane-attacking complex, the end product of the complement system, plays a bystander role to kill neurons and oligodendrocytes (Tradtrantip et al., 2017; Duan et al., 2019). In addition, inhibition of C3 released from A1 astrocytes reduces neuronal injury (Hou et al., 2020; Gharagozloo et al., 2021). This implies that the neurotoxin from A1 astrocytes is membrane-attacking complex.

A2 astrocytes

A2 astrocytes are predominantly induced in ischemia and acute trauma models (Anderson et al., 2016). Recent studies reported that primary astrocytes subjected to oxygen/glucose deprivation are polarized to the A2 phenotype (Su et al., 2019). Interleukin (IL)-4 and IL-10 can also induce A2-like astrocytes in vitro (Chistyakov et al., 2020). Notably, A2 astrocytes can promote neuronal survival as well as tissue repair (Liddelow et al., 2017).

A2 astrocytes retain phagocytic function (Morizawa et al., 2017) and can secrete substances that exhibit neuroprotective effects. For example, transforming growth factor-β (TGF-β) is an anti-inflammatory cytokine involved in synaptic formation (Diniz et al., 2014; Kang et al., 2014; Norden et al., 2014). Thrombospondins, secreted by A2 astrocytes, play a crucial role in synaptic plasticity and brain repair (Liauw et al., 2008; Zamanian et al., 2012). Moreover, estrogen is also beneficial to synapses (Hu et al., 2007) and plays a neuroprotective role by decreasing formation of A1 astrocytes or promoting polarization to A2 astrocytes (Wang et al., 2020, 2021).

S100 calcium binding A10 (S100A10) is significantly upregulated in A2 astrocytes and often used as a specific marker for this phenotype (Zamanian et al., 2012). S100A10 is also called the Annexin II ligand (Marenholz et al., 2006), as it can form a heterotetramer with the peripheral membrane-binding protein, Annexin A2, fundamental to cell proliferation, membrane repair, and inhibition of apoptosis (Hsu et al., 1997; Li et al., 2011; Rezvanpour et al., 2011; Liddelow et al., 2017; Conley et al., 2019). It has been proposed that S100A10 may negatively regulate the toll-like receptor signaling pathway independent of S100A10-Annexin A2 heterotetramer to inhibit downstream signaling and cytokine production, thus playing an anti-inflammatory role (Lou et al., 2020).

Heterogeneity

Heterogeneity of microglia

Microglia, as an another glial cell involved in neuroinflammation, was introduced to the concept of polarization earlier than astrocytes. Reactive microglia were initially thought to exhibit two opposite activation states (Gordon, 2003). To distinguish these, the term “M1/M2” originally derived from the macrophage is used to describe classically activated pro-inflammatory (M1) and alternatively activated anti-inflammatory (M2) microglia. Further studies reported that the M2-like phenotype has at least three subtypes, namely, M2a, b and c, with unique functions (Franco and Fernández-Suárez, 2015). As research progressed, some transition subtypes were reported in addition to the clear dichotomy, such as disease-associated microglia characterized by ability to promote neurodegeneration in Alzheimer’s disease (AD), as well as SOD1G93A microglia implicated in significant induction of potentially neuroprotective and neurotoxic factors concurrently in amyotrophic lateral sclerosis (ALS) (Chiu et al., 2013; Keren-Shaul et al., 2017). Recent advances in transcriptomics have questioned whether the M1/M2 dichotomy is an oversimplification (Hume and Freeman, 2014; Ransohoff, 2016; Sousa et al., 2018).

Microglia are located in the retina and in the unmyelinated optic nerve. They undergo activation and proliferation during the early stages of chronic glaucoma models (Bosco et al., 2011, 2015). IOP elevation induces transformation of most microglia to their M1-like phenotype, whereas those in the contralateral retinas show multiple phenotypes after activation (Rojas et al., 2014). The function of microglia usually undergoes a dynamic process, manifested by changes in the ratio of M1 phenotype and M2 phenotype, even the existence of other transition subtypes. So function of microglia is mixed.

Heterogeneity of astrocytes: the possibility of subtypes other than A1/A2

A recent meta-analysis of astrocyte polarized transcriptomes was performed based on acute injury and chronic neurodegeneration (Das et al., 2020). The results elucidated that A1-related, A2-related, and pan-reactive genes are upregulated in astrocytes in most acute injury and chronic neurodegeneration mouse models. This indicates two different views: (1) A1 and A2 astrocytes may not be mutually exclusive and may exist as a continuum, their proportions associated with the degree and severity of damage. (2) There may be transition subtypes related to disease, requiring further clarification.

Astrocyte Polarization in Glaucoma

Presence of A1 and A2 astrocytes in glaucoma

There are several reports about astrocyte polarization in glaucoma (). A1 astrocytes are induced rapidly in the optic nerve crush model, as well as in the microbead high intraocular pressure model (Liddelow et al., 2017; Guttenplan et al., 2020). C3-expressed astrocytes have also been reported to exist in other models, such as laser cauterization and genetic glaucoma models (Kuehn et al., 2006; Harder et al., 2017). Moreover, the three cytokines from microglia that induce A1 polarization, IL-1α, TNF-α, and C1q, are maintained at high levels to ensure existence of A1 astrocytes after the IOP returns to normal (Sterling et al., 2020). Persistent death of RGCs depends on neuronal injury and the presence of reactive A1 astrocytes (Guttenplan et al., 2020).

A2 astrocytes have not yet been reported in glaucoma. However, quantitative polymerase chain reaction results of A2-specific genomes including Clcf1, Ptx3, B3gnt5, Cd14 and Ifi202b, indicate that A2 astrocytes may be present in the retina and ONH in the chronic hypertension model. This finding provides a basis for further studies on A2 astrocytes in glaucoma (Guttenplan et al., 2020).

Subtype classification of scar-forming astrocytes

In general, activated astrocytes are classified into two categories: reactive astrocytes (RAs) and scar-forming astrocytes (SAs). Reactive astrogliosis is a hierarchical continuum of progressive changes in gene expression and cellular changes. Resting astrocytes undergo reactive astrogliosis to form RAs. Subsequently, RAs overlap their process and form glia scar with SAs (Sofroniew, 2009). Further, these two cell types exhibit different characteristic phenotypes (Hara et al., 2017). Studies report that RAs express Plaur, Mmp2, Mmp13, Axin2, and genes associated with the β-catenin-MMP migration signaling pathway. In contrast, SAs upregulate expression of Cdh2, SOX9, axon-directed rejection gene (Slit2), and chondroitin sulfate proteoglycan related genes (CSPG), including Xylt1, Chst11, Csgalnact1, Acan, and Pcan.

Several studies have emphasized that it is inadequate to describe only RAs in astrocyte polarization exploration. Initially, SAs were thought to inhibit axon regeneration by physically forming glial scars with regenerated axons forming so-called dystrophic endbulbs or by chemically blocking axon regeneration (Liuzzi and Lasek, 1987; Silver and Miller, 2004). Further studies explored the beneficial effects of SAs, which can repair the blood-brain barrier, prevent overwhelming inflammatory response and white blood cell infiltration, and limit cellular degeneration (Bush et al., 1999; Faulkner et al., 2004; Sofroniew, 2015). In addition, SAs up-regulate two CSPGs rRNAs (CSPG4 and CSPG5) that supports axon growth, thus contributing to axon regeneration (Li et al., 2018).

Furthermore, A2 astrocytes show stronger expression of collagen (Col6a1 and Col12a1) than A1 astrocytes and signal transducer and activator of transcription 3 (STAT3). The latter is a key regulator of reactive astrogliosis and scar formation, implying that A2 astrocytes are more likely to seal dying tissue and form glial scars. In addition, ablation of SAs aggravates neurodegeneration (Bush, et al., 1999; Anderson, et al., 2016).

Several studies using various glaucoma models reported increased STAT3 expression in astrocytes in the retina and ONH (Zhang et al., 2013; Wong et al., 2015; Sun et al., 2017; Lozano, et al., 2019). In glaucoma, proliferative astrogliosis occurs mainly in the ONH (Lozano et al., 2019), and the expression of STAT3 is higher in the optic nerve head than in the retina (Wong et al., 2015; Sun et al., 2017). Therefore, after ablating SAs, ONH remodeling was not apparent, and increased loss of RGCs and worse visual function were observed (Sun et al., 2017). Further studies should be conducted to categorize SAs.

Astrocyte Polarization Regulation as a Potential Therapeutic Direction

Current therapeutic strategies for neuroinflammation mainly target microglia and are aimed at reducing infiltrating immune cells. Regulation of astrocyte polarization to promote the protective effect of A2 astrocytes and reduce the adverse effect of A1 astrocytes is a potential alternative therapy approach for treatment of neuroinflammation. Regulation of astrocyte polarization can be achieved via the extracellular microglia-related pathway and the intracellular pathway ().

Extracellular microglia-related pathway

Recent research has indicated that the cascade between microglia and astrocytes strongly influences the course of disease, such as Parkinson’s disease (PD) and glaucoma (Liddelow and Barres, 2017; Lee et al., 2019; Wei et al., 2019). Microglia indirectly kill neurons in a neuroinflammatory response by activating downstream astrocytes to form the A1 phenotype, which secrete unknown neurotoxins (Liddelow et al., 2017). Glucagon-like peptide-1 receptor (GLP1R) agonist, NLY01, acts on astrocyte polarization via the extracellular microglia-related pathway in PD and glaucoma models (Yun et al., 2018; Sterling et al., 2020). This finding implies that regulation of this pathway may inhibit formation of A1 astrocytes to achieve a neuroprotective outcome. Moreover, M1 microglia mainly induce A1 astrocytes through three main factors, whereas M2 microglia induce A2 astrocytes (Wei et al., 2019) so the ratio of M1 to M2 microglia also directly affects astrocyte polarization. Current neuroprotective approaches in other diseases via the extracellular microglia-related pathway mainly focus on three strategies: (1) Inhibition of M1 microglia activation to reduce the three factors and thus the number of A1 astrocytes; (2) Polarization of microglia to M2 subtype so that astrocyte polarization is skewed toward the A2 subtype; (3) Inhibition of inflammatory factors from microglia alone, such as NLRP3-inflammasome, to reduce A1 astrocytes ().

Intracellular pathway

It has been demonstrated that A1 astrocyte activation does not always take place through the classic extracellular microglia pathway, and that microglia activation in some models occurs after the polarization of A1 astrocytes (Clark et al., 2019; Carroll et al., 2020). This implies that intracellular pathways may be more inclusive targets (Sofroniew, 2009).

The nuclear factor kappa-B (NF-κB) signaling pathway is involved in expression of GFAP and pro-inflammatory mechanisms in astrogliosis (Sofroniew, 2009). Notably, astrocytes activated through this pathway in neurodegenerative diseases such as AD, ALS, multiple sclerosis, and Huntington’s disease exert harmful effects in neuroinflammation (Migheli et al., 1997; Brambilla et al., 2012; Carrero et al., 2012; Hsiao et al., 2013). In human glaucoma, the immunoreactivity of NF-κB occurs in the GFAP-positive astrocytes in the ONH and retina (Agapova et al., 2006), whereas A1 astrocytes have significantly increased levels of NF-κB gene (Zamanian et al., 2012; Lian et al., 2015). This finding indicates that activation and function of A1 astrocytes are highly correlated with the NF-κB pathway.

Previous studies reported that inhibition of the NF-κB signal plays a protective role following injury (Brambilla et al., 2005, 2012). In addition, pharmacological therapies such as MFG-E8 for AD, Rolipram for SCI and NF-κB inhibitor (Bay11-7082) for abrogation of infection, as well as cellular therapies including mesenchymal stem cell transplantation in spinal cord injury, can reduce A1 astrocytes by inhibiting expression of NF-κB (Wang et al., 2018; Xu et al., 2018; Jin et al., 2019; Vismara et al., 2020). Regulation of astrocyte polarization through the intracellular pathway focuses on two strategies, namely inhibition of astrocyte activation to A1 subtype and induction of astrocyte polarization to A2 subtype (), hinting at the mechanism of astrocyte polarization in glaucoma.

Limitations

Although astrocyte polarization has been widely researched in recent years, it remains relatively poorly understood in glaucoma. In this review, the characteristics of astrocyte polarization in the visual nervous system are inferred based on knowledge of astrocyte polarization in the central and peripheral nervous systems. The study has some limitations.

First, while the visual nervous system and the central nervous system (brain and spinal cord) have broadly similar cell populations, including neuronal cells and glial cells, their organizational structures, cell connections and interactions differ. Therefore, astrocyte polarization in glaucoma and the central nervous system may not be directly comparable. Second, although glaucoma is considered a neurodegenerative disease, its pathogenesis differs from that of neurodegenerative diseases in the central nervous system. Therefore, the role of astrocyte polarization in glaucoma may be disease-specific.

Conclusions

Astrocyte polarization is a novel concept that has not fully been reported in glaucoma. Further studies are required to fully explore the concept of astrocyte polarization as a new potential therapeutic strategy for the treatment of glaucoma. The following research directions may help to further understand the role of astrocyte polarization in glaucoma.

(1) Drug development programs should target astrocyte polarization. It is imperative to explore whether known astrocyte polarization-related drugs in other diseases are also effective in glaucoma, or whether anti-glaucoma drugs exert their effect by regulating astrocyte polarization.

For example, omega (n)-3 polyunsaturated fatty acids (PUFAs) reportedly attenuate adverse outcome in other diseases by balancing the A1/A2 phenotype of astrocyte through both extracellular microglia-related pathway and intracellular increased mitophagy (Cao et al., 2021; Gu et al., 2021). In glaucoma, dietary deficiency of omega-3 PUFAs contributes to RGC damage in glaucoma (Nguyen et al., 2013). Dietary supplementation with omega-3 PUFAs can reduce IOP by increasing aqueous outflow and thus has neuroprotective effects on RGCs (Nguyen et al., 2007; Kalogerou et al., 2018; Lafuente et al., 2021). Although its specific mechanism of neuroprotection remains unclear in glaucoma, regulating astrocyte polarization may be an important pharmacological mechanism.

(2) The unknown toxins secreted by A1 astrocytes have to date exhibited harmful effects only on damaged neurons. Therefore, studies should explore the mechanism by which A1 astrocytes identify damaged neurons.

They may recognize damaged RGCs via transmitography, a process of transcellular mitochondria degradation (Davis et al., 2014).

Aged mitochondria appear in the process of powering signal transmission. Then they are retrograde transport from distant synapses to the cell body and finally degraded by mitophagy (Sheng, 2014). Those that exceed the retrograde transport capacity gather in the axons near the astrocyte membrane sites. At this place they form membrane-binding excreta, which are internalized and degraded by astrocytes.

Elevated IOP increases mitochondria fission in RGCs (Ju et al., 2008). Optineurin is an autophagy receptor (Wong and Holzbaur, 2014; Lazarou et al., 2015), mutations of which in normal tension glaucoma can lead to impaired mitophagy and increased mitochondrial fission (Rezaie et al., 2002; Shim et al., 2016). The increased mitochondrial fragments and the impaired mitophagy in RGCs may lead to the increased mitochondrial fragments in transmitophagy, which may be the signal for astrocytes to recognize damaged RGC.

(3) Astrocyte polarization is a potential target for the prevention of glaucoma in people with OHT and without glaucomatous neurodegeneration. Activation of astrocytes precedes axonal injury and A1 astrocytes do not cause damage to unimpaired RGCs, thus it is possible that A1 astrocytes are present in these patients (Guttenplan, et al., 2020) and may explain their predisposition to glaucoma. Prophylactic downregulation of A1 astrocytes or adjustment of the ratio between activated A1 and A2 astrocytes may effectively reduce the incidence of glaucoma in patients with high IOP. This would be a major breakthrough since currently there is no effective preventive therapeutic target for glaucoma.

Table 1

The timeline of studies related to astrocyte polarization in glaucoma

Studies	Model	Conclusion
Zamanian et al., 2012	Neuroinflammation model (intraperitoneal injection with LPS) and stroke model (middle cerebral artery occlusion)	Reactive astrocytes show harmful effects in neuroinflammatory models, whereas they tend to be beneficial in ischemic models
Liddelow et al., 2017	In vitro and ONC model	(1) Neuroinflammation-induced reactive astrocytes were firstly named “A1” and ischemia-induced reactive astrocytes named “A2”.
		(2) The neurotoxin secreted by A1 astrocytes has an RGC mortality rate at or close to 100%
Sun et al., 2017	Pressure sensing model (acute injury), microbead occlusion model (chronic injury) and ONC model	(1) Reactive astrocytes in the ONH show a protective response.
		(2) STAT3 signaling is important in protective reactive astrocytes in the ONH.
Livne-Bar et al., 2017	Excitotoxic damage model (acute injury) and circumlimbal suture model (chronic injury)	Protective reactive astrocytes secrete lipoxins LXA4 and LXB4 to rescue RGCs in both acute and chronic injury
Guttenplan et al., 2020	ONC model (acute injury) and microbead occlusion model (chronic injury)	(1) A1 astrocytes kill RGCs in both acute and chronic injury of glaucoma.
		(2) Only a combination of injury and toxin from A1 astrocytes can initiate the RGC death procedure.
Sterling et al., 2020	Microbead occlusion model	GLP1R agonist NLY01 inhibits three factors (IL-1 α, TNF-α, and C1q) to reduce A1 formation for neuroprotective purposes.
Yang et al., 2020	Microbead occlusion model	NF-κB signaling is important in neuroinflammation in glaucoma and has potential as an astrocyte treatment target to protect RGCs.
Guttenplan et al., 2021	ONC model	Saturated lipids contained in APOE and APOJ lipoparticles from A1 astrocytes kill RGCs.

APOE: Apolipoprotein E; APOJ: apolipoprotein J; C1q: complement component 1q; GLP1R: glucagon-like peptide-1 receptor; IL-1α: interleukin-1α; LPS: lipopolysaccharide; NF-κB: nuclear factor kappa-B; ONC: retro-orbital nerve crushes; ONH: optic nerve head; RGCs: retinal ganglion cells; TNF-α: tumor necrosis factor-α.

Table 2

Influence of astrocyte polarization through the extracellular microglia-related pathway

Mechanism	Disease	Animal	Method	Reference
Inhibition of NLRP3-inflammasome released from microglia	MS	Mice	MCC950 an inhibitor of the NLRP3 inflammasome	Hou et al., 2020
Inhibition of microglia activation (to M1 subtype)	MS	Mice	ACDT	Kuo et al., 2018
	TBI	Mice	Bexarotene a highly selective RXR agonist	He et al., 2018
	SCI	Rats	BMSCs-Exos	Liu et al., 2019
	CPSP	Rats	Minocycline a non-specific microglial inhibitor	Li et al., 2020
	Ischemic stroke injury	Rats	Cottonseed oil	Liu et al., 2020
	TBI	Mice	Estrogen	Wang et al., 2021
	SCI	Mice	Methylprednisolone	Zou et al., 2021
Regulation of microglia polarization to M2 subtype	Stroke	Rats	TMS	Hong et al., 2020; Zong et al., 2020a, b
	SCI	Mice	The NEAT1/miR-224-5p/IL-33 axis	Liu et al., 2021

ACDT: 5-Amino-3-thioxo-3H-(1, 2) dithiole-4-carboxylic acid ethyl ester; BMSCs-Exos: exosomes derived from bone mesenchymal stem cells; CPSP: chronic post-surgical pain; MS: multiple sclerosis; NLRP-3: nod-like receptor protein 3; SCI: spinal cord injury; TBI: traumatic brain injury; TMS: transcranial magnetic stimulation.

Table 3

Mechanisms other than the The nuclear factor kappa-B (NF-κB) signal pathway that can regulate astrocyte polarization through the intracellular pathway, including physical and physiological changes such as cytokine increase and regulatory protein expression

Mechanism	Disease	Animal	Method	Reference
Inhibit astrocytes activation to A1 subtype	Infrasound Exposure	Rats	FGF2/FGFR1 pathway ↑	Zou et al., 2019
	MS	Mice	TUDCA directly prevents A1 polarization in a dose-dependent manner	Bhargava et al., 2020
	Depression	Mice	IL-10 ↑	Zhang et al., 2020
	ALS	Mice	TDP-43 ↑	Peng et al., 2020
	In vitro	Rats	TGFβ3 ↑	Gottipati et al., 2020
	Pain	Rats	Ac2-26 ↑ a mimetic peptide of Annexin-A1	Luo et al., 2020
	PD	Mice	kir6.2 ↓	Song et al., 2020
	Subarachnoid hemorrhage	Mice	Ponesimod ↑ an S1PR1 specific modulator	Zhang et al., 2021
Induce astrocyte polarization to A2 subtype	PD	Mice	Cx30 ↑	Fujita et al., 2018
	PD and Subarachnoid hemorrhage	Rats	PK2 ↑	Ma et al., 2020
	TBI	Mice	Pentraxin 3 ↑	Zhou et al., 2020
	Chronic cerebral hypoperfusion	Mice	Trkβ signaling ↑	Miyamoto et al., 2020
	Rats	Physical exercise ↑	Jiang et al., 2021
	SCI	Mice	miR-21 ↓	Su et al., 2019; Zhang et al., 2021

↑: Up-regulation; ↓: down-regulation. ALS: Amyotrophic lateral sclerosis; Cx30: connexin 30; FGF2: fibroblast growth factor 2; FGFR1: fibroblast growth factor receptor 1; IL-10: interleukin 10; miR-21: microRNA 21; PD: Parkinson’s disease; PK2:prokineticin-2; S1PR1: sphingosine-1-phosphate receptor 1; SCI: spinal cord injury; TBI: traumatic brain injury; TDP-43: TAR DNA binding protein 43; TGFβ3: transforming growth factor beta 3; Trkβ: tyrosine receptor kinase β.