Reactive Astrocytes and Emerging Roles in Central Nervous System (CNS) Disorders

Abstract

In addition to their many functions in the healthy central nervous system (CNS), astrocytes respond to CNS damage and disease through a process called “reactivity.” Recent evidence reveals that astrocyte reactivity is a heterogeneous spectrum of potential changes that occur in a context-specific manner. These changes are determined by diverse signaling events and vary not only with the nature and severity of different CNS insults but also with location in the CNS, genetic predispositions, age, and potentially also with “molecular memory” of previous reactivity events. Astrocyte reactivity can be associated with both essential beneficial functions as well as with harmful effects. The available information is rapidly expanding and much has been learned about molecular diversity of astrocyte reactivity. Emerging functional associations point toward central roles for astrocyte reactivity in determining the outcome in CNS disorders.

Astrocytes exert many essential functions in the healthy central nervous system (CNS) as reviewed and discussed in other articles in this collection and elsewhere (Allen and Eroglu 2017; Verkhratsky and Nedergaard 2018; Khakh and Deneen 2019).⁶ In addition, astrocytes respond to all forms of CNS damage, infection, and disease with a variety of potential changes in molecular expression, cellular structure, and function—commonly referred to as astrocyte “reactivity.” Emerging evidence demonstrates that astrocytes play important roles in most if not all CNS disorders.

The last decade of astrocyte research has led to exciting findings related to the complexity of astrocyte reactivity. The application of transcriptomics in preclinical models and human brain to understand the diverse astrocyte response to insult and new genetic tools to target astrocytes have shepherded in a new era of astrocyte research. Modern culture and organoid systems have further provided for exciting discoveries about function and evolutionary conservation of astrocyte reactivity. An amazing array of tools and data are now available to the budding astrocyte biologist. These advances in technology have brought new researchers to the field, seeded and expanded exciting cross-discipline collaborations that have uncovered a great deal about astrocytes in health and disease. Here, we update a previous report in this series (Sofroniew 2015b) and provide an overview of currently available information about the mechanisms, functions, and impact of astrocyte reactivity, with a particular focus on recent advances in defining the heterogeneity and regulation of astrocyte reactivity and their diverse roles in different CNS disorders.

WHAT ARE REACTIVE ASTROCYTES?

Astrocyte reactivity can now be defined as a spectrum of potential molecular, cellular, and functional changes in astrocytes that occur in response to all forms and severities of pathology in surrounding CNS tissue, as evidenced by numerous studies in both experimental animals and human pathological specimens, as well as in vitro cell-based assays. These pathological insults include microbial infection, ischemia, traumatic injury, autoimmune attack, seizure activity, exposure to environmental toxins, peripheral metabolic disorders, foreign bodies including medical implants, neurodegenerative disease, and neoplastic growth (Sofroniew and Vinters 2010; Pekny et al. 2016; Robel and Sontheimer 2016; Liddelow and Barres 2017; Campbell et al. 2020; O'Shea et al. 2020; Sofroniew 2020; Escartin et al. 2021; Han et al. 2021; Krawczyk et al. 2022). Box 1 summarizes the key features of astrocyte reactivity, which are discussed below. Various terms are sometimes used to refer to astrocyte responses to CNS damage or disease, and their usage may vary among authors. In this article, we will use “astrocyte reactivity” and “reactive astrocytes” as general, all-inclusive descriptors of all forms of astrocyte responses to non-cell-autonomous cues associated with CNS damage or disease. As discussed in more detail later, these terms encompass astrocyte responses of considerable diversity and heterogeneity, and considerable effort is now aimed at identifying different forms of astrocyte reactivity.

BOX 1. WORKING DEFINITION AND KEY FEATURES OF ASTROCYTE REACTIVITY.

• Astrocyte reactivity occurs in response to all forms of CNS injury, infection, and disease.

• Astrocyte reactivity is a diverse set of changes to normal physiological function. This is largely defined by measurements of changes in gene expression and protein levels, morphology, and other molecular changes to lipids, metabolites, etc.

• Astrocyte reactivity-related changes can alter astrocyte activities through both loss and gain of functions.

• Astrocyte reactivity-related changes are context dependent and may vary with the type and severity of insult, the time span after an insult, the types and combinations of molecular triggers, the CNS location, and may depend on the heterogeneous nature of individual astrocytes. This may occur within a single cell, or across a population of cells in a specific CNS region.

• Astrocyte reactivity-related changes are regulated in a context-specific manner by combinatorial interactions of diverse inter- and intracellular signaling molecules.

• Astrocyte reactivity-related changes can be homeostatic and adaptive and can influence disorder outcomes in beneficial ways.

• Astrocyte reactivity-related changes have the potential to be maladaptive and influence disorder outcomes in detrimental ways.

• Diseased astrocytes should be differentiated from reactive astrocytes. One should refer to either “diseased” astrocytes (caused by genetic mutations), or to “reactive” astrocytes (responding to an external stimuli).

ASTROCYTES AS A NEXUS FOR MULTICELLULAR RESPONSES TO CNS INSULTS

While this article focuses on astrocyte reactivity, it is essential to note that astrocyte responses occur in the context of coordinated multicellular responses to CNS insults. Different types of glia, including microglia, astrocytes, NG2-positive oligodendrocyte progenitors, respond most robustly. Non-neural cells intrinsic to the CNS, such as endothelia, perivascular fibroblasts, pericytes, and meningeal cells also respond to insults. Additionally, blood-derived cells and molecules such as leukocytes, platelets, and blood serum proteins, fatty acids, and high concentrations of other serum components like glutamate and K⁺ enter the CNS and contribute to these responses. Together, these cells produce intercellular signaling molecules that can influence the activities and functions of different cell types, including astrocytes. Thus, the response to CNS insults is a complex mixture of events involving multiple cell type interactions that change over time. Astrocytes take part in these interactions both by receiving and sending instructive signals that influence other cells (Tables 1 and 2).

Table 1.

Examples of extracellular triggers of astrocyte reactivity

Categories	Source	Molecules
Blood and serum proteins and molecules	Bloodstream	Albumin, fibrinogen, thrombin, complement, fatty acids, K+, glutamate
Cytokines and growth factors	Other glia and local nonneural cells, infiltrating leukocytes, tumor cells	CNTF, C1Q, EGF, EDN1, FGFs, IL-1α, IL-1β, IL-6, IL-10, INF-β, INF-γ, LIF, SHH, TGF, TNF
Damage-associated molecular patterns (DAMPs)	Cell damage	ATP, HMGB1, nitric oxide, reactive oxygen species (ROS)
Degeneration-associated proteins	Neurodegenerative disorders	α-Synuclein, β-amyloid, mutant-Huntingtin, prions, Tau, among others
Environmental toxins	External environment	Amphetamine, herbicides, insecticides, methamphetamine, MDMA, MPTP
Foreign bodies	Medical implants, traumatic injuries	Cationic surfaces
Hypoxia and metabolic stress	Ischemia	Oxygen deprivation, glucose deprivation
Inorganic molecules	Systemic metabolic toxicity (liver failure)	Ammonium (NH4+)
Mechanical stretch	Acceleration, blast, compression deceleration	Stretch receptors
Oxidative stress	Ischemia, metabolic stress	H2O2, Free radicals, NO
Pathogen-associated molecular patterns (PAMPs)	Microbial infections	Bacterial (e.g., lipopolysaccharide [LPS]), prions, toxoplasma, viral (e.g., dsRNA), yeast (e.g., zymosan)
Transmitters	Local neurons	Glutamate, noradrenalin

Abbreviations for growth factors, cytokines, chemokines, receptors, and transcription factors are per standard nomenclature (Human Gene Compendium, GeneCards, www.genecards.org). See main text for literature references.

Table 2.

Examples of effector molecular released by reactive astrocytes

Categories	Molecules
Chemokines	CCL2, CCL5, CCL7, CCL8, CXCL1, CXCL9, CXCL10, CXCL12, CXCL16
Cytokines	CNTF, IFN-γ, IL-1β, IL-6, IL-11, IL-15, LIF, TGF, TNF
Extracellular matrix	BCAN, collagens, MMP3, NCAN, SEMA4A, TIMP1
Growth factors and other proteins	BDNF, BMPs, CRYAB, FGF2, GDNF, GPC4, GPC6, NGF, PTX3, THSP1, VEGFA
Small molecules	ATP, lipids, nitric oxide (NO), prostaglandin E (PGE)
Transmitters	Glutamate, kynurenic acid (KYN), D-serine

Abbreviations are as per standard nomenclature (Human Gene Compendium, GeneCards, www.genecards.org). See main text for literature references.

HETEROGENEITY OF ASTROCYTES AND REACTIVE ASTROCYTES

The concept of astrocyte heterogeneity is not new. Classical neuroanatomists of the late nineteenth and early twentieth centuries described multiple types of astrocytes based on morphologies that were common across mammals, including the well-known bushy astrocytes in gray matter, fibrous astrocytes in white matter, and glia limitans astrocytes along meninges, as well as others (Verkhratsky and Nedergaard 2018). Recent and ongoing studies are expanding and correlating information about structural, genetic, and functional diversity of astrocytes across the healthy CNS (Khakh and Sofroniew 2015; Srinivasan et al. 2016; Zhang et al. 2016; Chai et al. 2017; Haim and Rowitch 2017; John Lin et al. 2017; Wu et al. 2017; Khakh and Deneen 2019; Batiuk et al. 2020; Falcone et al. 2021; Torres-Ceja and Olsen 2022) and are beginning to identify their developmental origins (Molofsky et al. 2014; Clavreul et al. 2019). In addition, there is a steadily growing interest in characterizing the diversity of astrocyte reactivity and understanding how it is regulated.

Historically, astrocyte reactivity has been treated as a uniform entity, but this is not the case. It is now appreciated to be a spectrum of potential changes, ranging from reversible alterations in gene expression and protein levels, to altered morphology, cell proliferation, and tissue rearrangement. Common features exist across different forms and intensities of astrocyte reactivity, but even these occur along a gradient that can vary immensely. Thus, astrocyte reactivity is complex, context-dependent, and multivariate phenomenon that depends on various factors such as the type, severity, and time frame of insult, the CNS location, and the types and combinations of molecular triggers driving the heterogeneity of individual reactive astrocytes.

Information is gradually accumulating that allows for the subdivision of astrocyte reactivity into different categories, which may exhibit common features or definable differences. It remains unclear to what extent specific substates of astrocyte reactivity are programmatically induced and share common molecular and functional features across multiple types of insults. Experimental measures to characterize differences among reactive astrocytes include discriminating proliferative from nonproliferative reactivity, characterizing differences in molecular signatures identified by transcriptomic and proteomic analyses, and identifying differences in physiological and functional changes as discussed in the following sections.

DISCRIMINATING ASTROCYTE DISEASE FROM ASTROCYTE REACTIVITY

There is an emerging concept that diseased astrocytes contribute to neural dysfunction and degeneration. As discussed in more detail below, genetic mutations or polymorphisms can cause cell-autonomous astrocyte dysfunction, leading to neuronal dysfunction and neurodegeneration. This type of disease-precipitated cell-autonomous astrocyte dysfunction should be distinguished from astrocyte reactivity, which is triggered by external non-cell-autonomous signals generated by dysfunctioning and/or degenerating neural tissue (Box 1). Current molecular markers used to study astrocytes in disorders, such as GFAP, do not readily differentiate between the two. Therefore, it is important to avoid conflating non-cell-autonomously triggered astrocyte reactivity with cell-autonomously-precipitated astrocyte disease based simply on the common expression of certain markers.

NONPROLIFERATIVE AND PROLIFERATIVE ASTROCYTE REACTIVITY

In the healthy adult CNS, astrocytes are postmitotic cells that rarely divide (Wanner et al. 2013). Nonproliferative astrocyte reactivity typically occurs in neural tissue that is not overtly damaged and retains its basic tissue architecture such that the reactive astrocytes maintain many of the fundamental features of cell structure, cellular interactions, and basic functions that they have in healthy tissue, together with additional context-specific gain- or loss-of-function changes as discussed in more detail below. They exhibit variable degrees of cellular hypertrophy and reorganization of their processes, but generally retain the discrete, nonoverlapping cellular domains exhibited by astrocytes in healthy gray matter (Bushong et al. 2002; Wilhelmsson et al. 2006; Han et al. 2021).

In response to overt tissue damage, such as stroke, severe trauma, infection, foreign bodies, autoimmune inflammation, neoplasm, or severe neurodegeneration, some substates of reactive astrocytes can re-enter the cell cycle (O'Shea et al. 2020; Sofroniew 2020; Krawczyk et al. 2022). Proliferating reactive astrocytes form borders that separate damage, inflamed, and fibrotic tissue from adjacent viable neural tissue (Sofroniew 2020). Emerging evidence indicates that reactive astrocytes that proliferate exhibit markedly different features and functions compared with reactive astrocytes that do not proliferate, or astrocytes in healthy tissue. Newly proliferated reactive astrocytes form borders that separate damaged, inflamed, and fibrotic tissue from adjacent viable neural tissue (see section on traumatic injury and stroke below). During this border formation, newly proliferated astrocytes adopt new cellular interactions with nonneural cells (Sofroniew 2020; Han et al. 2021; Burda et al. 2022). During this border formation, newly proliferated astrocytes interact with many other CNS cells, adopting new interactions (Sofroniew 2020). Although most border-forming reactive astrocytes in narrow zones immediately abutting tissue lesions are newly proliferated, proliferation drops off rapidly with increased distance from lesions, which are surrounded by large areas of nonproliferative reactive astrocytes as well (Wanner et al. 2013). Loss or attenuation of these newly proliferated reactive astrocytes leads to increased spread of inflammation and serum proteins, increased loss of neural tissue, and decreased functional recovery (Bush et al. 1999; Wanner et al. 2013; Sofroniew 2015a; Frik et al. 2018; Williamson et al. 2021; Burda et al. 2022).

Thus, nonproliferative and proliferative astrocyte reactivity represent two broad categories that are easily differentiated and are associated with molecular and functional differences. Notably, these categories should not be regarded as homogenous as each appears comprised of substates that likely have different functional states.

TRANSCRIPTIONAL AND PROTEOMIC PROFILING OF REACTIVE ASTROCYTES

Transcriptomic investigation of astrocytes in the context of infection, injury, and disease have generated much information about diverse molecular expression profiles of reactive astrocytes. These studies have employed bulk analyses of whole tissue samples, pooled purified astrocytes, or single-cell/nucleus analysis (examples include Haumont et al. 1989; Zamanian et al. 2012; Sirko et al. 2015; Anderson et al. 2016; Boisvert et al. 2018; Clarke et al. 2018; Barbar et al. 2020; Carroll et al. 2020; Pan et al. 2020; Zhou et al. 2020; Diaz-Castro et al. 2021; Hasel et al. 2021; Wei et al. 2021; Burda et al. 2022; Sadick et al. 2022). Multiple online databases are available that provide searchable expression profiles of multiple reactive astrocytes subtypes (Table 3) that have led to the identification of molecular markers up-regulated in different forms of reactive astrocytes (i.e., stroke, traumatic injury, inflammation, and neurodegeneration). We provide a short list of reactive astrocyte genes that are commonly up-regulated in astrocytes across multiple disease/injury conditions (Table 4). Notably, many of these genes are not expressed in astrocytes at baseline but become up-regulated in the context of reactivity and are, in many cases, up-regulated across multiple CNS cell types. At the single-cell/nucleus level, transcriptomic heterogeneity has been identified within an individual insult/disease (or disease model), in addition to similarities across certain insults of some, but not all, astrocyte reactivity substates. Examples of this include similarities in inflammatory reactive astrocytes substates identified by transcriptional analyses that are present following systemic injection of lipopolysaccharide (LPS) as well as mouse models of AD, MS, and acute stab wound trauma (Hasel et al. 2021; Castranio et al. 2023).

Table 3.

Online resources and data sets for probing “omics” changes in reactive astrocyte substates

URL	Data type	Species	Data set variables
Development, aging, different brain regions
www.brainrnaseq.org	RNA-seq (sorted cells and TRAP)	Mouse, human	Normal, aging
igc1.salk.edu:3838/astrocyte_aging_transcriptome	RNA-seq (TRAP)	Mouse	Aging
dropviz.org	scRNA-seq	Mouse	Different brain regions
Reactivity and disease
www.gliaseq.com	RNA-seq, sc/snRNA-seq, scATAC-seq, nanostring, proteomics, lipidomics, metabolomics, spatial transcriptomics	Rodent, human, mESC, hiPSC	Normal, different brain regions, inflammation, disease
www.gliaweb.net a	All	All	All
astrocyte.rnaseq.sofroniewlab.neurobio.ucla.edu	RNA-seq	Mouse	Spinal cord injury, inflammation
astrocyternaseq.org	RNA-seq	Mouse	Different brain regions, disease models, inflammation
cellxgene.cziscience.com/collections	scRNA-seq, scATAC-seq, snmC-seq2, spatial transcriptomics	Mouse, human	Normal (different brain regions), disease
seqseek.ninds.nih.gov	scRNA-seq	Mouse	Normal, spinal cord injury
singlecell.broadinstitute.org/single_cell	sc/snRNA-seq, scATAC-seq	Mouse, human	Different brain regions, disease,
cells.ucsc.edu	scRNA-seq	many	Development, aging, disease
adsn.ddnetbio.com	scRNA-seq	Human	Disease
ki.se/en/mbb/oligointernode	sc/snRNA-seq, scATAC-seq	Mouse, human	Disease
tr.astrocytereactivity.com/home	snRNA-seq, scATAC-seq	Mouse	Disease, trauma
General CNS data sets
www.proteinatlas.org	scRNA-seq, proteomics, pathology	Mouse, human	Disease
gtexportal.org/home	scRNA-seq, RNA-seq, QTL, eGTEx	Human	CNS regions, peripheral tissues

Note that many data sets have low capture counts for astrocytes. Additional aggregation/meta-analyses are always recommended. Not all data sets have validated reactive astrocyte data, but many have data from astrocyte states with different physiology (e.g., development, aging, or multiple brain regions/species. See individual websites for publication references. This is by no means an exhaustive list, as more resources are being produced.

^agliaweb.net is an umbrella repository for all glia-sequencing efforts across the field. It allows for individual laboratories to link and share their independent data sets with ease.

Table 4.

Examples of molecular markers used to identify reactive astrocytes in different contexts

Gene ID	Expressed in astrocytes at baseline	Astrocyte enricheda	Protein function	Subcellular location
C3	Yesa,b,c	No	Activator of the complement system	Cytosol, secreted
Cd44	Yesb	No	Cell-surface glycoprotein	Membrane
Cp	No	No	Metalloprotein	Secreted
Cxcl10	No	No	Chemokine	Secreted
Gfap	Yesa,b	Yesa	Intermediate filament, structure	Cytoskeleton
Hspb1	Yesb	No	Molecular chaperone	Cytoplasm, nucleus
Lcn2	No	No	Lipid transport	Secreted
Lgals3	Yesb	Noa	Cell–cell and cell–matrix interactions	Cytoplasm, membrane, extracellular
Osmr	Yesb	No	Cytokine receptor	Membrane
S100b	Yesb	No	Ca2+-binding protein	Cytoplasm, nucleus
Serpina3n	Yesa,b	No	Serine protease inhibitor	Secreted
Steap4	No	No	Metalloreductase	Golgi
Timp1	No	No	Matrix metalloproteinase	Secreted
Vim	Yesa,b,c	Noa	Intermediate filament, structure	Cytoskeleton

Abbreviations are as per standard nomenclature (Human Gene Compendium, GeneCards, www.genecards.org). See main text for literature references.

^aWith caveats.

^bIn astrocyte subpopulations.

^cPresent in astrocytes developmental in CNS stem cell niches.

Transcriptomic analysis of reactive astrocytes has also revealed diverse gene expression level changes driven by type of insult, the duration from onset, and distinct CNS region. Yet, how these transcriptomic changes correlate with changes in protein, lipid, or metabolite levels and subsequent alterations in astrocyte function are only beginning to be evaluated. For example, recent advances in proteomic technologies, including higher sensitivity and lower protein input requirements, have enabled large scale screening of the astrocyte proteome across contexts. At present, most studies exploring the proteome of reactive or diseased astrocytes are limited to analysis of astrocyte enriched or common astrocyte reactive molecules in bulk tissue (examples include Diaz-Castro et al. 2019; Johnson et al. 2020; Heaven et al. 2022; Bac et al. 2023). These studies have confirmed commonly identified astrocyte dysregulated genes and enriched pathways and provided additional evidence of diverse astrocyte populations. Far fewer studies have evaluated the proteome of isolated astrocyte populations with evaluation limited to cultured astrocytes, including immunopanned rodent astrocytes, human fetal astrocytes, or iPSC-derived astrocytes, in response to cytokines and other inflammatory mediators or disease conditions (Levine et al. 2016; Dozio and Sanchez 2018; Guttenplan et al. 2021; Labib et al. 2022). Ultimately, proteomic approaches that allow for regional and temporal control of cell-type-specific metabolic labeling of newly synthesized proteins in vivo followed by isolation of tagged proteins using click chemistry (Prabhakar et al. 2023), cell-type-specific biotin labeling (Rayaprolu et al. 2022), or similar evolving technologies will be necessary to correlate transcriptomic and proteomic data.

As the field continues toward characterizing the functional outcomes of reactive astrocyte substates, the number of transcriptomically defined substates that can feed such studies is rich. A current bottleneck at the functional testing level is that in vivo methods provide insufficient fidelity to properly measure individual functions within a heterogeneous population of astrocytes, while some in vitro methods are often unable to properly recapitulate in vivo gene expression. This can be due to inclusion of serum in culture media, which drives a reactive phenotype at baseline (Foo et al. 2011), or omission of other CNS cells required for proper astrocyte gene expression signatures (Hasel et al. 2017). Future endeavors to label or capture astrocytes at the substate-specific level in vivo, such as the newly described FIND-seq approach, which enables capture of small subsets of cells based on nucleic acid detection rather than cell surface markers (Clark et al. 2023), may enable purification of these populations for protein/lipid/metabolomic analysis, further expanding our understanding of the nuances of astrocyte reactivity.

CONTEXT-SPECIFIC REGULATION OF ASTROCYTE REACTIVITY

Astrocyte reactivity represents a broad spectrum of potential changes that occur in context-specific manners as determined by a wide variety of different potential extracellular signals that can change over time and can vary within tissue microdomains. These external signals can derive from a wide variety of sources and activate a broad spectrum of potential intracellular signal transducers and result in the release of a wide variety of potential effector molecules (Tables 1, 2, and 5).

Table 5.

Examples of astrocyte intrinsic signaling pathways and transcriptional regulators of astrocyte reactivity

Categories	Molecules
Calcium signaling	Ca2+
Chromatin regulators	DNMT3B, EP300, MECP2, MECOM, MEF3C, SMARCA4, SMARCE1
MicroRNAs	DICER (ribonuclease), miR-21, miR-124, miR-146a, miR-153, miR-155, miR-181a, miR-200a-3p, miR-218, miR-223, miR-330, miR-326, miR-3099
Signal transducers	ADAM8, Camp, ERK, G-proteins, IRAK1, JAK2, MAPK1, MAP3K13 (LZK), mTOR, PKA, PKC
Transcription regulators	ARNT, ATF4, BCL3, BCL6, CEPBA, CLOCK, CREB1, FOS, HIF1A, HTT, IRF1, IRF5, IRF8, IRF9, JUN, KLF4, NFE2L2 (NRF2), NFKB, NOTCH1, NURR1, MYD88, OLIG2, RUNX1, SMARCAa4, SMAD3, SMAD4, SOCS3, SOX9, SP1, SPI1, SREDBF1, STAT2, STAT3, STAT5, TCF4 (TCF7L12), TP53, WT1, YAP1, ZBTB16, and many others

See main text for literature references.

Extracellular signals that can induce or modulate astrocyte reactivity include many molecules that are also used in physiological activities, such as purines, transmitters, steroid hormones, growth factors, and serum proteins, as well as pathological signals such as molecules derived from microbial infections, tissue damage, inflammation, or neurodegeneration-associated events. In addition, mechanical stress can trigger astrocyte reactivity. These diverse instructive signals can derive from many different non-cell-autonomous sources, including (1) local neural and nonneural cells intrinsic to CNS tissue such as neurons, microglia, oligodendrocyte lineage cells, other astrocytes, endothelia, pericytes, and fibroblasts; (2) nonneural cells that gain entry into the CNS, such as bone marrow–derived leukocytes, fibrocytes, and microbial infectious agents; (3) foreign materials introduced by trauma or medical implants; or (4) tissue stretching or compression caused by trauma or neoplasm (Burda and Sofroniew 2014; Liddelow and Barres 2017; O'Shea et al. 2020; Lantoine et al. 2021). Examples of different types of triggers of astrocyte reactivity and their potential sources are presented in Table 1.

These diverse extracellular signals can induce changes associated with different substates of reactive astrocytes via a wide range of potential intracellular signaling mechanisms. These intracellular mechanisms include signal transducers that act via phosphorylation, acetylation or SUMOylation, Ca²⁺ signaling, chromatin modulators, DNA-binding transcription factors, transcriptional regulators that bind protein or RNA, and microRNAs (Chen et al. 2018; Shigetomi et al. 2019; Bai et al. 2021; Burda et al. 2022). Notably, different intrinsic properties of individual astrocytes associated with regional or local differences (Makarava et al. 2023), aging, molecular memory of previous events, genetic mutations or polymorphisms can cell-autonomously influence intracellular signaling mechanisms and modulate reactivity responses. Examples of different types of intracellular regulators of astrocyte reactivity are presented in Table 5.

Comparisons both within and across disorders are revealing remarkable heterogeneity of reactive astrocyte transcriptional signatures, and dissecting the underlying molecular signaling mechanisms indicates this heterogeneity derives from highly combinatorial and context-specific interactions among the many different types of regulators. For example, modulating astrocyte Jak2-Stat3 signaling, which typically occurs via cytokine or growth factor exposure, can have different effects in different contexts. Deletion of Jak2-Stat3 signaling from reactive astrocytes in traumatic injury or infection increased inflammation and tissue loss and worsened outcome (Drögemüller et al. 2008; Wanner et al. 2013), whereas doing so reduced plaque load and improved outcome in the β-amyloidopathy APP/PS1ΔE9 AD model (Ceyzériat et al. 2018; Reichenbach et al. 2019), but had little effect in the 3xTg amyloidopathy AD model (Guillemaud et al. 2020). In Huntington's disease (HD) models, activating astrocyte Jak2-Stat3 signaling increased proteolytic degradation of mutant Huntingtin and slowed disease progression (Abjean et al. 2023).

Some of these differences may be explained by diverse transcriptional responses regulated by Stat3 in different reactivity contexts. Multiple lines of evidence show that reactivity-associated transcriptional changes are regulated in a highly combinatorial manner. ATAC-seq analysis highlights genes regulated by Stat3 also have DNA-binding sites for multiple other transcription factors (Burda et al. 2022). In addition, Stat3 can regulate the expression of chromatin modulators and thereby indirectly influence gene expression (Burda et al. 2022). Such combinatorial interactions of multiple regulators can result in different effects on the same functional systems. For example, Stat3 signaling can either promote or inhibit different inflammatory signatures in reactive astrocytes by interacting with interferon or Il6 signaling pathways (Leng et al. 2022). Thus, Stat3 cannot be regarded as a master switch that stereotypically activates a particular signature or substate of astrocyte reactivity. Nor can the effects of modulating Stat3 be extrapolated directly from one context to another.

Similar observations have been made for other reactivity regulators as well. The induction and maintenance of diverse astrocyte reactivity signatures and states is regulated in a context-dependent manner by the combinatorial and complex interactions of a vast array of mechanisms including intracellular signal transducers, chromatin regulators, transcription regulators, and microRNAs (Table 5). These findings highlight the need to understand the complex and combinatorial signaling regulation of astrocyte reactivity in specific contexts to develop rational therapeutic approaches based on modulating signaling pathways. Importantly, interpreting the effects of manipulating broad transcriptomic regulators that influence multiple reactive substates should now be done with caution. As the field moves forward, there is a need to develop and engage more specific tools, such as FIND-seq (Clark et al. 2023), to disentangle molecular drivers of specific responses associated with individual reactivity substates.

FUNCTIONS AND EFFECTS OF ASTROCYTE REACTIVITY

In response to the many signaling events described above, reactive astrocytes have the potential to release large variety molecules that impact on nearby cells (Table 2), as well as alter their physical contacts with adjacent cells, which are important for physiological functions like synapse stabilization, neurotransmitter reuptake and recycling, and maintenance of the glymphatic space. Reactive astrocyte-secreted molecules can exert many different functions and effects that are substate dependent, and that may be beneficial but can also give rise to maladaptive effects, and that modulation of these opposing effects is under tight temporal and spatial control. In the following sections, we highlight representative examples of major categories of insults that astrocytes respond to.

REACTIVE ASTROCYTES IN TRAUMATIC INJURY AND STROKE

Astrocytes respond to tissue acute damage like trauma and stroke in a graded manner (Sofroniew and Vinters 2010; Gleichman and Carmichael 2014; Burda et al. 2016). Scar border–forming astrocytes exhibit substantial transcriptional reprogramming and proliferate immediately abutting areas of severe tissue damage to wall off damaged and fibrotic areas and protect adjacent neural tissue (Wanner et al. 2013; Gleichman and Carmichael 2014; Sofroniew 2015a; Burda et al. 2022). Such reactive astrocytes were previously thought to be detrimental scars that prevent recovery (Sofroniew 2018), but recent studies have shown that astrocyte borders are necessary for axon regeneration (Anderson et al. 2016). Moreover, substantive axon regeneration through lesions can be achieved by providing growth-stimulating and chemoattractive factors, and this growth is attenuated by disrupting astrocyte borders (Anderson et al. 2016, 2018). In spared neural tissue adjacent to overt lesions, there is a gradient of nonproliferative astrocyte reactivity that diminishes with distance from the lesion (Wanner et al. 2013; Burda et al. 2016). This nonproliferative reactive astrocyte population can influence synaptic plasticity (Overman et al. 2012; Burda et al. 2016; Carmichael et al. 2017; Sozmen et al. 2019; Brennan et al. 2021; Lawal et al. 2022).

REACTIVE ASTROCYTES AND INFLAMMATION

Inflammation is a natural response of the body to protect cells from infection and disease-associated molecules. In the periphery, white blood cells, platelets, and other components mount the response, leading to swelling, heat production, and cytokine release. In the CNS, astrocytes and microglia, along with infiltrating peripheral immune cells, play a key role in the inflammatory response (Han et al. 2021). The interaction between astrocytes and microglia is crucial for appropriate responses and resolution of inflammation. This interaction could be manipulated for developing new therapeutic strategies in neurodegenerative diseases.

Astrocytes have been shown to produce both pro- and anti-inflammatory molecules, and specific molecular regulators can modulate their functions to either increase or limit CNS inflammation. Astrocytes produce a wide range of chemokines not only that open the blood–brain barrier (Argaw et al. 2012) and attract inflammatory cells (Hamby et al. 2012; Zamanian et al. 2012; Liddelow et al. 2017; Hasel et al. 2021) but also molecules that can exert potent suppressive effects on inflammatory cells (Kostianovsky et al. 2008; Hasel et al. 2021). Proliferative reactive astrocyte scar borders can limit the spread of inflammatory cells but may also limit the infiltration of anti-inflammatory modulators (Voskuhl et al. 2009; Wanner et al. 2013; Sofroniew 2015a). Modulating astrocyte functions can either attenuate or exacerbate CNS inflammation (Herrmann et al. 2008; Brambilla et al. 2009; Sofroniew 2015a), opening the door for gain or loss of astrocyte functions to impact CNS inflammation.

Clinical evidence suggests that disrupting astrocyte function worsens outcomes in CNS autoimmune diseases. Neuromyelitis optica (NMO), which specifically affects astrocytes, is a severe CNS autoimmune disease that causes vision loss and paralysis. Autoantibodies to aquaporin-4 (AQP4) on astrocytes drive an autoimmune response leading to complement-mediated astrocyte lysis (Lennon et al. 2005; Roemer et al. 2007), and patients with autoimmune-mediated CNS demyelination caused by AQP4 autoantibodies tend to have a worse outcome (Kitley et al. 2014; Sato et al. 2014), suggesting that AQP4 is not just a passive autoimmune antigen in the CNS, but that disrupting astrocyte function with AQP4 antibodies worsens outcomes. Experimental evidence suggests that astrocytes play a critical role in preventing the spread of inflammation during an autoimmune attack on the CNS (Voskuhl et al. 2009; Haroon et al. 2011), and loss of astrocytes exacerbates the spread of autoimmune tissue loss in NMO (Sofroniew 2015a).

Together, these findings provide compelling clinical and experimental evidence that astrocytes play a crucial role in CNS inflammatory responses, both as regulators and targets of inflammation. The interaction between astrocytes and other CNS resident cells, namely microglia, is vital for an appropriate response to such noxious stimuli. Manipulating this interaction has therapeutic potential for neurodegenerative diseases. Astrocytes have the ability to produce both pro- and anti-inflammatory molecules, and their functions can either exacerbate or attenuate CNS inflammation. Clinical studies have shown that CNS autoimmune diseases that affect astrocytes tend to be more severe, suggesting that disrupting astrocyte function can worsen outcomes. Changes in function in specific substates of reactive astrocytes can exacerbate CNS inflammation and tissue damage. Understanding the role of astrocytes in CNS inflammation is crucial for developing effective therapeutic strategies for neurodegenerative diseases.

REACTIVE ASTROCYTES AND CANCER

In response to brain cancer, reactive astrocytes can play a significant role in promoting tumor growth, angiogenesis, and invasion. Tumor cells themselves, or altered vascular permeability, and infiltrating peripheral immune cells, are likely drivers of these reactive transitions that have been measured by hypertrophy, proliferation, and increased GFAP protein or Gfap gene levels. At their most basic, tumor-associated reactive astrocytes form a barrier around the tumor and release various cytokines, chemokines, and growth factors that contribute to tumor progression and pathophysiology (John Lin et al. 2017; Krawczyk et al. 2022; Perelroizen et al. 2022). In other reports, breast-to-brain metastasis cancer cells infiltrate the synaptic cleft and exclude astrocyte processes (Zeng et al. 2019), while other studies report carcinoma–astrocyte gap junctions that further promote metastasis (Chen et al. 2016). However, it remains unclear whether astrocytes in these pathologies are reactive per se, or simply displaced and unable to perform physiological functions. Important recent discoveries suggest that metastatic cancer–astrocyte introductions do induce reactivity, with tumor-associated reactive astrocytes releasing lipocalin 2 (LCN2) to further drive inflammation in the CNS (Adler et al. 2023). Global deletion of Lcn2 in mice attenuated this neuroinflammation and inhibited the seeding of metastases in this study. Interestingly, studies using complete depletion of proliferating astrocytes (using Gfap-TK mice) have observed a regression of glioblastoma and prolonged mouse survival, presumably due to a lack of newly proliferated reactive astrocytes (Perelroizen et al. 2022). Nevertheless, targeting all Gfap-positive cells may remove several reactive substates of astrocytes that may have opposing functions. Melanoma metastases, among the most aggressively CNS penetrant of peripheral cancers, cleave amyloid precursor protein thereby forming and secreting amyloid β. This mounts a cascade involving prometastatic anti-inflammatory astrocytes, which in turn inhibit microglia phagocytosis of the initial melanoma (Kleffman et al. 2022). Thus, while complicated, at a cell biological level, some reactive astrocytes create a unique microenvironment that may simultaneously promote tumor growth while facilitating immune suppression and resistance to therapeutic interventions, while other reactive astrocytes are driven by microglia-chemotherapy responses to actively kill CNS-derived tumors like oligodendroglioma (Gibson et al. 2019). Targeting specific aspects of reactive astrocyte interactions with cancer cells has been proposed as a promising therapeutic strategy for the treatment of brain cancer, but much remains unknown about the specific substate functional changes in reactive astrocytes in these contexts.

REACTIVE ASTROCYTES AND PERIPHERAL INFECTION

Peripheral infections can result in high levels of circulating cytokines and inflammatory mediators, including microbial pathogen-associated molecular pattern molecules (PAMPs), that can access the CNS and drive astrocyte transcriptome profiles toward a proinflammatory and cytotoxic state. This response is essential to limit the spread of infection, and diversion from this tightly controlled CNS inflammatory response will alter the molecular expression and function of reactive astrocytes (and microglia). Studies have shown a direct correlation between peripheral immune response signals, microglia–astrocyte communication, and neurological diseases. The GLP1R signaling pathway has been found to reduce astrocyte reactivity and mitigate neurodegeneration in part by reducing neurotoxic astrocyte reactivity in a number of in vivo mouse neurodegeneration models (Yun et al. 2018; Sterling et al. 2020, 2023).

Recent research has also begun to uncover the interactions between peripheral immune cells and the microglia–astrocyte metabolic signaling cascade in the CNS. Microbiome-reprogrammed natural killer cells have been found to be sufficient to drive reactive states in microglia and induce neurotoxic reactive astrocytes (Sanmarco et al. 2021). Additionally, mechanisms have been identified that are initiated by altered interferon signaling and regulated by dietary tryptophan metabolites, microglial aryl hydrocarbon receptor signaling, and VEGFb/TGF-α, which tune astrocyte reactivity in CNS autoimmunity (for reviews, see Linnerbauer et al. 2020; Han et al. 2021).

Exposure of astrocytes to PAMPs such as LPS and immune cell- and peripherally derived cytokines can drive astrocytes toward proinflammatory states with the potential for neurotoxicity (Zamanian et al. 2012; Liddelow et al. 2017; Barbar et al. 2020; Guttenplan et al. 2021; Hasel et al. 2021). Although the potential for neurotoxicity by reactive astrocytes was first discussed in the context of SOD1 mutations associated with amyotrophic lateral sclerosis (ALS) (Di Giorgio et al. 2007, 2008; Nagai et al. 2007), the identity of molecular mediators of this neurotoxicity has eluded researchers for decades and has often been attributed to glutamate excitotoxicity or lack of trophic support (Han et al. 2021). Recent investigations using in vitro experiments validated with in vivo acute injury paradigms (Liddelow et al. 2017), and analysis of gene expression, protein, and lipid secretion changes (Guttenplan et al. 2020a) highlight that fully saturated very long chain fatty acids are the most likely driver of this neurotoxicity. These saturated lipids drive PERTK-ATF3-mediated lipo-apoptosis in neurons. Importantly, previous work described that healthy neurons themselves are not susceptible to this astrocyte-mediated cell death, and instead these cells must be susceptible (modeled using axotomy or inflammatory stressors) for lipo-apoptosis to occur (Guttenplan et al. 2020a). Comorbid peripheral infections may alter the molecular expression and function of reactive astrocytes in ways that exacerbate tissue damage and compromise neural repair (Failli et al. 2012; Heintz and Mair 2014; Sanmarco et al. 2021). Further research is needed to explore the potential impact of concurrent low-grade infections or alterations in the microbiome on the induction of reactive astrocytes and consequent interactions with CNS cells driving pathology.

DISEASE-ASSOCIATED GENE MUTATIONS THAT ALTER ASTROCYTE FUNCTION

Genetic mutations, like those in Huntington's disease, familial ALS, and Alexander disease, can cause cell-autonomous astrocyte dysfunction that leads to neuronal dysfunction and neurodegeneration. These “diseased” astrocytes exhibit alterations in transcriptomics, proteomics, or function due to the mutation. This should be differentiated from astrocyte reactivity, which is triggered by external non-cell-autonomous signals from degenerating neural tissue (Fig. 1).

Figure 1.

Changes in astrocyte states during development, disease, and reactive responses. Astrocytes have varying gene expression profiles and functions during development, disease, and in response to external stimuli. Aging can cause changes in gene expression, which may not be a true reactive state but rather a change in physiological demands. Diseased astrocytes exhibit altered gene expression due to mutations in astrocyte-specific genes (e.g., GFAP or SOD1). Reactive astrocytes result from external factors such as immune responses, pathogenic proteins, or traumatic injury, and may occur at any stage or in diseased astrocytes. Transitions between states may occur bidirectionally (e.g., acute insult, solid arrows), but some may be irreversible (dashed arrows). (SCI) Spinal cord injury, (TBI) traumatic brain injury.

Alexander disease (AxD) is a well-studied example of cell-autonomous disease in astrocytes, caused by a dominant gain of mutation of the GFAP gene. Patients with early-onset type 1 AxD exhibit macrocephaly, motor and cognitive delay, seizures, psychomotor disturbances and premature death, while those with late-onset type II AxD present with bulbar symptoms, autonomic dysfunction, and spinal cord and brainstem atrophy (Brenner et al. 2001; Prust et al. 2011). In murine models of type II AxD, AxD astrocytes down-regulate expression of key homeostatic proteins including those involved in extracellular glutamate (Glt1) and K⁺ regulation (Kir4.1) (Minkel et al. 2015), providing a mechanistic link between astrocyte and neuronal dysfunction and disease-associated phenotypes. A recent proteomic study confirmed these functional changes and revealed additional disruption of the typical “mature astrocyte proteome,” fatty acid metabolism, and alterations in protein expression associated with other CNS cell populations (Heaven et al. 2022), indicating diseased astrocytes can drive neurological disorders.

Diseased astrocytes may not be reactive at baseline but may react to pathological stimuli and become reactive during disease progression. Often the mutated gene is highly expressed in astrocytes as well as other CNS cell populations, including several neurodegenerative disease-associated mutations (Huntington's disease: HTT; ALS: SOD1; AD: PSEN1, PSEN2, APP; Parkinson's disease (PD): LRRK1, PINK1, CD38). A rich literature now exists indicating diseased astrocytes alter their functional properties in ways that may contribute significantly to neurological symptoms and neurodegeneration. For example, in Huntington's disease, astrocytes as well as neurons accumulate nuclear inclusions of mutant huntingtin protein (mHTT). Astrocytes with mutant huntingtin protein down-regulate Kir4.1, leading to increased extracellular potassium and neuronal excitability. These effects occur at early stages of symptom onset and in the absence of astrocyte changes associated with reactivity. Both genetically mutated and abnormal “physiological” astrocytes likely contribute to the detrimental effects on neuronal function and neurotoxicity (Tong et al. 2014).

The role of diseased astrocytes in neuronal dysfunction and disease progression has been well characterized in ALS, where mutant SOD1-expressing astrocytes contribute to neuronal dysfunction and degeneration. Mutant SOD1 produced in neurons also contributes to neuronal degeneration in the full disease (Lobsiger and Cleveland 2007; Nagai et al. 2007; Yamanaka et al. 2008). Reactive astrocyte formation is drastically reduced by mutant SOD1, and gene expression during inflammation is altered (Guttenplan et al. 2020b). Investigation into the interplay between altered baseline astrocyte gene expression/function and secondary responses to external stimuli is ongoing and will likely become a major focus of reactive astrocyte research.

ASTROCYTE MOLECULAR POLYMORPHISMS

An intriguing and potentially important question is the degree to which genetic polymorphisms alter functions of physiologically normal and reactive astrocytes. Available evidence suggests that this is the case, with the most studied example represented by human polymorphisms of apolipoprotein E (APOE). APOE is a major lipid transporter in the brain and represents the strongest genetic risk factor for late-onset AD. In Western European populations, carrying two alleles of the APOE4 isoform increases risk for AD eight- and 12-fold, while APOE3 is considered neutral and APOE2 considered a protective variant. It should be noted, however, that APOE4 is protective in disorders of the eye like glaucoma (Margeta et al. 2020). APOE is highly expressed in astrocytes, representing one of the most abundant protein-coding transcripts. Studies using various transgenic mouse models show that astrocyte-secreted APOE4, but not APOE3 or APOE2, is associated with early seeding of amyloid pathology, increased blood–brain barrier leak, reduced astrocyte end-foot coverage of blood vessels, and increased neuronal degeneration (Bell et al. 2012; Liu et al. 2017; Jackson et al. 2022). In contrast, selective reduction of APOE4 in mice leads to significantly reduced amyloid β plaque load, and reduced overall cortical Gfap levels (Liu et al. 2017; Mahan et al. 2022). APOE4 is also reported to alter astrocyte transcriptional responses to proinflammatory stimuli and result in maladaptive immune and metabolic responses (Lee et al. 2023). As new genetic polymorphisms associated with neurologic disease are identified, the field must consider the notion that these may impact astrocyte function and alter reactive astrocyte substates in disease-relevant ways. For example, a polymorphism of CD38, which is highly expressed in astrocytes (Margeta et al. 2020), has recently been identified as a risk factor for PD (Guerreiro et al. 2020; Kia et al. 2021).

ASTROCYTE-MEDIATED NON-CELL-AUTONOMOUS DYSFUNCTION OR DEGENERATION

There are many ways that loss of astrocyte functions or disruption of reactive astrocytes could lead to non-cell-autonomous neuronal dysfunction or degeneration. In healthy neural tissue, astrocytes play critical roles for normal neuronal function including homeostasis of extracellular fluid, ions and transmitters, regulation of blood flow, energy provision, and interactions with synapses (Barres 2008). In damaged neural tissue, reactive astrocytes play critical roles in neuroprotection, blood–brain barrier repair, and regulation of inflammation (Sofroniew 2020; Han et al. 2021). Thus, perhaps not surprisingly, experimentally induced loss of specific functions of astrocytes or reactive astrocyte substates can cause neuronal dysfunction and degeneration and worsen outcome after CNS trauma, ischemia, or autoimmune attack. A pioneering example is that selective deletion of the astrocyte glutamate uptake transporter, GLT1 (Eaat2) will lead to seizures and excitotoxic neuronal death (Rothstein et al. 1996). Further examples include (1) astrocyte-selective expression of mutant SOD causes neuronal dysfunction and degeneration (Lobsiger and Cleveland 2007; Nagai et al. 2007; Yamanaka et al. 2008), (2) astrocyte-selective deletion of the endoribonuclease, Dicer, cases severe ataxia, progressive cerebellar degeneration, seizures, and premature death (Tao et al. 2011), and (3) astrocyte-selective deletion of the Wnt signaling molecule adenopolyposis coli causes delayed degeneration of cerebellar Purkinje neurons (Wang et al. 2011). Experimental disruption of various astrocyte signaling molecules can alter reactive astrocytes and cause neuronal dysfunction and tissue degeneration during traumatic injury, ischemia, and autoimmune attack (Okada et al. 2006; Drögemüller et al. 2008; Herrmann et al. 2008; Li et al. 2008; Haroon et al. 2011).

THERAPEUTIC TARGETING OF REACTIVE ASTROCYTES

There is increasing recognition that molecular mechanisms associated with specific functions of astrocytes in physiological and pathological/reactive states may be potential targets for novel therapeutic strategies for CNS disorders. As emphasized throughout this article, reactive astrocytes exert both essential beneficial functions as well as harmful effects in specific contexts as determined by specific signaling events—often by the same cell. Thus, useful therapeutic strategies will need to be specifically targeted, likely to individual functions or substates and not to global drivers of reactivity, so as to preserve or augment beneficial effects of reactive astrocytes while blocking or reducing harmful ones. The once prevalent view that wholesale blockade of reactive astrocytes could be a therapeutic strategy is no longer tenable and would likely do more harm than good. This is well reported in the literature by bulk blockade of reactivity using Stat3 deletion, which can have both net positive and net negative effects in the same disease models (e.g., in AD; Sadick and Liddelow 2019; Smit et al. 2021). Specific aspects of reactive astrocytes that are being explored as potential targets for therapeutic manipulations include mechanisms that regulate extracellular glutamate and K⁺, enzymes that generate or neutralize reactive oxygen species, and production of certain cytokines; however, a clear understanding of which transcriptomically defined substates of reactive astrocytes have alterations in these functions, and whether they are associated with specific regions of dysfunction/pathology remains to be unearthed.

CONCLUDING REMARKS

Astrocyte reactivity is a complex and multifaceted response to various forms and degrees of CNS pathology. The response can range from subtle, reversible changes to the formation of long-lasting scar border. Specific responses are controlled by various extracellular and intracellular signaling mechanisms, which are context-dependent and influenced by factors such as location, genetics, age, and previous reactivity events. Reactive astrocytes can have both beneficial and harmful effects, and changes in their function can cause or contribute to CNS disorders. As such, they are potential targets for novel therapies aimed at augmenting or attenuating their context-specific functions. Effective therapies will need to be directed at these specific functions.