Early works have primarily studied spinal cord astrocytes in the context of the first genetic abnormality associated with ALS—missense point mutations in the Superoxide Dismutase 1 (SOD1) gene. SOD1-ALS accounts for ∼12% of familial and ∼1% of sporadic cases (Renton et al., 2014). Although this genetic variation of ALS presents with progressive motor neuron degeneration, there is no TDP-43 pathology present. In SOD1-ALS, SOD1 variants undergo misfolding that causes structural and functional defects in catalytic activity and metal binding leading to a toxic gain-of-function (Huai and Zhang, 2019). Misfolded SOD1 variants have been observed to localize not only in motor neurons of the spinal cord, but also in the nuclei of astrocytes (see Figure 1), microglia and oligodendrocytes in the ventral horn (Forsberg et al., 2011). Early studies demonstrated that when SOD1 variants are expressed only in motor neurons of a chimeric mouse model, this is not sufficient to initiate neurodegeneration in the spinal cord (Clement et al., 2003). However, primary mouse astrocytes expressing SOD1^G93A variant reduced the survival of both wild type and SOD1^G93A motor neurons in co-culture conditions, eliciting a direct effect of neuronal degeneration (Di Giorgio et al., 2007). Additionally, primary mouse motor neuron cultures treated with preconditioned media from primary astrocytes from SOD1^G93A mice displayed a selective toxic effect on neuronal morphometry and survival (Nagai et al., 2007). When conditioned media from SOD1^G93A cortical neurons, microglia, myocytes and fibroblasts were added to the primary motor neurons, there was no effect on survival (Nagai et al., 2007). However, when SOD1^G37R variant was selectively deleted in mouse astrocytes in vivo, disease onset was not altered, but instead astrocyte-specific deletion of SOD1^G37R delayed microglial activation and significantly slowed later disease progression (Yamanaka et al., 2008).

More recently, the knockout of pro-inflammatory cytokines, IL-1α, TNFα and C1q, was found to significantly delay the death of motor neurons in the SOD1^G93A mouse model (Guttenplan et al., 2020). In these IL-1α^−/− TNFα^−/− C1q^−/− SOD1^G93A mice, lower levels of the microglia-mediated reactive astrocyte marker C3, identified in Liddelow et al. (2017), was observed (Guttenplan et al., 2020). Additionally, these triple knockout SOD1^G93A mice had a 50% lifespan extension compared to strain matched SOD1^G93A mice (Guttenplan et al., 2020). Together these studies provide evidence that astrocytes expressing ALS-linked SOD1 variants have the ability to elicit direct and indirect toxic effects selective to motor neurons, while exacerbating microglial activation and disease progression.

Tau is encoded by the gene MAPT and due to alternative splicing in exons 2, 3, and 10, six isoforms of the protein are produced. These isoforms make-up various combinations of three or four microtubule-binding repeat-regions in the C-terminal, and zero to two inserts in the N-terminal; resulting in 0N3R, 1N3R, 2N3R, 0N4R, 1N4R and 2N4R tau species (Buée et al., 2000). It is expected that the isoforms have various specific functions, but the primary function of tau is to promote the assembly of microtubules (Buée et al., 2000). Although tau isoforms are expressed differentially among neuronal subpopulations, tau is not expressed in astrocytes under physiological conditions (Muller et al., 1997).

Neurons are the primary cell type affected by FTLD-TAU, but astrocytes have been identified as a contributor of disease pathology. There is evidence indicating that astrocytic degeneration is associated with the degree of tissue atrophy in the frontal and temporal lobes in FTLD (Broe et al., 2004). FTLD-TAU accounts for 36%–50% of all FTLD cases with Pick’s disease, corticobasal degeneration, and progressive supranuclear palsy as the most common subtypes (Bang et al., 2015). Tufted astrocytes and astrocytic plaques (see Figure 1) are common pathologies in FTLD-TAU (Bang et al., 2015). When iPSC-derived astrocytes expressing the MAPT ^N279K variant were co-cultured with control neurons, there was a direct toxic effect decreasing neuronal survival during a rotenone-induced oxidative stress assay (Hallmann et al., 2017). Furthermore, these MAPT ^N279K FTD astrocytes displayed increased cell death and astrogliosis, defined by hypertrophy and increased glial fibrillary acidic protein (GFAP) levels (Hallmann et al., 2017). These results suggest that the MAPT ^N279K variant elicited a detrimental reactive astrocyte transformation that mediated an astrocyte cell-autonomous and non-cell autonomous neuron pathology. It is important to note that tau-related inclusions overlap with Alzheimer’s disease and enact similar astrocyte pathology, which is further discussed below.

The C9orf72 HRE variant is responsible for 30%–50% of familial ALS cases, 10% of sALS cases, and up to 50% of FTD cases (DeJesus-Hernandez et al., 2011; Renton et al., 2011; Cooper-Knock et al., 2015). The discovery of this variant provided the strongest genetic evidence to date of the significant overlap between ALS and FTD. In addition to TDP-43 pathology (see below), there are three primary non-exclusive mechanisms hypothesized to cause C9orf72-ALS/FTD that necessitate the cooperation of both toxic loss and gain-of-functions. The repeat expansion suppresses the production of C9orf72 protein by inhibiting transcription, causing haploinsufficiency and loss-of function of the C9orf72 protein (DeJesus-Hernandez et al., 2011; Renton et al., 2011). The toxic gain-of-function from sense and antisense C9orf72 repeat RNA, generated by bidirectional transcription, form sense and antisense HRE-RNA foci (DeJesus-Hernandez et al., 2011; Donnelly et al., 2013). The secondary structures of the repeat RNA, G-quadruplexes and DNA/RNA hairpins, bind and sequester RNA-binding proteins and cause DNA damage, respectively (Haeusler et al., 2014). Protein products of the repeat RNA, dipeptide repeat proteins (DPRs), are generated via repeat-associated non-ATG (RAN) translation forming five species: poly-GA, poly-GP, poly-GR, poly-PA and poly-PR (Ash et al., 2013; Tran et al., 2015). DPRs are aggregation prone and are thought to have some level of pathogenic effect. However, DPR pathology does not coincide with TDP-43 pathology and neurodegeneration observed in these patients. Nucleocytoplasmic trafficking deficits have also been associated with C9orf72 pathogenesis (Zhang et al., 2015; Boeynaems et al., 2016; Moore et al., 2020). These pathological features of the disease have been observed in spinal cord, motor, and frontal cortex neurons of C9orf72-ALS/FTD patient postmortem tissue (Murray et al., 2011; Mackenzie et al., 2015; DeJesus-Hernandez et al., 2017).

In C9orf72-ALS, few studies have described a cell-autonomous astrocyte pathology and an astrocyte-mediated non-cell autonomous motor neuron pathology (Madill et al., 2017; Allen et al., 2019; Birger et al., 2019; Taga et al., 2019; Varcianna et al., 2019; Zhao et al., 2020). One study demonstrated that C9orf72-ALS human induced pluripotent stem cell (iPSC)-derived spinal cord astrocytes develop RNA foci and DPR pathology (Zhao et al., 2020). DPRs have led to mis-splicing of the SLC1A2 gene causing dysfunction of EAAT2 (Lin et al., 1998; Kwon et al., 2014). Consequently, DPR pathology in astrocytes reduces their ability to uptake glutamate from their surroundings, leading to glutamate excitotoxicity in motor neurons already made vulnerable by C9orf72 haploinsufficiency (Shi et al., 2018). In co-culture, C9orf72-ALS astrocytes with confirmed RNA foci and poly-GP presence, triggered adverse effects on the electrophysiological properties of motor neurons (Zhao et al., 2020). Additionally, it has been found that DPRs can be transmitted from neurons to astrocytes through both contact dependent and independent mechanisms (Westergard et al., 2016). The transmission of different DPR species leads to nuclear and cytoplasmic aggregates in astrocytes co-cultured with DPR transfected neurons (Westergard et al., 2016).

Furthermore, when exposed to C9orf72-ALS astrocyte derived extracellular vesicles (EVs) and conditioned media, motor neurons experienced a 50%–70% increase in death compared to control astrocyte EVs and conditioned media (Varcianna et al., 2019). In this same study, C9orf72-ALS astrocytes were observed to secrete fewer EVs compared to controls (Varcianna et al., 2019). This contrasts studies describing astrocytes expressing SOD1 variants having an increase in EV secretion, therefore suggesting a characteristic difference between SOD1 and C9orf72 ALS astrocytes (Basso et al., 2013; Varcianna et al., 2019). Together, these studies indicate a role for C9orf72 astrocytes in the neurodegeneration of motor neurons observed in C9orf72-ALS/FTD; however, there is little to no information on the role of astrocytes in the neurodegeneration of cortical neurons leading to dementia symptoms in this disease spectrum.

The hallmark pathology of approximately 97% of ALS cases and 50% of FTD cases is the simultaneous nuclear depletion and cytoplasmic accumulation of TDP-43 (TAR-DNA binding protein 43), encoded by the TARDBP gene (Arai et al., 2006; Neumann et al., 2006). TDP-43 is an RNA and DNA binding protein, serving multiple functions in the regulation of gene expression at both the transcription and translation levels (Ratti and Buratti, 2016). Similar to other heterogeneous nuclear ribonucleoproteins (hnRNPs), TDP-43 consists of an N-terminus region followed by two RNA recognition motifs (RRMs) that bind RNA and DNA, and a prion-like domain containing C-terminus that aids in its ability to participate in protein-protein binding, phase separation and aggregation. Predominantly a nuclear protein, TDP-43 can shuttle to and from the cytoplasm via active and passive transport under physiological conditions. However, the aberrant nuclear depletion and cytoplasmic mislocalization of TDP-43 causes two toxic mechanisms—a toxic loss-of-function in the nucleus and toxic gain-of-function in the cytoplasm that work in conjunction (Tziortzouda et al., 2021). The resulting pathology in affected neurons and glia are cytoplasmic inclusions containing TDP-43, phosphorylated TDP-43 (pTDP-43), and cleaved C-terminal fragments. This pathology primarily occurs in neurons in the spinal cord, frontal, temporal and primary cortices, as well as subcortical regions, such as the hippocampus, basal ganglia, amygdala, thalamus and regions within the brainstem (Arai et al., 2006; Neumann et al., 2006; Murray et al., 2011; Mackenzie et al., 2015). TDP-43 inclusions are found in the cytoplasm and radiating processes of astrocytes (see Figure 1), but are more common in oligodendrocytes in the spinal cord and some cortical regions (Nishihira et al., 2008; Cooper-Knock et al., 2012; Heo et al., 2022).

The achievement of TDP-43 pathology in vitro and in vivo models most often requires either biological stressors or overexpression of specific TDP-43 variants. Astrocytic models require TDP-43 (full length, phosphorylated or the C-terminal fragment) to be selectively overexpressed or seeded into astrocytes (Serio et al., 2013; Hebron et al., 2014; Heyburn et al., 2016; Peng et al., 2020; Velebit et al., 2020). In a recent study investigating TDP-43 proteinopathy in sALS, hiPSC-derived motor neurons and astrocytes were exposed to seeded pathological TDP-43 aggregates from serially passaged sALS postmortem tissue (Smethurst et al., 2020). Through this seeding model, iPSC-motor neurons were observed to readily transfer TDP-43 aggregates to iPSC-astrocytes and vice versa. However, TDP-43 aggregates in iPSC-astrocytes were cleared more readily and there were less aggregates transferred from astrocytes to motor neurons (Smethurst et al., 2020). This study suggests a cell-type specific vulnerability and protein clearance of TDP-43, whereby astrocytes are less vulnerable and more efficient at clearing TDP-43 than motor neurons.

Another study, using a conditional TDP-43 knockdown mouse model to investigate the effects of nuclear depletion of TDP-43 in spinal cord astrocytes, revealed enhanced GFAP immunoreactivity and a molecular signature resembling that of the “A1-astrocytes” identified in Liddelow et al. (2017) (Peng et al., 2020). These TDP-43 depleted astrocyte mice presented with motor deficits, upregulation of C1 complement expression in microglia, and reduction of mature oligodendrocytes without directly affecting motor neuron survival (Peng et al., 2020). This indicates that nuclear loss-of-function of TDP-43 in affected astrocytes has both a cell autonomous effect and a non-cell autonomous effect on other glial types. Furthermore, a study that transfected primary rat cortical astrocytes with the C-terminal fragment of TDP-43, found that affected cells developed cytoplasmic TDP-43 inclusions and nuclear depletion of endogenous TDP-43 (Velebit et al., 2020). This is consistent with findings in neurons and cell lines where expression of the C-terminal fragment affects endogenous TDP-43 nucleocytoplasmic trafficking (Geser et al., 2008; Winton et al., 2008). Transfected astrocytes with TDP-43 cytoplasmic inclusions also presented with a stress state that involved an increase in lipid droplets, signifying an impaired lipid metabolism and disruptions in ß-adrenergic mediated aerobic glycolysis and lactate production (Velebit et al., 2020). This study further supports the idea that astrocytes affected by TDP-43 pathology in ALS and FTD may undergo a reactive transformation that disables an astrocyte’s ability to metabolically support surrounding neurons.

Given the dominant role of TDP-43 in ALS pathogenesis, ALS-linked TDP-43 variants have also been studied in astrocytes. In iPSC-derived astrocytes, carrying the TDP-43^M337V variant, cells displayed cytoplasmic mislocalization of TDP-43, but not simultaneous nuclear depletion (Serio et al., 2013). Motor neurons with the same TDP-43^M337V variant presented with high levels of insoluble TDP-43 (Bilican et al., 2012), whereas astrocytes did not (Serio et al., 2013). Notably, this suggests a cell-type specific function of TDP-43 between neurons and astrocytes. In vivo, the same TDP-43^M337V variant expressed only in rat astrocytes led to the progressive loss of spinal motor neurons, skeletal muscle atrophy, and eventual paralysis (Tong et al., 2013). However, a separate study where the TDP-43^A315T variant was overexpressed in astrocytes, found no motor neuron death in either an in vitro or transplant mouse model (Haidet-Phillips et al., 2013). The lack of consensus among these studies regarding whether astrocytes harboring a TDP-43 disease related-variant can cause motor neuron death may be due to variant-specific astrocyte-associated TDP-43 dysfunction. Additionally, variations in protein processing responses between cell types and models could also account for these differences.

Another RNA binding protein commonly associated with both ALS and FTD is the fused in sarcoma (FUS) protein. This predominantly nuclear protein also has a large role in regulating RNA metabolism events (Lagier-Tourenne and Cleveland, 2009; Dormann et al., 2010). Similar to TDP-43 pathology, FUS pathology is characterized by a simultaneous nuclear depletion and cytoplasmic mislocalization (Deng et al., 2010; Mackenzie et al., 2011). However, ALS and FTD cases with FUS inclusions do not present with TDP-43 inclusions; vice versa for cases with TDP-43 inclusions (Ling et al., 2013). ALS and FTD related variants in FUS lead to a dysfunctional protein. The ubiquitous knock down and conditional knock out of FUS in mice does not lead to ALS related motor deficits, but rather behavioral impairments (Kino et al., 2015; Udagawa et al., 2015). FUS inclusions are observed in neurons and glia in human and model tissue (see Figure 1), with a specific set of variants causing wide-spread FUS pathology triggering cell death (Kwiatkowski et al., 2009; Blair et al., 2010; Hewitt et al., 2010; Suzuki et al., 2012). Primary astrocytes transduced with FUS^R521G variant, presented with cytoplasmic mislocalization in 90% of the cultured cells and a substantial increase in GFAP protein expression (Kia et al., 2018). Additionally, motor neuron death occurred in cultures treated with conditioned media from the FUS^R521G astrocytes (Kia et al., 2018). This study indicated that the presence of FUS^R521G in astrocytes was not detrimental to astrocytes, rather it caused a toxic phenotype that triggered motor neuron death through secreted factors like TNFα (Kia et al., 2018).

The RNA binding protein adenosine deaminase acting on RNA 2 (ADAR2), was recently found to also undergo nucleocytoplasmic mislocalization (Moore et al., 2019). ADAR2 is a member of the ADAR family whose primary role is to deaminate adenosine residues, specifically in double-stranded RNA (Melcher et al., 1995; O'Connell et al., 1995). This catalytic process converts a single adenosine to an inosine resulting in A-to-I editing (Melcher et al., 1995; Hwang et al., 2016). In C9orf72-ALS/FTD human tissue from the motor cortex and spinal cord, ADAR2 was found to mislocalize and form cytoplasmic accumulations, and co-localize with TDP-43 inclusions in the spinal cord (Moore et al., 2019). Nucleocytoplasmic mislocalization of ADAR2 was also observed in C9orf72-ALS/FTD hiPSC-motor neurons and in an AAV9-(G₄C₂)₁₄₉-transduced WT mouse model (Moore et al., 2019). Substantial RNA editing aberrations associated with A-to-I editing were identified in spinal cord, cerebellum, motor and frontal cortices from C9orf72-ALS/FTD patients (Moore et al., 2019). Although a rigorous characterization of ADAR2 mislocalization and its transcriptional consequences in motor neurons were performed, there is no evidence yet supporting the mislocalization of ADAR2 in astrocytes or any other glial cells.

Collectively, these studies suggest astrocytes have a critical involvement in the degeneration of neurons through the loss of supportive and/or gain of toxic functions in various genetic subtypes of ALS and FTD. It is important to note that a C3 ⁺ reactive astrocyte transformation has been observed in C9orf72, SOD1 and sALS human tissue samples in the spinal cord and motor cortex, indicating a neuroinflammatory astrogliosis as a common pathology in ALS (Guttenplan et al., 2020). This is further supported by several studies utilizing RNA-sequencing data from ALS mouse models and iPSC-derived astrocytes, indicating that ALS astrocytes undergo a cell autonomous reactive transformation and pro-inflammatory profile in early disease stages (Ziff et al., 2021; Taha et al., 2022). However, the notion that astrocytes are directly the cause of motor neuron death is unlikely, instead it seems that diseased astrocytes exacerbate the vulnerable condition of diseased neurons. Whether this aggravation is due to a compensatory mechanism initiated by a loss of homeostatic functions and/or a toxic reactivity state is unclear. What is clear is that astrocytes do develop similar protein accumulations to neurons such as TDP-43 cytoplasmic inclusions, SOD1 nuclear accumulations, and DPRs. Nevertheless, the incidence of these protein accumulations in astrocytes is reduced compared to neurons, which suggests astrocytic clearance is either more efficient or not completely affected by underlying disease mechanisms (Yoshiyama et al., 2003; Forman et al., 2005; Serio et al., 2013; Westergard et al., 2016; Kia et al., 2018; Shi et al., 2018; Wang and Ye, 2021).

Amyloid-ß (Aß) is the product of the proteolytic cleavage of the amyloid precursor protein (APP). Cleaved sequentially by ß-secretase and γ-secretase, the Aß protein ranges from 40–42 amino acids in length (Hardy and Allsop, 1991). Identification of genetic variants in the APP, PSEN1 and PSEN2 genes associated with familial AD, were found to favor the longer Aß₄₂ peptide and enhance its production (Chartier-Harlin et al., 1991; Goate et al., 1991; Murrell et al., 1991; Citron et al., 1992; Hendriks et al., 1992; Mullan et al., 1992; Cai et al., 1993; Scheuner et al., 1996). These findings resulted in the conception of the controversial “Amyloid Cascade Hypothesis” that asserts the imbalance of Aß₄₂ production and clearance is the cause of AD (Beyreuther and Masters, 1991; Hardy and Allsop, 1991; Hardy and Higgins, 1992). The pathological distribution of Aß begins in pre-clinical stages, where deposits can be found in neocortical areas of high metabolic demand, then extending into the allocortex and brainstem whereby patients have transitioned into clinical stages (Thal et al., 2000; Buckner et al., 2005). However, failures in therapeutics against Aß have resulted in a refocus from improving the clearance of Aß plaques to also improving the clearance of NFTs and other misfolded aggregate prone proteins, and alleviating neuroinflammation—all of which are believed to contribute to neuronal death and the clinical manifestations of AD.

Surrounding Aß plaques are activated microglia and astrocytes (see Figure 1) (Pike et al., 1995; Bates et al., 2002). Astrocytes have been shown to be important mediators of amyloid-ß clearance (Thal et al., 2000). Specifically, astrocytes can internalize Aß_1-42 in vitro, with aggregate size and conformation limiting astrocytic Aß uptake (Nielsen et al., 2010). In a 2010 study, primary adult human astrocytes from the temporal lobe were shown to internalize smaller size Aß_1-42 oligomers more efficiently than fibrillar Aß_1-42 species (Nielsen et al., 2010). Furthermore, amyloid associated proteins, like ApoE, significantly altered the astrocytic uptake of oligomeric Aß, suggesting a mechanism that could ultimately effect clearance and deposition at different disease stages (Nielsen et al., 2010). Reactive astrocytes found in postmortem AD tissue of the entorhinal cortex have a significant Aß load (Nagele et al., 2003), however astrocytes with large phagocytic burden become inefficient at degradation of cell debris (Loov et al., 2012). One study found that Aß accumulation in astrocytes resulted in enlarged endosomes, where engulfed Aß₄₂ was stored instead of being actively degraded (Söllvander et al., 2016). These intracellular Aß₄₂ accumulations further caused endosomal/lysosomal defects, which led to microvesicle-induced apoptosis of neurons (Söllvander et al., 2016).

A recent study aimed at characterizing activation states within AD pathogenesis used single-nuclei RNA-sequencing to identify differences in astrocyte states in the hippocampus of the 5xFAD transgenic AD mouse model (Habib et al., 2020). This mouse model expresses five familial AD-linked variants in the human APP and PS1 transgenes, causing severe extracellular amyloid deposition, synaptic and cortical neuron loss in layer V and the subiculum, and impaired spatial and working memory deficits (Oakley et al., 2006). A continuous range of astrocyte profiles were identified and divided into six transcriptional sub-clusters. Both AD and WT groups had clusters ranging from a homeostatic Gfap-low state, to an intermediate state, into a Gfap-high state (Habib et al., 2020). The AD group had an additional Gfap-high cluster that was designated as the diseased-associated astrocyte (DAA) cluster and an intermediate cluster between the Gfap-high and DAA clusters, described as a transitional DAA state. The DAAs were associated with different molecular pathways involved in amyloid accumulation, metabolism and clearance, and the proteolytic processing of APP. The authors determined that the DAAs were only present in the hippocampus of AD mice using co-staining of GFAP⁺SERPINA3N⁺VIM⁺ astrocytes. DAAs were also found to be adjacent to Aß plaques, appearing before the onset of cognitive decline and continuing to develop in parallel with disease progression in the AD model. The DAAs were found to develop from homeostatic Gfap-cells in WT mice, who acquired these with age, suggesting a similar phenotypic DAA-like stage that also occurs in normal aging. In human postmortem tissue, DAA-like cells were identified with higher frequency in AD patients along a spectrum with the intermediate transitional DAA cluster, which suggests a dynamic activation process in AD. This study supports the notion that astrocytes take on pathogenic changes with a role in early stages of AD disease pathogenesis.

In the pathological context of AD, the 3R and 4R ratio of tau isoforms is disturbed, leading to aggregation and NFT formation. Tau pathology begins in the entorhinal cortex, spreading to limbic regions and then finally to the neocortex (Braak and Braak, 1991; Braak et al., 2011). Because tau is not expressed in astrocytes (Muller et al., 1997), the source of tau in astrocytic accumulations in AD is unknown. Nevertheless, a recent study described tau accumulation in hilar astrocytes (see Figure 1) of the dentate gyrus in patients (Richetin et al., 2020). The presence of 3R tau⁺ inclusions were more prevalent in astrocytes in the hippocampus and dentate gyrus compared to controls, with incidence increasing with disease state (Richetin et al., 2020). The occurrence of 3R tau⁺ inclusions were exacerbated when hyperphosphorylated tau and Aß plaques were observed, which correlated with synaptic alterations. Astrocytic 4R tau⁺ inclusions were present, but only in cases without hyperphosphorylated tau and Aß plaques. Furthermore, when 1N3R tau was overexpressed in astrocytes in vivo, these cells experienced impaired mitochondrial motility and function, and decreased adult neurogenesis and inhibitory synapses, ultimately contributing to impaired spatial memory in the animals (Richetin et al., 2020). In a recent study examining the transmission of tau aggregates released by neurons and its effects on receptive astrocytes, it was observed that filamentous tau aggregates bind to the integrin receptor complex ITG⍺V/ß1 located on immune-panned primary mouse astrocytes (Wang and Ye, 2021). It was observed that this integrin complex facilitated the uptake of tau fibrils while also regulating tau fibril-induced inflammation and conversion to a reactive astrocyte state similar to that of an “A1 astrocyte” (Wang and Ye, 2021). The authors proposed that astrocytic integrin receptors act as a sensing mechanism for extracellular misfolded proteins that help with clearance. However, when overburdened by these extracellular misfolded proteins astrocytes then shift towards an inflammatory reactivity state. Providing further evidence that the overexpression or presence of tau in astrocytes may lead to disruptions in their trafficking network of cellular cargo leading to degeneration of not only neurons, but also astrocytes themselves (Yoshiyama et al., 2003; Forman et al., 2005). The role of astrocytes in tauopathies, other than AD, has been extensively reviewed in Kovacs (2020).

In 57% of AD cases there is evidence of TDP-43 pathology similar to what is observed in ALS and FTD (McAleese et al., 2017). Recently updated staging schemes describe TDP-43 pathology occurring first within limbic brain regions such as the amygdala, hippocampal formation and entorhinal cortex, with early distribution in the occipitotemporal cortex and ventral striatum (stages 1–3), and finally extending into neocortical areas such as the mid-temporal and mid-frontal cortex, as well as the basal ganglia (stages 4 and 5/6) (Josephs et al., 2016; Nag et al., 2018). TDP-43 inclusions are predominantly found in neurons, but there are reports of glial cytoplasmic inclusions (GCI) of TDP-43 in AD tissue (see Figure 1). However, GCIs are found only in transferrin-positive oligodendrocytes (Higashi et al., 2007).

A recent study examined whether TDP-43 in AD represents a coexistence of AD and FTD-TDP or if the two TDP-43 pathologies are discrete. They identified distinct molecular patterns of TDP-43 species in AD human tissue of hippocampus, entorhinal, frontal, temporal and occipital cortex, amygdala (Tome et al., 2020). They found a range of TDP-43 inclusion patterns from the absence of inclusions to inclusions only positive for pTDP-43 (serine 409/410), and inclusions positive for pTDP-43 (serines 403/404 and 409/410) and full length TDP-43 (N and C-terminal domains), similar to that of FTLD-TDP. About 20% of AD cases with only pTDP-43 inclusions, presented with language deficits, and about 36% of AD cases positive for both pTDP-43 and full length TDP-43 experienced FTD-like behavioral or language problems. This further supported the hypothesis that distinct phosphorylation sites have an impact on the type of TDP-43 pathology and its relation to FTD-like symptoms observed in AD. However, glial TDP-43 inclusions were not individually assessed, but were considered in the classification of samples.

Limbic-predominant age-associated TDP-43 encephalopathy-related neuropathological changes (LATE-NC), is a newly defined TDP-43 related proteinopathy with inclusions primarily in limbic regions of the brain (Nelson et al., 2019). LATE-NC affects up to 50% of individuals past the age of 80 years old and individuals with or without AD (Nelson et al., 2019). LATE-NC clinically manifests with amnestic cognitive syndrome affecting emotional lability and global cognitive function, which is independent of co-existing pathologies (Nelson et al., 2019). TDP-43 proteinopathy extends from the amygdala to the hippocampus, then to the middle frontal gyrus; this distribution can occur with or without hippocampal sclerosis (Nelson et al., 2019). Individuals with LOAD with TDP-43 pathology in limbic regions are now recognized as having simultaneous LATE-NC. A diagnosis of ALS and/or FTLD-TDP excludes a diagnosis of LATE-NC. Unlike ALS and FTD, there are rare mentions of glial TDP-43 inclusions associated with LATE, however there is no clarification on glial cell type (Hunter et al., 2020).

Finally, there are a handful of studies implicating TDP-43 interactions with Aß and tau. Studies investigating the relationship between Aß and TDP-43 pathology in AD have found that Aß₄₂ expression in vivo upregulated endogenous TDP-43 and caused cytosolic pTDP-43 inclusions (Herman et al., 2011; Herman et al., 2012). Using an AD mouse harboring APP and PS1 variants (APP/PS1 model), the injection of TDP-43 oligomers increased Aß plaques and further impaired spatial memory (Shih et al., 2020). The relationship between tau and TDP-43 has indicated that TDP-43 has a role in assisting with the MAPT mRNA splicing leading to a decrease of 4R isoforms by exclusion of exon 10 (Gu et al., 2017). Furthermore, ALS/FTD related TDP-43 variants led to inclusion of exon 10, resulting in an increase and disruption of the 3R:4R ratio (Gu et al., 2017). Specificity of phosphorylation sites on TDP-43 dictate its ability to suppress exon 10 in MAPT mRNA (Gu et al., 2018). Additionally, TDP-43 and tau have been shown to have a collaborative seeding process in the brain whereby TDP-43 oligomers could serve as a nucleation site for tau aggregates (Montalbano et al., 2020). However, these interactions have not been identified or studied in astrocytes or any other glia.

The role of astrocytes in AD is ultimately important to maintain the integrity of neurons; however, there is evidence that AD pathology has an early and eventual impact on astrocytic integrity caused by a reactive transformation. Both Aß and tau pathology have been observed in relation to astrocytes in both models of AD and AD human tissue. In a recent study investigating how the characteristic AD proteinopathies influence astrocytes, TRAP-seq data from mouse models of tauopathy and ß-amyloidopathy revealed that neither of these proteinopathies triggered “A1” or “A2” reactivity profiles in these mice, but rather model specific deleterious and adaptive-protective profiles (Jiwaji et al., 2022). Specifically, these pathologies induced activation of inflammatory pathways, alongside protein degradation pathways, all the while exhibiting an induction of a set of genes normally elevated in aged mouse brains (Jiwaji et al., 2022). Thus, furthering the argument that astrocyte reactivity is context and disease dependent. TDP-43 pathology has also been observed in glia, but the identification of TDP-43 in an AD context in astrocytes has yet to be properly described. There is much more research needed in describing AD-related pathology in astrocytes and identifying astrocyte specific perturbations that occur due to the accruement of Aß, tau, and TDP-43 pathologies.