Astrocytes play a major role in glutamate uptake from the surrounding neuronal synapsis and its posterior recycling into glutamine, which can then be reused by neurons as a substrate for glutamate synthesis. To this end, astrocytes present several glutamate transporters on their cell surface in direct contact with tripartite synapsis space, including the Na⁺-dependent transporters EAAT1 and EAAT2 (human gene names), also known as GLAST and GLT-1 (mouse gene names). It has been reported that the astrocyte-dependent glutamate buffering system becomes altered shortly after ischemia in several ways. These include epigenetic modulation of GLT-1 and GLAST promoters, resulting in lower gene expression; aberrant histone methylation giving rise to dysfunctional, but not increased, expression ^[1][72]; and S-Nitrosylation of GLT-1 with a concomitant reduction in its activity ^[2][73]. At a later stage, a decrease in ATP levels in astrocytes, mainly in the ischemic core, induces glutamate transporters reversal, further contributing to glutamate excitotoxicity and neuronal damage ^[3][74]. In vivo upregulation of GLT-1 using ceftriaxone ^[4][75] or by targeted overexpression with adeno-associated viral vectors ^[5][76] was neuroprotective and reduced brain damage. Similar results were obtained in cultured astrocytes subjected to OGD ^[6][77]. Carnosine was also proved to preserve GLT-1 activity in astrocytes after pMCAO, improving neurological function and decreasing infarct size ^[7][78]. A compound similar to ceftriaxone, sulbactam, was efficiently used to prevent hippocampal neuronal damage in a rat global brain ischemia model, while this effect was suppressed by either antisense knockdown or pharmacological inhibition of GLT-1 using dihydrokainate ^[8][79]. Furthermore, antisense knockdown of astrocytic GLT-1, but not EAAC1 neuronal glutamate transporter, increased neuronal damage induced in a transient focal cerebral ischemia (tMCAO) rat model ^[9][80], highlighting the special relevance of astrocytes in buffering ischemia-induced excitotoxicity. Nonetheless, astrocytes are also known to exacerbate excitotoxicity upon ischemia by releasing glutamate into the synapsis through volume-sensitive outwardly rectifying anion channels (VRACs) and connexin hemichannels. In fact, knocking out Swell1 (Lrrc8a), the only obligatory subunit of astrocytic VRACs, results in reduced brain damage and better neurological outcome upon I/R ^[10][81]. In accordance with this, pharmacological inhibition of VRAC using tamoxifen in MCAO-subjected rats reduced infarct area and improved neurological outcome ^[11][82]. Additionally, astrocytic P2X7 receptors (P2X7Rs) are also able to release glutamate upon ATP binding ^[12][83]. Given the increase in ATP extracellular concentration that takes place in the early phases of ischemic injury, P2X7Rs could be significantly contributing to excitotoxicity.

Connexin 43 (Cx43) is the predominant connexin in astrocytes and localizes to the cell surface where it conforms gap junctions and hemichannels. It has been described that, early after an ischemic insult (1–30 min), Cx43 gap junctions are phosphorylated by several kinases, like MAPK, PKC, pp60Src, and casein kinase 1δ, which triggers internalization of Cx43 hexamers. The remaining Cx43 hemichannels are dephosphorylated in a subsequent stage (after 60 min), increasing their opening probability and allowing harmful molecules into the extracellular space (ECS), like ATP and glutamate ^[13][84]. Leptin was found to suppress Cx43 rise after I/R reducing brain damage in a mouse model of MCAO, and it also blocked Cx43 hemichannels in cultured U87 cells ^[14][85]. Mimetic peptides are another way to block hemichannels. More precisely, Gap19 is a highly selective blocker that spares gap junctions at the initial moments of treatment (first 30 min), while effectively blocking hemichannels. At higher exposure times Gap19 slightly inhibits gap junctions. Suppression of Cx43 by Gap19 showed beneficial effects in MCAO mouse models ^[13][84].

Neurons heavily depend on glucose oxidative metabolism for their normal functioning, which makes them selectively vulnerable to hypoxia/hypoglycaemia ^[15][86]. For this reason, the role of astrocytes as metabolic supporters is key for neuronal survival during an ischemic insult. Neuronal energy demands can be mirrored by vascular regulation through astrocytic signaling pathways in what is known as neurovascular coupling. Released glutamate upon synaptic activity is bound by mGluR5, which then leads to a rise in intracellular Ca²⁺ concentration ([Ca²⁺]_i) through PLCβ1 activation. This event triggers the release of arachidonic acid (AA) from the membrane. In the presence of low surrounding oxygen pressure, as happens upon ischemia, AA is preferentially transformed into PGE2 by COX-1, and is then exported inducing vasodilation and diffusion of oxygen and glucose from the nearby blood vessels into brain parenchyma ^[16][87].

Astrocytes provide neurons with lactate as a precursor for the tricarboxylic acid (TCA) cycle, in a proposed mechanism known as “astrocyte-neuron lactate shuttle hypothesis” ^[17][88], which remains controversial ^[18][89]. As previously stated, synaptic glutamate uptake by astrocytes triggers Na⁺-K⁺ ATPase, which in turn stimulates glycolysis and glycogenolysis to produce lactate ^[19][90]. Metabolism of lactate inside neurons implies that the pyruvate dehydrogenase complex (PDHC) is susceptible to inactivation through oxidative stress generated upon ischemic insults ^[20][91]. Energy depletion in an ischemic scenario leads to an increase in AMP levels and activation of AMPK, which in turn phosphorylates and thus inactivates acetyl-coA carboxylase, with the subsequent decrease in malonyl-CoA, a natural inhibitor of mitochondrial carnitine palmitoyltransferase I (CPT-I) ^[21][92]. Consequently, hypoxia/hypoglycaemia lead to increased activity of CPT-I and higher production of ketone bodies (KBs) through mitochondrial β-oxidation of free fatty acids (FAs) obtained from the bloodstream ^[20][91]. Astrocytes release KBs into the ECS, captured by neurons through MCTs and used as precursors for TCA cycle. Given that PDHC becomes partially inhibited in the presence of I/R-derived ROS, KBs become the most important source of energy over lactate after brain ischemia ^[22][93].

Exogenous KBs could constitute an interesting therapy for I/R injury. KBs increase mitochondria health and activity, reduce ROS and astrogliosis ^[23][94], and increase neurotrophin secretion (BDNF, bFGF) ^[24][95]. It has been recently reported that the axis SIRT3–FoxO3a–SOD2 becomes upregulated upon treatment with KBs, increasing mitochondria complex I activity and reducing protein oxidation, with a concomitant improvement in neurological outcome after an ischemic insult ^[25][96]. Adiponectin could also constitute a good therapy as it promotes oxidation of FAs and production of KBs through AMPK activation ^[26][97].

Endogenous ROS in the CNS is generated by the mitochondrial electron transport chain and NADPH-oxidized pathway, while reactive nitrogen species (RNS) mainly proceed from L-arginine metabolism by nitric oxide synthase (NOS) ^[27][98]. Clearance methods can be classified into enzymatic and non-enzymatic. The former group includes Nrf2-controlled ones, catalases, SODs, and glutathione peroxidase. Non-enzymatic methods consist of molecules able to scavenge ROS/RNS including glutathione (GSH), bilirubin, uric acid, melatonin, and vitamins C and E. Another non-enzymatic method is the thioredoxin (Trx) system, where NADPH is used to reduce cysteine residues on Trx, making it a potent antioxidant ^[28][99].

Neurons are less efficient than astrocytes in dealing with oxidative stress owing to a continuous repression of Nrf2, which is a master regulator of redox genes including GCLC, glutathione reductase, NQO-1, and HO-1 ^[29][100]. Increased levels of Nrf2 have been observed in I/R injury mainly in the penumbra, both in mouse and human ^[30][101]. It has recently been described that Nrf2 activation in astrocytes relies on glutamate binding to NMDA receptors (NMDARs), which suffer subunit composition changes in different models of ischemia ^[31][102]. GluN3A is known to increase in MCAO mice ^[32][103], rendering lower [Ca²⁺]_i elevations, which is expected given the inhibitory effect of GluN3A on NMDARs ^[33][104]. This could negatively impact GSH production and global antioxidant capacity of astrocytes.

One detrimental effect of ROS accumulation in astrocytes during ischemia is the activation of NLRP3 inflammasome. This process may depend on a two-step event (priming and activation) or on a single event (activation). Detection of ischemia-related DAMPs by TLRs can prime NLRP3 transcriptionally through NF-kB activation, which induces expression of NLRP3 and pro-inflammatory cytokines in a process partially dependent on mtROS ^[34][105]. Another way of NLRP3 non-transcriptional priming is through its deubiquitination by BRCC3, which can be triggered by mtROS and is a crucial step for NLRP3 activation ^[35][106]. ROS accumulation can directly activate NLRP3, promoting the release of TXNIP from Trx, to facilitate inflammasome polymerization ^[36][107].

Adiponectin (APN) is an adipose tissue-derived hormone released into the bloodstream that increases upon ischemia ^[37][108] and presents neuroprotective properties ^[38][109]. A very recent study proves that APNp, an APN-derived peptide able to cross the BBB, reduces ROS and NLRP3-mediated inflammation. APNp was shown to increase AMPK activation, Nrf2 nuclear translocation, and Trx1 levels ^[39][110]. Ascorbic acid is another molecule able to scavenge ROS directly. It is produced in astrocytes by GSH-mediated reduction and then transported into neurons ^[40][111]. Oral administration of nanocapsuled ascorbic acid has been shown to reduce ROS-mediated mitochondrial damage ^[41][112]. Peng et al. ^[42][113] recently described that DJ-1, which is an important antioxidant molecule mainly produced by reactive astrocytes, exerts a neuroprotective function upon ischemia through upregulation of Nrf2 and a concomitant increase in GSH levels. The AMPK-PGC-1α axis, which is induced upon ischemia owing to an increase in AMP levels, drives expression of GCLM specifically in astrocytes, and thus facilitates GSH synthesis. Accordingly, those AMP analogous molecules, like metformin and AICAR, improve neuroprotection and are good candidates for therapies ^[43][114]. Astroglia-specific ROS scavengers metallothionein(MT)-I and MT-II presented increased mRNA levels early after brain ischemia and deficient mice for these two proteins presented larger infarct sizes after ischemic injury compared with control mice ^[44][115].

Given the conspicuous relevance of ROS-mediated neurotoxicity in I/R, promoting the aforementioned astrocyte-related mechanisms to scavenge these toxic species represents a promising therapeutic approach in stroke, especially those Nrf2-centered strategies.

Astrocytes play a prominent role in the maturation and maintenance of the BBB by controlling water abundance, ion homeostasis, and other osmotically-active molecules (Figure 2). Astrocytes endfeet cover almost all the vessel surface stablishing closed contacts between them as the main glial component of the NVU. This structure strictly controls the diffusion of molecules into the brain parenchyma.

At early phases of ischemia, astrocytes become reactive and swell as a result of increased uptake of glutamate, K⁺, and lactate at the endfeet, but also due to Na⁺/K⁺ ATPase failure. Both these factors induce a change in morphology in astrocytes that cannot retain their normal functions and lose physical connections with endothelial cells (ECs) ^[45][71]. AQP4 is the main way in which water goes into astrocytes upon ischemia, resulting in the dysfunction of the endfeet, and deletion of this transporter improves the outcome after the insult, reducing swelling and edema ^[46][116]. Additionally, Na⁺/H⁺ exchanger isoform 1 (Nhe1) has been shown to be abnormally activated upon ischemia, which provokes an overload of intracellular Na⁺ and a concomitant astrocyte swelling ^[47][117]. Astrocytic-selective Nhe1 KO improves BBB integrity after tMCAO and reduces astrocyte activation, pro-inflammatory cytokine secretion, and hemispheric swelling, improving the neurological outcome ^[48][118].

Prompted by an ischemic insult, astrocytes secrete several factors with dual effects over permeability and integrity of the BBB. Some of those factors are classified here into negative effectors and positive effectors.

VEGF is a factor that promotes vascular permeability through a downregulation of TJPs and angiogenesis. In different models of I/R, inhibition of VEGF signaling proved to be beneficial for BBB integrity maintenance, either by blocking VEGF with specific antibodies ^[49][119] or VEGFR2 with SU5416 or by genetic ablation ^[50][120].

Another group of negative effectors is MMPs, in charge of degrading TJPs and extracellular matrix in a physiological process that facilitates angiogenesis and BBB permeability. It has been described that reactive astrocytes produce and secrete MMP-2 and MMP-9 ^[51][121]. Several studies reported an increase in the levels and activity of MMP-2 and MMP-9 after an ischemic insult in animal models and cell lines (for a review, see Michinaga and Koyama, 2019). Zhang et al. ^[52][122] showed that MMP-2 and MMP-9 contributed to degradation of ZO-1 protein, leading to BBB disruption, while the MMPs’ inhibitor SB-3CT reduced BBB permeability. Another study showed beneficial effects of MMPs’ inhibitor BB-1101 administration after brain ischemia ^[53][123]. Genetic ablation of MMP-9 also proved to prevent proteolysis of BBB after ischemia ^[54][124].

Astrocytes can produce NO by iNOS upon several stimuli like IL-1β and ROS ^[55][125]. NO has a known negative impact over TJPs on the BBB ^[56][126]. In line with this, NOS inhibitor Nitro-L-arginine methyl ester (L-NAME) was able to prevent BBB disruption after focal ischemia ^[57][127]. It is also known that astrocyte-released glutamate can induce NO synthesis by eNOS through NMDAR activation in endothelial cells, resulting in vasodilation and higher permeability ^[58][128].

Astroglia are also an important source of endothelin 1 (ET-1), which promotes BBB permeability ^[59][129]. In fact, ET-1 overexpression in astrocytes aggravates neurological outcome of I/R in several animal models, leading to higher brain edema, BBB disruption, neurodegeneration, and mortality that can be partially corrected by ABT-627, an ETA receptor antagonist ^[60][130]. Other ET receptor antagonists were also proved to have beneficial effects, as was the case for S-0139 in I/R rats, which reduced BBB permeability, edema, and infarct size ^[61][131].

Sonic Hedgehog (Shh) is a factor expressed mainly by astrocytes in the CNS, which drives expression of anti-apoptotic genes and promotes cell proliferation and progenitor self-renewal through activation of transcription factor Gli1a. It has been shown that astrocyte-released Shh promotes BBB formation and integrity through Shh receptors in ECs ^[62][132]. Recombinant Shh diminished BBB leakage after I/R by activating angipoietin-1, which promoted an increase in ZO-1 and occludin expression ^[63][133]. Shh signaling pathway was also protective against ECs’ apoptosis ^[64][134].

Astrocytes also increase their basal expression of RALDH2 in different brain inflammation paradigms, giving rise to higher retinoic acid levels, which was shown to be beneficial for BBB integrity ^[65][135]. Although not investigated yet, retinoic acid could be beneficial against I/R-induced BBB damage provided it is an inflammatory scenario.

Additionally, astrocyte-derived IGF-1 is known to exert a neuroprotective role against brain damage ^[66][136]. Actually, human IGF-1 gene transfer controlled by GFAP promoter showed to improve neurological outcome after I/R in aged rats ^[67][137]. Furthermore, IGF-1 reduced ECs’ apoptosis, BBB permeability, and infarct volume in rats after I/R ^[68][138].

Finally, to fight edema, astrocytes are also able to release osmotically active molecules like taurine upon ischemia, in what is called “regulatory volume decrease”. Release of taurine seems to be regulated by intermediate filaments-controlled channels, given that astrocytes from GFAP^−/−Vim^−/− mice showed reduced secretion of this factor ^[69][139].

Activated astrocytes are key players in the response to many different brain damages. This activation is mainly characterized by hypertrophy, increased proliferation, and newly acquired specific functions driven by gene expression regulation. This process is known as “reactive astrogliosis” ^[70][71][140,141].

Early after a brain stroke, there are several stimuli triggering astrocytes’ activation, like the release of neurotransmitters from nearby dying neurons, blood extravasation, hypoxia, and cell death, as well as release of cytokines from injured neurons and microglia, including TGF-α, CNTF, IL-1, IL-6, and KLK6 ^[45][71].

Within minutes after activation, reactive astrocytes produce and secrete many different pro-inflammatory molecules with detrimental effects over neuronal viability, like the cytokines IL-6, TNF-α, IL-1α, IL-1β, and IFN-γ, as well as ROS/RNS ^[72][142]. P2Y1 astrocytic receptors are known to induce release of pro-inflammatory cytokines and chemokines, aggravating brain damage after stroke ^[73][14]. In fact, it has been described that P2Y1 agonist treatment has a deleterious effect, increasing infarct area ^[74][143], while antagonists over P2Y1 have a positive effect on astrocyte viability upon ischemic damage ^[75][144]. TLR4 stimulation by ischemia-generated DAMPs on astrocytes’ cell surface induces secretion of IL-15, which acts as a chemoattractant for CD8+ T cells and natural killer cells, aggravating brain damage ^[76][145]. Consistently, antibodies directed against IL-15 improve the disease outcome ^[77][146]. On the other hand, astrocytic TGF-β cytokine has a neuroprotective function and its specific inhibition in astrocytes results in excess inflammation and higher damaged brain area ^[78][147]. This evidence highlights the dual role of reactive astrocytes upon ischemia. In fact, two main reactive astrocyte phenotypes have been proposed (Figure 2). A1 astrocytes are mainly induced by IL-1α, TNF-α, and C1q secreted by microglia and present a neurotoxic phenotype. By contrast, A2 phenotype is neuroprotective through neurotrophic factor release ^[70][79][140,148]. Ischemia-induced reactive astrocytes tend to neuroprotective phenotypes, as stated by RNA-seq data analysis of MCAO mice, which showed a predominance of A2-specific genes ^[80][149].

Days after an ischemic insult, a glial scar can be detected, displaying a Janus-faced role. Glial scar formation requires astrocyte multiplication in the penumbra and a posterior migration to the core border, where they secrete extracellular matrix proteins to avoid immune cells’ infiltration towards healthy tissue ^[81][150]. Astrocytes secrete chondroitin sulphate proteoglycans (CSPGs) as the main integrating factor of the glial scar with a well described role in axonal sprouting inhibition. Interestingly, cholinesterase ABC significantly reduces the inhibitory effect of CSPGs over axonal growth after brain ischemia in rats ^[82][151]. It has been described that GFAP^−/−Vim^−/− mice show less organized and less compact glial scar after brain injury ^[83][152] with larger infarct areas ^[84][153], suggesting a key role of intermediate filaments in proper astrocyte activation and scarring. On the contrary, knockout mice for CD36 astrocytic receptor, which mediates activation and glial scar formation, present 49% less infarct volume after an ischemic injury than their WT counterparts ^[85][154], and an attenuation of GFAP induction and glial scar formation ^[86][155]. It has been recently described that reactive astrocytes also acquire phagocytic activity in the penumbra of MCAO mice a few days after the insult, and this process depends on the upregulation of ABCA1, MEGF10, and GULP1 ^[87][156]. It is known that knockdown or knockout of any of the previously mentioned factors decreases astrocytes’ ability to phagocyte in vitro. Moreover, ABCA null mice present larger damaged brain area after ischemia ^[88][157], suggesting that the previously unsuspected role of astrocytes in debris clearance is key for ischemic injury resolution.

Promoting neurogenesis is a promising therapeutic approach after an ischemic insult, where astrocytes, alike in other processes, play a dual role. As previously mentioned, glial scar is a good example of this behavior. It is known that glypican treatment, which is normally deposited by astrocytes in the peri-infarct region, reduced GFAP immunoreactivity and scar formation, but it also improved neurite outgrowth and behavior outcome ^[89][158]. In line with this, inhibition of the glial scar-forming enzymes chondroitin polymerizing factor and chondroitin synthase-1 by RNAi techniques improved neurite outgrowth ^[90][159].

Astrocytes secrete several factors able to differentially influence neurogenesis and plasticity after stroke. Ephrin-A5, which has a well-known inhibitory effect over axonal growth, is secreted by astrocytes in the peri-infarct zone. In fact, its inhibition promotes axonal outgrowth and functional recovery after stroke ^[91][160]. In contrast to this, astrocytes increase their basal expression of thrombospondins (Thbs1/2) upon ischemic injury, and those are essential for synaptic plasticity and functional recovery ^[92][161]. Additionally, it has been described that astrocytes secrete the chemokine stromal cell-derived factor-1 after MCAO, which functions as a neuroblast attractor to the ischemic lesion ^[93][162]. It was also shown that HMGB1, which is secreted by astrocytes upon ischemia, stimulates neural stem/progenitor cells differentiation into neurons in a PI3K/Akt-dependent manner ^[94][163].

Astrocyte-derived neurotrophic factors are crucial for neuronal survival after stroke. It is known that astrocyte-conditioned medium applied to MCAO brain reduces infarct volume ^[95][164]. BDNF is upregulated hours after stroke and its artificial overexpression in ischemic rat brain promotes neurite outgrowth ^[96][165]. Treatment with EPO also improves neurogenesis, brain remodeling, and neurorestoration after stroke in the perilesional zone and in the contralesional zone ^[97][166]. Activity-dependent neuroprotective protein (ADNP) ^[98][167] is a neurotrophic factor secreted by astrocytes with relevant roles in axonal transport stimulation, autophagy, neuronal sprouting, cell survival, learning, and memory ^[99][168]. An eight-amino acid ADNP-derived peptide named NAP, which is the minimal active peptide retaining neuroprotective properties, has been successfully used against brain stroke (MCAO) ^[100][101][169,170].

Interestingly, it has been recently described that striatal astrocytes can differentiate into neurons guided by a latent neurogenic program repressed by Notch signaling in basal conditions. Upon ischemia, Notch receptors 1, 2, and 3 along with their ligands DII1, Jagged1, and Jagged2 are downregulated, thus triggering astrocyte-dependent neurogenesis ^[102][171]. Direct reprogramming of glial scar astrocytes into neurons has also been achieved in brain-injured mice using viral-delivered NeuroD1 transcription factor ^[103][172].

Ischemic tolerance (IT) induction or ischemic preconditioning (PC) consists of reducing the damage caused in a severe ischemic episode by provoking a previous mild ischemic insult and has recently appeared as an exciting therapeutic approach for I/R injury ^[104][173]. Astrocytic activation seems to be essential for induction of ischemic tolerance as previously preconditioned brains of GFAP^−/−Vim^−/− mice show greater infarct volume upon ischemia compared with preconditioned brains of WT mice ^[84][153]. Hirayama et al. ^[105][174] showed that inhibition of initial microglial activation with minocycline did not prevent IT, while astrocyte inhibition with fluorocitrate showed no effective PC, suggesting that astrocytes are the relevant glial cell type in this process. They also showed that the axis P2X7R-HIF-1α is crucial for an effective IT induction. Moreover, it was described that a selective KO of neuronal HIF-1α does not prevent IT, indicating that the process does not depend on neuronal HIF-1α ^[106][175]. Another study has recently shown that exercise-mediated PC upregulates P2X7R and HIF-1α ^[107][176], reinforcing the role of these two factors in IT induction.

After a brief ischemic insult, reactive astrocytes increase glutamate transport, connexins, ZO-1, aquaporins, and neurotrophic factor release, all of which are supposed to mediate neuroprotection, but at the same time, ischemic PC is also protective for astrocytes, an effect that seems to be mediated by Nrf2 in primary astrocytes subjected to OGD ^[108][177]. These PC-treated astrocytes are more efficient, protecting neurons upon OGD possibly due in part to higher lactate release ^[109][178]. In line with this result, cross-tolerance mechanisms like epileptic PC seem to strongly depend on lactate transport into the mitochondria ^[110][179]. Other studies suggest that exosomes released by PC-treated astrocytes, which are then engulfed by neurons, could contain neuroprotective molecules such as miR-92b-3p ^[111][180].

Remote ischemic preconditioning (RIPreC), usually induced by blocking blood flow to the limbs, has repeatedly proved to induce tolerance against subsequent I/R damage in animal models and clinical trials ^[112][181]. In fact, it is known that patients with a history of peripheral vascular disease (PVD) present lower infarct volumes, better outcomes, and lower mortality rates after an ischemic insult ^[113][182]. Although there is no direct evidence supporting a role for astrocytes as the mediators of the mentioned protective effect, it could be interesting to test whether inhibition of astrocyte activation abrogates tolerance induction by RIPreC.