The brain is highly vulnerable to ischemia in part due to its high metabolic demands. Although it constitutes only about 2% of total body mass, the brain consumes approximately 20% of the body’s oxygen and 25% of its basal glucose intake [27]. These high levels of oxygen and glucose are essential for generating the ATP required to maintain and restore membrane potential and to synthesize neurotransmitters [28]. It is estimated that glutaminergic synapsis accounts for most (~80%) of the brain’s energy expenditure, highlighting the close relationship between brain activity, glutamatergic neurotransmission, and energy requirements [29]. The brain lacks the ability to store oxygen and has only limited reserves of carbohydrate substrates and high-energy phosphate compounds. Since neurons have high energy demands, relying almost exclusively from ATP generated through mitochondrial oxidative phosphorylation, they require a continuous oxygen supply, making the brain particularly susceptible to ischemia [12].

The first effect of ischemia following its onset is a severe decrease of oxygen and glucose supply to neurons, leading to insufficient ATP synthesis by the electron transport chain (ETC) to sustain Na⁺/K⁺-ATPase normal activity. Within minutes of AIS onset, neurons affected by ischemia lose the ability to maintain Na⁺ and K⁺ gradients across the cell membrane. This disruption activates voltage-sensitive Ca²⁺ channels, causing excessive Ca²⁺ influx into neurons and triggering the excessive release of neurotransmitters, particularly glutamate, into the extracellular space [30–34]. Simultaneously, neurotransmitter reuptake from the extracellular space is reduced [35, 36], further contributing to their accumulation in the extra-neuronal space (Figure 1).

In 1969, the term excitotoxicity was coined to describe the condition in which excessive glutamate acts on excitatory receptor, leading to cell death [37]. Excitotoxicity refers to neuronal death mediated by the excessive activation of ion channel-linked glutamate receptors, particularly N-methyl-D-aspartate (NMDA) receptors, resulting in cellular Ca²⁺ overload [32, 38–41], although astrocytes may also suffer damage by excessive glutamate levels [42]. Within the complexity and heterogeneity of mechanisms leading to neuronal death, glutamate excitotoxity is the central process underlying delayed neuronal death in AIS [39, 43–45]. Excitotoxicity disrupts cellular Ca²⁺ homeostasis, promotes free radical generation, induces mitochondrial dysfunction and triggers apoptosis [40]. Although each of these mechanisms can independently cause neuronal death, they act synergistically.

In the brain, glutamate is the most abundant neurotransmitter, playing essential roles in presynaptic and postsynaptic neurons as well as glial cells. It exhibits a Janus effect: While appropriate glutamate release is crucial for physiological functions and neuronal survival, excessive glutamate release contributes to pathological changes, including cell death [32, 34, 39, 46, 47]. Glutamate concentration at excitatory synapses is exquisitely fine-tuned regulated, as disruption of its homeostasis can lead to brain pathology, as observed in several neurodegenerative diseases. In the resting state, the extracellular glutamate concentration in neurons is mantained below 1 μM, while its cytoplasmic concentration is significantly higher (approximately 2 mM) and its concentration within synaptic vesicles reaches 100 mM. Upon the arrival of an action potential at presynaptic terminals, glutamate is released into the synaptic cleft, transiently increasing its concentration to approximately 1 mM [48]. However, this elevated concentration lasts only a few milliseconds, as glutamate is rapidly removed through enzymatic degradation and reuptake by high-affinity glutamate transporters (excitatory amino acids transporters, EAAT) located in pre- or post-synaptic neurons, but primarily on astrocytes, from which it is recycled into pre-synaptic neurons by the glutamate-glutamine cycle [49, 50]. Since astrocytes are responsible for the majority of extracellular glutamate uptake, whereas only a small fraction is recovered by neuronal uptake [46], they play a central role in the glutamate cycle in the brain. In astrocytes, glutamate is converted into glutamine by the enzyme glutamine synthetase, which is exclusively expressed in these cells [51]. The synthesized glutamine is then released by astrocytes and taken up by presynaptic neurons via glutamine transporters located on the neuronal membrane. Inside the neuron, glutamine is deaminated into glutamate, which then accumulates in synaptic vesicles, ready to be released into the synaptic cleft during neurotransmission [49, 50].

After an ischemic stroke, within a few minutes, neurons release high levels of glutamate, which can exceed the capacity of the glutamate-glutamine cycle to maintain homeostatic glutamate levels. The presence of elevated glutamate in the extracellular medium leads to the activation of both synaptic and extra-synaptic receptors in neighboring neurons.

In the brain, the neurons have two kinds of the glutamate receptors: a) ionotropic glutamate receptors (iGluRs), which include NMDA, AMPA, and kainite receptors, are tetramers forming a central pore allowing ion flow upon activation; b) metabotropic glutamate receptors (mGluR), which are coupled with a G-protein [52–54]. Although all glutamate receptors contribute to the excitotoxic cascade [53], the NMDA glutamate receptor (NMDAR), located in the synaptic and extra-synaptic neuron membranes of the CNS [55, 56], is considered the primary mediator of excitotoxic damage in AIS [54, 57, 58]. NMDAR is a tetrameric protein complex composed of two essential GluN1 subunits and two GluN2 and/or GluN3 subunits. More than a dozen NMDAR subtypes have been characterized and this diversity is critical for its role in cell survival and death [58, 59]. Unlike other receptors, such as AMPA and kainite, NMDAR is a voltage-dependent channel that remains blocked by Mg²⁺ under resting conditions. AMPA receptor (AMPAR) activation by glutamate leads to membrane depolarization, which removes the Mg²⁺ block from NMDAR. This allows glutamate binding to NMDAR, opening the channel and permitting Ca²⁺ influx into the postsynaptic neuron due to its high permeability to Ca²⁺ and other cations [54].

Depending on the type of subunit that forms the tetramer, NMDAR exhibits different functions [60]. The properties of NMDAR depend on the type of GluN2 subunit that makes up the GluN1/Glun2 complex [61, 62]. It has been proposed that NMDARs containing GluN2B are excitotoxic and are predominantly located at the extra synaptic sites, while NMDARs containing GluN2A are neuroprotective and are primarily localized at the synaptic sites [55, 63]. In AIS, specific GluN2 subtypes (A–D) appear to play a pivotal role [64]. In neurons, NMDARs are part of a structure known as postsynaptic density (PSD), which appears electron-dense under an electron microscope [65, 66]. The PSD includes members of the membrane-associated guanylate kinase family (MAGUKs), which are essential structural proteins that anchor and organize NMDARs, governing the overall molecular organization of PSD. These MAGUK proteins consist of three PSD 95/large/zonula occludens-1 (PDZ) domains and one SH3-GK supermodule [66]. The PDZ domains of MAGUK proteins bind to the GluN2 subunits of the NMDARs [67–69]. Increased Ca²⁺ input induces binding of neuronal nitric oxide synthase (nNOS) to NMDAR via NMDAR-PSD95-nNOS complex [70].

In ischemic brain, excessive activation of the NMDAR-PSD-nNOS complex leads to excessive Ca²⁺ influx and activation of nNOS, resulting in the overproduction of nitric oxide (NO^•) [65]. This leads to an accumulation of Ca²⁺ and NO^• in the neuronal cytosol [71–74]. Excess of NO^•, together with hydrogen peroxide (H₂O₂), which is produced from superoxide radical (O₂^•–) by the enzyme superoxide dismutase (SOD), triggers the formation of peroxynitrite (ONOO^–), a highly reactive free radical that promotes cellular damage and ultimately neuron death [71, 72]. In AIS, Ca²⁺ entry through NMDARs activates cell death pathways more effectively than Ca²⁺ influx through other channels [75]. However, the efficiency of Ca²⁺ in triggering the excitotoxic signaling cascade via NMDAR can be reduced by disrupting the interaction between components of the NMDAR-PSD-95 or nNOS-PSD-95 complexes. In experimental animals, altering the interaction of NMDAR-PSD-95 or nNOS-PSD-95 complexes using small peptides has been shown to reduce the excitotoxic effects of Ca²⁺ and increase neuronal resistance to focal cerebral ischemia [76, 77]. The importance of the NMDAR/PSD-95 complex interaction in cerebral ischemia is further supported by findings showing that inhibiting PSD-95 binding to the NMDAR decreased brain damage without affecting NMDAR function [77]. In neuronal cultures, the binding of low-molecular-weight peptides to PDZ domains in PSD-95 induces structural changes in the PSD-95 complex, decreases NO^• production, and reduces excitotoxicity [65].

It is well established that when excessive Ca²⁺ input surpasses the ability of mitochondria and the endoplasmic reticulum to uptake Ca²⁺ and regulate Ca²⁺ homeostasis in the cytosol, the resulting Ca²⁺ overload activates a cascade of mechanisms leading to neuronal death in the AIS [9, 73, 78]. There is strong evidence that loss cellular Ca²⁺ homeostasis is a key factor in neuronal death following an AIS [75, 78, 79]. As in other organs, Ca²⁺ is essential for proper brain function, so maintaining Ca²⁺ homeostasis in neurons is strictly regulated and any disruption can have severe consequences for the brain. Neurons possess specialized homeostatic mechanisms to control cytosolic Ca²⁺ concentration [78].

In the resting state, there is a significant difference in free Ca²⁺ levels between the extracellular medium, where Ca²⁺ concentration is estimated at 1–2 mM, and the cytosol, where Ca²⁺ concentration is approximately 100 nM. The maintenance of homeostatic Ca²⁺ levels in the cytosol results from a complex balance between Ca²⁺ influx, buffering, internal storage, and efflux [73, 80]. An overload of Ca²⁺ in the cytosol can saturate the buffering capacity of mitochondria and the endoplasmic reticulum, leading to mitochondrial damage and, ultimately, neuronal death [81, 82]. This Ca²⁺ overload triggers the activation of various enzymes, including lipases (phospholipase A2), proteases (calpain), phosphatases (calcineurin), Ca²⁺-dependent endonucleases and nNOS [38, 75, 78, 83–85]. Together with excessive free radical formation, these enzymatic activities cause mitochondrial dysfunctioin, cell membrane rupture and, ultimately, neuronal death [9, 10, 78, 86] (Figure 2).

Other pathways independent of glutamate may contribute to the increase in intracellular Ca²⁺ levels. It has been reported that transient receptor potential (TRP) membrane channels, such as TRPM7 and TRPM2, are activated during ischemia contributing to elevated neuronal Ca²⁺ levels, as TRP channel blockage reduces Ca²⁺ accumulation in neurons exposed to excitotoxicity [87, 88]. Additionally, voltage-gated Ca²⁺ channels and Ca²⁺-permeable acid-sensing ion channels also contribute to increased cytosolic Ca²⁺ levels in ischemic neurons [89]. These findings indicate that Ca²⁺ accumulation in neurons within the infarcted area disrupts Ca²⁺ homeostasis, which is the primary cause of ischemic damage, rather than the specific pathway of Ca²⁺ entry into neurons. This may explain why drugs that reduce Ca²⁺ inflow through NMDARs have not been successful in clinical AIS treatments. However, since that NMDARs are the primary route for Ca²⁺ accumulation in ischemic neurons, reducing Ca²⁺ entry via NMDA receptor results in lower Ca²⁺ buildup in neurons and a subsequent reduction in brain damage.

In AIS, excessive NMDAR stimulation leads to excessive Ca²⁺ influx into the cytosol, activating nNOS and increasing NO^• levels, as well as promoting Ca²⁺ accumulation in mitochondria. Excessive mitochondrial Ca²⁺ disrupts oxidative phosphorylation, leading to increased reactive oxygen species (ROS) production, heightened oxidative stress and mitochondrial dysfunction, ultimately resulting in increased apoptosis cell death in the ischemic area [90–95]. Mitochondrial dysfunction, oxidative/nitrosative stress, calpain activation and inflammation are among the key events triggered by Ca²⁺ overload due to excessive NMDAR activation in AIS neurons (Figure 3).

One mechanism involved in the excitotoxic cascade in AIS is oxidative stress, a term used to describe a state in which “an imbalance between prooxidants and antioxidants in favor of the oxidants, that leads to a disruption of cellular redox homeostasis” [96]. Redox processes play a fundamental role in nearly all biological functions, from bioenergetics and metabolism to vital cellular activities [97]. Cell physiology is closely linked to maintaining a delicate balance between oxidant production and elimination. Redox homeostasis results from the equilibrium between oxidant systems and an antioxidant system, and the normal functioning of cells depends on their ability to maintain this redox state. Neurons continuously produce superoxide radicals, mainly via the ETC during mitochondrial oxidative phosphorylation. While the production of radicals was once thought to pose a threat to cellular survival and contribute to aging, it is now understood that oxidizing species, particularly hydrogen peroxide, act as second messengers involved in essential cellular functions, primarily redox signaling. Under physiological conditions, a sophisticated antioxidant system maintains oxidative activity within homeostatic levels [98]. The antioxidant system include: i) low-molecular-weight molecules, such as ascorbic acid, glutathione, thiols, α-tocopherol, carotenoids, and polyphenols; ii) enzymes that directly neutralize oxidants, such as SOD, catalase, and glutathione peroxidase (GPx); iii) enzymes that remove peroxides, such as glutaredoxins, thioredoxins, and peroxyredoxins; iv) enzymes that repair oxidative damage, such as methionine sulfoxide reductase, disulfide reductases/isomerases, and sulfiredoxins; v) systems responsible for removing damaged material, such as proteasomes, lysosomes, proteases, phospholipases, and DNA repair enzymes [99].

The normal functioning of cells depends on a delicate balance between the production and elimination of oxidants. However, when antioxidant systems fail to maintain redox homeostasis, either due to excessive oxidant production and/or reduced antioxidant production, a state of oxidative stress occurs. The most common oxidants induced by excitotoxicity are derived from oxygen and nitric oxide, known as ROS and reactive nitrogen species (RNS), respectively. ROS can be classified into two main groups; free radicals and non-radical molecules. In mammals, the primary free radical ROS are the superoxide radical (O₂^•–) and hydroxyl radical (OH^•), while the main non-radical ROS is hydrogen peroxide (H₂O₂). The principal RNS are the nitric oxide (NO^•) and peroxinitrite (ONOO^–) [100].

In mammals, ROS and RNS are continuously generated by various biochemical processes involved in cellular metabolism, both under physiological as pathological conditions. Under homeostatic conditions, hydrogen peroxide and nitric oxide are essential for normal neuronal function due to their roles in cell signaling processes [100]. However, under oxidative stress conditions, their harmful effects on lipids [101] and proteins [102] in cell membranes, as well as on DNA [103], make them key contributors to cellular damage, ultimately leading to cell death. The recognition of the significance of ROS in the physiology of living organisms has led to the emergence of a new discipline in Biology, Redox Biology, which studies the oxidation and reduction processes associated with life [97].

Several factors make neurons particularly vulnerable to oxidative stress, exposing the CNS to oxidative damage [104]. High ROS production results from a high metabolic activity and energy demand of neurons, which is primarily met through mitochondrial oxidative phosphorylation. When neuronal membranes contain high levels of polyunsaturated fatty acids and have reduced levels of endogenous antioxidant enzymes, particularly catalases, lipid oxidation of the neuronal membrane occurs [105].

In neurons, there are two main physiological sources of ROS: NADPH oxidases (NOX) and mitochondria. NOX family enzymes are transmembrane proteins that transfer electrons across biological membranes, using oxygen as the electron acceptor and producing the superoxide radical as a reaction product [106]. NOX was initially discovered in phagocytic cells, where it was identified as the first system to generate ROS as its primary function. In phagocytic cells, NOX is responsible for the “respiratory burst”, a mechanism by which NOX activation leads to the production of large amounts of ROS in the phagolysosome, oxidizing lipids and proteins and ultimately destroying the engulfed bacterium. Later, it was found that NOX is expressed in nearly all tissues, where its main function is ROS generation [106]. In neurons, NOX activity plays a crucial role in altering cell fate and modulating neuronal activity. NOX enzymes are involved in synaptic potentiation and long-term learning, although the cognitive impairment observed in NOX-deficient mice is relatively mild, suggesting that NOX has a modulatory rather than essential role. ROS generation by NOX enzymes has been implicated in various CNS diseases. In stroke, increased NOX expression in rats has been associated with greater brain injury [107], whereas in NOX-deficient mice exhibit a significant reduction in ischemic damage size [108].

ROS production derived from mitochondria is particularly important in high-energy demanding tissues. In humans, the brain has significantly higher metabolic activity compared to other organs. It consumes nearly ten times more of oxygen and glucose. Despite comprising only 2% of body weight, the brain accounts for 20% of the body’s oxygen consumption and 25% of its glucose consumption [27]. More than 90% of this oxygen is used by the mitochondrial ETC, where oxygen serves as the final electron acceptor. Under quiescent conditions, not all electrons travel through the ETC efficiently; some and are prematurely reduced by oxygen. Between 0.2% and 2% of the electrons supplied by NADH and FADH₂ to the ETC leak from complexes I and III, directly reacting with oxygen to form the superoxide radical O₂^•– [109, 110]. The continuous requirement for an ATP production via oxidative phosphorylation makes mitochondria a significant source of O₂^•–. The lifespan of O₂^•– is 1 nanosecond, and it can rapidly undergo dismutation either spontaneously or enzymatically, primarily through SOD, which converts O₂^•– into O₂ and H₂O₂ [111, 112]. Depending on the redox state, H₂O₂ can participate in cellular redox signaling, serve as a substrate in the Fenton reaction (in the presence of ferrous ion, Fe²⁺) to produce hydroxyl radicals (OH^•), or to be reduced to H₂O by glutathione [113–115]. The production of mitochondrial ROS is influenced by factors such as the mitochondrial membrane potential (Δψm), matrix pH, and the oxygen tension [116]. Mitochondria can also generate ROS through enzymes of the tricarboxylic acid (TCA) cycle, including aconitase, pyruvate dehydrogenase and α-ketoglutarate dehydrogenase, as well as through outer mitochondrial membrane (OMM) enzymes, such as cytochrome P450 and enzyme monoamine oxidase [117]. In ischemic regions, elevated ROS levels play a critical role in the pathogenesis of AIS [118].

As a key component of oxidative stress, RNS, including NO^• and ONOO^–, are involved in brain damage in AIS. In neurons, nNOS is part of the NMDAR complex via PDZ domains and its activation by Ca²⁺ leads to the production of NO^• through the catalytic conversion of arginine to citrulline. In AIS, excessive Ca²⁺ influx through NMDAR triggers the activation of both neuronal and inducible NOS, resulting in excessive NO^• production [72, 119]. Accumulating evidence suggests that excessive NO^• production from nNOS and iNOS significantly contributes to brain damage in AIS [72, 110, 120]. High levels of O₂^•– and NO^• rapidly react to form OH^• and ONOO^–, both highly toxic species due to their strong reactivity with biomolecules such as lipids and proteins. These reactive species can also activate pathways that disrupt the BBB and exacerbate brain damage [72, 121]. In AIS, increasing ROS/RNS levels in the brain [118, 122, 123] indicate that oxidative and nitrosative stress play a crucial role in neurological damage following AIS [118, 124–126].

Mitochondria and NOX family are the main sources of ROS in AIS. NOX contributes to oxidative stress and ischemic brain damage by oxidizing NADPH to NADP⁺ and reducing O₂ to O₂^•– [106, 127]. NO^• and ONOO^– have been shown to inhibit mitochondrial respiratory complexes, particularly complex IV, leading to increased O₂^•– production [90]. High levels of RNS can lead to ATP depletion through a mechanism initiated by DNA oxidation caused by RNS. Damaged DNA molecules activate poly(ADP-ribose) polymerase (PARP) leading to the depletion of its substrate, NAD⁺, which is a crucial electron donor for the mitochondrial ETC [128]. ROS and RNS can: a) directly cause the neuronal death by oxidazing lipids and proteins, irreversibly altering the architecture of essential cellular structures; b) indirectly activate pathological processes such as mitochondrial dysfunction [122, 129], the caspase cascade, and ultimately, neuronal apoptosis [109]. Excessive ROS production is a key factor in ETC impairment within mitochondria, leading to decreased ATP synthesis. This, in turn, exacerbates O₂^•– formation, further increasing mitochondrial dysfunction, which can trigger the intrinsic mitochondrial apoptotic pathway [93, 130] and reduce antioxidant activity [131].

Another key mechanism induced by excitotoxicity, mediated by the accumulation of Ca²⁺ in neurons within infarcted areas, is the activation of calpains, a family of Ca²⁺-dependent cysteine proteases abundant in the CNS, with essential roles in synapses and memory [91]. Under physiological conditions, Ca²⁺ regulates the activity of a limited number of calpain molecules. However, when cytosolic Ca²⁺ levels rise to the micromolar range, calpains become overactivated, contributing to excitotoxic damage [91]. This occurs, in part, due to their ability to impair the activity of Na⁺/Ca²⁺ exchangers, which are essential for maintaining Ca²⁺ homeostasis in neurons [132]. The activation of calpain by increased cytosolic Ca²⁺ levels may also trigger the intrinsic mitochondrial apoptosis pathway. Calpains promote the cleavage of the pro-apoptotic component BH3 (Bid) to its truncated active form (tBid), which subsequently activates Bax. Bax then translocates to the OMM, where it induces pore formation, facilitating the release of cytochrome c (cyt c) into the cytosol, where it binds to the apoptosis-inducing factor (AIF), leading to increased caspase cascade activation and neuronal death [74]. In animal models of AIS, the use of calpain inhibitors has demonstrated neuroprotective effect [133].

Mitochondria, especially in neurons, serve as the center of various biological processes, including proliferation, differentiation, Ca²⁺ signaling regulation, redox homeostasis and cell death control [134]. Under certain conditions, such as increased Ca²⁺ levels and oxidative stress, mitochondria lose their ability to produce sufficient ATP to sustain essential functions. This condition, known as mitochondrial dysfunction, is a key factor in many neuropathologies, including AIS [94, 135].

For mitochondria to perform their normal functions, the creation of an electrochemical gradient between the mitochondrial intermembrane space (IMS) and the inner mitochondrial membrane (IMM) is essential. This gradient generates a force known as the protonmotive force (PMF), which drives the protons to flow through complex V (F₁F₀-ATP synthase) for ATP synthesis. This process that requires the re-entry of protons into the matrix via complex V, leading to a decrease of the Δψm [136]. Uncoupling proteins (UCPs) located in the IMM regulate the proton gradient by facilitating proton flow into the mitochondrial matrix [137]. Alterations in Δψm occur when changes in mitochondrial membrane permeability take place and often represent the decisive event that determines neuronal survival or death. The permeability of the mitochondrial membrane is primarily regulated by the opening and closing of the mitochondrial permeability transition pore complex (mPTP) [130]. This complex is a non-selective, solute-permeable channel (< 1.5 kDa) present in the IMM. Although its exact molecular composition remains uncertain, F1F0-ATP synthase, adenine nucleotide translocase (ANT), and matrix cyclophylin D (CypD) are known components of the mPTP. Mitochondrial Ca²⁺ concentrations and redox balance play crucial roles in maintaining Δψm by regulating mPTP opening and closing. Under physiological conditions, transient mPTP opening helps regulate mitochondrial bioenergetics, cellular metabolism and ROS production. However, excessive Ca²⁺ accumulation in the mitochondrial matrix and oxidative stress lead to prolonged mPTP opening [138, 139], resulting in the leakage of protons, pyridine nucleotides (NAD⁺, NADP) and TCA cycle substrates from the mitochondrial matrix. This renders both the TCA cycle and the ETC inoperable, ultimately activating pathways that cause neuronal death [92, 117]. Overall, mPTP opening in the IMM results Δψm loss, increased ROS production and the excessive opening of voltage-dependent anion channels (VDAC) in the OMM. This allows substrates up to 5 kDA to pass through VDAC [140], further contributing to Δψm dissipation and severe mitochondrial dysfunction.

Loss of control over mitochondrial membrane permeability is a critical event in cell death. In AIS, Ca²⁺ overload, combined with oxidative and nitrosative stress in the infarcted area leads to the opening of the mPTP, resulting in the loss of Δψm and the activation of pathways that ultimately lead to neuronal death [92, 117, 141, 142]. The regulation of mitochondrial outer membrane permeabilization (MOMP) is essential for proper mitochondrial activity [143]. Sustained mPTP opening causes mitochondrial swelling and rupture of the OMM [144], triggering pathways that lead to neuronal death [92, 117]. In most cases, prolonged mPTP results in a state where the mitochondria rapidly hydrolyze both cytosolic and mitochondrial ATP. This ATP depletion causes the death by necrosis. However, if ATP synthesis is maintained, transient mPTP opening leads to apoptosis, a highly regulated and controlled process of programmed cell death [74, 145].

The maintenance of MOMP depends on the relationship between the levels of antagonist (Bcl-2 and Bck-xl) and agonists (Bax, Bad, and Bid) apoptotic proteins of the B-cell lymphoma 2 (Bcl-2) family. Activation of the intrinsic mitochondrial apoptotic pathway begins with the activation of the proapoptotic BH3-only protein and cleavage of Bid into its truncated form (tBid) by caspase-8. This process leads to the activation of inactive Bax in the cytosol, converting it to its active form. Once activated, Bax translocates to the OMM, where it forms pores that facilitate the release of cyt c into the cytosol. In the cytosol, cyt c binds to apoptosis protease-activating factor-1 (Apaf-1), leading to the formation of the apoptosome. This structure promotes the activation of procaspase-9, which in turn triggers the caspase cascade ultimately resulting in the activation of caspase-3. Caspase-3 then promotes the cleavage of enzymes such as PARP, which activates deoxyribonuclease, which leads to DNA fragmentation and neuronal death by apoptosis [146, 147].

Mitochondrial dysfunction can contribute to brain damage after a stroke through a mechanism other than apoptosis or necrosis, involving iron metabolism and oxidative stress. Ferroptosis is a form of caspase-independent programmed cell death caused by excessive lipid peroxidation of cell membranes, driven by iron and leading to oxidative stress through the Fenton reaction. This process ultimately results in cellular disruption and neuronal death [113–115].

Because the AIS is caused by a decrease in blood flow, the primary goal for treatment is the rapid restoration of blood flow to the affected brain area. For this reason, research has primarily focused on identifying suitable thrombolytics. For years, the only treatment authorized by the U.S. Food and Drug Administration (US-FDA) for AIS has been the administration of the tissue plasminogen activator (tPA), a thrombolytic drug that dissolves the clot. Treatment with tPA has proven effective, with significant improvement largely depending on the time elapsed between stroke onset and initiation of tPA therapy. Administering tPA as early as possible remains the most effective strategy for achieving a favorable outcome after stroke. tPA treatment should begin within 4.5 h of stroke symptom onset. Treatment outside this therapeutic window can result in intracerebral hemorrhage, where the risks outweigh the benefits [148, 149].

Several factors limit the use of tPA, including the need for an accurate diagnosis of stroke type (which requires imaging techniques that are not always available), knowledge of the stroke onset, and the risk of intracerebral hemorrhage. As a result, only 1–3% of stroke patients receive this treatment, with a success rate of less than 35% and full recovery in only about half of treated patients [150]. Despite significant advances in acute-phase stroke therapy over the past five years, including ongoing developments in thrombolytic strategies [150, 151] and mechanical thrombectomy [152, 153], most patients remain untreated due to limited therapeutic windows and strict selection criteria.

Similarly, endovascular thrombectomy, another US-FDA-approved procedure that mechanically restores blood flow, is restricted to a subset of patients [153]. Mechanical clot retrieval must be performed within 6–24 h of stroke symptom onset. Although thrombectomy extends the therapeutic window, the number of eligible ischemic stroke patients remains limited due to the risk of intracranial hemorrhage [154].

In response to the large number of AIS patients who are not eligible for thrombolytic and/or endovascular thrombectomy therapy, significant efforts have been directed toward developing a drug capable of protecting the brain from the damage caused by AIS over time [155]. While several compounds have shown success in treating ischemic stroke in animal models, most have failed in clinical trials [156]. Several factors may explain this failure, including differences in the characteristics of AIS models used in animals versus those used in clinical trials, the heterogeneity of AIS in humans compared to the relative homogeneity in animal models, and the fact that preclinical studies are conducted on healthy young animals, whereas clinical trial participant are typically elderly individuals [157, 158].

EPO is a sialoglycoprotein that plays a crucial role in regulating erythropoiesis. It is synthesized and secreted by type I peritubular interstitial fibroblasts near the corticomedullary border in the kidney and released into the bloodstream [158–161]. EPO controls red blood cell production in the bone marrow by binding to its receptor on the cell membrane of red blood cell precursor cells. This binding activates a signaling pathway initiated by the phosphorylation of Janus tyrosine kinase 2 (JAK2), which prevents apoptosis of colony-forming unit-erythroid (CFU-E) cells and promotes their proliferation and differentiation into mature erythrocytes [162]. In the human body, EPO regulates the production of approximately 200 billion new red blood cells every day.

Human EPO was purified from the urine of patients with aplastic anemia in 1977 [163] and its gene was cloned in 1985 [164, 165]. In 1986, treatment with recombinant human EPO (rHuEPO), produced using recombinant DNA technology from Chinese hamster ovary cells, was administered to 12 patients diagnosed with end-stage renal disease and anemia, resulting in a significant increase in hemoglobin levels in 9 of the patients [166]. Since its approval by US-FDA in 1989, EPO has been used clinically to treat anemia caused by insufficient EPO production [167]. Shortly after, it was approved in Europe and other parts of the world. Since then, more than 1 million people have benefited from the use of rHuEPO [168].

For a time, the only clinical application of EPO was in treating various types of anemia, particularly those associated with chronic kidney disease. However, studies beginning in the late 20th century revealed that, in adult humans, EPO is also expressed in other tissues, including the brain [169, 170], where it does not serve an erythropoietic function [171–175]. The first animal studies on EPO’s effects in stroke demonstrated its potent neuroprotective properties [176–182]. Subsequent studies [179–181] confirmed EPO’s neuroprotective effects in ischemic stroke treatment. Alongside these neuroprotective effects, the extensive clinical use of EPO in thousands of patients with due to chronic kidney disease [159] has paved the way for exploring EPO as a potential treatment for AIS in humans [182–187].

The fact that EPO exerts different functions in different tissues implies the existence of different molecules acting as EPO receptors (EPOR) [168, 179, 183]. In CFU-E cells, EPO binds to a receptor formed by the homodimerization of two identical 66 kDa monomers expressed on the cell membrane. The binding of EPO to EPOR leads to the formation of EPOR-EPOR (EPOR₂) complex, known as the canonical isoform. The interaction induces the phosphorylation of the JAK2, which is constitutively bound to the EPOR, thereby triggering the signaling pathway that promotes the expression of anti-apoptotic proteins, facilitating precursor cells in their transition to erythrocytes. Different isoforms of EPOR have been identified in the CNS. In neurons, along with EPOR₂, the beta common receptor (βcR), also known as CD131, has been identified [200]. The binding of EPOR monomer to βcR leads to the formation of the EPOR/βcR heterodimer, whose activation initiate the same JAK2 signaling pathway as EPOR₂ [201]. In the CNS, neurons, astrocytes, microglia, and endothelial cells express EPOR/βcR [202]. EPO derivatives, including neuro-EPO, carbamylated erythropoietin (CEPO), and asialo-EPO, preferentially bind to EPOR/βcR [202].

Different studies confirm that the brain synthesizes EPO, which can carry out non-erythroid biological functions [171–173, 203]. In the brain, EPO has various effects, acting both as an inhibitor (of inflammation, apoptosis, and oxidative stress) and as a stimulator (promoting neuroprotection, tissue protection and repair, neuronal plasticity, memory enhancement, cell proliferation, and neurogenesis) [175, 179, 183, 204, 205].

The diversity of functions exerted by EPO in the CNS may, in part, be due to the different isoforms of EPOR and the various intracellular signaling pathways activated by the phosphorylation of JAK2 [202]. Phosphorylation of JAK2, in turn, leads to phosphorylation of phosphatidylinositol 3-kynase (PI3K), Ras-mitogen-activated protein kinase (MAPK), and signal transducer and activator of transcription (STAT5), which modulates the expression of various genes involved in proliferation, differentiation and cell survival [205, 206]. EPO may protect neurons through a combination of mechanisms, including limiting the production of ROS, enhancing the expression of anti-apoptotic proteins, inducing anti-inflammatory cytokines and promoting neurogenesis [207].

Abundant evidence confirms the potential of EPO for neuroprotection in the treatment of stroke. The neuroprotective efficacy of rHuEPO has been tested in several animal species (mouse, rat, gerbil, and rabbit) using models of focal and global cerebral ischemia, primarily by inducing stroke through permanent or transient occlusion of the middle cerebral artery (MCAO). This artery supplies blood to a large portion of the cerebral hemispheres and is responsible for approximately 80% of strokes. The results indicate that EPO promotes a reduction in neuronal death, improves outcomes, and significantly decreases brain infarct volumes [208–212].

Since 1998, exogenous EPO has been shown to exert potent neuroprotective effects in experimental models of focal and global cerebral ischemia [177–179, 182, 186, 208, 213, 214], yielding encouraging results in the treatment of ischemic stroke, particularly in preventing ischemic neuronal injury [176–179, 186, 215]. In mice, intracerebroventricular injection of EPO 24 hours prior to MCAO significantly reduced infarct volume [177]. The survival rate after 14 days in adult male C57 BL/6 mice treated with EPO (5,000 IU/kg) increased to 80%, compared to only 50% in untreated mice. EPO also improved neurobehavioral outcomes on days 3 and 7 post-MCAO and attenuated axonal injury caused by AIS [212]. Similarly, in a gerbil model of global ischemia, infusion of EPO into the lateral ventricles prevented hippocampal CA1 neuron death and increased synapse density in the same region [176]. In rats subjected to MCAO, treatment with EPO (5,000 IU/kg) reduced stroke size by 75% and suppressed apoptosis in ischemic penumbra [178]. Intravenous administration of a single dose of 1,000 IU/kg rHuEPO before the onset of a 60-minute transient MCAO protected the brain from ischemic damage [211]. A meta-analysis of sixteen studies exploring the efficacy of EPO in experimental focal cerebral ischemia concluded that administration of EPO AIS reduced stroke size by 32% and significantly improved neurobehavioral deficits [182]. Administration of EPO within 6 hours of stroke onset was more effective was more effective than administration after 6 hours [180]. Additionally, in mice undergoing transient cerebral ischemia, increased EPO expression was associated with a reduction in infarct size [216]. Other studies have demonstrated that, beyond reducing infarct volume, EPO also improves learning capacity in gerbils and prevents navigation disability in rats [176].

These results indicate that EPO is a potent neuroprotector agent capable of mitigating the various mechanisms that contribute to brain injury in stroke. Preclinical research demonstrates that EPO attenuates excitotoxicity, promotes intracellular Ca²⁺ homeostasis, reduces oxidative stress and inflammation, supports Δψm maintenance, and plays a crucial role in preventing mitochondrial dysfunction and the activation of the apoptotic cascade [177, 179, 183]. Additionally, EPO increases the levels of the anti-apoptotic protein Bcl-2 and facilitates the formation of tBid-Bad channels, which release cyt c from mitochondria, promoting neuronal recovery in ischemic penumbra [146, 177, 217].

Since the beginning of its therapeutic use, rHuEPO has been administered via subcutaneous, intraperitoneal, intravenous and intranasal at high doses to ensure that it crosses the BBB and reaches sufficient levels to activate brain tissue cells expressing EPOR [218]. However, there is no convincing evidence demonstrating that, under conditions of BBB structural integrity, EPO can reach the brain parenchyma. Only molecules that are highly lipid-soluble and have a molecular mass of less than 400 Da, including amino acids, can pass through the BBB [219]. In contrast, EPO is a large (34 kDa), highly glycosylated and negatively charged protein, making it poorly suited to cross the BBB. Normally, proteins and peptides require a receptor-mediated translocation system to traverse the BBB. In the case of EPO, it has been suggested that it may cross via a receptor-mediated transcytosis mechanism [220]. The presence of EPOR in cerebral capillaries, along with findings that injection of labeled EPO is blocked by the administration of unlabeled EPO, led to the hypothesis that EPO accesses the brain through the BBB [180]. However, studies using radiolabeled EPO in rodents and primates showed that its entry into the brain is very slow and occurs via a non-specific mechanism [221].

In early studies on the efficacy of EPO in the brain, EPO was administered via intraventricular or intracerebral routes in animal models [101, 102, 222]. However, these invasive administration methods require surgical expertise and can cause brain damage and infections, making them unsuitable for clinical use. More recent studies in several species, including humans, have confirmed that when EPO is administered in high doses, both intravenously and intraperitoneally, a sufficient amount crosses the BBB to mediate neuroprotection against AIS [180, 182, 221].

Since only 1% of the administered EPO reaches the brain parenchyma [180, 221], effective AIS treatment would require the injection of a high dose of EPO. Although a relatively small fraction crosses the BBB and exerts its neuroprotective effects, the majority remains in systemic circulation, where it binds to EPOR in CFU-E cells, promoting their proliferation and differentiation into mature erythrocytes [162]. This leads to an increase in hematocrit levels, raising the risk of cerebral thrombosis and hypertension [223], which ultimately limits the clinical use of exogenous EPO in AIS treatment.

There is a great deal of variability in the dose of EPO to be used for the treatment of AIS. In rats, intravenous doses of rHuEPO between 500 to 5,000 IU/kg injected at 6, 24, and 48 h after initiation of AIS caused a significant reduction in infarct volume and neurologic deficit after 24 days [181]. Treatment with intra-arterial doses of 800 IU/kg of rHuEPO at the onset of AIS attenuated infarction volume and improved neurobehavioral outcomes at 2 h and 24 h after MCAO [224]. In humans, intravenous injection of a bolus of 16,000 IU of EPO, followed by additional injections of 8,000 IU every 12 h up to a total dose of 56,000 IU of EPO for 3 days, showed an improvement in neurological damage caused by AIS after 14- and 18-day evaluation [225]. In humans, subcutaneous administration of 5,000 IU 48 h and 72 h after stroke significantly recovered from neurological damage after 90-day evaluation [186], while in other studies the concentration supplied was between 3,500 to 10,500 IU of EPO per day [223, 226].

In an attempt to minimize the side effects associated with high-dose EPO administration, the intranasal route has been explored. This approach allows to bypass the BBB while reducing systemic exposure. The intranasal delivery of rHuEPO is an effective, safe and non-invasive method for introducing rHuEPO into the brain while avoiding side effects [227–229]. In rats with MCAO-induced AIS, intranasal administration of rHuEPO at doses of 4, 8, 12, and 24 IU, 10 minutes after occlusion, has been shown to effectively reduce infarct volume and cell damage in the ischemic hemisphere while improving behavioral dysfunction 24 hours after cerebral ischemia [230]. Comparatively, the intranasal route is 10 times faster, requires smaller doses of rHuEPO than the intraperitoneal route, and presents a lower risk of collateral damage [231, 232]. Thus, intranasal administration represents a promising method for efficiently delivering EPO and its derivatives to the brain.

The mechanism by which EPO exerts its function in the brain is similar to the one by which it induces erythropoiesis [206]. The difference lies in the EPOR isoform: while hematopoietic cells express the homodimer (EPOR₂), non-hematopoietic organs present a heterodimeric receptor [205, 206]. In both cases, the binding EPO to its receptor triggers the phosphorylation of JAK2, which subsequently activates three main signaling pathways: phosphatidylinositol 3-kinase (PI-3K)/AKT (protein kinase B), MAPK, and signal transducer and activator of transcription 5 or 3 (STAT5 or STAT3) [206, 216]. STAT5 induces the expression of anti-apoptotic proteins (e.g., Bcl-xl), while PI-3K and MAPK inhibit pro-apoptotic proteins (Bax, Bad) [168]. The action of EPO concludes with the dephosphorylation of the EPOR complex and the internalization of the EPO/EPOR complex. After internalization, approximately 40% is degraded by the proteasome, while 60% of the EPO is recycled into the extracellular medium [233].

Results from experimental studies on the neuroprotective effects of EPO prompted the initiation of clinical trials to evaluate rHuEPO in patients with AIS. The first clinical study on the effects of EPO treatment in AIS was conducted by the Göttingen EPO Stroke Study, which demonstrated that rHuEPO is safe and has beneficial outcomes in AIS patients [215]. Safety was analyzed in 13 patients who received 33,000 IU of rHuEPO once daily for the first 3 days after the stroke. Efficacy was evaluated in 40 patients who received the same dose of rHuEPO within 5 hours of stroke onset. Treatment with rHuEPO did not raise safety concerns, as treated patients rHuEPO exhibited cerebrospinal fluid (CSF) levels of rHuEPO from 60 to 100 times higher than untreated, along with a reduction in infarct size. The study concluded that high doses of NeuroEPO were well tolerated by AIS patients and that, after 1 month, clinical outcomes had improved [215]. However, a subsequent trial conducted by the same group failed to demonstrate any beneficial effects of EPO in AIS [234]. The multicenter phase II/III clinical trial involved 460 AIS patients who received rHuEPO or a placebo within six hours of symptom onset. The results did not support a beneficial effect of rHuEPO treatment. Moreover, the mortality rate in the rHuEPO-treated group was 1.8 times higher than the placebo group. These negative findings dismissed the generated by earlier clinical trials. However, a thorough analyses of the methodology revealed significant differences between the phase I and phase II/III clinical trials. Surprisingly, the same protocol was not followed in both trials: in phases II and III, 63% of patients had previously been treated with tPA, whereas tPA was not used in phase I. A subsequent study analysing a subgroup of phase II/III patients who had not received tPA confirmed the results observed in phase I [184]. Further human studies have reaffirmed the neuroprotective effects of EPO, concluding that long-term EPO treatment significantly reduces the adverse neurological effects of AIS [186, 215, 225].

Along with the beneficial effects of EPO in stroke treatment, its therapeutic use is limited by the difficulty of crossing BBB, which requires the administration of high doses of rHuEPO. Three hours after an intravenous injection of 5,000 IU, less than 2% of rHuEPO crosses BBB [174]. While rHuEPO exerts the neuroprotective effect, its presence rHuEPO in the bloodstream leads to increased hematocrit, blood viscosity, and blood pressure, all of which significantly raise the risk of thrombus formation and secondary stroke [224, 235]. These undesirable effects make rHuEPO unsuitable as a treatment for AIS. Recognizing both the beneficial and adverse effects of EPO in stroke treatment, research has focused on developing EPO derivatives with neuroprotective activity but without erythropoietic effects.

To date, NeuroEPO has not been used in any clinical trials for the treatment of AIS, as pharmaceutical companies have shown little interest. Only a study in phase I has been conducted in healthy volunteers (Trial registration: Cuban Public Registry of Clinical Trials RPCEC00000157, June 10, 2013). In this study, intranasal NeuroEPO was administered to 25 healthy volunteers, 14 women, with a mean age of 27 years. NeuroEPO quickly reached the CSF and the brain. The results showed that NeuroEPO is a safe, well-tolerated product that does not stimulate erythropoiesis in healthy volunteers [241]. NeuroEPO was also been tested in patients with Parkinson’s disease (stages 1 and 2 on the Hoehn and Yahr scale). Nasal administration of NeuroEPO was found to be safe and well tolerated [248]. Currently, NeuroEPO is undergoing phase II-III clinical trials for Alzheimer’s and Parkinson’s diseases, with very promising results [249].