There are over 12 million stroke cases each year worldwide and over 6.5 million people die from stroke annually [1]. According to DALY (disability-adjusted life year), which is an indicator that measures the overall burden of the disease [2], more than 143 million years of healthy life are lost annually due to stroke-related death and disability [1]. As reported in the thirty-year projections of stroke incidence, prevalence, deaths, and disability-adjusted life years, there was a 27% increase in the number of people living with a stroke estimated between 2017 and 2047 in the European Union [3].

Current methods of reperfusion therapy include intravenous thrombolysis and mechanical thrombectomy. However, they are associated with several adverse effects, such as the risk of intracranial hemorrhage, allergic reactions, hypotension, risk of bleeding, acute kidney injury [4], emboli, vessel dissections, and vasospasms [5], while also having narrow therapeutic time windows [6], which means that only a small percentage of stroke patients are able to benefit from such treatment [7]. Therefore, there is a great need for the development of neuroregenerative and neuroprotective methods.

The occlusion of cerebral and precerebral arteries caused primarily by either atherosclerotic plaques in large vessels, microatheromatosis, or cardioembolism [8] leads to anoxia and activation of anaerobic metabolism, which causes the formation of lactic acid and contributes to the dysregulation of the acid–base balance and cell destruction [9]. Reduced adenosine triphosphate (ATP) production leads to the impaired function of ion pumps, the outflow of K${}^{+}$ ions from cells, and the influx of Na${}^{+}$ and Ca${}^{2+}$ ions into cells. Depolarization of neurons contributes to the release of glutamic acid from synaptic terminals, thereby causing excitotoxicity by increasing the influx of calcium ions into the cell, which in turn leads to the activation of enzymes that digest proteins, lipids, and nucleic acids. Oxygen free radicals that are generated as a result of lipid degradation of the cell membrane and mitochondrial dysfunction contribute to the destruction of DNA, proteins, and lipid peroxidation, thereby causing cell death [10]. The increased concentration of Na${}^{+}$ ions inside the cell entails the influx of water, which leads to cell edema, increased pressure on vessels and brain tissue, increased permeability of the blood-brain barrier [11, 12], and the infiltration of immune cells that release proinflammatory cytokines [13]. Microglia play an important role in the development of neuroinflammation, by, on the one hand, removing damaged cells and, on the other, releasing cytokines and cytotoxic substances, which makes them a potential target for neuroprotective therapy [14].

Until the development of photon imaging, genome-wide transcription analysis, and epigenetic analysis, microglia were classified into two main phenotypes [15]: classically activated M1 with proinflammatory properties that release tumor necrosis factor alpha (TNF-$\alpha$), interleukin 1$\beta$ (IL-1$\beta$), IL-6, IL-12, IL-23, and inducible nitric oxide synthase (iNOS); alternatively activated M2 releasing anti-inflammatory cytokines: IL-4, IL-10, IL-13, transforming growth factor beta (TGF-$\beta$), and growth factors—vascular endothelial growth factor (VEGF), brain-derived neurotrophic factor (BDNF), platelet-derived growth factor (PDGF), and nerve growth factor (NGF) [16]. In addition to their proangiogenic and anti-inflammatory properties, M2 microglia were associated with removing damaged neurons [17], stimulating the repair of the extracellular matrix [18], and regulating neurogenesis by modulating synapse maturation, forming dendritic spines, and producing key trophic factors for migration of new neurons [16, 19]. However, such a division into M1 and M2 microglia is now considered as an oversimplified one since microglial cells were discovered to be a significantly more heterogeneous group, where every individual cell exerts a different function. Their surface markers were found to be insufficient to define their functions because different states of microglia are dynamic and depend on the changes in the local environment [20]. In the pathological stages, microglia were observed to change their molecular profile, morphology, ultrastructure, motility, and function [15, 20, 21]. The main purpose of microglia cell therapy in stroke treatment is to prevent their excessive activation and the production of proinflammatory molecules [22], while also inducing their protective phenotype.

Neurogenesis involves the proliferation of neural stem cells, migration of neuroblasts, differentiation of neuroblasts into neurons, development of synaptic connections with other neurons, and survival of immature and mature neurons [7, 23]. In an adult brain, neurogenesis takes place mainly in two regions: in the subgranular zone of the hippocampal dentate gyrus (SGZ) and in the subventricular zone (SVZ) along the lateral ventricles [24].

Neurogenesis can be stimulated by an ischemic episode. Unfortunately, most neurons formed in this way die within about 2 weeks of a stroke. The low survival rate of new neurons is thought to be due to a lack of trophic factors and chronic neuroinflammation [24].

One of the factors stimulating neurogenesis in an adult brain is vascular endothelial growth factor (VEGF), which inhibits apoptosis of hippocampal neurons, stimulates stem cell proliferation and migration of newly formed cells via the vascular endothelial growth factor receptor 2 (VEGFR2) pathway, and stimulates angiogenesis and the repair of damaged neurons [25]. The stroke-induced blood–brain barrier disruption facilitates the contact between adult neural stem cells (NSC) and vascular cells, including VEGF. It was shown that VEGF induces the expression of the Notch ligand Delta-like 4 (DLL4) via its receptor VEGFR2, which leads to the proliferation and differentiation of NSCs into neurons [26]. Angiopoietin-like 4 (ANGPTL4) protein, which is released by vascular cells as a result of hypoxia, has been shown to stimulate neurogenesis in the dentate gyrus of the hippocampus and subventricular zone by stimulating Akt kinase activity and it also reduces the inflammatory response and neuronal death by inhibiting Fas expression and the Fas ligand (FasL) [27]. Fibroblast growth factor 2 (FGF-2) stimulates the proliferation and differentiation of neural progenitor cells derived from the subventricular zone. It was shown that the infusion of FGF-2 into the rats’ lateral ventricles increased the proliferation and migration of neurons from the subventricular zone to the olfactory bulb, and the injection of FGF-2 neutralizing antibodies contributed to the inhibition of their proliferation. Moreover, fibroblast growth factor also stimulates post-ischemic neurogenesis in classically non-neurogenic areas, such as the striatum, substantia nigra, and cerebral cortex [28]. The stromal cell-derived factor 1 and C-X-C chemokine receptor type 4 and 7 (SDF-1/CXCR4/CXCR7) signaling pathway stimulates axonal elongation and branching, remyelination, and the migration, proliferation, and differentiation of neuronal progenitor cells [29]. Moreover, insulin-like growth factor 1 (IGF-1) and brain-derived neurotrophic factor (BDNF) also enhance stem cell proliferation. Monocyte chemotactic protein (MCP-1) and matrix metalloproteinases 2, 3, and 9 (MMP) stimulate the migration of neuroblasts [30].

Although initially it was proposed that transplanted stem cells directly replace neurons in ischemic regions, currently, it is postulated that their therapeutic effect is mainly the result of their paracrine action (the “bystander effect”) since most of the systemically injected stem cells are trapped in the lungs and do not reach the affected tissues [31]. The paracrine action involves the secretion of factors that stimulate endogenous neurogenesis (BDNF, FGF, angiopoietin 2), angiogenesis (VEGF, angiopoietin 2), and neuroplasticity (integrin $\beta$1) [7]. In addition, it has been shown that transplanted stem cells have immunomodulatory properties and, by modulating the levels of TNF-$\alpha$, IL-1$\beta$, IL-6, and monocyte chemoattracting protein 1 (MCP-1) [32, 33], they reduce the post-ischemic inflammatory response and contribute to the reduction in nerve tissue damage. The main mechanism through which mesenchymal stem cells exhibit their paracrine properties is exosome secretion [34]. Exosomes are extracellular vesicles, which are secreted by almost every cell type and play a key role in intercellular communication [35, 36, 37]. Their applications have been extensively studied in many medical fields, including primarily oncology and cardiology, where microRNA (miRNA) can modulate angiogenesis and tumor progression [38], in addition to inhibiting inflammation in cardiac ischemic diseases [39]. They contain various proteins, including cytokines, chemokines, growth factors, and membrane receptors as well as miRNAs through which they promote neurogenesis, angiogenesis, and cell growth and reduce inflammation, oxidative stress, and cell death [35].

In 2013, a new mechanism through which transplanted stem cells exert their therapeutic functions was proposed. Exogenous cells were found to form “biobridges” between the neurogenic area (subventricular zone, SVZ) and the ischemic area, thereby facilitating the successful migration of endogenous stem cells, which is one of the key limitations in the endogenous repair system. Biobridges, consisting of metalloproteinases (MMP) and an extracellular matrix (ECM), form a pathway that helps direct the migration of endogenous stem cells to the damage zone through non-neurogenic brain areas [40]. Increased activity of MMP-9 along the formed biobridges was demonstrated, and its inhibition was shown to impair cell migration from the SVZ to the cerebral cortex, suggesting a key role of metalloproteinase 9 in ECM remodeling. Interestingly, once the exogenous stem cells form biobridges, their concentration decreases and they are replaced by endogenous cells derived from neurogenic areas of the brain; thus, making their long-term administration potentially unnecessary [41]. Moreover, an increase in endogenous cell proliferation and neural differentiation in the peri-injured cortical areas was demonstrated, which further suggests that the transplantation of exogenous stem cells and biobridge formation can facilitate endogenous repair mechanisms. As mentioned before, ischemia-induced endogenous post-stroke neurogenesis itself is insufficient because of low stem cell survival and migration rates, incomplete integration in neural circuits, and increased differentiation to glial cells [42]. However, more studies explaining the mechanisms underlying the migration pathways and their implications in stroke therapy are still needed.

The stem cells most frequently used in medical research are mesenchymal stem cells [43], which exhibit several properties that make them suitable for cell transplants in stroke therapies [44]. They are multipotent, meaning they can differentiate into more than one cell type, including mesodermal lineage adipocytes, chondrocytes, osteocytes, and ectodermal lineage cells, such as neurons and glial cells [45]. They are relatively easy to harvest because they can be obtained from various body tissues, such as adipose tissue, bone marrow, umbilical cord blood, umbilical cord tissue, dental tissue, and olfactory mucosa, while their isolation and amplification are not expensive [43]. They can be injected in several ways: intracerebrally, cerebroventricularly, intravenously, intra-arterially, or intranasally [43]. MSCs exhibit immunomodulatory properties by reducing the expression of proinflammatory cytokines, such as TNF-$\alpha$, IL-1, interferon-$\gamma$ (IFN-$\gamma$), and MCP-1; by reducing astrogliosis and microglia activation via atypical JAK-STAT signaling pathway [46, 47]; increasing the expression of anti-inflammatory cytokines, such as IL-4, IL-10, and TNF-$\beta$ [48]. Moreover, by stimulating the secretion of neurotrophic and growth factors, they promote angiogenesis (VEGF, angiogenin-1, and PDGF), cell proliferation, differentiation, migration and survival (PDGF, NGF, brain-derived growth factor, neurotrophin-3, and FGF), axonal growth (PDGF), synaptic plasticity (synaptophysin), and myelination [43]. However, their proliferation decreases over time in long-term cultures [49, 50].

The current microglia research focuses on preventing the excessive activation and production of proinflammatory molecules [51]. One of the therapeutic strategies for ischemic stroke uses oxygen–glucose deprivation (OGD), whereby an optimal ischemia event is hypothesized to induce the protective phenotype in microglia [52]. Intravascular administration of OGD-preconditioned microglia in animal models was shown to promote angiogenesis, axonal outgrowth, and functional recovery. Since the main outcome of the microglial activity is considered to coincide with the result of the secreted neurotrophic factors rather than the microglial cells themselves, the effect on the neurological recovery by extracellular vesicles (EVs) derived from OGD preconditioned microglia was investigated. EVs from OGD-preconditioned microglia were found to be high in TGF-$\beta$1, which activates the Smad2/3 signaling pathway that plays a role in angiogenesis and neuronal injury repression [53]. Moreover, therapy with OGD-preconditioned peripheral blood mononuclear cells was also shown to promote angiogenesis, axonal outgrowth, and functional recovery in stroke [54]. The underlying mechanisms involved a reduction in miR-155-5p, via the hypoxia-inducible factor 1$\alpha$ (HIF-1$\alpha$) axis [55], which increased the expression of VEGF and played a crucial role in neurovascular repair. Moreover, higher levels of anti-inflammatory cytokines, TGF-$\beta$1 and TGF-$\beta$2, and lower levels of proinflammatory cytokines, IL-1$\beta$ and TNF-$\alpha$, were found after OGD-preconditioning than under normoxic conditions [56]. Overall, oxygen–glucose deprivation has therapeutic potential in ischemic stroke as it was shown to promote protective phenotypic conversion and functional recovery. In addition, the usage of extracellular vesicles derived from OGD-preconditioned microglia presents several advantages over cell transplantations, such as a lack of immunogenicity, no risk of cell embolism, and lower costs. However, more research elucidating the signaling pathways mechanisms and cell-to-cell communication is still needed.

Other factors affecting microglia activation phenotypes include IL-4 and IFN-$\gamma$ [57, 58]. It was demonstrated that, via the PI3K-Akt pathway, the secretome of microglia induced by IL-4 promoted the proliferation, survival, and differentiation of neural stem/progenitor cells (NSPCs) into neurons and oligodendrocytes, while the induction of IFN-$\gamma$ inhibited neurogenesis and oligodendrogliogenesis and led to the differentiation of NSPCs into astrocytes and induction of apoptosis. However, it remains unknown whether the induced microglia can maintain their protective phenotype after the removal of the stimulus. Recent studies [57] have demonstrated a decreased plasticity in terms of functions and phenotypic characteristics of induced microglia with time.

A variety of other factors that promote anti-inflammatory properties by microglia are being researched. Minocycline, an antibiotic from the tetracycline group was shown to increase the survival of neurons, stimulate neurogenesis, inhibit reactive gliosis, and promote functional recovery via the STAT1/STAT6 pathway in a rodent study model [59]. IL-4 was associated with improved functional recovery after stroke and a deficit in endogenous IL-4 promoted the proinflammatory phenotypic conversion [60]. Recently, tetramethylpyrazine, used in treating cerebrovascular disorders, was shown to modulate microglial polarization via the JAK2–STAT1/3 and GSK3–NF$\kappa$B pathways and to stimulate the expression of anti-inflammatory cytokines, IL-10 and TGF-$\beta$, in addition to downregulating the expression of IL-6 and alleviating axonal and myelinated sheath injuries [61]. Following the development of proteomics, RNA sequencing [62], epigenetics, cell-targeted deletion [63], and an increased understanding of the inflammatory and immunological processes occurring during stroke, new therapeutic targets can be identified. However, as the role of microglia is far from being binary [21], their intercellular communication, dynamic molecular profile, and signaling pathway mechanisms still need to be elucidated to find an effective neuroprotective treatment.

Stem cell-based therapies for ischemic stroke using mesenchymal stem cells (MSCs), neural stem cells (NSCs), induced pluripotent stem cells (iPSCs), and multilineage-differentiating stress-enduring cells (Muse cells) have been extensively researched recently. Indeed, stem cells can exert immunomodulatory, proangiogenic, and proneurogenic functions through their paracrine action, exosome secretion, and biobridge formations.

Because the main mechanism through which mesenchymal stem cells exhibit their paracrine properties is exosome secretion [34] and extracellular vesicles (EVs) derived from MSCs have been associated with promising therapeutic results in rodent stroke models [64, 65, 66, 67, 68, 69, 70, 71], current research has been focusing on elucidating the mechanisms of their action. Exosomes contain various proteins, including cytokines, chemokines, growth factors, and membrane receptors as well as miRNAs through which they promote neurogenesis, angiogenesis, and cell growth and reduce inflammation, oxidative stress, and cell death [72, 73, 74, 75]. Different miRNAs are carried in MSC-derived exosomes and target different mechanisms involved in stroke. It has been demonstrated that exosomes enriched with miRNA-17–92 increased the proliferation of neural and oligodendrocyte progenitor cells and increased neural plasticity via the PI3K/Akt/mTOR/GSK-3$\beta$ signaling pathway [76, 77]. Indeed, miRNA can inhibit an inflammatory response by inducing the microglia protective phenotype by either inhibiting cysteinyl leukotriene receptor 2 (CysLT2R) (miRNA-223) [78], suppressing the IRAK1/TEAF6 pathway (miRNA-146a) [79], regulating toll-like receptor 4 (TLR4) (miRNA-542) [80], or inhibiting the iron transporter—lipocalin-2 (LCN2) (miRNA-221) [81]. Moreover, they can promote angiogenesis by increasing the expression of VEGF through miR-210 [82, 83] and miRNA-21-5p [84] and by targeting the transient receptor potential melastatin 7 (TRPM7) (miRNA-181b) [85]. Furthermore, they were also shown to promote cell growth by modulating the KDM6B/BMP2/BMF axis [86] and inhibiting the apoptotic pathway [87]. It was demonstrated that serum-derived exosomes helped maintain the integrity of the blood–brain barrier by inhibiting apoptosis of endothelial cells via the upregulation of B-cell lymphoma 2 (Bcl2) and the inhibition of caspase-3 activation. Moreover, by inhibiting MMP-9 and microtubule-associated protein 1 light chain 3B (LC3B)-mediated autophagy, they help maintain tight junction proteins—zonula occludens-1 (ZO-1) and claudin 5 [88]. In addition, it is worth emphasizing that exosomes may also be used as biomarkers: their miRNA content varies in relation to the progress of stroke, they elude degradation due to the vesicular structure, and they can be found in all bodily fluids, including blood plasma, which makes them easy to isolate [89, 90, 91]. Their small size, lack of immunogenicity [92], the ability to pass through the blood–brain barrier [93], and escape phagocytosis and lysosome degradation [32] make them excellent candidates for stroke therapy. In addition, using exosomes has some important advantages over any therapy that uses mesenchymal stem cells, such as no risk of blood vessel occlusion and the ease of storage and modification [94, 95] (Fig. 1). However, although various mechanisms of action by the MSC-derived exosomes have been researched, exosome-based therapy still presents several translational challenges. Preclinical studies have been performed mainly on healthy animals [32]; however, stroke patients usually present many comorbidities [96], whose effects on stroke treatment should also be considered. Similarly, not enough studies have been performed on old animal stroke models [97]. Moreover, the appropriate administration methods, dosage, and time windows have not yet been established [98], while purification, large-scale production [99], and targeting [34] remain the challenges to overcome. However, potential solutions include applying microfluidic technologies for exosome isolation [100], myelocytomatosis oncogene (MYC) transformation for large-scale production [101], and peptide conjugation on the exosome surface to improve their targeting [102].

Fig. 1.

Exosome characteristics. Exosomes contain various proteins, including cytokines, chemokines, growth factors, and membrane receptors as well as microRNAs through which they promote neurogenesis, angiogenesis, and cell growth and reduce inflammation, oxidative stress, and cell death. They have the ability to pass through the blood–brain barrier and escape phagocytosis and lysosome degradation. Their advantages over cell therapies include ease of storage and modification and no risk of blood vessel occlusion. As their miRNA content varies in relation to the progression of stroke, they can be used as biomarkers.

While stem cell therapies were assessed to be relatively safe, with low risks of tumorigenesis [103] and the association of only minor adverse effects [104], the results in terms of their efficacy are mixed. Several studies did not show any significant difference between the treatment and control group [105, 106, 107]. Moreover, although many clinical trials demonstrated improvements in neurological functions [108, 109, 110, 111, 112], the studies yielding positive results presented high risks of selection, performance, and publication bias and had small sample sizes [113]. In addition, in most cases, they lacked an effective study design, including randomization, blinding, and statistical comparison [104]. Interestingly, the unsatisfactory results for efficacy can be explained by differences between the preclinical and clinical protocols [114]. In several trials, doses below the efficacious dose previously established in preclinical studies were used [106, 115]. Taking into consideration the above limitations of recent clinical trials, more studies with larger sample sizes, longer follow-ups, improved methodological designs, and better adherence to preclinical outcomes are needed.

In addition to mesenchymal stem cells, neural stem cells (NSCs) are also used in stroke research. However, their harvesting methods are problematic. Their sources are limited and transplantation from the adult brain would require complicated surgery. Derivation from neuroectoderm of the fetal tissue or from embryonic stem cells (ESCs) raises major ethical concerns and such transplantations may result in tumor formation [116, 117]. Generating NSCs from induced pluripotent stem cells (iPSCs) [118] or via direct neuronal reprogramming by omitting the PSC stage would eliminate any ethical issues and the risk of immune rejection because they can be obtained from the patient’s own cells [119, 120]. However, the reprogramming process is long and time-consuming and iPSC transplantations in animal models have been associated with tumor formation [121]; thus, improving their safety remains the primary issue [122]. Moreover, as their engraftment efficiency is low [123], their therapeutic effects are achieved mainly by the paracrine action, including secreting factors enhancing neurogenesis, angiogenesis, and reducing inflammatory responses [119]. Thus, the current research involving neural stem cells focuses primarily on the extracellular vesicles derived from NSCs that present lower tumorigenicity, improved blood–brain barrier (BBB) permeability, and biodistribution. The comparison of extracellular vesicles (EV) derived from neural stem cells and mesenchymal stem cells (both derived from the same pluripotent stem cell line) in animal stroke models demonstrated a higher effectiveness of NSC treatment. Therapy using NSC EVs resulted in a larger reduction in infarct size, greater improvement in somatosensory function, and smaller neurological deficits than after treatment with MSC EVs. Moreover, therapy using NSC EVs was associated with an increase in macrophages with protective phenotypes and a decrease in proinflammatory T helper 17 cells (Th17) [124]. Although the initial results are encouraging, more research studies elucidating the downregulation of inflammatory responses are still needed.

Interestingly, mesenchymal stem cells used in combination with neural stem cells have been proven more effective in animal models than the use of individual therapies [125], while the co-transplantation of MSCs and NCSs in stroke patients has been shown to be a safe and feasible method [112], making it a potential new therapy for ischemic stroke patients [126]. Furthermore, stem cell therapy has also been shown to be effective in combination with gene therapy [127, 128], tissue engineering scaffolds [129, 130, 131, 132], and reperfusion therapy [114].

Cell therapy using multilineage-differentiating stress-enduring cells (Muse cells) appears to overcome a large number of limitations by MSCs, NSCs, iPSCs, and ESCs (Table 1). Muse cells, first reported in 2010 [133], are found in a variety of tissues, such as bone marrow, peripheral blood, connective tissue, and the umbilical cord [134]. They can differentiate into three germ layers, including spontaneous in vivo differentiation into neuronal cells [135] and they can integrate into the neural network [136]. Moreover, they exhibit paracrine functions by secreting a variety of neurotrophic, proangiogenic, anti-inflammatory, and antiapoptotic factors. They are immune-privileged, meaning they inhibit the inflammatory immune response by, as suggested, expressing human leukocyte antigen G (HLA-G) molecules, meaning HLA-matching or immunosuppressant treatment is not required [134]. Due to the high expression of sphingosine-1-phosphate receptor 2 (S1PR2), as part of the S1P-S1PR2 axis, they can selectively migrate to the damaged site [137]. In addition, owing to their high capacity for DNA repair and lower telomerase activity and gene expression of tumorigenic factors than in ESCs and iPSCs, Muse cells are considered non-tumorigenic [134]. Muse cells have been researched in several animal stroke models [135, 138, 139, 140, 141] and were shown to differentiate into neuron cells, integrate into the cortex, improve motor functions and survival rates, and be assessed to be a safe treatment option. Moreover, in-human transplantations of allogenic Muse cells demonstrated safety and efficacy in clinical trials on myocardial infarction and dystrophic epidermolysis bullosa [142, 143]. However, there are several challenges associated with culturing Muse cells, whereby expanding their small populations is time-consuming and their culture cost is higher than that of other stem cell types. Moreover, golden standards with regard to cell sources, sorting methods, and donor age still have to be established [137] alongside more preclinical and clinical studies that investigate and illustrate their mechanisms of action.