Microbiota, systemic immunity, and extracellular vesicles in stroke: peripheral nodes as therapeutic leverage points

Abstract

Background

Stroke is increasingly understood as a systemic disorder rather than a brain-only lesion. Beyond the initial cerebral ischemic insult, rapid autonomic and neuroendocrine stress responses destabilize peripheral organ homeostasis and promote widespread immune and metabolic remodeling. Subsequent barrier failure and peripheral immune dysregulation can generate a sustained “second hit” in which circulating microbial products, damage-associated signals, and inflammatory mediators feedback to amplify neuroinflammation in a blood–brain barrier–vulnerable state. Meanwhile, post-stroke immunity is temporally plastic: inflammatory programs that worsen acute injury can later support resolution and repair, indicating that outcomes depend on immune balance and timing, not simply inflammatory magnitude.

Main body

Stem cell–derived extracellular vesicles (EVs) are emerging as multi-cargo biologics with consistent preclinical benefit across functional, histological, and inflammatory endpoints. However, clinical translation has progressed slowly, in part because development has largely prioritized strategies to enhance central nervous system delivery even though systemically administered vesicles typically show low exposure in brain parenchyma. Here, we propose a “periphery-first” therapeutic strategy that reframes this pharmacokinetic profile as an advantage. By leveraging the natural sequestration of systemically delivered vesicles by reticuloendothelial and barrier-associated organs—particularly the liver, spleen, and gut—this approach aims to reprogram peripheral immune trajectories, strengthen barrier integrity, and suppress humoral amplification loops that sustain secondary brain injury. We synthesize evidence for stroke-driven multi-organ dysfunction and phase-dependent immune remodeling and integrate mechanistic plausibility for EVs acting through complementary routes: peripheral immune and metabolic rebalancing, actions at the blood–brain barrier interface and limited but potentially meaningful effects within central nervous system immune niches. We also summarize the emerging clinical landscape of EV interventions in stroke and highlight key translational constraints, including product heterogeneity and potency-linked quality control, comorbidity-relevant modeling aligned with systemic pathology, dosing and safety limitations imposed by hepatic clearance, and the need for artifact-resistant biodistribution methods and causal necessity/sufficiency study designs to quantify route-to-efficacy.

Conclusion

A periphery-first framework positions EV therapy as a systems-level intervention that targets peripheral drivers of secondary brain injury. Establishing quantitative causal mechanisms and translation-ready manufacturing and dosing principles will be essential to accelerate clinical development beyond a primarily brain-delivery paradigm.

Background

Stroke remains a leading cause of death and long-term disability worldwide, with a persistently high global burden that is unevenly distributed across regions and populations, underscoring major disparities in epidemiology, prevention, and access to care [1]. In acute ischemic stroke, intravenous (IV) thrombolysis and endovascular thrombectomy have improved outcomes in selected patients, yet their overall impact is constrained by strict eligibility criteria and time sensitivity in real-world practice [2, 3]. Importantly, treatment sequencing and timing continue to matter: a meta-analysis evaluating time to IV thrombolysis before thrombectomy linked earlier treatment to better functional outcomes, reinforcing that benefit is highly time-dependent and may still be limited by downstream injury processes even with reperfusion [2, 3].

Despite decades of effort, effective neuroprotection has remained elusive in the clinic. Large-scale reviews and consensus recommendations have highlighted a long history of preclinical “successes” that failed to translate, emphasizing issues such as model heterogeneity, bias, insufficient reproducibility, and gaps between experimental designs and clinical reality [4–6]. Collectively, these lines of evidence suggest that central nervous system (CNS)-centric neuronal protection alone is unlikely to be sufficient in many patients, particularly when stroke initiates complex systemic responses that can shape injury evolution and recovery trajectories [4, 7].

Growing evidence supports reframing stroke, particularly ischemic stroke, where most mechanistic and translational evidence has accumulated, as a systemic disorder characterized by coordinated neuroinflammatory and peripheral immune remodeling. Reviews of post-stroke immune-modulation describe rapid, phase-dependent changes in innate and adaptive immunity that contribute both to secondary brain injury and to systemic complications [8]. Beyond immune-cell trafficking, stroke induces broad whole-body pathophysiological responses—spanning neurohumoral stress signaling, vascular and metabolic perturbations, infection susceptibility, and multi-organ dysfunction—supporting the concept that brain injury and peripheral organs engage in bidirectional crosstalk [7, 9]. These systemic responses are not merely downstream manifestations of brain injury; rather, through mechanisms such as barrier disruption, the release of circulating inflammatory mediators, immune cell recruitment, and organ dysfunction, they can serve as secondary drivers of persistent CNS damage. Within this framework, the gut–brain axis has emerged as a major node: work synthesizing animal and clinical observations links acute ischemic stroke to microbiota shifts, intestinal barrier disruption, and peripheral inflammatory signaling that may feed back to influence neuroinflammation and outcomes [10]. Recent reviews further consolidate how neuroinflammation intersects with peripheral immunity, arguing that peripheral inflammatory programs are not merely bystanders but can actively modulate BBB integrity, inflammatory cell recruitment, and post-stroke recovery [9, 11].

These insights motivate therapeutic strategies capable of multimodal immunomodulation and repair, including extracellular vesicle (EV)-based approaches. Stem cell–derived EVs have been reviewed as a rapidly advancing modality for cerebral ischemic stroke, with proposed advantages including cell-free delivery of bioactive cargos that can influence inflammation, vascular remodeling, and tissue repair pathways [12]. Delivery route is a key translational variable: intra-arterial administration has been discussed as a strategy to improve CNS targeting for cells and exosomes, while also raising practical considerations around safety, microvascular risk, and distribution [13]. In parallel, EV surface modification/functionalization technologies are expanding, aiming to enhance targeting, stability, and therapeutic precision—yet they also heighten manufacturing complexity and regulatory demands [14]. Across these EV-focused reviews, recurrent translational challenges include pharmacokinetics (PK) and biodistribution, dose and route optimization, scalable manufacturing, and potency-linked quality control (QC)—factors that will likely determine whether promising preclinical signals can be converted into clinically reliable benefit [12–14].

In this review, we investigate stroke-induced brain–periphery crosstalk as a mechanistic foundation for peripheral therapeutic strategies. Most mechanistic evidence regarding microbiota–immune interactions and EV-based interventions have been derived from studies on ischemic stroke or ischemia–reperfusion models; thus, ischemic stroke serves as the primary focus of this review. However, emerging research on hemorrhagic stroke, particularly intracerebral hemorrhage and subarachnoid hemorrhage, suggests that inflammatory activation, blood-brain barrier (BBB) disruption, and systemic immune remodeling also occur in these subtypes [15], although the supporting evidence is less developed and not as directly associated with EV-based interventions. Therefore, we argue that stroke should be viewed as a whole-body disorder rather than a purely cerebral event. We first synthesize evidence that brain injury rapidly triggers dysfunction in peripheral organs—including the gut, liver, spleen, heart, lung, and kidney—and that neurohumoral signaling creates self-reinforcing inflammatory loops that can aggravate neurological outcomes. We then examine how barrier breakdown and time-dependent immune remodeling unfold after stroke, emphasizing bidirectional immune trafficking and stage-specific programs that shape both secondary injury and repair, and explaining why stem cell–derived extracellular vesicles (SC-EVs) are well positioned to promote a shift from persistent inflammation toward resolution [7, 8, 10]. Next, we map the evolving clinical landscape of EV-based therapies in stroke and place it in context with CNS-focused neuroprotection and microbiota-directed approaches, highlighting where EVs may provide practical advantages—and where uncertainties around safety, dosing, scalable production, and endpoint selection still limit translation [13, 14]. Finally, we outline the key hurdles that must be addressed to move EVs from promise to practice, including potency-linked QC, models that reflect clinical comorbidities, and rigorous causal testing of peripheral mechanisms using robust biodistribution and necessity/sufficiency frameworks. Overall, adopting a systemic immune–metabolic perspective supports a periphery-informed therapeutic strategy and offers a clearer roadmap for translating EV interventions beyond traditional brain-centric paradigms.

The double-hit framework: how the periphery becomes a self-reinforcing driver

Stroke is increasingly recognized as a bidirectional “central–peripheral” pathophysiological cascade that can be conceptualized as a dynamic double-hit model: an initial brain-to-periphery signal wave is followed by peripheral organ dysfunction that feeds back to the CNS through inflammatory mediators, barrier disruption, immune-cell trafficking, and organ dysfunction, thereby reinforcing neuroinflammation, secondary CNS injury, and poor recovery (Fig. 1) [16].

Fig. 1.

Systemic “double-hit” framework in ischemic stroke. This model depicts how stroke-induced injury spreads throughout the body, contributing to worsened brain damage. Hit 1 (Neurogenic Stress): When cerebral ischemia occurs, it triggers the SNS and the HPA axis, disturbing the balance of peripheral organs, especially the gut, spleen, and liver. This results in a compromised gut barrier, allowing harmful molecules like LPS and PAMPs to enter the bloodstream. The spleen releases immune cells, while the liver produces inflammatory signals such as HMGB1 and various cytokines. Hit 2 (Immune and Humoral Feedback): These factors circulate back to the brain, increasing BBB permeability, promoting immune cell infiltration, and activating microglia. This feedback loop intensifies neuroinflammation, leading to expanded brain damage. The bold arrow shows the ongoing feedback process. BBB, blood–brain barrier; HPA, hypothalamic–pituitary–adrenal; HMGB1, high mobility group box 1; IL-17, interleukin-17; LPS, lipopolysaccharide; PAMPs, pathogen-associated molecular patterns; SNS, sympathetic nervous system; NE, norepinephrine; ZO-1, zonula occludens-1; γδ T, gamma delta T cell

The first hit is an immediate neurogenic stress response: acute cerebral ischemia engages autonomic outflow (particularly sympathetic activation) and the hypothalamic-pituitary-adrenal axis (HPA) axis, consistent with the broader stress/adaptation framework in which the brain coordinates systemic “stability through change” (allostasis) via catecholamines and glucocorticoids, but can incur harmful downstream consequences when responses are excessive or prolonged (allostatic load) [17–19]. In the stroke setting, these neuroendocrine–autonomic signals rapidly reshape immune trafficking and disturb peripheral barriers and organ physiology—notably across the gut, spleen, and liver—thereby creating a permissive environment for secondary complications and systemic inflammatory dysregulation [17–19].

The second hit should be refined not only as a delayed but self-sustaining systemic immune phase, but as a peripheral-to-CNS feedback phase in which gut barrier failure, dysbiosis, and immune-metabolic organ reprogramming reinforce CNS inflammation. As the gut microbiota becomes dysregulated and intestinal barrier integrity deteriorates, microbiota-derived products and metabolites increasingly influence systemic inflammation and infection susceptibility, creating a permissive milieu for secondary injury [20]. In parallel, damage-associated molecular patterns (DAMPs) released from dying brain cells initiate and amplify inflammatory cascades that spill over into the circulation (a “cytokine storm”), contributing to peripheral organ dysfunction and further systemic inflammation [21]. These circulating signals reprogram peripheral immune compartments, producing a biphasic state of systemic activation followed by immunodepression and infection risk, which is increasingly recognized as a key determinant of outcome [22]. At the CNS interface, ischemia rapidly erodes the functional separation between the brain and peripheral immunity [23, 24]. Oxidative stress and inflammatory mediators disrupt BBB integrity, enabling infiltration of circulating leukocytes (neutrophils, lymphocytes, monocytes) into the ischemic brain and accelerating secondary injury [24]. Thus, peripheral dysfunction can worsen CNS outcomes through four converging routes: humoral amplification by cytokines and DAMPs/PAMPs, cellular feedback through mobilized leukocytes, barrier-mediated entry across a vulnerable BBB, and physiological stress imposed by distal organ dysfunction.

Importantly, post-stroke immunity is phase dependent: immune cells contribute to acute injury but later participate in repair by clearing debris, secreting trophic factors, and supporting neurogenesis and vascular remodeling [23]. This temporal plasticity provides a mechanistic entry point for SC-EVs, which may reprogram immune trajectories systemically and locally, potentially shifting post-stroke inflammation toward repair. However, peripheral EV uptake should be interpreted as a testable therapeutic mechanism rather than assumed efficacy, and should be validated through biodistribution, target-engagement, and sufficiency studies.

Phase-dependent immune programs after stroke

Post-stroke inflammation is often “double-edged,” with the same cell class exerting opposing effects depending on timing, location, and activation state. Adaptive immune responses illustrate this clearly. In experimental stroke, T cells exert time-dependent effects: during the acute stage, CD4⁺, CD8⁺, and γδ T cells can aggravate injury through cytokine release, platelet interactions, and cytotoxicity; accordingly, T-cell–deficient models often exhibit smaller infarcts [25–27]. In contrast, regulatory T cells (Tregs) are often protective during subacute/chronic phases by secreting IL-10, limiting MMP-9 activity, stabilizing BBB integrity, and restraining neuroinflammation, thereby supporting repair programs [27–29]. B cells are similarly context dependent: in the acute phase, they can enhance inflammation via cytokines (e.g., interleukin [IL]-6, interferon [IFN]-γ) and reactive oxygen species (ROS), yet they may also suppress injury through IL-10–dependent pathways that reduce neutrophil infiltration and infarct volume [30–32]. During recovery, certain B-cell programs support neurogenesis and functional improvement, whereas persistent B-cell infiltration and antibody production can contribute to cognitive decline and post-stroke dementia phenotypes [33–35].

Innate immune cells also show temporal divergence: neutrophils acutely promote BBB breakdown, thrombosis, and oxidative injury [36–39], yet later neutrophil programs can aid resolution by degrading DAMPs and releasing growth-promoting mediators [40, 41]. Microglia and infiltrating monocyte-derived macrophages likewise span pro-inflammatory injury (tumor necrosis factor [TNF], IL-1, IL-6; cytotoxic mediators) and pro-repair functions (IL-10, transforming growth factor-β [TGF-β]; efferocytosis; remodeling) [42–47].

Superimposed on CNS inflammation, stroke induces a systemic transition from acute inflammation to stroke-induced immunosuppression, characterized by lymphoid organ atrophy (spleen/thymus), peripheral lymphopenia, impaired monocyte/granulocyte function, and a sharply increased infection risk [48–51]. Peripheral lymphocytes (T, B, NK) fall within hours after stroke, and experimental ischemia triggers marked splenic atrophy and lymphocyte apoptosis, indicating that peripheral immunity is actively remodeled rather than passively affected [28, 49].

Gut microbiota-immune-brain coupling after stroke

With the gut–brain axis (GBA) converting intestinal dysfunction from a downstream consequence into an active driver of systemic inflammation and secondary brain injury, gut is regarded as an early and highly vulnerable peripheral target after stroke [20, 52, 53]. For example, in a middle cerebral artery occlusion (MCAO) mouse model, mast cells, granule-packed immune cells mainly in tissues—especially at “barrier” sites like the skin, airways, and gut, and around blood vessels and nerves, were identified as early responders that exacerbate systemic inflammation and worsen gut-barrier disruption [54]. Pharmacological mast-cell stabilization via cromolyn reduced circulating histamine and IL-6, preserved intestinal barrier integrity, attenuated stroke-associated dysbiosis, and improved functional recovery despite minimal change in infarct volume—supporting the concept that gut immune activation can shape neurological outcomes independently of lesion size [54].

Bidirectional neural and endocrine pathways, primarily the autonomic nervous system (ANS) and HPA axis, enables rapid central-to-peripheral signaling immediately after ischemia and sustains longer-term immunometabolic remodeling during recovery [10, 52, 55]. In the acute “neurogenic rapid-response” phase (hours after onset), ischemic stroke acts as a potent systemic stressor that induces excess sympathetic nervous system (SNS) activation and HPA-axis signaling, thereby reducing gut perfusion and precipitating intestinal ischemia–reperfusion–like injury, which directly compromises epithelial integrity and barrier function [10, 55]. Within this window, peripheral immune effectors can amplify both gut and brain inflammation. Neural control of the GBA is not limited to sympathetic pathways. The vagus nerve (VN) represents a major hub for microbiome–gut–brain communication and a key modulator of post-stroke inflammation [56, 57]. Vagal afferents (the majority of VN fibers) sense microbial metabolites (including short-chain fatty acids, SCFAs) and enteroendocrine mediators (e.g., 5-HT, CCK) and relay these signals to the nucleus tractus solitarius (NTS) and central autonomic network (CAN), which then shapes sympathetic/parasympathetic output to regulate gut motility, permeability, and immune tone [56]. Conversely, vagal efferents engage the cholinergic anti-inflammatory pathway (CAP), in which acetylcholine signaling through macrophage α7 nicotinic acetylcholine receptors (α7nAChR) restrains inflammatory cytokine release (e.g., TNF-α, IL-1β), thereby forming a neuroimmune negative-feedback reflex [56, 58]. Beyond cytokine suppression, vagal activity has been linked to barrier preservation through promotion of anti-inflammatory macrophage polarization and upregulation of tight-junction proteins (e.g., occludin, Zonula occludens-1 [ZO-1]), supporting a mechanistic rationale for gut-centric neuroimmune interventions during stroke recovery [56, 57].

As stroke progresses into the subacute-to-chronic phase, gut dysfunction frequently evolves from transient neurogenic injury into a persistent state characterized by dysbiosis, reduced epithelial repair capacity, and sustained inflammatory signaling [20, 53, 59]. Mechanistically, post-stroke dysbiosis and barrier breakdown facilitate the systemic spillover of microbial components such as LPS and other pathogen-associated molecular patterns (PAMPs), which amplify inflammation through innate immune sensors, notably the toll-like receptor 4 (TLR4)/MyD88 axis, and can secondarily compromise BBB integrity, thereby reinforcing neuroinflammation [53, 60–62]. At the epithelial level, stroke is repeatedly associated with downregulation of tight-junction proteins (ZO-1, occludin, claudins), producing a “leaky gut” state that promotes bacterial translocation and circulating inflammatory mediator accumulation; importantly, interventions that restore tight-junction expression can improve both gut and brain pathology, supporting intestinal barrier integrity as a tractable therapeutic target [63–67]. Preclinical interventional studies indicate that this stage is modifiable and mechanistically relevant. For instance, the prebiotic Puerariae Lobatae Radix–resistant starch (PLR-RS) alleviated brain injury and restored gut-barrier integrity in ischemic stroke rats while reshaping microbial ecology [68]. Fecal microbiota transplantation (FMT) experiments supported microbiota-dependent causality, and mechanistic analyses implicated a gut microbiota–driven increase in melatonin as a mediator of neuroprotection [68]. These data highlight an emerging paradigm in which diet–microbiome–metabolite pathways can be therapeutically leveraged to stabilize gut homeostasis and dampen post-stroke inflammation [20, 68]. Clinical observations are consistent with the relevance of gut dysregulation to stroke pathophysiology. In cryptogenic stroke, patients exhibit gastrointestinal dysfunction accompanied by biomarkers of barrier disruption and systemic inflammation (e.g., elevated CRP, LPS, LBP, leukocytosis), alongside distinct microbial signatures enriched for pro-inflammatory taxa (including Enterobacteriaceae and related genera) that correlate with stroke severity indices and infarct burden [69]. Consistent with causality, multiple studies demonstrate that transferring a post-stroke or stroke-prone dysbiotic microbiota to naïve recipients worsens infarct severity and functional deficits, whereas FMT or microbiota modulation can reverse dysbiosis and improve outcomes—positioning the gut as both a target of the early neurogenic insult and a key engine of the later humoral “second hit” [53, 55, 70].

In addition to soluble LPS, PAMPs, and microbial metabolites, gut microorganisms can release vesicle-associated signals. These particles are broadly referred to as bacterial membrane vesicles (MVs), a heterogeneous group of lipid-bilayer vesicles released by both Gram-negative and Gram-positive bacteria [71]. Outer membrane vesicles (OMVs) are the classical Gram-negative subtype generated mainly by outer-membrane blebbing [72], whereas Gram-positive bacteria can release cytoplasmic membrane vesicles (CMVs) derived from the cytoplasmic membrane and passing through the peptidoglycan-rich cell wall [73].Gram-negative OMVs are nanoscale outer-membrane-derived vesicles, typically approximately 20–250 nm, whereas Gram-positive EVs bud from the inner membrane and traverse the thick cell wall before release [74, 75]. Functionally, OMVs can carry LPS, peptidoglycan fragments, outer-membrane/periplasmic proteins, adhesins, toxins, virulence factors, and immunomodulatory molecules, whereas lysis-associated MVs and CMVs may additionally carry cytoplasmic material, DNA/RNA, or endolysins [76, 77]. In a post-stroke leaky-gut state, microbiota-derived MVs could provide a particulate route for delivering microbial cargos to epithelial, endothelial, and immune cells, thereby modulating Nuclear factor-κB (NF-κB)-, inflammasome-, cytokine-, endothelial-, and epithelial-barrier-related responses [78–80]. Consistent with this source-dependent view, Tian et al. found that Bacteroides fragilis OMVs carrying polysaccharide A promoted dendritic-cell-mediated Treg/IL-10 responses and ameliorated DSS-induced colitis, whereas pathogenic bacterial EVs such as E. coli, Salmonella, or Pseudomonas OMVs could promote inflammatory cytokine release or inflammasome activation [81]. However, direct evidence that gut-derived bacterial MVs drive post-stroke inflammation or CNS injury remains limited; thus, they should be viewed as a plausible, hypothesis-generating extension of the gut-derived humoral “second hit,” rather than an established causal mechanism.

Beyond humoral signaling, the gut also acts as an immunological reservoir that can supply pathogenic effector cells to CNS border tissues after stroke. In mice, ischemic stroke promotes the expansion of IL-17–producing intestinal γδ T cells that subsequently traffic to the meninges and aggravate neuroinflammation; microbiota depletion with antibiotics reduced meningeal IL-17 + γδ T-cell responses and decreased infarct volume, implicating microbiota–T cell coupling as a causal axis [70, 82]. Complementarily, increases in gut-derived Treg responses can restrain Th17 differentiation and limit pro-inflammatory trafficking to the brain, consistent with a competitive balance between pathogenic and regulatory intestinal T-cell programs during recovery [83, 84]. More broadly, immune cell-tracing studies support directed redistribution of intestinal immune populations into peripheral lymphoid tissues and CNS-associated compartments, reinforcing the concept that the gut actively participates in post-stroke immune deployment rather than functioning solely as a passive barrier [85–87]. Additional systemic disease models suggest EVs can influence microbiota-immune coupling and Treg/Th17 balance, further supporting a gut–immune mechanism compatible with stroke-associated dysbiosis and systemic inflammation [88].

Beyond immune-cell trafficking, microbiota-derived metabolites constitute a parallel humoral route through which gut dysbiosis may influence post-stroke neuroinflammation, vascular integrity, and recovery [89]. Stroke-specific metabolomic studies, together with integrated microbiome–metabolome analyses and mechanistic literature, suggest that ischemic stroke is accompanied by alterations in several microbial-related metabolic pathways, commonly involving SCFAs and TMAO-related choline/carnitine metabolism, with emerging evidence implicating tryptophan-derived metabolites, bile-acid signaling, and amino-acid metabolites [90–92]. These signatures have been associated with inflammatory cytokine profiles, insulin resistance, endothelial dysfunction, stroke severity, and recovery-related outcomes, suggesting that microbial metabolites may serve as biomarkers and candidate mediators of post-stroke immune–vascular remodeling. However, most available evidence remains observational and heterogeneous with respect to sampling time, biofluid sources, and outcome definitions; thus, causal interpretation will require longitudinal microbiome–metabolome–immune–outcome profiling.

SCFAs provide a representative example linking microbiota composition to immune regulation and recovery, but their interpretation is time- and compartment-dependent. Depletion of SCFA-producing taxa and reduced SCFA availability can weaken epithelial repair programs and anti-inflammatory homeostasis, while expansion of opportunistic Gram-negative bacteria (e.g., Enterobacteriaceae) increases the intestinal endotoxin burden, thereby potentiating a self-reinforcing cycle of barrier dysfunction and systemic inflammation [54, 59, 93]. Clinically, Tan et al. found that reduced SCFA levels, especially acetate, were associated with an increased risk of poor 90-day functional outcome [94], whereas Henry et al. reported that plasma SCFAs measured at the time of thrombectomy were associated with inflammatory markers and worse symptoms at discharge [95]. These apparently divergent findings suggest that fecal and circulating SCFAs may reflect different biological states during acute and recovery phases. Consistent with a mechanistic role, experimental SCFA supplementation improves post-stroke recovery through immunological mechanisms, including modulation of inflammatory cytokine production and barrier-related pathways [96].

In parallel, TMAO—generated by microbial metabolism of dietary choline/carnitine—has been associated with cerebrovascular risk and poorer stroke-related indices in multiple clinical datasets, although its contribution may vary by disease stage and microbial context, emphasizing the need to interpret metabolite associations within a dynamic, time-resolved framework [20, 97–100]. Mechanistically, TMAO may connect gut dysbiosis to vascular dysfunction through endothelial activation, platelet hyperreactivity, vascular inflammation, and atherothrombotic risk. Experimental studies summarized in recent reviews suggest that TMAO may promote reactive astrogliosis and glial scar formation after MCAO/R and may activate NLRP3 inflammasome–IL-1β signaling, linking microbial metabolism to neuroinflammatory remodeling [101]. Other microbial-related pathways, including tryptophan-derived kynurenine indole metabolites, bile-acid signaling, and amino-acid metabolites, may provide additional links to immune–glial activation, gut–liver–vascular signaling, and metabolic stress; however, compared with SCFAs and TMAO, their stroke-specific causal roles remain less established [92, 102, 103].

Together, these findings support a coherent model in which the gut is an immediate victim of neurogenic injury after stroke and a dominant source of sustained inflammatory reinforcement via barrier failure, dysbiosis, immune-cell trafficking, and metabolite-mediated signaling—thereby making gut stabilization a rational, mechanistically grounded strategy to mitigate secondary brain injury and improve recovery.

Peripheral immune-metabolic hubs as amplifiers: liver and spleen

Stroke initiates a systemic loss of homeostasis that extends well beyond the BGA. The neurogenic burst (autonomic/HPA-driven) and the subsequent humoral wave (circulating DAMPs/PAMPs, cytokines, and dysregulated metabolites) propagate injury signals across peripheral networks, engaging immune–metabolic hubs (liver and spleen) and ultimately compromising life-support organs (heart, lung, kidney). Together, these processes form the mechanistic substrate of stroke-associated multi-organ dysfunction syndrome (MODS), in which peripheral injury is not merely comorbidity but an active contributor to neurological deterioration and poor outcome [104–106].

As the body’s dominant metabolic organ and a major immune filter, the liver sits at the center of the neuro–humoral network activated by stroke. In the early neurogenic phase, central stress signaling directly reprograms hepatic immunity and metabolism via sympathetic innervation and neuroendocrine cues. Using intravital imaging in the MCAO mouse model, Wong, Jenne et al. demonstrated that ischemic stroke triggers noradrenergic signaling in the liver, inducing marked behavioral and functional changes in hepatic invariant NKT (iNKT) cells and promoting an IL-10–biased response consistent with systemic immunosuppression and heightened infection susceptibility [104]. This hepatic neuroimmune circuit illustrates that the liver is not a passive bystander but a direct neurogenic target capable of shaping systemic immune tone after brain ischemia [104]. In parallel, stroke can evoke hepatic danger signaling. Davaanyam, Seol et al. reported elevated hepatic high mobility group box 1 (HMGB1) after cerebral ischemia, consistent with DAMP-mediated amplification of peripheral inflammation and liver injury [107]. As the disease evolves, the liver becomes a primary target of the second hit, receiving continuous exposure to gut-derived inflammatory stimuli through the portal circulation. In the setting of post-stroke dysbiosis and barrier disruption, PAMPs such as LPS enter portal blood and activate hepatic innate immune pathways, potentially sustaining an acute-phase response and reinforcing systemic cytokine production [55, 60, 107]. This gut–liver axis is particularly relevant because pre-existing metabolic liver vulnerability can predispose to worse systemic inflammation and impaired recovery capacity. In stroke-prone hypertensive rats, Kanezawa, Moriyama et al. showed that a high-fat/high-cholesterol diet induces steatohepatitis-like pathology accompanied by pronounced microbiota shifts (e.g., altered Firmicutes/Bacteroidetes ratio and changes in Lactobacillus/Streptococcus in the small intestine), highlighting a disease context in which hepatic–microbiome coupling may increase baseline susceptibility to inflammatory amplification after ischemic events [108]. Mechanistically, this peripheral immune remodeling includes liver-specific neuroimmune circuitry: stroke induces noradrenergic signaling–dependent behavioral changes in hepatic iNKT cells, driving immunosuppression, and targeted iNKT modulation can change post-stroke infections—illustrating how peripheral immune organs are not merely bystanders but active nodes in a feed-forward loop [104]. Clinically, hepatic dysfunction has measurable prognostic relevance in severe stroke care pathways. In a multicenter thrombectomy cohort (n = 420), Huang, Zha et al. reported that post-procedural abnormal liver function tests (ALFT) were common and independently associated with worse 90-day functional outcomes and higher mortality, indicating that peri-stroke hepatic stress responses correlate with clinical trajectory even when pre-procedural liver tests are not predictive [109].

Notably, emerging work also suggests that the liver can generate protective endocrine-like signals that influence brain recovery. In a rat MCAO model, Li, Lian et al. showed that delayed recanalization enhanced brain delivery of liver-derived TFF3, and that TFF3 signaling through a LINGO2–EGFR/Src axis reduced apoptosis and improved functional outcomes, emphasizing that hepatic responses may be bidirectional—capable of both amplifying injury and enabling repair depending on context and timing [110]. Taken together, the liver functions as a phase-specific integrator and amplifier in the double-hit model: early neurogenic signaling rapidly reshapes hepatic immunity, while later gut-derived humoral inputs sustain inflammatory programming and systemic toxicity—yet selected liver-derived mediators can also support neuroprotection under defined conditions [104, 107, 109, 110].

The spleen is a dominant peripheral immune reservoir that rapidly transforms cerebral injury into systemic immune deployment after stroke. Splenectomy studies provide causal support that the spleen is not merely a biomarker of systemic inflammation but a functional contributor to secondary brain injury [111]. A substantial body of rodent evidence indicates that pre-stroke splenectomy can reduce infarct volume and attenuate stroke-associated immune activation, supporting a detrimental role for excessive splenic immune-cell mobilization [112, 113]. More quantitatively, Sternak et al. conducted a systematic review and meta-analysis of murine ischemic stroke models and found that splenectomy reduced ischemic lesion volume, with an SMD of − 1.42 (95% CI, − 1.98 to − 0.85), and improved neurological deficit scores, with an SMD of − 1.20 (95% CI, − 2.20 to − 0.20) [114]. However, the same meta-analysis highlighted timing effects, study heterogeneity, and risk of bias; therefore, splenectomy should be interpreted as a mechanistic probe of spleen-dependent inflammation rather than as a clinically viable therapy [114].

Importantly, splenic involvement is context dependent rather than uniformly deleterious. In a rat MCAO model combined with limb remote ischemic preconditioning, Chen et al. showed that RIPC increased splenic volume and splenic lymphocyte populations, reduced infarct volume and cerebral edema, and reshaped circulating lymphocyte composition [115]. Splenectomy before RIPC and MCAO attenuated these protective effects, suggesting that splenic immune responses may be injurious or protective depending on timing and intervention context.

In preclinical models, stroke induces an early splenic contraction that coincides with immune-cell mobilization, and related splenic volume dynamics are also detectable in human patients. In rats subjected to MCAO, Seifert, Hall et al. demonstrated a transient decrease in spleen size at 24–48 h after MCAO in rats followed by recovery by ~ 96 h, and cell-tracking experiments indicated that splenic leukocytes redistribute into the circulation during this contraction phase [116]. Consistent with these preclinical dynamics, Sahota et al. conducted a prospective observational study using daily abdominal ultrasound in 30 patients with suspected acute ischemic stroke and 20 healthy individuals. Longitudinal modeling showed a significant change between baseline spleen volume measured within 6 h of stroke onset and the last measured time point up to 7 days, with a volume difference of 51.9 cm³ (P = 0.04) [117]. Zha et al. further analyzed serial abdominal ultrasound data from acute ischemic stroke patients without infection at admission and linked spleen contraction to systemic inflammatory response syndrome (SIRS), age, race, and leukocyte remodeling; at 48 h, spleen contraction was associated with higher lymphocytosis and lower neutrophil percentages, whereas SIRS at 72 h was associated with worse discharge mRS [118]. Complementarily, Chiu, Kaiser et al. found a biphasic pattern in splenic volume with an early nadir around up to 48 h and subsequent rebound, correlating with peripheral leukocyte composition (inverse with lymphocyte percentage and positive with neutrophil percentage), supporting splenic remodeling as a potential biomarker of systemic immune events [119]. Together, these ultrasound and imaging studies indicate that splenic volume is not a static anatomical marker but a dynamic readout of post-stroke systemic immune remodeling.

Mechanistically, the spleen contributes to post-stroke pathology through two tightly linked functions: (i) rapid immune-cell deployment and (ii) inflammatory polarization by circulating mediators. After mobilization, splenic monocytes/macrophages can traffic toward the ischemic brain and constitute a substantial fraction of infiltrating myeloid cells, supporting a spleen-to-brain axis operating alongside other reservoirs (e.g., gut-derived immune populations) [53, 70, 120]. In parallel, the secondary humoral hit shapes splenic immune programming: circulating LPS (from leaky gut) can activate splenic innate immune pathways (e.g., TLR4-dependent signaling), promoting systemic production of inflammatory cytokines such as TNF-α, IL-6, and IL-1β, while dysregulated metabolites (including altered SCFA availability) may modulate helper T-cell polarization and peripheral immune tone [55, 121, 122]. Moreover, liver-derived danger signals such as HMGB1 can further stimulate splenic immune activation, reinforcing a gut–liver–spleen inflammatory circuit [107].

At the clinical level, persistent systemic inflammation correlates strongly with poor outcomes. Elevated cytokines—particularly IL-6, but also TNF-α, IL-1β, and IL-8—have repeatedly been associated with greater stroke severity and worse functional recovery in meta-analyses and longitudinal cohorts [123–129]. Functionally, these inflammatory programs can increase leukocyte recruitment and parenchymal injury, thereby expanding infarct evolution beyond the initial ischemic insult [49, 50, 116, 130]. Collectively, these findings position the spleen as an amplifier node that converts neurogenic and humoral inputs into systemic immune deployment and cytokine escalation, which then feeds back to the CNS and shapes the neuroinflammatory and repair landscape [55, 116, 119, 120].

For the periphery-first EV framework proposed here, these spleen studies are particularly relevant because the spleen is both a stroke-responsive immune reservoir and a major site of systemically administered EV capture. Splenic volume dynamics, leukocyte composition, and cytokine signatures may therefore serve as candidate target-engagement readouts when testing whether EVs reprogram peripheral immune trajectories rather than acting solely through direct brain delivery.

To sum up, the liver and spleen convert gut-derived and brain-derived danger signals into systemic cytokine tone and immune-cell deployment; once the BBB is destabilized, these peripheral outputs can re-enter the CNS inflammatory circuit and expand secondary neurovascular injury.

Distal organs under a neuro-humoral storm: heart, lung, and kidney

Once immune hubs and systemic mediators are engaged, distal organs become vulnerable targets of the same two-hit architecture. Clinical syntheses indicate a substantial burden of cardiac complications after stroke, supporting the concept of a stroke–heart syndrome [105]. Cardiac dysfunction after stroke includes overlapping phenotypes such as myocardial injury with cardiac troponin elevation, electrocardiographic (ECG) abnormalities including QT interval (QTc) prolongation and ST-segment–T-wave (ST–T) changes, atrial or ventricular arrhythmias, left ventricular dysfunction, acute heart failure, and stress cardiomyopathy/Takotsubo-like presentations [131, 132]. Fan et al. summarized this spectrum of acute ischemic stroke (AIS)-related cardiovascular dysfunctions and emphasized their association with adverse outcomes and mortality [133, 134]. Representative studies support these phenotypes: Alkhachroum et al. reported that type II myocardial infarction with cardiac troponin elevation was associated with stroke severity, discharge disposition, and mortality in patients with ischemic stroke, intracerebral hemorrhage, and subarachnoid hemorrhage [135]; Hromadka et al. found prolonged QTc in 65.2% of patients at baseline and 26.1% after 48 h [136]; and Adeoye et al. reported systolic dysfunction in approximately one-third of 1020 acute stroke patients, associated with increased 1-month mortality [137]. These phenotypes align with the neuro-humoral storm framework, as neurogenic stress cardiomyopathy after acute brain injury involves sympathetic hyperactivity, HPA-axis activation, and immune/inflammatory pathways [138, 139]. Stroke can precipitate “stroke–heart” interactions through acute sympathetic storm and subsequent immune-mediated injury. Early mechanistic work demonstrated that catecholamine surges after stroke can be dose-linked to myocardial injury patterns, highlighting the cardiotoxic potential of neurogenic sympathetic discharge [140]. In the subsequent humoral stage, immune trafficking becomes prominent. In the MCAO model, macrophage infiltration into the heart contributes to dysfunction, and blocking inflammatory chemotaxis can alleviate cardiac impairment [141]. In addition, reduced plasma miR-126 after cerebral ischemia has been implicated in aggravated cardiac fibrosis, supporting the existence of specific soluble “messengers” connecting brain injury to cardiac remodeling, indicating that circulating regulatory factors such as miRNAs can act as cross-organ mediators [142].

Lung injury shows a temporally staged shift from neurogenic vascular effects to infection driven by immune dysregulation. In the acute phase, intense sympathetic activation can promote neurogenic pulmonary edema via hemodynamic stress and capillary leak [143]. In subacute stages, stroke-induced immune suppression—compounded by gut barrier disruption and microbial translocation—markedly increases susceptibility to pneumonia, which is a leading cause of non-neurological mortality after stroke [144, 145]. Experimental infection challenge studies provide direct evidence that post-stroke host defense collapses in a quantifiable manner, consistent with a mechanistic pathway linking neurogenic immune reprogramming to lethal pulmonary infection risk [144].

Renal vulnerability reflects combined neurogenic hemodynamic changes and humoral inflammatory injury. Stroke-associated SNS/HPA activation can reduce renal perfusion and alter renal vascular tone, setting the stage for ischemic tubular stress, while systemic cytokines and endothelial dysfunction contribute to longer-term injury and fibrotic remodeling [106]. In the broader context of resistant hypertension and chronic kidney disease, angiotensin II–driven inflammation provides a mechanistic bridge [146]. Zhang, Wang et al. showed that IL-6 is a key mediator linking angiotensin II to hypertension and renal fibrosis, and IL-6 deletion attenuated both hypertensive responses and kidney injury, supporting inflammatory IL-6 signaling as a tractable pathway in cardio–renal–cerebrovascular crosstalk [146].

In sum, post-stroke MODS reflects a structured biological cascade rather than parallel comorbidities. Neurogenic autonomic/HPA outputs rapidly perturb peripheral physiology. Humoral mediators (PAMPs/DAMPs, cytokines, metabolites) then sustain immune and metabolic reprogramming, and immune hubs such as the liver and spleen convert these signals into systemic inflammatory tone and immune-cell deployment that ultimately damages heart, lung, and kidney, feeding back to worsen CNS inflammation and recovery [104–106, 116, 119]. This framework motivates therapeutic strategies that are time-resolved and organ-informed, targeting early neurogenic drivers (e.g., autonomic modulation) and later humoral amplification nodes (e.g., gut permeability, hepatic/splenic immune reprogramming) to break the feed-forward cycle and improve outcomes [59, 130, 147].

Collectively, these organ-axis observations substantiate a unified periphery-to-CNS amplification model. Following the initial neurogenic stress response, the gut, liver, spleen, heart, lung, and kidney do not merely reflect stroke severity, they produce secondary signals capable of reshaping the CNS lesion environment [15, 148]. Leaky gut and dysbiosis provide microbial products and metabolites, while the liver and spleen convert these inputs into cytokine production and immune cell mobilization [149]. Additionally, dysfunction in distal organs contributes to hypoxemic, hemodynamic, endothelial, and metabolic stress. In a state vulnerable to blood-brain barrier (BBB) compromise, these humoral, cellular, barrier-related, and physiological feedback loops converge on the injured brain, thereby amplifying neuroinflammation and hindering recovery. This mechanistic framework supports the perspective of peripheral nodes as therapeutic leverage points rather than mere passive complications of stroke.

Central–peripheral immune responses in hemorrhagic stroke

Although the current body of evidence is still largely derived from ischemic stroke and ischemia–reperfusion models, stroke should not be viewed as a single disease entity. Hemorrhagic stroke differs fundamentally from ischemic stroke and mainly includes subarachnoid hemorrhage (SAH) and intracerebral hemorrhage (ICH). These subtypes are characterized by distinct initiating mechanisms, followed by subtype-specific patterns of central inflammation, peripheral immune remodeling, and central–peripheral immune feedback [15, 149, 150].

The inflammatory trigger of hemorrhagic stroke differs from that of ischemic stroke. Ischemic stroke is primarily initiated by vascular occlusion, hypoperfusion, hypoxia, and ischemia–reperfusion injury. Jiang et al. proposed that, in ischemic stroke, BBB dysfunction involves tight-junction disruption, oxidative stress, inflammatory mediators, and matrix metalloproteinase-related mechanisms, thereby facilitating immune-cell entry into injured brain regions [151]. By contrast, the primary event in hemorrhagic stroke is vascular rupture, followed by the extravasation of blood into the brain parenchyma or subarachnoid space. In ICH and SAH, blood accumulation within the CNS initiates inflammatory and immune responses involving resident CNS cells and infiltrating peripheral immune cells; these responses may contribute to tissue repair but can also aggravate secondary injury [15, 150]. In ICH, thrombin, fibrinogen, and erythrocyte-derived degradation products enter the brain parenchyma and induce robust perihematomal immune-inflammatory responses, contributing to cytotoxic and vasogenic edema [152]. Thus, the key inflammatory stimuli in hemorrhagic stroke are not simply ischemia, but rather the combined effects of blood-component extravasation, hematoma formation, BBB disruption, and perihematomal inflammation.

Inflammation after hemorrhagic stroke is also highly dynamic over time. Wan et al. reported that, in an experimental mouse ICH model, microglia are activated early after hemorrhage and show increased M1-like phenotypic markers [153]. Evidence summarized in hemorrhagic-stroke inflammation reviews further indicates that early post-ICH inflammation is accompanied by increased pro-inflammatory cytokine production, NF-κB pathway activation, and cell-death-associated secondary injury [150]. Jiang et al. reported that, in human perihematomal tissue, infiltration of major immune-cell populations can already be detected on day 1, with a cellular composition partially resembling that of peripheral blood [154]. Using a collagenase-induced ICH model in male mice, Shi et al. found that inflammatory-cell infiltration and cytokine production peak during the early post-ICH phase and then gradually decline, although residual immune cells and chronic inflammatory changes may persist for weeks [155]. These findings indicate that, similar to ischemic stroke, hemorrhagic stroke does not comprise a single acute inflammatory phase, but instead evolves through a continuum from acute injury to subacute immune remodeling and later repair or residual inflammation.

BBB injury also follows a staged trajectory after ICH. Jia et al. showed that, in mouse ICH models, BBB leakage is associated with reduced tight-junction protein expression, increased MMP-9 activity, and ultrastructural tight-junction damage; collagenase-induced ICH produced earlier and more pronounced BBB disruption than the autologous whole-blood model [156]. Yang et al. used multimodal MRI and histological assessment to characterize grey- and white-matter injury as well as BBB disruption after ICH in mice [157]. These observations suggest that BBB injury and repair occur within temporally defined windows after ICH, providing a mechanistic rationale for therapeutic strategies targeting immune modulation and barrier protection.

Hemorrhagic and ischemic stroke also share convergent pathways in central–peripheral immune communication. Duan et al. emphasized that the CNS and peripheral immune system are not isolated compartments. After stroke, the CNS can activate and recruit peripheral immune components through autonomic, neuroendocrine, and lymphatic-related routes. These immune components may then interact with resident CNS cells through interfaces such as the BBB and blood–cerebrospinal fluid barrier, thereby participating in secondary injury or tissue repair [149].

However, the initiating triggers and dominant features of peripheral immune remodeling differ between ischemic and hemorrhagic stroke. In ischemic stroke, peripheral immune alterations are mainly centered on the ischemic penumbra, reperfusion injury, BBB breakdown, immune-cell infiltration, and post-stroke immunosuppression [11]. Experimental ischemic-stroke data indicate that infiltrating immune cells can persist in infarcted and remote brain regions even at two months after stroke [158]. Ischemic stroke can also induce systemic immune alterations and contribute to secondary injury in peripheral organs, including the heart, lung, spleen, kidney, and gastrointestinal tract [159].

By contrast, hemorrhagic stroke is more prominently characterized by blood-component-driven inflammation and perihematomal immune-cell infiltration. In ICH, hemoglobin- and heme-related toxicity can promote ferroptotic neuronal injury, thereby exacerbating hemorrhage-associated secondary brain damage [160]. Bilirubin, hemoglobin, heme, and other blood-derived products can also activate innate immune cells and shape microglia/macrophage responses after ICH [161]. In SAH, delayed inflammatory injury involves microglial activation, chemokine release, BBB disruption, peripheral immune-cell recruitment into the subarachnoid space, oxidative stress, and subsequent brain injury [150, 162].

In hemorrhagic stroke, the BBB represents a critical interface linking central inflammation with peripheral immune feedback. Li and Chen proposed that damage-associated molecular patterns activate microglia, while pro-inflammatory mediators further activate astrocytes and disrupt BBB integrity. Infiltrating peripheral neutrophils and macrophages can release IL-6, TNF-α, reactive oxygen species, and MMP-9, thereby promoting BBB damage and amplifying neuroinflammation [15]. MMP-2 and MMP-9 degrade the basement membrane and tight-junction structures and are implicated in acute ICH-related BBB injury [163]. Clinical biomarker data further suggest that serum MMP-9 is associated with perihematomal edema progression and early clinical outcome in primary ICH patients [164].

In experimental hemorrhagic stroke, astrocyte-specific deletion of CCL5 increases the expression of tight-junction-associated proteins, including ZO-1 and occludin, reduces CD8⁺ T-cell infiltration, decreases Evans blue leakage, and improves neurobehavioral outcomes [165]. These findings indicate that astrocyte-derived CCL5 can promote CD8⁺ T-cell infiltration and BBB disruption after hemorrhagic stroke. Collectively, these data demonstrate that peripheral immunity after hemorrhagic stroke is not merely a secondary consequence of CNS injury, but can actively worsen secondary brain injury through chemokine signaling, T-cell infiltration, MMP activity, and BBB breakdown. Therefore, although ischemic and hemorrhagic stroke differ in their initiating pathological triggers, both may converge on shared secondary injury pathways involving inflammatory-cell infiltration, BBB disruption, and systemic immune imbalance.

Peripheral feedback in hemorrhagic stroke is also reflected in extracerebral immune nodes, particularly the spleen and gut. Chiu et al. showed that acute stroke is associated with splenic volume changes in patients, supporting the relevance of the brain–spleen axis in human stroke [119]. In an experimental autologous blood-induced ICH model, Illanes et al. found that hematoma size modulates systemic immune-cell populations after ICH [166]. Li et al. further showed that spleen-associated immune responses mediate brain–heart interactions after ICH; in mice, splenectomy reduced inflammatory responses and immune-cell infiltration into the brain and heart, while modulating macrophage polarization and improving neurological and cardiac outcomes [167]. Therefore, the spleen should be viewed not simply as a passive peripheral organ, but as an immune node that can shape post-ICH brain and systemic injury.

Similarly, Yu et al. found that ICH induces gut microbiota dysbiosis, characterized by reduced microbial diversity, bacterial overgrowth, impaired gut motility, and increased gut permeability. Recolonization with healthy microbiota ameliorated functional deficits and neuroinflammation after ICH [168]. Additional experimental work suggests that targeting NLRP3 inflammasome signaling may modulate gut microbiota and improve neurobehavioral deficits after ICH [169]. These findings suggest that peripheral organs should not be viewed merely as sources of post-stroke complications, but may act as immune amplifiers of secondary CNS injury in hemorrhagic stroke.

Taken together, ischemic and hemorrhagic stroke have distinct initiating CNS triggers, but they converge mechanistically at the levels of BBB disruption, neuroinflammation, peripheral immune-cell infiltration, remodeling of peripheral immune nodes such as the spleen and gut, and systemic immune feedback. In ischemic stroke, where spleen–liver–gut immune circuits have been more extensively characterized, the peripheral biodistribution of extracellular vesicles may represent a potential therapeutic advantage. However, this hypothesis remains to be independently validated in hemorrhagic stroke models. Although a periphery-first strategy is mechanistically plausible for hemorrhagic stroke, given the presence of peripheral immune amplification and BBB injury, hemorrhagic stroke also involves hematoma expansion, coagulation status, vascular stability, and bleeding-related safety concerns. Therefore, the therapeutic safety, optimal timing, peripheral-node targeting efficacy, and potential effects of a periphery-first extracellular vesicle strategy on coagulation and hematoma evolution require dedicated validation in hemorrhagic stroke-specific models.

Extracellular vesicles as systemic modulators in stroke

Definition, biogenesis and main features of extracellular vesicles

EVs are lipid-bilayer nanoparticles released by most cell types and present in virtually all biofluids. They mediate intercellular communication by transferring proteins, lipids, and nucleic acids. Consistent with MISEV2023 guidance, EVs are often described using biogenesis-based terms—exosomes (∼30–150 nm; endosomal origin), microvesicles/ectosomes (∼100–1000 nm; plasma membrane budding), and apoptotic bodies (> 1000 nm; apoptosis-associated)—but best practice is to report EV preparations using operational descriptors (e.g., size range, density, marker profile, and isolation method) rather than relying only on subtype labels (e.g., size, density, markers, and isolation method) rather than relying solely on subtype labels [170, 171]. Recent stroke-EV studies increasingly follow this operational framework, emphasizing EV source, isolation method, particle size, marker profile, cargo composition, and biological potency rather than assuming functional equivalence across “exosome” preparations. This distinction is particularly important in stroke, where EV preparations from different stem or precursor cell sources may differ in immunomodulatory, angiogenic, neurotrophic, and barrier-stabilizing activities [170, 172, 173].

Exosome-enriched SC-EV fractions are produced through the endosomal pathway: early endosomes mature into multivesicular bodies (MVBs) containing intraluminal vesicles, which are either degraded via lysosomes or released when MVBs fuse with the plasma membrane, a process regulated by Rab GTPases and membrane-fusion machinery [174–176]. EV yield and cargo are not fixed; they shift with microenvironmental cues such as hypoxia, which can alter both vesicle abundance and molecular composition—features that may matter in post-stroke repair settings [177].

For characterization, SC-EV preparations typically show enrichment of membrane tetraspanins (e.g., CD9, CD63, CD81), and exosome-associated cytosolic proteins such as TSG101 and ALIX are commonly detected in exosome-enriched fractions [178–180]. Functionally, SC-EVs deliver selective “signal cargo,” including regulatory RNAs (miRNAs, lncRNAs, circRNAs) and proteins that can reprogram recipient cells post-transcriptionally and engage multiple pathways in parallel—an advantage in ischemic stroke, where injury evolves through coupled processes (cell death, inflammation, vascular dysfunction, and remodeling) rather than a single pathway [181].

SC-EVs are also heterogeneous across source cell types, with reproducible functional biases described in the literature. iPSC-derived EVs are often positioned as customizable platforms; mechanistic reports link specific miRNA cargos (e.g., miR-100-5p) to cytoprotective programs such as Ca²⁺ homeostasis in ischemia/reperfusion contexts [182]. Bone marrow MSC-EVs are frequently highlighted for immunomodulation, including reports of miR-27b–linked suppression of NF-κB signaling [183]. Adipose-derived SC-EVs are repeatedly associated with pro-angiogenic effects involving PI3K/AKT signaling [184], while neural stem cell EVs are often discussed for neurotropic features relevant to neural repair [185]. After uptake via receptor–ligand interactions, endocytosis, or membrane fusion, EV cargo can influence neuronal survival, mitochondrial integrity, glial activation states, and vascular remodeling, aligning with the multi-factorial biology of stroke [186, 187].

MSC-derived EVs should not be interpreted as interchangeable therapeutic products. Comparative analyses of BM-MSC- and AD-MSC-derived EVs indicate that MSC tissue source and in vitro expansion history shape EV protein/miRNA cargo, angiogenic activity, and immunomodulatory capacity [172]. BM-MSC-EVs have been associated with stronger acute immunomodulatory profiles in some comparative systems, including suppression of inflammatory signaling and regulation of macrophage/T-cell responses, whereas AD-MSC-EVs often show more prominent pro-angiogenic activity in endothelial assays [173, 188]. UC/WJ-MSC-EVs may provide higher yield and broad regenerative or immunomodulatory cargo profiles, making them attractive for scalable production, although their source-specific potency still requires indication-matched validation [189, 190]. These differences are particularly relevant in stroke because therapeutic priorities shift over time: early reperfusion injury may require vascular–immune stabilization and BBB protection, whereas subacute recovery may depend more on angiogenesis, synaptic remodeling, and immune resolution [191]. Thus, MSC-EV efficacy should be discussed in relation to the intended mechanism and disease phase rather than treated as a uniform class effect. This source-dependent view also helps explain why recent EV studies increasingly align the EV product with the disease phase: vascular–immune stabilization and blood–brain barrier protection may be prioritized during acute reperfusion injury, whereas angiogenesis, synaptic remodeling, glial reprogramming, and immune resolution may be more relevant during the subacute and recovery phases [181, 191].

Compared with cell transplantation, SC-EVs reduce concerns such as microvascular obstruction and tumorigenicity and may be more compatible with standardized production, storage, and formulation [192]. BBB penetration is often cited as an advantage, but in practice, systemic EV delivery usually results in limited brain exposure and strong context dependence [177]. This motivates a more pragmatic framing: in stroke—where peripheral immune organs contribute materially to outcome—dominant peripheral biodistribution may be therapeutically useful rather than purely “off-target” [193].

Endogenous EV responses after stroke: circulating profiles, cellular origin, and temporal dynamics

Before considering therapeutic EV administration, it is important to recognize that stroke itself reshapes the endogenous EV landscape. Clinical profiling studies show that circulating EV signatures change after acute ischemic stroke, although the direction and magnitude of these changes depend on sampling time, comparator population, and analytical platform. Buntsma et al. analyzed plasma EVs by flow cytometry in an emergency cohort of 155 patients with suspected stroke and found that activated platelet-derived and leukocyte-derived EVs were lower in ischemic stroke patients than in patients with non-ischemic diagnoses; in their multivariable model, leukocyte-derived EVs combined with diastolic blood pressure independently indicated ischemic stroke [194]. By contrast, Eyileten et al. compared 28 acute ischemic stroke patients with coronary heart disease controls and measured platelet-derived EVs, leukocyte-derived EVs, endothelial EVs, and plasma miRNAs at 24 h and 7 days; they reported increased platelet- and leukocyte-derived EVs in the acute phase, with diagnostic performance improved when EV markers were combined with miR-19a-3p and let-7f [195]. These divergent findings highlight that endogenous EVs should be interpreted as context-dependent vascular and immune readouts rather than uniformly increased disease markers. Accordingly, circulating EVs are better framed as phase- and context-sensitive indicators of thromboinflammation, endothelial stress, platelet activation, leukocyte activity, and systemic immune remodeling after stroke, rather than as a single-direction biomarker class [194–196].

Cell-of-origin assignment provides a second layer of biological interpretation, although marker specificity remains an important methodological limitation. Platelet-derived EVs, often identified by markers such as CD41 or CD62, have been linked to platelet activation, thromboinflammation, and high platelet reactivity [197, 198], whereas endothelial marker-positive EVs, including CD31-, CD105-, or CD144-positive populations, have been associated with endothelial activation, vascular injury, BBB dysfunction, NIHSS score, infarct volume, and long-term cardiovascular risk [197]. Maciejewska-Renkowska et al. prospectively quantified platelet-derived EVs in ischemic stroke patients and reported clinically relevant associations between platelet EV phenotypes and stroke-related outcomes [199]. Ollen-Bittle et al. summarized evidence that endothelial-, platelet-, erythrocyte-, leukocyte-, and monocyte-derived EVs are altered in ischemic stroke, supporting the concept that EV subtypes map onto distinct thrombotic, vascular, inflammatory, and oxidative/coagulant axes [196]. Machado et al. further used transcriptomic analysis of EVs from plasma and thrombus samples and showed that EV-associated molecular profiles may reflect stroke etiology and thromboinflammatory mechanisms [200]. Depending on cellular origin and disease phase, endogenous EVs may therefore carry pathogenic cargo, such as procoagulant or inflammatory signals, or reparative signals related to immune resolution and tissue remodeling.

Brain- and glia-associated EVs may extend this concept into subacute injury monitoring and recovery prediction, although marker specificity remains a methodological challenge. Manwani et al. performed small RNA sequencing of L1CAM-positive, putatively neuron-enriched EVs and developed brain-enriched miRNA signatures that distinguished acute ischemic stroke with AUC values of 0.83–0.93 [201]. Edwardson et al. measured CD9⁺EAAT1⁺ putative astrocyte-derived EVs over the first month after stroke and found persistent elevation at 5, 15, and 30 days compared with controls; higher levels at day 15 were observed in patients who developed hemorrhagic transformation, suggesting a potential relationship with BBB injury and vascular instability [202]. Jödicke et al., in the BAPTISe study of 110 patients with subacute ischemic stroke randomized to aerobic exercise or relaxation, found that higher baseline platelet-derived and neuron-derived EV levels were associated with better Barthel Index outcomes over 6 months, whereas smaller increases in neuron- and leukocyte-derived EVs were associated with poorer functional recovery [197]. Picciolini, Mangolini et al. demonstrated that exogenous SC-EVs are introduced into a pre-existing, phase-dependent EV signaling environment shaped by thrombosis, endothelial dysfunction, systemic inflammation, BBB injury, and CNS remodeling [203]. Recent human studies therefore support a dual role for EVs in stroke: endogenous EV profiles may help define disease phase, vascular instability, and recovery trajectories, while exogenous EV products may need to be timed and selected according to this evolving EV background rather than administered as phase-independent biologics [197, 201, 202].

Pharmacokinetics and biodistribution: limitations for CNS targeting, leverage for periphery-first therapy

Across in vivo tracking studies, systemically administered EVs exhibit rapid blood clearance and preferential uptake by the mononuclear phagocyte system, particularly the liver and spleen. Quantitative whole-body imaging using luciferase-labeled engineered EVs showed short circulating half-lives with rapid accumulation in liver and spleen, while brain signal remained low even when alternative routes were examined [204]. Follow-up biodistribution work similarly emphasizes dominant reticuloendothelial capture and limited CNS exposure after systemic dosing [205]. This biodistribution pattern does not necessarily invalidate systemic EV therapy in stroke. Instead, it supports a more realistic therapeutic model in which EVs modulate liver-, spleen-, gut-, and barrier-associated immune circuits that feed back onto neurovascular inflammation and recovery [193, 204, 205]. While this profile complicates a purely CNS-delivery paradigm, it aligns naturally with a “periphery-first” model in which EVs are used to modulate immune–metabolic hubs and barrier systems that feed into neuroinflammatory trajectories after stroke [193, 206]. We summarize experimental models and EV interventions across biodistribution, neuroinflammation, and stroke-relevant organ-axis studies in Tables 1, 2.

Table 1.

Stroke and stroke-comorbidity models of EV-related peripheral mechanisms

Model	Interventions/Source of EVs	Methods	Key Results	References
Mice (Intracerebral Hemorrhage, ICH)	MSC-EVs	FMT/16S rRNA	Gut microbiota remodeling is crucial for neuroprotection; microbial community alone after transplant treatment can replicate therapeutic effects.	[207]
Mice (MCAO)	iPSC-EVs (+Electroacupuncture)	16 S rRNA/ELISA	Improved stroke outcomes by modulating the brain-gut microbiota and immune response (reducing IL-17/increasing IL-10).	[206]
Stroke-Associated Pneumonia Mice (MCAO + Pneumonia)	BM-MSC-derived Migrasomes	Pulmonary macrophage targeting / Bacterial clearance assay	Enhanced pulmonary macrophage phagocytosis to resolve pneumonia and interrupt systemic inflammatory cascades.	[208]
Type 2 Diabetic Stroke Mice (T2DM-MCAO)	CD133 + Hematopoietic Stem Cell Exosomes	Liver pathology / Metabolic function assessment	Ameliorated liver inflammation and fibrosis, restoring systemic metabolic homeostasis and supporting neurological recovery.	[209]
Aged Mice (MCAO)	iPSC-sEVs	Proteomics/Immunohistochemistry	Targeted repair of peripheral interface (BBB), blocking peripheral leukocyte infiltration into the brain.	[210]
Mice (MCAO)	Antibiotics	Cell tracing	Blocked migration of intestinal immune cells (e.g., γδ T cells) to the meninges, alleviating brain damage.	[70]

Table 2.

Non-stroke organ-axis models informing the periphery-first framework

Model	Interventions/Source of EVs	Methods	Key Results	References
Rats (Mild Liver Injury, MLD)	MSC-EVs	16 S rRNA/Behavioral tests	Restored peripheral homeostasis via regulation of gut microbiota and metabolites, indirectly improving central nervous system function.	[211]
Mice (Septic Liver Injury)	BMSC-EVs	Immunofluorescence/qPCR	Inhibited liver macrophage STING pathway, reducing liver injury and systemic inflammation levels.	[212, 213]
Mice (DSS Colitis)	MSC-EVs	Immunofluorescence/qPCR	Upregulated tight junction proteins (Claudin-1/ZO-1), repairing damaged intestinal physical barrier.	[214]
NMRI Mice	Engineered EVs (Luciferase-labeled)	In vivo bioluminescence imaging (BLI)	EVs are primarily enriched in the liver and spleen (90%+), not the brain, confirming their natural peripheral targeting characteristics	[204]

Immunomodulation and repair

Direct, stroke-specific proof that SC-EVs independently reprogram the spleen–liver–gut axis to disrupt systemic feed-forward loops remains limited. Nevertheless, mechanistically adjacent inflammatory models provide consistent evidence that SC-EVs can shift peripheral immune setpoints and barrier function through multi-cargo regulation [171, 215–218].

Within stroke-relevant preclinical literature, systemic SC-EV administration has been associated with improved functional recovery and reduced infarct pathology, alongside decreased apoptosis and lower systemic inflammatory cytokines [219]. Mechanistically, these benefits have been associated with suppression of canonical inflammatory signaling (e.g., NF-κB, JAK2/STAT3, and AMPK-related inflammatory programs) and post-transcriptional regulation via EV-delivered miRNAs [220–222]. Beyond systemic cytokine reduction, SC-EVs also influence CNS innate immune phenotypes: multiple studies report inhibition of M1-like microglial programs and promotion of M2-like, repair-associated polarization with decreased TNF-α/IL-1β/IL-6 and increased IL-10/TGF-β, consistent with coordinated peripheral and central immune rebalancing [223–225]. Importantly, several studies provide more granular immune readouts supporting immune “trajectory reprogramming” rather than non-specific suppression. MSC-EVs were reported to reverse post-stroke lymphopenia across B cells, NK cells, and T cells, mitigating stroke-induced peripheral immunosuppression and potentially restoring a permissive environment for repair [226]. MSC-derived sEVs reduced leukocyte (especially neutrophil) infiltration, decreased microvascular ICAM-1, and stabilized BBB function in acute reperfusion injury, indicating an early vascular–immune protective window [227]. EV-driven expansion of Tregs has also been described: MSC-EVs promoted CD4⁺CD25⁺Foxp3⁺ Treg differentiation via APC-dependent mechanisms [228], and ESC-sEVs reduced peripheral leukocyte infiltration, lowered TNF-α/IL-1β/IL-6, and reduced infarct volume in MCAO rats—effects abolished by Treg depletion, implying that Treg expansion can be causal for neuroprotection in at least some EV platforms [229]. EV studies implicate pathways including TLR4/NF-κB, mTOR, and AMPK–Nrf2 in microglial state control, reinforcing the concept that EVs can steer neuroinflammation toward a repair phenotype rather than simply “turning it off” [224, 230]. In ischemic brain injury models, EVs from pluripotent sources have been reported to expand regulatory immune programs (e.g., Treg-associated responses) and modulate inflammatory mediators after systemic delivery, supporting the plausibility that peripheral immunoregulation contributes to observed neuroprotection [229, 231]. Likewise, MSC-EVs have been linked to shifts toward reparative phenotypes in innate immune cells (e.g., microglia M2-like polarization), consistent with broad anti-inflammatory and pro-repair capacity [232].

Central-direct delivery versus peripheral-first reprogramming

Despite robust preclinical efficacy, the dominant “route-to-efficacy” remains debated because systemic delivery yields low parenchymal exposure [193, 204]. Two non-mutually exclusive models are commonly considered. The CNS-direct hypothesis posits that a biologically meaningful fraction crosses the BBB and directly engages neurons, glia, and endothelium, motivating efforts to enhance BBB penetration and intraparenchymal delivery [233, 234]. In practice, limited brain accumulation, narrow therapeutic windows, and delivery inefficiency constrain this strategy [204, 235].

The peripheral-first hypothesis proposes that SC-EVs act primarily in peripheral tissues—restoring immune and barrier homeostasis in organs that drive secondary inflammatory feedback—thereby reducing inflammatory invasion and stabilizing BBB integrity (Fig. 2). Axis-level studies in related models support the plausibility of indirect CNS benefit: MSC-EVs improved microbiota and metabolic abnormalities in liver injury models and were accompanied by reduced microglial activation and behavioral improvement [211], while in intracranial hemorrhage, FMT from EV-treated animals reproduced much of the observed neuroprotection, implicating microbiota remodeling as a key driver [207]. A “BBB-interface” mechanism may bridge these views: iPSC-sEVs were reported to target brain endothelial cells and restore BBB function by delivering AKT1 and calmodulin proteins to activate the eNOS–Sirt1 axis, reducing leukocyte infiltration and inflammatory cytokines in an aging stroke model [210]. Collectively, available data support a mixed mechanism in which extensive parenchymal penetration may not be required if EVs can modulate barrier systems and peripheral immune drivers that amplify injury.

Fig. 2.

Extracellular vesicles as systemic modulators: a “periphery-first” strategy. This diagram illustrates the mechanistic approach of the “periphery-first” strategy using SC-EVs in stroke treatment. Following systemic administration, SC-EVs exhibit limited brain penetration but are predominantly taken up by peripheral organs like the liver and spleen. These organs act as key points for immune and metabolic modulation. Spleen: SC-EVs promote immune regulation by enhancing Tregs and anti-inflammatory cytokines (IL-10, TGF-β), while decreasing pro-inflammatory Th1/Th17 cells, possibly through the cholinergic anti-inflammatory pathway (ACh–α7nAChR). Liver: EV cargo, such as miRNAs, reduces inflammatory pathways, curbs acute-phase proteins like HMGB1, and helps balance systemic cytokines. Gut–Immune Interface: EVs improve gut barrier function by upregulating tight junction proteins (ZO-1, claudin-1), support the production of beneficial metabolites (e.g., SCFAs), and expand gut-derived Tregs, connecting to the central autonomic nervous system via the vagus nerve/NTS. Indirect CNS Effects: These peripheral actions contribute to improved brain outcomes by reducing immune cell infiltration, decreasing neuroinflammation, stabilizing the BBB, shifting microglial activation from a pro-inflammatory M1 state to a reparative M2 state, and reducing infarct volume. ACh, acetylcholine; α7nAChR, alpha7 nicotinic acetylcholine receptor; BBB, blood–brain barrier; CAP, cholinergic anti-inflammatory pathway; EVs, extracellular vesicles; HMGB1, high mobility group box 1; miRNAs, microRNAs; NF-κB, nuclear factor kappa B; NTS, nucleus tractus solitarius; PK, pharmacokinetics; SC-EVs, stem cell–derived extracellular vesicles; SCFAs, short-chain fatty acids; STING, stimulator of interferon genes; Tregs, regulatory T cells; Th, T helper; ↑, increased; ↓, decreased

Organ-axis modulation beyond stroke models

Mechanistic support for periphery-first development also comes from systemic inflammatory models. In LPS-driven septic liver injury, BMSC-EVs were reported to suppress liver macrophage activation through inhibition of the cGAS–STING pathway, reducing liver injury and systemic inflammation [212, 213]. In a rat experimental autoimmune prostatitis model, Peng, Guo et al. reported that IV injection of EVs derived from human iPSC-MSCs reduced splenic Th1/Th17 proportions while increasing Treg representation, indicating a systemic tilt from inflammation toward tolerance measurable within the spleen [236]. In experimental autoimmune myocarditis, EVs from cardiac stromal/progenitor cells increased IL-10⁺ Treg cells in the spleen, again supporting spleen-level immune reprogramming as a mode of action [237]. At the gut barrier, IV administration of MSC-EVs improved epithelial tight junction protein expression (e.g., Claudin-1, ZO-1) in DSS colitis models, supporting a mechanism by which EVs can restore barrier integrity and reduce inflammatory translocation—processes directly relevant to post-stroke “leaky gut” biology [214]. Across inflammatory paradigms, SC-EVs have been associated with reductions in circulating pro-inflammatory cytokines. For example, MSC-EVs reduced circulating pro-inflammatory cytokines in LPS-driven systemic inflammation models, and placental MSC-EVs reduced IL-1β, IL-6, and TNF-α in myocardial infarction models [238, 239]. A combined electroacupuncture with iPSC-EV approach remodeled gut microbiota, reduced colonic inflammation (decreased IL-17, increased IL-10), and coincided with reduced central neuroinflammation and improved recovery; while causality for EVs alone is limited by the combination design, the work provides proof-of-concept that EV-containing regimens can jointly improve gut immunity and brain inflammation [206].

Engineering strategies and the cost of complexity

Efforts to engineer EVs for improved brain delivery and amplified on-target potency fall into two categories [204, 240]. First, enhanced therapeutic loading aims to concentrate bioactive cargos so that even the small fraction of EVs reaching CNS interfaces delivers a pharmacologically meaningful signal. Representative examples include BDNF-enriched MSC-EVs that engage the BDNF/TrkB axis to support neurogenesis, angiogenesis, and synaptic plasticity [241]; BMSC-EVs loaded with lncRNA KLF3-AS1 that modulate a miRNA/USP22/Sirt1 pathway to reduce ischemic injury [242]; and EVs enriched with the miR-17–92 cluster that promote axonal extension and myelination by suppressing PTEN and activating PI3K/Akt/mTOR signaling [243, 244]. Second, enhanced brain-targeted delivery seeks to increase lesion-site enrichment by surface functionalization or physical guidance [14], including arginine-glycine-aspartic acid (RGD)-based targeting with downstream attenuation of MAPK-associated inflammatory signaling [245], hydrogel-based encapsulation enabling sustained local release and vascular regeneration [246], magnetic guidance of SPION-loaded vesicles to improve regional accumulation [247], and intranasal delivery paradigms that reduce pro-inflammatory cytokines while activating neurotrophic programs [248]. While these approaches demonstrate strong technical feasibility, they also raise translational hurdles—added manufacturing complexity, higher cost of goods, batch-to-batch variability, and potential immunogenicity or off-target interactions—factors that may constrain scalability and regulatory tractability [249].

Key translational priorities: causality, dose-exposure, and mechanism mapping

A central translational task is to establish causal, quantitative evidence for where and how EVs work in stroke. This includes identifying which peripheral nodes (gut, liver, spleen, lung immune compartments) are necessary and/or sufficient for benefit, rather than relying on correlative peripheral readouts [193, 235]. It also requires quantifying the relative contribution of peripheral immunometabolic reprogramming, BBB-interface effects, and any direct parenchymal actions to functional recovery, and defining how timing, dose–exposure relationships, and EV source/cargo composition shift the dominant route-to-efficacy across disease phases [193, 235].

Preclinical studies support the general principle that interrupting peripheral pathological nodes can yield meaningful neuroprotection: microbiota-directed approaches (FMT, probiotics) improve experimental outcomes, and interventions targeting liver and spleen can mitigate injury severity [59, 130, 147]. Selected probiotic regimens can also reshape splenic immune tone, underscoring the mechanistic link between peripheral immune programming and central inflammation [122].

SC-EVs offer a credible multi-pathway therapeutic logic for stroke by combining immunoregulatory and barrier-repair capacities with a biodistribution profile that naturally targets peripheral immune–metabolic hubs. The field faces a strategic choice: continue prioritizing engineered CNS delivery to maximize direct parenchymal action, or develop a periphery-first approach that leverages EV tropism to disrupt systemic inflammatory amplification and stabilize barrier systems—potentially a simpler route to scalable translation if supported by causal mechanism mapping and clinically grounded PK/PD relationships [193, 204, 205]. Recent stroke-EV literature therefore supports a shift from proof-of-concept efficacy toward mechanism-resolved translation. The next step is not simply to show that EVs reduce infarct volume or improve behavioral scores, but to define which EV source, cargo profile, dose, route, timing, and target compartment are responsible for benefit in a given stroke phase [181, 193, 235].

Clinical trials of extracellular vesicles and other strategies for stroke therapy

Extracellular vesicles in clinical trials

Registered interventional EV studies in stroke remain largely early-phase programs focused on safety/tolerability, feasibility, and dose escalation, predominantly using stem/precursor cell–derived EV products (shown in Table 3, ClinicalTrials.gov). Current examples include allogeneic Wharton’s jelly MSC-EVs (SNE-101; NCT06995625, not yet recruiting), human induced neural stem cell (hiNSC)-EVs (NouvSoma001; NCT06612710, recruiting), intranasal hUC-MSC small EVs (hUC-MSC-sEV-001; NCT07232563, not yet recruiting), human induced NSC-derived exosomes (NCT07143786, not yet recruiting), hiPSC-derived exosomes (GD-iExo-003; NCT06138210, recruiting), and intranasal umbilical-cord MSC exosomes (NCT05158101, recruiting). Across these records, endpoints emphasize safety and preliminary efficacy signals; however, definitive peer-reviewed clinical outcome reports directly attributable to these specific interventional registrations are not yet clearly available in the sources identified.

Table 3.

Ongoing and registered clinical studies evaluating extracellular vesicles (EVs) as therapeutics or biomarkers in stroke (ClinicalTrials.gov)

NCT ID	Study type	EV source/products	Country	Conditions	Study status
NCT06995625	Therapeutic (interventional)	Allogeneic Wharton’s jelly MSC-EVs(SNE-101)	South Korea	Ischemic Stroke	Not yet recruiting
NCT06612710	Therapeutic (interventional)	Human-induced NSC-EV(NouvSoma001)	China	Ischemic Stroke	Recruiting
NCT07232563	Therapeutic (interventional)	Human UC-MSC small EVs(hUC-MSC-sEV-001)	China	Ischemic Stroke	Not yet recruiting
NCT07143786	Therapeutic (interventional)	Human induced NSC-Exosomes	China	Ischemic Stroke	Not yet recruiting
NCT06138210	Therapeutic (interventional)	Human iPSC-exosomes(GD-iExo-003)	China	Ischemic Stroke	Recruiting
NCT05158101	Therapeutic (interventional)	Allogeneic adult umbilical cord- MSC-exosomes	Antigua and Barbuda, Argentina	Stroke	Recruiting
NCT05370105	Biomarker/ observational	Circulating EVs profiling (EXO4STROKE)	Italy	Stroke	Completed
NCT06319742	Biomarker/ observational	EV surface-marker signatures (ElViS-ACS)	Switzerland	Acute Cerebrovascular Syndromes	Recruiting
NCT06257823	Biomarker/ observational	Vascular phenotyping + plasma EV profiling (ENIGMA)	Denmark	Ischemic Stroke	Completed

A parallel and more mature clinical direction treats EVs as circulating biomarkers rather than therapeutics. EXO4STROKE (NCT05370105, completed) evaluates blood EV detection and characterization—potentially including subpopulation profiling—in relation to stroke and rehabilitation trajectories; a recent preprint further suggests that specific circulating microglia-derived EV measures may associate with, and help predict, functional recovery following rehabilitation [250]. ElViS-ACS (NCT06319742, recruiting) assesses EV surface-marker signatures for diagnosis/prognosis in acute cerebrovascular syndromes, conceptually aligned with prior evidence that EV-surface antigen profiling can support TIA diagnostic modeling [251]. ENIGMA (NCT06257823) is structured as an observational framework integrating vascular dysfunction phenotyping with plasma EV profiles to predict post-stroke cognitive outcomes; to date, it appears primarily represented in the literature as a protocol/design report rather than final outcome data [252]. At present, clinical EV work in stroke is bifurcating into early-stage interventional development and biomarker programs aiming to improve risk modeling, prognostication, and ultimately trial stratification. These biomarker-oriented studies complement therapeutic EV trials by leveraging the endogenous EV response as a readout of thromboinflammation, endothelial dysfunction, BBB injury, CNS damage, and recovery trajectory. In this sense, endogenous EV profiling may help define disease phase, patient heterogeneity, and treatment windows for future SC-EV interventions.

Pharmacotherapy targeting the central nervous system

Modern CNS-directed stroke pharmacotherapy includes approved/regionally available agents, thrombectomy/thrombolysis adjuncts, and a growing pipeline of mechanism-targeted or pleiotropic candidates (Table 4). Many agents face predictable translational barriers—narrow therapeutic windows, heterogeneous physiology, and the dominant effect of reperfusion timing on outcomes—driving a shift toward “reperfusion-plus” paradigms designed to mitigate excitotoxicity, oxidative stress, BBB disruption, edema, and maladaptive inflammation while also supporting subacute repair [253, 254].

Table 4.

Representative clinical trials of CNS-/neurovascular unit–targeting pharmacotherapies for ischemic stroke

Drug	Status/Trial	Key mechanism	Limitation	References
Anti-excitotoxicity agents
Nimodipine	Phase IV/ NCT01220622	L-type Ca2 + channel blocker; vasodilation and CBF support; limits Ca2 + overload.	Evidence strongest in aneurysmal subarachnoid hemorrhage; ischemic stroke benefit uncertain.	[253, 255]
Verapamil	Phase I-II/ NCT03347786	L-type Ca2 + channel blocker; reduces Ca2+-mediated excitotoxic injury.	Early-phase data; efficacy not established.	ClinicalTrials.gov
Memantine	Phase III /NCT02535611	NMDA receptor antagonist; limits glutamate excitotoxicity.	Clinical benefit unconfirmed; Phase III results pending.	[256]
ZL006-05	Phase II/ CTR20231033	Disrupts NMDAR-PSD95-nNOS coupling; reduces NO-mediated excitotoxicity.	Phase II ongoing; efficacy and safety not established.	[257]
Salfaprodil	Phase III/ NCT05041010	NR2B-selective NMDA receptor antagonist; anti-excitotoxicity.	Phase III results pending; definitive benefit not confirmed.	[258]
Nerinetide(NA-1)	Phase III/ NCT02930018	PSD-95 inhibitor; blocks NMDAR downstream neurotoxic signaling.	Overall benefit not demonstrated; efficacy attenuated with alteplase; context-dependent effect.	[259, 260]
SY-007	Phase I/ NCT04891913	PTEN pathway modulator; blocks PTEN nuclear translocation downstream of GluN2B signaling.	Phase I only; efficacy not established.	[261]
Oxidative stress
Edaravone	Launched	Free-radical scavenger; antioxidant neuroprotection.	Efficacy heterogeneous across studies and care settings.	[262, 263]
Edaravone Dexborneol	Launched	Antioxidant and anti-inflammatory combination; supports BBB protection.	Evidence mainly regional; limited placebo-controlled data; generalizability uncertain.	[264, 265]
LT3001	Phase II/ NCT05686642;NCT05403866	Antioxidant strategy; free-radical scavenging.	Phase II ongoing; efficacy and safety not established.	ClinicalTrials.gov
Uric Acid	Phase III/ NCT00860366	Peroxynitrite scavenger; antioxidant adjunct therapy.	Benefit may depend on reperfusion context and timing; patient selection required.	[266, 267]
NXY-059	Phase III/ NCT00061022; NCT00119626	Free-radical scavenger; antioxidant strategy.	Did not demonstrate clinical benefit in Phase III trials.	[268]
Neuroinflammation & BBB/NVU protection
Fingolimod	Phase II/ NCT02002390; NCT04675762	S1P receptor modulator; limits lymphocyte egress and neuroinflammation.	Limited clinical evidence; potential immunosuppression; larger trials needed.	[269], ClinicalTrials.gov
Otaplimastat (SP-8203)	Phase III/ NCT06660719	MMP pathway modulation; BBB protection; reduces oxidative injury.	Phase II studies underpowered; Phase III results pending.	[270], ClinicalTrials.gov
ApTOLL	Phase II/ NCT04734548	TLR4 antagonist aptamer; innate immune modulation.	Underpowered for efficacy; mortality signal requires confirmation in larger trials.	[271]
SA4503	Phase II/ NCT00639249	Sigma-1 receptor agonist; supports recovery and neuroplasticity.	Phase II; clinical benefit not established.	[272]
Glyburide (IV)	Phase III/ NCT02864953	SUR1-TRPM4 channel inhibitor; reduces cerebral edema and midline shift.	Functional outcome benefit not definitive; confirmatory evidence needed.	[273]
3K3A-APC	Phase III/ NCT05484154	Activated protein C variant; PAR1 biased signaling supports cytoprotection and BBB stabilization.	Efficacy unproven; Phase III results pending.	[274], ClinicalTrials.gov
Imatinib	Phase III/ NCT03639922	PDGFR alpha inhibition on perivascular astrocytes; stabilizes BBB and reduces edema and hemorrhagic transformation.	Phase II signal; Phase III confirmation needed.	[240, 275]
Multi-target neuroprotection
NBP	Launched	Multi-target neurovascular protection; improves microcirculation and mitochondrial function; supports BBB integrity.	Regional approval; limited external validation and standardized evidence across populations.	[276, 277]
Glyceryl trinitrate	Phase III/ ISRCTN26986053	Nitric oxide donor; modulates neurovascular tone via NO-cGMP-PKG signaling.	Time-window sensitive; efficacy inconsistent across trials.	[278]
Ginkgolides	Launched	Platelet activating factor antagonist; supports BBB protection and anti-inflammatory effects.	Evidence mainly regional and adjunctive; limited high-quality multicenter trials.	[279, 280]
Vinpocetine	Phase II-III/ NCT01400035	PDE inhibition and cGMP modulation; improves cerebral blood flow; antioxidant support.	Efficacy not established; confirmatory large trials needed.	ClinicalTrials.gov
GM6 (MNTF)	Phase II/ NCT01221246	Motoneuronotrophic factor peptide; supports neuronal survival and repair.	Early-phase evidence; efficacy and dosing require validation.	ClinicalTrials.gov
Neurorepair / membrane support
Cerebrolysin	Launched	Neurotrophic peptide mixture; supports neuroplasticity and recovery.	Heterogeneous studies and populations; high-quality randomized trials needed.	[281, 282]
Citicoline	Phase III/ NCT00331890	CDP-choline; supports phospholipid synthesis and membrane repair.	Large trial showed no overall benefit; subgroup signals inconsistent.	[283]

Nimodipine illustrates a potential niche beyond its clear role in aneurysmal subarachnoid hemorrhage. In acute ischemic stroke patients with mild cognitive impairment, a double-blind randomized trial reported higher rates of cognitive improvement in the nimodipine arm versus placebo (e.g., ΔADAS-cog ≤ − 2), suggesting a possible benefit signal for selected post-stroke cognitive phenotypes rather than broad acute neuroprotection [255]. In contrast, highly targeted anti-excitotoxic strategies have shown context-dependent limitations: NA-1/nerinetide was developed to uncouple NMDAR–PSD95–nNOS signaling to attenuate downstream NO/ONOO⁻-linked toxicity [259], yet the phase 3 ESCAPE-NA1 trial found no overall improvement in 90-day functional outcome when nerinetide was added to endovascular thrombectomy-era care, with evidence consistent with treatment-background interactions [260]. Together, these data exemplify a key theme: efficacy signals may be subgroup- and context-specific, particularly in the setting of thrombolysis/thrombectomy workflows [254, 260].

A second major axis targets oxidative stress, inflammation, and BBB integrity. Edaravone’s proposed mechanisms include free-radical scavenging and suppression of inflammatory signaling relevant to both acute ischemia and barrier injury [262]. In clinical practice–adjacent evidence, early edaravone use in patients undergoing endovascular reperfusion has been associated with improved in-hospital outcomes in observational analysis, supporting its positioning as a pragmatic adjunct in some systems [263]. Newer formulations and combinations aim to improve delivery and pleiotropic impact: phase 3 evidence supports benefit of edaravone dexborneol versus edaravone alone, and subsequent randomized work evaluating alternative delivery routes continues to indicate clinical feasibility and efficacy signals in appropriately defined windows [264, 265]. Parallel antioxidant approaches include uric acid, supported by preclinical infarct reduction and neurological improvement in transient ischemia models [266], and by clinical analyses highlighting that the success of neuroprotective adjuncts may depend on physiological context such as collateral status—reinforcing the principle that “one-size-fits-all” neuroprotection is unlikely [267].

Multi-target small molecules with neurovascular actions represent another pragmatic class. dl-3-n-butylphthalide (NBP) has a long development trajectory and is positioned as a pleiotropic neurovascular protective agent [276]. A randomized clinical trial reported improved functional outcomes and acceptable safety with NBP in acute ischemic stroke, supporting its continued use and further evaluation, while also underscoring the importance of external validation across healthcare systems and reperfusion pathways [254, 277]. Natural-product–derived strategies such as ginkgolide have mechanistic support from preconditioning and oxidative stress pathway studies (e.g., HO1/CRMP2-related signatures) [279] and have shown adjunctive benefit signals when combined with intravenous alteplase in a multicenter cluster-randomized design, although generalizability beyond specific regions and combination contexts remains a key question [280]. Finally, neurotrophic and repair-oriented formulations such as cerebrolysin are conceptually aligned with subacute recovery biology; evidence syntheses support potential benefit across neurological disorders [281] and prospective single-center data suggest that cerebrolysin add-on therapy after mechanical thrombectomy may improve 90-day functional recovery, though confirmatory multicenter randomized trials are needed [282].

ClinicalTrials.gov registrations point to ongoing diversification of “reperfusion-compatible” adjuncts: BBB/edema modulation (imatinib, NCT03639922), intra-arterial vasomodulation during thrombectomy (verapamil, NCT03347786), growth factor/peptide candidates (GM602, NCT01221246), inflammatory signaling modulators (vinpocetine, NCT02878772), dose-finding/efficacy programs (LT3001, NCT05403866; NCT05686642), immunomodulation combined with reperfusion (fingolimod + alteplase bridging thrombectomy, NCT04675762), and neurovascular unit protection as add-on to standard reperfusion (otaplimastat SP-8203, NCT06660719; 3K3A-APC, NCT05484154). Collectively, these programs reflect a strategic shift from single-target neuronal rescue toward multimodal, workflow-compatible protection and recovery enhancement [253, 254].

Across CNS-targeted therapy classes, the trade-offs are consistent. Single-pathway agents offer clean pharmacology and mechanistic precision [259], but are vulnerable to effect dilution in heterogeneous populations where reperfusion timing, collateral physiology, and multi-compartment inflammation dominate outcomes [254, 260]. Antioxidant/anti-inflammatory adjuncts are scalable and clinically implementable [262], yet benefits often depend on timing and care-pathway integration; stronger randomized evidence for edaravone dexborneol contrasts with variability in observational adjunct data [263–265]. Pleiotropic small molecules and neurotrophic formulations may better match stroke complexity, but reproducibility across healthcare systems, populations, and reperfusion contexts remains a key barrier [254, 277, 280, 282]. Finally, collateral-dependent analyses underscore that biology at the tissue-at-risk level can gate efficacy, strengthening the case for biomarker- and imaging-guided stratification [284], but commonly face limitations in cross-population reproducibility and regulatory generalization, especially when evidence is region-concentrated or dependent on combination settings [254]. Repair-oriented formulations (e.g., cerebrolysin) are attractive for subacute recovery windows but currently rely on mixed-quality evidence and require rigorous multicenter confirmation [281, 282]. Importantly, uric acid and collateral-dependent analyses emphasize that the same neuroprotective strategy can appear effective or ineffective depending on tissue-at-risk physiology, supporting the need for biomarker- and imaging-guided stratification in future trials [267].

Microbiota-modulating therapies

Dietary patterns and nutrient exposures are most consistently supported for long-term vascular risk modulation rather than acute functional rescue. Here, we summarize clinical trials of microbiota-modulating interventions for stroke in Table 5. Mediterranean dietary patterns are associated with reduced cardiovascular events and mortality in primary and secondary prevention settings [285], and mechanistic clinical studies link Mediterranean diet adherence to improved vascular health signatures and circulating adipokine/ceramide profiles plausibly connected to inflammatory–metabolic risk reduction [286]. Circulating polyunsaturated fatty acid (PUFA) profiles also associate with clinically relevant phenotypes: ω3/ω6 fatty acids relate to atrial fibrillation risk in acute ischemic stroke [287], and broader cohort data connect ω3/ω6 profiles to cognitive decline and dementia trajectories [288]. Dietary fiber—conceptually upstream of microbiota fermentation and SCFA production—shows a dose-associated inverse relationship with stroke risk in meta-analytic synthesis [289]. These datasets support prevention-oriented plausibility but remain limited by adherence heterogeneity and residual confounding for stroke-specific functional endpoints. Recent stroke-microbiome work has also moved beyond vascular risk association toward subtype-specific and recovery-related signatures. In intracerebral hemorrhage, clinical microbiome profiling has associated gut dysbiosis with neurological function recovery, supporting a role for microbiota–gut–brain interactions beyond ischemic stroke alone. Comparative analyses across ischemic and hemorrhagic stroke further suggest that both stroke types show altered microbial communities, although the direction and functional significance of these shifts may vary by subtype [290, 291].

Table 5.

Clinical trials of microbiota-targeted interventions for stroke

Intervention	Status/Trial	Key mechanism	Limitation	References
Dietary/nutrition optimization
Mediterrane	Observational/ Meta	Anti-inflammatory; improves vascular–metabolic profile	Observational/meta; adherence & confounding; limited acute-stroke interventional evidence	[285, 286]
Omega-3/6 PUFA-rich diet	Observational	Modulates inflammation; cardiometabolic support; potential gut–metabolite interactions	Observational; causality/dose/target population unclear	[287, 288]
Dietary fiber intake	Observational/ Meta	Fermentation → ↑SCFAs; cardiometabolic/anti-inflammatory support	Mostly prevention/long-term; fiber types/endpoints heterogeneous	[289]
Probiotics supplementation
Probiotics (e.g., Bifidobacterium/Lactobacillus)	Stroke clinical (small trials) + Non-stroke RCT (mechanistic)	Microbiota/barrier support; anti-inflammatory; IPA–BDNF link (indirect, non-stroke)	Neurologic endpoints inconsistent; strain/dose/duration heterogeneity; limited stroke-core outcome evidence	[292, 293]
Antibiotic strategy
Prophylactic antibiotics (acute phase)	IPD meta / RCT-level evidence	Infection prevention (e.g., UTI).	Infection reduction ≠ functional benefit; resistance and dysbiosis/immune risks; not for routine use	[294, 295]
D. Fecal microbiota transplantation (FMT)
FMT	Regulatory-approved (non-stroke indication); Stroke: no confirmatory trial	Microbiota replacement; engraftment-dependent immune–metabolic modulation	Stroke trials lacking; standardization/safety and engraftment variability; unclear benefit–risk boundary	[296–299]

For probiotics, preclinical studies suggest plausible mechanistic pathways but comparatively limited evidence for core stroke outcomes. Preclinical work demonstrates that specific Lactobacillus strains can modulate barrier integrity and microbiota composition in metabolically stressed animals [300], supporting a rationale for barrier-focused interventions that could be relevant to post-stroke gut dysfunction. Human trials outside stroke show that probiotic intake can increase microbiota-derived indole-3-propionic acid (IPA) and correlate with neurotrophic markers (e.g., BDNF), and in vitro IPA can dampen microglial inflammatory signaling, providing biologically plausible “gut–immune–brain” links [292]. In stroke, clinical evidence remains comparatively limited and heterogeneous; a synthesis suggests possible benefits for complications and inflammatory markers, with weaker and less consistent effects on neurological outcome scales [293]. However, across probiotic studies the major disadvantages are intervention heterogeneity (strain, dose, duration, co-interventions), population heterogeneity, and endpoint mismatch, which collectively make reproducibility and clinical translation to stroke-functional endpoints difficult. Thus, probiotic studies currently support biological plausibility for gut–immune–brain modulation, but they do not yet establish a reproducible stroke-recovery intervention. Future trials will need strain-level definition, standardized dosing, microbiome and metabolite readouts, and stroke-specific functional endpoints [292, 293, 300].

Prebiotic and postbiotic interventions are emerging as complementary microbiota-targeted strategies in stroke. Prebiotics, such as fermentable fibers and resistant starches, may enhance SCFA-producing taxa and support gut-barrier and immune-metabolic homeostasis; however, stroke-specific interventional evidence remains limited and is largely extrapolated from broader microbiome and metabolic studies [301, 302]. Postbiotics, defined as preparations of inanimate microorganisms and/or their components that confer health benefits, may offer practical advantages over live probiotics or FMT in product stability, standardization, and safety, but stroke-specific evidence remains preliminary and their efficacy in acute or recovery-phase stroke has not been established [303, 304]. Therefore, both strategies are promising but should still be interpreted as mechanistically plausible rather than clinically validated stroke therapies, with effects likely shaped by baseline microbiome composition, host diet, intestinal barrier integrity, immune state, and stroke phase.

Antibiotic strategies illustrate the clearest example of “biological plausibility without core endpoint gain.” Systematic evidence indicates that prophylactic antibiotics can reduce infections after stroke in some settings [294], and individual patient data meta-analysis confirms a robust reduction in infections but no consistent improvement in 3-month functional outcomes or reduction in poor functional endpoints (modified Rankin Scale [mRS] 3–6) [295]. Observational work suggests that targeted prophylaxis may reduce specific infectious endpoints such as symptomatic urinary tract infection arising from asymptomatic bacteriuria [305], but this remains prone to confounding and—critically—does not address the central problem that broad-spectrum antibiotic perturbation is non-targeted, carries antimicrobial resistance risk, and may disrupt immune–microbiome homeostasis in ways that could counteract recovery biology [294, 295, 305]. Thus, antibiotics are best viewed as infection-control tools in selected scenarios rather than microbiota therapies that improve neurological outcome. This evidence also illustrates a broader translational lesson for microbiota-targeted stroke therapy: modifying the microbiome or infection risk is not sufficient unless the intervention improves predefined neurological recovery endpoints without disrupting immune–microbiome homeostasis [295, 305].

FMT provides the strongest causal manipulation of the microbiome in principle, and the literature highlights both its translational promise and the current evidence gap in stroke. Foundational clinical gastroenterology work frames FMT as a potent ecosystem-level intervention with therapeutic potential and growing standardization in non-neurological indications [296]. In stroke-relevant preclinical models, FMT enriched for SCFA-producing communities can improve microbiota composition and metabolic profiles, reduce infarct volume, and ameliorate gut permeability in MCAO paradigms [297]. Human microbiome profiling further supports a prognostic “dysbiosis signature”: the Stroke Dysbiosis Index associates gut community patterns with brain injury severity and outcome, and “human-to-mouse” FMT experiments show that microbiota from high-dysbiosis donors can worsen brain injury and augment proinflammatory gut immune signatures (e.g., IL-17⁺ γδ T cells) [298]. Yet the updated review literature emphasizes that, for stroke, FMT remains largely investigational with insufficient multicenter, standardized trials anchored to stroke core endpoints and rigorous safety characterization [299]. Its advantages are strong biological leverage and potential for mechanistic discovery; its disadvantages are donor/recipient variability, standardization challenges, and the need for careful risk management in medically fragile stroke populations. At present, FMT is most useful as a causal tool for testing whether dysbiotic microbial communities can worsen or rescue stroke phenotypes, rather than as a clinically established stroke therapy. Future studies should define donor criteria, microbial and metabolite potency markers, timing relative to stroke onset, safety monitoring, and stroke-specific functional endpoints [298, 299].

Across microbiota-modulating approaches, advantages and limitations follow a consistent pattern. Diet is scalable and low risk, with the strongest evidence for long-term risk reduction rather than acute outcome modification [285, 289]. Probiotics offer plausible barrier/immune benefits but are limited by intervention heterogeneity and inconsistent neurological endpoints [293]. Antibiotics reliably reduce infections yet do not improve core functional outcomes and carry ecological harms [295]. FMT is the most direct microbiome intervention with compelling preclinical stroke signals but faces the steepest hurdles in standardization, donor variability, and safety in medically fragile populations [299].

Positioning EV therapeutics relative to CNS drugs and microbiota strategies

Across both CNS-directed pharmacotherapy and microbiota-modulating interventions, a central challenge is the mismatch between stroke’s phase-evolving, systems-level pathobiology and the single-node structure (or high heterogeneity) of many interventions [253, 254]. EVs can be framed as a modality that addresses several of these gaps (Fig. 3). This positioning is strengthened by recent evidence from both fields. Stroke-microbiome studies indicate that gut dysbiosis, intestinal barrier dysfunction, microbial metabolites, and peripheral immune tone may shape post-stroke inflammation and recovery, whereas stroke-EV studies indicate that EVs can carry multi-cargo signals capable of modulating endothelial stability, immune programming, BBB integrity, and repair responses [181, 193, 290, 299].

Fig. 3.

Therapeutic strategies for stroke: strengths, limitations, and where EVs fill gaps by targeting peripheral nodes. This figure compares three therapeutic approaches targeting the post-stroke “double-hit” brain–periphery axis. CNS-directed pharmacotherapy provides pathway-specific precision and can be dosed in a standardized manner, but it is limited by narrow time windows for treatment, restricted brain exposure due to BBB and PK constraints, and its dependence on reperfusion and collateral circulation status. Microbiota-modulating strategies (including diet, probiotics, and fecal microbiota transplantation, or FMT influence immune responses and metabolites like SCFAs, but these therapies often show slow effects, suffer from host/strain/donor variability, and face challenges with standardization and alignment with clinical endpoints. SC-EVs offer multi-target immunomodulation, naturally accumulating in peripheral organs such as the liver, spleen, and gut. This peripheral-first advantage helps reduce systemic cytokine levels, thereby decreasing neuroinflammation. However, key gaps for SC-EVs include the need for potency-linked QC and batch consistency, PK/PD-guided dosing strategies, and stronger causal evidence supporting the necessity and sufficiency of proposed peripheral mechanisms. BBB, blood–brain barrier; EVs, extracellular vesicles; SC-EVs, stem cell–derived extracellular vesicles; FMT, fecal microbiota transplantation; IL, interleukin; PK/PD, pharmacokinetics/pharmacodynamics; QC, quality control; SCFAs, short-chain fatty acids; TNF, tumor necrosis factor; tPA, tissue plasminogen activator; EVT, endovascular therapy

First, EVs are intrinsically pleiotropic biologics: multi-cargo payloads (RNAs, proteins, lipids) can, in principle, modulate endothelial stability, BBB integrity, neuroinflammation, and repair programs in parallel—features that may be more robust in heterogeneous thrombectomy-era populations where single-pathway agents can be diluted [260, 267]. Second, EVs align naturally with “reperfusion-plus” concepts: they can be deployed as adjuncts aimed at both neurovascular-unit protection and immune trajectory shaping, consistent with the direction of current pipelines [254]. Third, relative to microbiota interventions, EV platforms may offer more controllable dosing, manufacturing standardization, and mechanistic tractability, while still interfacing with the same biological bottlenecks (barrier function, immune programming) that diet/probiotics/FMT influence more indirectly [295, 299]. In this sense, EVs can be positioned as an attempt to convert broad systemic associations into testable, mechanism-forward interventions that are more compatible with biomarker-guided stratification—an explicit need highlighted by collateral- and background-therapy–dependent efficacy signals across neuroprotection trials [254, 260, 267]. However, EVs should not be presented as a solved alternative to microbiota interventions. Their advantages in controllability and engineering must be weighed against unresolved issues in source heterogeneity, batch reproducibility, potency assays, biodistribution, target engagement, and long-term safety. Conversely, microbiota-directed approaches may have broader ecological effects but remain constrained by donor or strain variability, slower biological onset, and inconsistent stroke-specific functional endpoints [193, 235, 295, 299].

Limitations and perspectives

Despite robust neuroprotective signals in preclinical studies, clinical translation of SC-EVs has been slow. One reason is conceptual: the dominant CNS-centric development paradigm treats peripheral biodistribution as a defect to be engineered away (“force EVs across the BBB”), rather than a biological feature that could be leveraged to reprogram systemic immunometabolic circuits that shape post-stroke injury and recovery [204, 205, 240]. This bias tends to increase manufacturing and regulatory complexity by favoring heavily engineered products, and underweight the therapeutic leverage of actions in peripheral immune hubs and barrier organs. Moving a periphery-first strategy from an appealing hypothesis to a clinically testable paradigm will require coordinated progress across product definition, model realism, and causal mechanism validation [170, 306].

A practical constraint remains the pharmacokinetic reality of IV delivery: systemically administered EVs are rapidly cleared from blood and are preferentially captured by mononuclear phagocyte system (MPS) organs such as the liver and spleen, while only a small fraction reaches the brain parenchyma [204, 205]. Many current programs respond by engineering EVs to improve brain targeting [13, 14]. While technically reasonable, this route can increase manufacturing burden and may miss a simpler therapeutic logic: if peripheral inflammation and barrier failure drive the humoral “second hit,” then modulating the peripheral immune microenvironment and gut–immune interfaces could be both biologically consistent and clinically tractable. A periphery-first framework therefore reframes “low brain exposure” not as the sole bottleneck, but as an invitation to target high-leverage peripheral nodes—liver, spleen, and gut-associated immune compartments—to attenuate systemic inflammatory amplification and protect the CNS indirectly.

Product consistency and potency: moving from particle metrics to mechanism-linked quality control

A major bottleneck for clinical translation is the intrinsic heterogeneity of EV preparations, which varies with donor cell state, passage, culture conditions, and isolation workflows, producing substantial batch-to-batch divergence in size distributions, cargo composition, and bioactivity. Widely used isolation methods (e.g., ultracentrifugation or polymer precipitation) can yield preparations contaminated with non-vesicular particles and may impair vesicle integrity and function, while routine storage practices such as freeze–thaw cycles can alter surface features and biological performance [307–309]. These variables complicate reproducibility and regulatory standardization, particularly when “purity” metrics do not predict therapeutic effect [310].

Culture and preconditioning paradigms further complicate, but may also rationally enhance, MSC-EV potency. In vitro expansion is a key variable. Liu et al. compared EVs derived from BM-MSCs and AD-MSCs across passages and showed that tissue source and expansion history reshaped EV protein/miRNA cargo as well as angiogenic and immunomodulatory activity; importantly, higher-passage EVs could exhibit reduced functional potency despite comparable or increased particle/protein output [172]. This reinforces the principle that equivalent particle counts do not necessarily represent equivalent biological products.

Preconditioning can further reshape EV function in a mechanism-specific manner. Hypoxic conditioning is the most frequently studied example. Zhang et al. reported that hypoxia-stimulated human umbilical-cord MSC-derived microvesicles promoted human umbilical vein endothelial cell (HUVEC) tube formation and angiogenesis in vivo, supporting a pro-angiogenic effect relevant to neurovascular repair [311]. Yuan et al. further showed in a renal ischemia–reperfusion injury model that hypoxia altered the MSC transcriptional profile, increased EV production, and enhanced therapeutic effects [189], while Gupta et al. demonstrated tissue-source-specific hypoxic responses in WJ-, BM-, and AD-MSCs. Together, these studies support a stroke-relevant rationale for hypoxia-conditioned EVs in ischemic tissue repair, although direct head-to-head validation in ischemic stroke models remains limited [312].

Inflammatory priming may preferentially enhance immunoregulatory potency. Kaur et al. compared MSC-EVs generated under monolayer versus microcarrier culture conditions in experimental autoimmune uveitis and related in vitro potency assays, showing that microcarrier-derived EVs displayed stronger immunomodulatory activity and higher levels of immunosuppressive mediators such as TGFβ1 and let-7b [313]. Similarly, Losurdo et al. reported that EVs from TNF-α/IFN-γ–preconditioned hBM-MSCs exerted immunomodulatory and neuroprotective effects after intranasal delivery in a 3xTg Alzheimer’s disease model [314]. Although these findings are mainly derived from non-stroke inflammatory or neurodegenerative models, they support the broader principle that inflammatory priming can tune MSC-EV cargo toward immune regulation; for stroke translation, priming protocols must be standardized to avoid unwanted pro-inflammatory or pro-coagulant vesicle profiles.

Three-dimensional culture and bioreactor systems add another layer of potency and scalability control. Zhang et al. compared exosomes from 2D and 3D cultures of human BM-MSCs in a rat traumatic brain injury model and found that 3D-derived exosomes produced greater suppression of glial fibrillary acidic protein (GFAP)-positive astrocytes and CD68-positive microglia/macrophages, together with higher exosome yield [315]. In a stroke-oriented translational study, Son et al. used a 3D spheroid culture platform with hWJ-MSCs and reported improved EV production consistency across batches and donors compared with conventional 2D culture [316]. These data suggest that 3D or bioreactor-based systems may improve yield, cargo complexity, and manufacturing consistency, but higher yield alone should not be equated with superior stroke efficacy [317]. These examples illustrate that culture optimization should be treated as a testable potency-shaping variable rather than as a generic enhancement strategy.

More fundamentally, current QC pipelines often emphasize physical descriptors (particle number, modal size, total protein) that correlate poorly with functional potency, which is the clinically relevant attribute. For example, EVs derived from the same donor cell source at different passages can show markedly different immunoregulatory capacity, indicating that “equivalent” particle counts can reflect very different biological products [172]. This gap becomes especially consequential under a peripheral-first model, where the therapeutic hypothesis depends on measurable immune reprogramming rather than direct neuronal replacement. Therefore, future QC should be structured around mechanism-linked potency assays, implemented alongside MISEV2023-compliant characterization, rather than treated as optional add-ons [170, 236]. For source- or preconditioning-enhanced MSC-EVs, potency testing should be performed side-by-side against non-preconditioned EVs from the same MSC source, using matched dose units and identical isolation/storage workflows [172, 318]. Candidate assays should be selected according to the proposed stroke mechanism and may include mixed lymphocyte reaction-based immunosuppression, macrophage polarization, Treg induction, endothelial tube formation, BBB or gut epithelial barrier restoration, and cytokine suppression. Surrogate cargo markers such as TGFβ1 may help predict immunomodulatory activity [318], but no single molecule is likely to capture the full mechanism of action; therefore, an orthogonal potency panel is preferable for stroke applications.

Potency readouts should be chosen to match the proposed mechanism and IV pharmacokinetics, given preferential capture by liver/spleen and limited parenchymal delivery [204, 205]. Practical options include standardized co-culture platforms that quantify (i) induction of regulatory programs (e.g., Treg differentiation/expansion), (ii) suppression of pro-inflammatory cytokine output, and (iii) restoration of barrier-relevant signaling in immune–epithelial systems approximating gut and secondary lymphoid tissue biology [204, 205, 236, 284, 319]. Importantly, the field also needs consensus on which cargo or functional signatures define “active” lots for a given indication, and whether potency is best captured by single assays or a minimal orthogonal panel.

In addition, most reports still under-specify EV dose using mixed units (protein mass, particle counts, or donor-cell equivalents), limiting cross-study comparability. Harmonized dosing conventions and exposure–response analyses are prerequisites for credible clinical translation [170]. Immunogenicity may be lower than live-cell therapy, but EVs can still carry pro-coagulant or pro-inflammatory components depending on source and manufacturing; infusion reactions and off-target immunomodulation remain plausible and need systematic surveillance [170].

Aligning preclinical disease modeling with clinical reality and IV pharmacokinetics

A second translational gap is the mismatch between common preclinical stroke models and real-world patient biology. Many efficacy studies rely on otherwise healthy young animals with simple MCAO, whereas typical stroke patients are older and frequently present with hypertension, diabetes, dyslipidemia, and chronic low-grade inflammation—factors that reshape immune setpoints, gut microbiota composition, endothelial vulnerability, and systemic metabolic tone [306]. These comorbidities are not merely “background noise”; they can determine whether a therapy that looks effective in simplified models will fail in heterogeneous clinical populations. Under a peripheral-first framework, this issue is even more central because the hypothesized benefit depends on restoring peripheral homeostasis in systems already perturbed by cardiometabolic disease [108, 320]. Future studies should therefore prioritize aging and comorbidity-integrated stroke models, and they should explicitly quantify whether peripheral stabilization (gut/immune/metabolic indices) causally precedes and predicts brain protection and functional recovery.

Aging should also be modeled through an immune–microbiome lens rather than treated only as a chronological variable. Age-related dysbiosis and barrier fragility can amplify the infection-prone state that follows stroke-induced immunodepression [321]. Crapser et al. used young and aged mice subjected to ischemic stroke and orally gavaged them with GFP-tagged E. coli to track bacterial dissemination; they found that aged mice exhibited greater gut permeability, increased bacterial translocation into peripheral tissues, and a persistent sepsis-like response with immune dysfunction, higher mortality, and worse neurological deficits compared with young mice [322]. Stanley et al. further provided direct evidence that stroke can promote commensal bacterial translocation: in experimental tMCAO models, stroke induced gastrointestinal dysbiosis, altered gut autonomic innervation, increased intestinal permeability, and bacterial translocation within 24 h, while a prospective clinical cohort detected typical gut bacteria in blood, urine, or saliva samples from 22% of patients with ischemic stroke and stroke-associated infections [323]. Wen et al. combined transient MCAO in young and aged male C57BL/6J mice with analysis of a 509-patient clinical cohort, showing that aged post-stroke mice developed more spontaneous bacterial lung infections, gut inflammation, and barrier disruption, while age, stroke severity, and indwelling catheter use independently predicted post-stroke infection in patients [324].

Together, these studies support a gut-origin route for post-stroke infection, in which barrier disruption and autonomic/neuroendocrine imbalance permit selective commensal bacteria to disseminate to normally sterile sites [325, 326]. Beyond chronological age, “microbiome age” may independently influence stroke outcome: Spychala et al. used 16 S rRNA sequencing and microbiota-transfer experiments in young and aged mice and showed that aged microbiota worsened post-stroke mortality, behavioral deficits, and inflammatory responses when transferred to young mice, whereas young microbiota improved survival and functional recovery in aged mice after MCAO [327]. Thus, aged animals and microbiome-age–manipulated models should be prioritized in EV studies, with infection/sepsis endpoints, bacterial translocation, gut-barrier integrity, and immune–microbiome readouts incorporated alongside infarct and behavioral outcomes.

Biodistribution is context-dependent and disease-state dependent. Stroke alters vascular permeability, immune cell activation, and MPS behavior; biodistribution measured in healthy animals may not predict distribution in acute ischemia or in comorbidity models [204, 205]. Route-of-administration choices should also be reconsidered considering systemic pathophysiology. Intracerebral, intranasal, and intra-arterial routes can increase CNS exposure and have shown experimental efficacy, but they may underutilize the therapeutic opportunity to intervene at peripheral immune hubs that drive the humoral “second hit” [240, 248, 328–330]. From a peripheral-first perspective, IV delivery remains attractive because it provides broad immune-system coverage; however, it introduces a PK paradox: rapid hepatic sequestration can remove most of the dose, potentially forcing supra-physiological dosing to achieve meaningful exposure in secondary lymphoid organs and the gut, thereby escalating manufacturing burden and increasing risk of infusion-related adverse events [204, 205]. Accordingly, a realistic development path should not simply “default to IV,” but should seek solutions that optimize systemic coverage while mitigating hepatic sink effects. A rational direction is to develop surface chemistries or formulation strategies that reduce non-productive liver capture while preserving spleen/immune organ engagement, because indiscriminate “liver escape” could also remove the very organ-level immunometabolic leverage that makes peripheral-first attractive. Timing is often treated loosely in preclinical studies, yet post-stroke immune programs are phase-dependent. Trials should be designed around time-resolved biology (hyperacute/acute/subacute) with explicit hypotheses for which peripheral nodes and immune programs are being targeted at each stage [306].

In addition, endpoints of trials are often mismatched to mechanism. If the hypothesized benefit is immune trajectory reprogramming, studies should incorporate immune phenotyping, barrier integrity metrics, and organ-axis readouts alongside infarct volume and behavioral outcomes, ideally with preregistered primary endpoints. EV efficacy should be tested in “reperfusion-plus” settings and compared with relevant adjuncts to clarify additive vs. redundant effects, especially given modern thrombectomy-era care.

Mechanistic verification: eliminating tracing artifacts and establishing causality

Although many studies report reduced neuroinflammation and microglial shifts toward M2-like phenotypes after SC-EV administration, the mechanistic chain is often correlational rather than causal [231]. A critical reason is that the field’s biodistribution evidence base is frequently built on lipophilic dye labeling, which is vulnerable to non-specific dye transfer, micelle formation, and signal persistence independent of intact EV uptake, thereby inflating confidence in organ “accumulation” patterns without proving cargo delivery or intracellular function [204]. If the peripheral-first strategy is to be validated, future work must prioritize artifact-resistant reporter systems that demonstrate not only EV presence but functional cargo release in recipient cells (e.g., genetically encoded reporters, luminescence-based systems, or other approaches that resolve true uptake and activity rather than membrane labeling).

Beyond improved tracing, the field needs necessity and sufficiency designs that directly test peripheral mechanisms. Under a peripheral-first model, mechanistic studies should incorporate organ- or pathway-level perturbations that can falsify or support causal claims. For example, because the spleen is a major immune mobilization hub and a major site of EV capture after systemic dosing, splenectomy or selective splenic pathway blockade can test whether spleen-dependent immune reprogramming is required for CNS benefit [122, 320]. Similarly, if gut-mediated immune signals are proposed to be intermediaries, then experiments should combine EV administration with defined microbiota manipulations and barrier-function readouts, while ensuring that the EV effect cannot be replicated solely by confounding experimental changes unrelated to EV cargo.

Equally important, pharmacokinetic studies should expand from “where the EV goes” to what the effector cells do. Classical biodistribution labeling tells little about whether EV-treated peripheral immune cells adopt distinct phenotypes and then traffic directionally to CNS compartments. To address this, future studies should incorporate immune-cell tracking approaches capable of resolving migration from spleen or gut-associated lymphoid tissue into CNS border regions and parenchyma, thereby directly testing the predicted “periphery → immune intermediate → brain” route [85]. Cargo attribution is often underpowered. Many mechanistic claims are based on candidate miRNAs/proteins without demonstrating that they are necessary for efficacy (e.g., cargo depletion, RNase/protease controls with membrane integrity verification, or donor-cell genetic perturbations). Stronger causal designs will improve interpretability and accelerate regulatory confidence.

Testing neuroimmune signal transmission: CAP/vagus–spleen axis as a falsifiable pathway

A concrete and experimentally tractable mechanism for peripheral-to-central immunoregulation is the CAP, in which vagal circuits regulate cytokine production through acetylcholine signaling and α7nAChR–dependent macrophage responses in the spleen [331–333]. Functional evidence across inflammatory paradigms shows that vagotomy or interruption of vagus–spleen signaling diminishes anti-inflammatory effects, and α7nAChR deficiency eliminates acetylcholine-mediated suppression of TNF release [334–337]. Because SC-EVs repeatedly show systemic anti-inflammatory effects and neuroprotection after stroke, CAP provides a falsifiable route for testing whether peripheral immune reprogramming is actively transmitted to CNS outcomes through neuroimmune reflex circuitry [316, 338, 339].

However, mechanistic rigor will require moving beyond blunt interventions that introduce major confounds. Rather than relying solely on vagotomy—which can disrupt physiology broadly—future stroke studies should preferentially use genetic or pharmacologic specificity (e.g., α7nAChR knockout, selective antagonists, or targeted splenic nerve blockade where feasible) to test whether EV efficacy depends on CAP signaling [334, 337]. If SC-EV benefits are lost under α7nAChR blockade/knockout or under selective interruption of splenic signaling, while remaining intact under manipulations that preserve CAP, this would provide strong causal support for a peripheral-to-central transmission mechanism. Conversely, if EV benefits persist despite CAP interruption, then CAP should be deprioritized in favor of alternative pathways (e.g., direct BBB-interface effects or other immune trafficking routes). Importantly, these experiments should be embedded in comorbidity-relevant stroke models and paired with potency-linked QC assays, so that mechanistic conclusions are not confounded by batch variability or non-representative disease biology [170, 306].

A refined perspective: turning the pharmacokinetic paradox into a testable development strategy

Pathological and clinical evidence increasingly frames stroke as a systemic disorder driven by central–peripheral coupling, where an initial neurogenic insult perturbs gut, liver, and spleen homeostasis and subsequent humoral amplification exacerbates secondary brain injury. In that context, the pharmacokinetic profile of IV SC-EVs—dominant sequestration by liver and spleen with limited brain entry—need not be treated solely as a bottleneck. The periphery-first strategy advanced in this review reframes MPS capture as a feature that can be exploited to intercept systemic inflammatory amplification at its source, rather than forcing high-efficiency BBB crossing as the primary success criterion [204, 205].

Conceptually, this approach is strongest when framed as a set of falsifiable hypotheses rather than a broad narrative: (i) cellular pathway—SC-EVs induce tolerogenic programs in peripheral immune reservoirs (e.g., spleen) and alter immune-cell trafficking to CNS border regions; (ii) humoral pathway—SC-EVs dampen hepatic acute-phase and systemic cytokine programs to stabilize BBB integrity; and (iii) neural pathway—SC-EVs restore gut barrier/immune tone and engage vagal anti-inflammatory reflexes to limit inflammatory escalation. Each route can be tested with necessity/sufficiency designs, time-resolved sampling, and artifact-resistant biodistribution/functional delivery readouts.

At present, the main interpretive weakness is causal uncertainty: many studies show peripheral and CNS improvements together but do not prove that peripheral mechanisms are necessary independent drivers of benefit [231]. Closing this gap will require methodological upgrades involving replacing dye-based tracking with functional delivery reporters, pathway-level perturbations to test causality, comorbidity-aligned models that capture baseline systemic inflammation, and mechanism-linked potency assays that reflect the proposed immune/barrier mechanism, such as Treg induction or cytokine suppression, rather than particle counts alone [170, 306].

Conclusions

Stroke is increasingly understood as a phase-evolving, bidirectional CNS–periphery disorder rather than a brain-confined event: acute ischemia rapidly triggers neurogenic stress programs that disrupt immune trafficking and barrier function across the gut–liver–spleen network, and a subsequent humoral/immune phase can sustain systemic inflammation, BBB compromise, and secondary neurovascular injury, consistent with a “double-hit” cascade. Many CNS-centric neuroprotectants—often designed around single pathways and narrow time windows—have repeatedly failed to deliver robust benefit in heterogeneous, reperfusion-era populations, where collateral physiology, comorbid immune-metabolic states, and parallel injury mechanisms dilute effect sizes. Microbiota-modulating strategies (dietary patterns, probiotics, antibiotics, FMT) highlight the therapeutic relevance of the gut–immune–brain axis, but their clinical translation to stroke core outcomes is constrained by intervention heterogeneity, ecosystem-level variability, and endpoint mismatch despite strong mechanistic plausibility. Against this landscape, SC-EVs align naturally with stroke biology: as pleiotropic, multi-cargo biologics they can modulate inflammatory signaling, endothelial stability, barrier repair, and recovery programs in parallel, fitting “reperfusion-plus” concepts that seek adjunctive protection and repair rather than single-node neuronal rescue. Critically, the pharmacokinetic reality that IV EVs are rapidly cleared and preferentially sequestered by liver and spleen—often viewed as a CNS-delivery limitation—can be reframed as an advantage under a periphery-first strategy, enabling direct access to immune–metabolic hubs and barrier organs that drive the humoral “second hit”. The path to translation, however, hinges on rigor: mechanism-linked potency assays beyond particle counts, artifact-resistant biodistribution and cargo-delivery readouts, comorbidity- and age-relevant stroke models, and necessity/sufficiency experiments that causally test peripheral routes (including falsifiable CAP/α7nAChR–spleen signaling where appropriate). With these priorities, SC-EVs can mature into a stratification-ready, system-targeted therapeutic platform that complements reperfusion and converts an expanded understanding of stroke pathophysiology into clinically testable, phase-aware interventions.

Abbreviations

ADSC

Adipose-derived stem cell

Akt

Protein Kinase B

APC

Antigen-presenting cell

BBB

Blood-brain barrier

BDNF

Brain-derived neurotrophic factor

BMSC

Bone marrow mesenchymal stem cell

CAN

Central autonomic network

CAP

Cholinergic anti-inflammatory pathway

cGAS

Cyclic GMP-AMP synthase

CMVs

Cytoplasmic membrane vesicles

CNS

Central nervous system

DAMPs

Damage-associated molecular patterns

DSS

Dextran sulfate sodium

eNOS

Endothelial nitric oxide synthase

ESC

Embryonic stem cell

EV

Extracellular vesicle

FMT

Fecal microbiota transplantation

GBA

Gut-brain axis

GFAP

Glial fibrillary acidic protein

HMGB1

High mobility group box 1

HPA

Hypothalamic-pituitary-adrenal axis

hUMSC

Human umbilical mesenchymal stem cell

HUVEC

Human umbilical vein endothelial cell

ICAM-1

Intercellular adhesion molecule 1

ICH

Intracranial hemorrhage

IFN

Interferon

IL

Interleukin

iNKT

Invariant natural killer T cells

IPA

Indole-3-propionic acid

iPSCs

Induced pluripotent stem cells

IV

Intravenous

LPS

Lipopolysaccharide

MAPK

Mitogen-activated protein kinase

MCAO

Middle cerebral artery occlusion

MMP

Metalloproteinase

MNVs

Magnetic nanovesicles

MODS

Multi-organ dysfunction syndrome

MPS

Mononuclear phagocyte systems

mRS

Modified Rankin Scale

MSCs

Mesenchymal stem cells

mTOR

Mammalian target of rapamycin

NF-κB

Nuclear factor-κB

NIHSS

National Institutes of Health Stroke Scale

NSC

Neural stem cell

NTS

Nucleus tractus solitarius

OMVs

Outer membrane vesicles

PAMPs

Pathogen-associated molecular patterns

PI3K

Phosphoinositide 3-Kinase

PK

Pharmacokinetic

PLR-RS

Puerariae Lobatae Radix–resistant starch

PTEN

Phosphatase and Tensin Homolog

PUFA

Polyunsaturated fatty acid

QC

Quality control

RGD-C1C2

RGD-4 C peptide fused to C1C2

ROS

Reactive oxygen species

SC-EVs

Stem cell-derived extracellular vesicles

SCFAs

Short-chain fatty acids

sEVs

Small extracellular vesicles

Sirt1

Sirtuin 1

SPIONs

Superparamagnetic iron oxide nanoparticles

STING

Stimulator of interferon genes

TGF-β

Transforming growth factor-β

TLR4

Toll-like receptor 4

TMAO

Trimethylamine N-oxide

TNF-α

Tumor necrosis factor-α

Tregs

Regulatory T cells

TrkB

Tropomyosin receptor kinase B

USP22

Ubiquitin-specific protease 22

VN

Vagus nerve

ZO-1

Zonula occludens-1

α7nAChR

α7 nicotinic acetylcholine receptor

Authors’ contributions

CD, LZ, and XH conceived and designed the study. LZ, JL, FA, YL, and CD collected the data and drafted the original manuscript. CD and XH critically revised the manuscript for important intellectual content. All authors reviewed and approved the final version.

Funding

This research was supported by the National Natural Science Foundation of China under grant number 82302108.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.