Enhanced treatment of ischemic stroke by scutellarin loaded Angiopep-2-modified milk exosomes via multiple pathological pathways regulation

Abstract

Ischemic stroke is a major cerebrovascular disease with high morbidity and mortality. However, effective treatments remain limited due to the narrow thrombolytic window, ischemia-reperfusion injury, and restricted drug delivery across the blood-brain barrier (BBB). Due to multiple pathological processes including oxidative stress, inflammation, mitochondrial dysfunction, and microglial dysregulation involved in ischemic stroke, it is urgent to develop drug delivery systems capable of crossing the BBB and targeting multiple pathological pathways. Scutellarin (SCU) exhibits neuroprotective effects while its application is constrained by low bioavailability, rapid clearance, and poor stability. This study developed an Angiopep-2 (ANG-2)-modified milk exosome delivery system loaded with SCU (SCU@AMExo) for ischemic stroke therapy. SCU@AMExo improved the bioavailability and stability of SCU, facilitated efficient BBB penetration, and exerted neuroprotective effects by reducing oxidative stress, alleviating mitochondrial dysfunction, inhibiting NLRP3 inflammasome activation, and regulating microglial polarization. In vitro, SCU@AMExo increased the cell viability of PC12 cells with Oxygen-Glucose Deprivation/Reperfusion (OGD/R) injury from 48.1% to 92.9%. In transient middle cerebral artery occlusion/reperfusion (tMCAO/R) mice, SCU@AMExo reduced the cerebral infarct volume from 51.06% to 11.29%, inhibited neuronal apoptosis, and alleviated neurological deficits. These results demonstrate that SCU@AMExo is an effective brain-targeted drug delivery system for ischemic stroke through multi-pathway neuroprotection.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12951-026-04536-6.

Introduction

Currently, stroke remains the second leading cause of death worldwide and the third leading cause of disability, posing a serious threat to human life and health [1]. Among all strokes, ischemic stroke accounts for 87% [2]. At present, thrombolysis and neuroprotection are the two major therapeutic strategies for ischemic stroke [3–5]. However, the clinical benefit of intravenous thrombolysis is constrained by its strong time dependence, with proven efficacy requiring treatment within a narrow time window [6–8]. Moreover, reperfusion following vascular recanalization of the ischemic brain may lead to substantial cerebral tissue injury, thereby further aggravating neurological deficits [9]. On the other hand, the existence of the blood-brain barrier (BBB) severely hinders the entry of most therapeutic agents into the brain, limiting the efficacy of neuroprotective treatments [10]. The pathogenesis of ischemic stroke is highly complex, including oxidative stress, mitochondrial dysfunction, cellular excitotoxicity, neuroinflammation and excessive microglial activation, which collectively contribute to neuronal injury [11]. Therefore, there is an urgent need to explore novel drug delivery platforms that can overcome the BBB as well as target the pathophysiological features of ischemic stroke.

Natural bioactive components from traditional Chinese medicinal herbs have attracted extensive attention in stroke therapy research because of their multi-target effects and low toxicity [12, 13]. Scutellarin (SCU), the major bioactive component of Erigero breviscapus has been widely investigated in ischemic disease researches. Accumulating evidence indicates that SCU reduces reactive oxygen species (ROS) [14], rescues mitochondrial damage [15, 16], suppresses inflammation by regulating microglial polarization [17], improves the hemodynamic balance, and thereby exerting neuroprotective effects against cerebral ischemic injury. However, its application in brain diseases is limited by low bioavailability, short half-life, and poor stability [18]. Hence, developing an efficient delivery system to enhance its BBB penetration is critical for SCU clinical translation.

Exosomes have recently emerged as promising natural nanocarriers owing to their endogenous origin, high stability, low toxicity, and minimal immunogenicity [19, 20]. As nanosized extracellular vesicles secreted by diverse cell types, exosomes can encapsulate and transport a wide range of bioactive cargos, including nucleic acids, proteins, lipids, and small-molecule drugs, thereby serving as efficient mediators of intercellular communication and attractive delivery platforms for therapeutic applications [21]. Notably, exosomes have shown particular promise for central nervous system drug delivery because evidence indicates that they are capable of interacting with and, to some extent, traversing the BBB. Although some studies have shown the majority of exosomes bypassed the brain microvascular endothelial cells (BMECs) through transcellular route, followed by endocytosis, endosomal multivesicular bodies formation and exocytosis steps, the precise mechanisms of interaction between exosomes and BBB are still not elucidated and need further investigation [22, 23]. However, despite these advantages, native exosomes administered systemically are still prone to rapid sequestration by the reticuloendothelial system, resulting in limited accumulation in brain tissue [24–26]. Among them, milk-derived exosomes (MExo) are abundant, cost-effective, and highly biocompatible, and have been successfully used to deliver nucleic acids [27–29], protein [30], and small molecules [31]. To further improve BBB penetration and brain-targeted delivery, Angiopep-2 (ANG-2), a peptide that binds to low-density lipoprotein receptor-related protein 1 (LRP-1), has been extensively explored to promote transport across the BBB [32, 33]. Accordingly, ANG-2-modification of exosomes is expected to enhance brain-targeted drug delivery [34].

Based on the neuroprotective potential of SCU and the drug delivery advantages of MExo, we constructed an ANG-2-modified milk exosome-based drug delivery system (SCU@AMExo) (Fig. 1) to enhance the anti-ischemic stroke efficacy of SCU. In this system, SCU was encapsulated into MExo to improve its stability and systemic bioavailability, while surface modification with ANG-2 was introduced to facilitate LRP-1-mediated transport across the BBB. Following intravenous administration, SCU@AMExo is expected to efficiently traverse the BBB and preferentially accumulate in ischemic brain regions, thereby enhancing brain-targeted delivery of SCU. Once internalized by neural cells and microglia, SCU@AMExo may exert neuroprotective effects by attenuating oxidative stress, alleviating mitochondrial dysfunction and modulating neuroinflammatory responses. In particular, this delivery strategy is hypothesized to suppress excessive microglial activation, promote a shift toward an anti-inflammatory phenotype, and reduce inflammasome-related inflammatory signaling, ultimately contributing to neuronal protection and microenvironmental remodeling after ischemic injury. The aim of this study is to use engineered milk-derived exosomes as a brain-targeted carrier to enhance the brain delivery and anti-ischemic stroke efficacy of SCU.

Fig. 1.

Schematic illustration of the preparation of SCU@AMExo and therapeutic mechanisms of SCU@AMExo for ischemic stroke treatment

Results and discussion

Isolation and characterization of MExo

MExo were isolated from fresh milk using differential centrifugation and were further purified by a sucrose density gradient method (Fig. 2A). Dynamic light scattering (DLS) analysis showed that MExo exhibited a relatively uniform particle size distribution, with an average diameter of 137.3 nm and a polydispersity index (PDI) of 0.177 (Fig. 2B). The zeta potential of MExo was − 12.76 mV (Fig. 2H). To determine the concentration of MExo, nanoparticle tracking analysis (NTA) and a bicinchoninic acid (BCA) protein assay were performed, yielding a concentration of 4.04 × 10¹¹ particles per mg protein (Fig. 2C). Transmission electron microscopy (TEM) images revealed that MExo displayed a typical spherical vesicular morphology, with an approximately 100 nm diameter (Fig. 2D) [35]. In addition, Western blot analysis demonstrated that MExo expressed the typical exosome marker proteins HSP70, TSG101, and CD81, while showing little expression of the negative exosome marker protein calnexin (Fig. 2I), suggesting high purity of the isolated milk exosomes [36]. Collectively, these results confirm the successful isolation of high-purity milk-derived exosomes from fresh milk.

Fig. 2.

Isolation of MExo and Characterization of SCU@AMExo. (A) Schematic illustration of the isolation of MExo. (B) MExo determined by dynamic light scattering (DLS). (C) MExo determined by nanoparticle tracking analysis. (D) Representative TEM images of MExo. (E) SCU drug loading with different approaches (n = 3). (F) SCU encapsulation efficiency with different approaches (n = 3). (G) DLS size and PDI of different nanoparticles (n = 3). (H) Zeta potential of different nanoparticles (n = 3). (I) Western blot analysis of different nanoparticles, four proteins were detected, including Calnexin, HSP70, TSG101 and CD81. (J) The in vitro release profile of SCU (n = 3). Data are presented as mean ± SD; n = 3 means three independent experiments; ns, not significant, *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001

Preparation and characterization of SCU@AMExo

ANG-2-modified MExo (AMExo) were prepared by conjugating ANG-2 onto MExo using the crosslinker Maleimide-PEG3500-Succinimidyl Carboxymethyl Ester (Mal-PEG3500-SCM). To verify whether ANG-2 was successfully grafted onto MExo and to quantify the conjugation efficiency, rhodamine B-labeled ANG-2 was reacted with MExo. Then, the fluorescence intensity of rhodamine B was measured, and the amount of grafted ANG-2 was indirectly calculated based on a rhodamine B standard curve (Figure S1). The results indicated that approximately 321 ANG-2 molecules were conjugated to each MExo.

Subsequently, three different loading strategies, including sonication, stirring, and emulsification, were evaluated for SCU loading based on drug loading content and encapsulation efficiency. Sonication achieved the highest drug loading content and encapsulation efficiency (30.8% and 43.57%, respectively), followed by stirring (24.3% and 30.3%, respectively). However, considering that sonication may compromise the stability of the exosomal membrane, the stirring method was ultimately selected for SCU loading (Fig. 2E, F).

The finally prepared AMExo loaded with SCU was designated as SCU@AMExo. In parallel, SCU-loaded MExo without ANG-2 modification (SCU@MExo) were prepared as a control. The average particle size of SCU@AMExo was 165.2 nm, larger than that of MExo, possibly due to the existence of PEG3500 that increased the hydrodynamic diameter. Nevertheless, the PDI remained below 0.2, indicating good particle size uniformity (Fig. 2G). The zeta potential of SCU@AMExo was higher than that of MExo, possibly due to the positive charge of ANG-2 in the phosphate-buffered saline (PBS, pH 7.4) reaction medium (Fig. 2H), which also supported the successful conjugation of ANG-2 indirectly. SCU@AMExo also retained abundant exosome-related proteins, suggesting that drug loading and peptide modification had a negligible impact on exosome integrity (Fig. 2I).

The release behavior of SCU was evaluated in artificial cerebrospinal fluid. The cumulative release reached 78.9% within the first 8 h, which could ensure rapid drug action after reaching the lesion (Fig. 2J). Stability studies showed that both MExo and SCU@AMExo exhibited little change in particle size after storage in PBS at 4 °C for one week, with the PDI consistently remaining below 0.3, indicating uniform dispersion without apparent aggregation or degradation and thus good stability (Figure S2 and S3).

In vitro BBB-penetrability and targeting ability

Previous studies have demonstrated that ANG-2 can be internalized by brain capillary endothelial cells via the LRP-1. Therefore, the mouse brain microvascular endothelial cell line (bEnd.3) was selected for cellular uptake studies. The bEnd.3 cells were co-incubated for 2 h with coumarin-6-loaded MExo (C6@MExo) or coumarin-6-loaded AMExo (C6@AMExo) at different concentrations. As shown in Fig. 3A, the uptake of both nanoparticles by bEnd.3 cells appeared to be concentration-dependent. With the increase of coumarin-6 concentration, the intracellular green fluorescence was significantly enhanced. Notably, at the same concentration, the C6@AMExo group displayed visibly stronger green fluorescence compared with that of the C6@MExo group. Semi-quantitative analysis (Fig. 3B) revealed a consistent trend, showing that the cellular uptake of C6@AMExo was about twice that of the C6@MExo group at identical concentrations. These results indicate that ANG-2 modification could markedly promote the internalization of MExo by bEnd.3 cells and facilitate their intracellular distribution, which lays a foundation for improved BBB penetration.

Fig. 3.

In vitro BBB penetration and neuroprotective efficacy of SCU@AMExo. (A) Representative fluorescence microscopy images about cellular uptake behaviors of different C6-labeled nanoparticles in bEnd.3 cells. (Scale bars:100 μm) (B) Semi-quantification of cellular uptake for C6-labeled nanoparticles in bEnd.3 cells (n = 3). (C) Schematic diagram of the in vitro BBB model. (D) In vitro BBB permeability in transwell model (n = 5). (E) Cell viability of PC12 cells with OGD/R injury after treatment with different concentrations of SCU@AMExo (n = 5). (F) Live/Dead staining of PC12 cells with OGD/R injury following different treatments (Scale bars:100 μm). (G) The Semi-quantitative analysis of the ratio of live cells in live/dead cell staining (n = 3). Data are presented as mean ± SD; n = 3, n = 5 means three or five biological replicates; ns, not significant, *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001

To further verify the ANG-2-mediated BBB transport, an in vitro BBB model was established using a transwell system with bEnd.3 cells(Fig. 3C) [37]. C6@MExo or C6@AMExo was co-incubated with bEnd.3 cells in the upper chamber, and the extent of BBB penetration was evaluated by measuring the fluorescence intensity in the lower chamber. At all examined time points, the C6@AMExo group exhibited higher BBB permeability than the unmodified C6@MExo group. After 8 h, the BBB penetration rate of C6@AMExo reached 11.09%, which was twice that of the C6@MExo group (Fig. 3D). These findings demonstrate that in vitro AMExo could not only be efficiently internalized by bEnd.3 cells but also traverse the BBB, suggesting their potential capability to penetrate the BBB in vivo.

In vitro neuroprotective effects

An Oxygen-Glucose Deprivation/Reperfusion (OGD/R) model was established in PC12 cells to mimic stroke-related neuronal damage [38]. After 6 h of OGD followed by 24 h of reoxygenation, cell viability decreased to approximately 50%, and this concentration was therefore selected for subsequent experiments (Figure S4). As shown in Fig. 3E, SCU@AMExo exhibited a concentration-dependent neuroprotective effect in injured PC12 cells. At the highest concentration, SCU@AMExo increased cell viability from 48.1% to 92.9%, indicating a strong capability to rescue neuronal injury.

The neuroprotective effects of different formulations were further visualized using live/dead cell staining. As shown in Fig. 3F, live cells were stained green, whereas dead cells were stained red. The control group displayed intact cellular morphology with strong green fluorescence and minimal red fluorescence. In contrast, the OGD/R group exhibited intense red fluorescence, indicating severe neuronal damage. The free SCU group showed reduced red fluorescence compared with the OGD/R group, suggesting a certain protective effect. Notably, the SCU@AMExo group showed green fluorescence intensity closer to that of the control group, accompanied by relatively low red fluorescence and a generally preserved cellular morphology. The Semi-quantitative analysis showed that the SCU@AMExo group increased the ratio of live cells to 92.4%, significantly higher than that in the saline group (45.1%). These results indicate that compared with free SCU, SCU@AMExo has a stronger neuroprotective effect, possibly because MExo protects SCU by reducing degradation and enhancing cellular uptake.

Neuroprotective mechanisms

Previous studies have demonstrated that SCU exerts pronounced antioxidant effects in various models of neuronal injury and cardio-cerebrovascular diseases by scavenging excessive ROS, improving mitochondrial function and regulating microglial polarization [14–18]. Based on these findings, we evaluated the effects of SCU@AMExo on oxidative stress in LPS-induced PC12 cells. Under LPS stimulation, PC12 cells exhibited a pronounced oxidative stress response. As revealed by DCFH-DA staining, the LPS group displayed the strongest green fluorescence, indicating substantial ROS accumulation. Treatment with free SCU moderately reduced ROS levels, whereas the SCU@AMExo group showed a markedly attenuated green fluorescence signal, suggesting a superior ROS-scavenging capacity (Fig. 4A, B). Given the close association between oxidative stress and mitochondrial dysfunction, we further examined changes in mitochondrial membrane potential in LPS-injured PC12 cells. JC-1 staining results highly consistent with the ROS findings. Cells in the control group showed mainly red fluorescence with an intact mitochondrial membrane potential. Conversely, LPS treatment led to reduced red fluorescence and increased green fluorescence, indicating mitochondrial depolarization. Treatment with free SCU partially alleviated this alteration. Remarkably, cells treated with SCU@AMExo exhibited a more pronounced recovery of red fluorescence with relatively weaker green fluorescence, indicating a more effective restoration of mitochondrial membrane potential (Fig. 4C-E). Taken together, these findings suggest that the protective effect of SCU@AMExo on mitochondrial function might be related, at least in part, to its ability to alleviate oxidative stress. Given that excessive ROS has been reported to impair mitochondrial integrity and promote further ROS generation [39], reduced oxidative stress may contribute to the improved mitochondrial status observed after SCU@AMExo treatment.

Fig. 4.

Cellular mechanisms underlying SCU@AMExo-mediated neuroprotection. (A) Intracellular ROS generation in PC12 cells detected by DCFH-DA staining under different treatments. (Scale bars: 100 μm) (B) Semi-quantitative analysis of DCFH fluorescence intensity normalized to protein content. (Scale bars: 100 μm) (C) Mitochondrial membrane potential (ΔΨm) in PC12 cells evaluated by JC-1 staining. Red fluorescence indicates JC-1 aggregates (intact ΔΨm), while green fluorescence represents JC-1 monomers (depolarized mitochondria). (D, E) Semi-quantitative analysis of JC-1 aggregate (D) and JC-1 monomers (E), respectively. (F) Immunofluorescence staining of BV2 microglial cells showing M1 polarization marker CD16/32(red) and M2 polarization marker CD206(green); nuclei were counterstained with DAPI (blue). (Scale bars: 25 μm) (G) Semi-quantitative analysis of the CD206/CD16/32 ratio, reflecting microglial polarization status. Data are presented as mean ± SD; ns, not significant, *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001

In addition, the regulatory effect of SCU@AMExo on inflammatory phenotypes was further investigated in BV2 microglial cells. Immunofluorescence analysis revealed that LPS stimulation markedly increased the expression of the M1 polarization marker CD16/32 while significantly reducing the expression of the M2 polarization marker CD206. Treatment with free SCU partially reversed this trend by enhancing CD206 expression. In contrast, the SCU@AMExo group exhibited the strongest green fluorescence signal and the highest CD206 expression, indicating a robust ability to promote microglial polarization toward the M2 phenotype. These findings suggest that SCU@AMExo exerts a more potent immunomodulatory effect on microglial polarization than free SCU (Fig. 4F, G).

In vivo targeting and biodistribution

To investigate the in vivo biodistribution and brain-targeting capability of SCU@AMExo, equal doses of Dir-labeled MExo and AMExo were intravenously administered to C57 mice subjected to tMCAO/R surgery. The systemic distribution and brain accumulation of the formulations were monitored by in vivo imaging system (IVIS). As shown in the real-time fluorescence images and corresponding semi-quantitative analysis (Fig. 5A, B), the fluorescence intensity in the brains of mice treated with Dir@AMExo was higher than that in the Dir@MExo group at all observed time points. This enhanced brain accumulation was attributed to ANG-2-mediated BBB penetration, which corresponded with the results of in vitro BBB penetration results. Although the fluorescence intensity gradually decreased after 8 h, Dir@AMExo still displayed a relatively strong signal in the brain, indicating sustained accumulation. Ex vivo imaging of the brain and its semi-quantitative analysis further confirmed that the accumulation of Dir@AMExo in the brain was significantly higher than that of Dir@MExo (Fig. 5C, D). Moreover, in main organs other than the brain, Dir@AMExo exhibited reduced distribution, indicating a higher degree of accumulation in the brain (Fig. 5E). Collectively, Dir@AMExo possesses superior brain-targeting capability compared with Dir@MExo, which is in good agreement with the in vitro findings.

Fig. 5.

Targeting effects and in vivo distribution of SCU@AMExo. IVIS imaging (A) and corresponding quantification fluorescence (B) of tMCAO/R mice after injection of Dir@MExo and Dir@AMExo at 2 h and 8 h (n = 3). IVIS imaging (C) and corresponding quantification fluorescence (D, E) of the ex vivo major organs of tMCAO/R mice after injection of Dir@MExo and Dir@AMExo at 8 h (n = 3). Data are presented as mean ± SD; n = 3 means three biological replicates; *P < 0.05, **P < 0.01, ***P < 0.001

In vivo therapeutic efficacy against ischemic brain injury

To assess the in vivo therapeutic efficacy of different formulations against ischemic brain injury, TUNEL staining was performed to evaluate apoptotic cell death. A large number of TUNEL-positive cells were observed in the ischemic regions of saline-treated mice, whereas treatment with SCU or SCU@MExo reduced apoptosis to varying extents. Among all groups, SCU@AMExo led to the most pronounced decrease in TUNEL-positive cells, indicating a strong inhibitory effect on ischemia-induced neuronal apoptosis (Fig. 6A). The 2,3,5-triphenyltetrazolium chloride (TTC) staining results revealed extensive infarction in the ischemic hemisphere of saline-treated mice, with an average infarct volume of 51.06%. Treatment with free SCU or SCU@MExo reduced the infarct volume to 30.32% and 22.45%, respectively. The improved efficacy observed in the SCU@MExo group compared with free SCU may be attributed to the protective effect of MExo against rapid drug degradation. SCU@AMExo achieved the greatest effect with the infarct volume decreased to 11.29%, significantly lower than that observed in the other treatment groups (Fig. 6B, C). This enhanced therapeutic benefit may result from the combined effects of SCU together with ANG-2-mediated BBB penetration. Consistently, neurological deficit scores indicated that mice treated with SCU@AMExo exhibited relatively milder functional impairment and a more pronounced trend toward neurological recovery than the other treatment groups. (Figure S5).

Fig. 6.

In vivo therapeutic efficacy and mechanisms of SCU@AMExo. (A) TUNEL staining (green) of brain sections showing apoptotic cells (nuclei are counterstained with DAPI in blue). (B) Representative images of TTC-stained brain sections from mice receiving different treatments. (C) Semi-quantitative analysis of cerebral infarct volume based on TTC staining (n = 3). (D) Semi-quantitative analysis of reactive oxygen species (ROS) levels, measured by DCFH-DA assay, in brain homogenate supernatants (n = 4). Scale bar: 40 μm. (E-H) Western blot analysis (E) and corresponding semi-quantification (F-H) of protein expression levels in brain tissues (n = 3). (I) Representative immunofluorescence images showing microglial polarization in brain sections. Microglia are stained with Iba-1(red); M2 phenotype markers with CD206(green) and M1 phenotype markers with CD16/32(green); nuclei are stained with DAPI (blue). (J) TNF-α and (K) IL-10 levels in mouse brain homogenates measured by ELISA (n = 4). Data are presented as mean ± SD; ns, not significant, *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001

In vivo therapeutic mechanisms against ischemic brain injury

Based on our previous in vitro findings showing effective ROS scavenging by SCU@AMExo, intracellular ROS levels were further measured in mouse brain homogenates. Compared with the sham group, saline-treated mice displayed markedly elevated ROS levels in brain tissue. SCU@AMExo treatment significantly decreased ROS accumulation, suggesting attenuation of oxidative stress in vivo (Fig. 6D). Since ROS is recognized as an important contributor to NLRP3 inflammasome activation [40] and SCU has been reported to inhibit the activization of NLRP3 [41], we next examined the expression of proteins involved in the NLRP3-caspase-1-IL-1β pathway by Western blotting. The levels of NLRP3, caspase-1, and mature IL-1β were markedly increased after ischemia-reperfusion in saline-treated mice, whereas SCU@AMExo treatment reduced the expression of these proteins (Fig. 6E-H). These findings suggest that SCU@AMExo might suppress inflammasome-related signaling, possibly in association with its effect on reducing oxidative stress. In line with this, previous studies have shown that inhibition of NLRP3 activation can reduce infarct volume, alleviate neurovascular injury, and improve outcomes after cerebral ischemia [42].

Microglia are a type of highly plastic immune cells in the brain, which are activated during ischemia-reperfusion injury. There are two opposite phenotypes of microglia: pro-inflammatory M1-like microglia and anti-inflammatory M2-like microglia. Their effects on brain function recovery are completely different. M1-type microglia exacerbate brain injury after ischemic stroke by releasing pro-inflammatory cytokines, while M2-type microglia rescue local neuroinflammation by releasing anti-inflammatory cytokines and promote brain repair. Therefore, the polarization of microglia from M1 type to M2 type plays a crucial role in neuroprotection after ischemic stroke [43]. The influence of SCU@AMExo on microglial polarization was then investigated. Immunofluorescence analysis revealed a clear increase in Iba-1 expression within ischemic brain regions following MCAO/R, indicating pronounced microglial activation. Relative to the saline group, both SCU@MExo and SCU@AMExo markedly reduced the proportion of M1-type microglia (Iba-1⁺/CD16/32⁺) while increasing M2-type microglia (Iba-1⁺/CD206⁺). This regulatory effect was most evident in the SCU@AMExo group (Fig. 6I). Although free SCU exhibited a modest ability to promote M2 polarization, its effect was considerably weaker than that achieved with exosome-mediated delivery.

Because distinct microglial phenotypes secrete different inflammatory mediators, levels of TNF-α and IL-10 in brain tissue were further quantified. Saline-treated mice exhibited a typical pro-inflammatory profile, characterized by elevated TNF-α and reduced IL-10. In contrast, SCU@AMExo significantly decreased TNF-α expression while enhancing IL-10 production (Fig. 6J, K), further supporting its role in suppressing M1 polarization and favoring an M2 phenotype. Taken together, these findings indicate that SCU@AMExo markedly improves the inflammatory microenvironment within ischemic lesions by regulating microglial polarization and cytokine balance.

Behavioral assessment of ischemic stroke mice by SCU@AMExo treatment

Ischemic stroke frequently results in motor deficits, which are largely attributable to damage within the motor network, including cortical motor regions [44]. Affected mice typically show reduced spontaneous movement, poorer coordination and balance, and increased anxiety-like behaviors [45, 46]. Behavioral tests are therefore valuable tools for assessing functional recovery. Here, we used the open-field test and rotarod test to evaluate whether different SCU formulations could improve behavioral outcomes after stroke.

Mice received a total of seven doses following tMCAO/R surgery. Behavioral performance was assessed at three time points: 24 h, day 8, and day 21 post-stroke (Fig. 7A). The open-field test was performed during both the acute (24 h) and recovery (day 8) phases. At the acute stage, SCU@AMExo treatment already improved most behavioral measures compared to saline controls (Figure S6), suggesting a rapid therapeutic effect. During the second assessment (Fig. 7B-F), saline-treated mice displayed clear deficits: their movement paths were restricted, and they traveled less, moved more slowly, and entered the center zone less often, indicating impaired mobility and heightened anxiety. Treatment with free SCU or SCU@MExo resulted in modest improvements, but mice treated with SCU@AMExo performed much like sham controls, with no significant differences in distance, speed, or center entries. We further assessed coordination and balance using the rotarod test (Fig. 7G, H). Sham-operated mice stayed on the rod longer and completed more rotations. Saline-treated mice fell sooner and performed poorly on all measures. Although free SCU and SCU@MExo slightly extended running time, only SCU@AMExo brought performance close to sham levels, significantly improving latency to fall, total rotations, and distance traveled. Collectively, these behavioral tests demonstrate that SCU@AMExo promotes clear functional recovery after ischemic stroke, enhancing both spontaneous activity and coordinated movement. Notably, therapeutic benefits were detectable after a single dose and became more pronounced with repeated treatment. These results indicate that SCU@AMExo effectively restores normal exploratory behavior and reduces post-stroke anxiety.

Fig. 7.

Animal experimental protocols and behavioral assessment. (A) Schematic illustration of the experimental timeline, including tMCAO surgery, drug administration schedule, and behavioral assessments. Open-field tests were performed during the acute and recovery phases, and the rotarod test was conducted at the late recovery stage. (B) Representative motive path of mouse activity (C). Total distance traveled in the open-field test (n = 8). (D)Mean speed traveled in the open-field (n = 8). (E) Number of entries into the center zone in the open field test (n = 8). (F) Time spent in the center zone in the open field test (n = 8). (G) Latency to first fall from the rotating rod in the rotarod test (Sham and SCU@AMExo groups: n = 8; Saline group: n = 6; SCU and SCU@MExo groups: n = 7). (H) Total number of complete revolutions performed in the rotarod test (Sham and SCU@AMExo groups: n = 8; Saline group: n = 6; SCU and SCU@MExo groups: n = 7). Data are presented as mean ± SD; ns, not significant, *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001

Study on molecular mechanisms

To investigate the molecular mechanism of SCU@AMExo in promoting post-stroke recovery, RNA sequencing (RNA-seq) was performed on the ischemic hemisphere of tMCAO/R mice 72 h after treatment. A total of 2036 differentially expressed genes (DEGs) were identified between Sham and Saline groups, and 557 DEGs between Saline and SCU@AMExo groups (Fig. 8A-C).

Fig. 8.

RNA-seq analysis of the ischemic hemispheres after different treatments (Sham, Saline, SCU@AMExo). (A) Summary of differentially expressed genes (DEGs) in pairwise comparisons (Saline vs. Sham, SCU@AMExo vs. Saline, SCU@AMExo vs. Sham), with “up” and “down” indicating upregulated and downregulated DEGs respectively. (B) Volcano plot of DEGs in the Saline vs. Sham comparison; “no-DEGs” denotes non-differentially expressed genes. (C) Volcano plot of DEGs in the SCU@AMExo vs. Saline comparison. (D) KEGG enrichment analysis (top 20 pathways) for the Saline vs. Sham comparison; the dot size represents gene number, color represents Q value, and the x-axis represents Rich Factor. (E) KEGG enrichment analysis (top 20 pathways) for the SCU@AMExo vs. Saline comparison. (F) GO term enrichment analysis for the SCU@AMExo vs. Saline comparison. (G) Heatmap of clustered DEGs across different treatment groups (Sham, Saline, SCU@AMExo) and replicates (n = 3)

Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of DEGs between groups showed that pathways enriched in the Saline group compared to the Sham group were primarily related to inflammation and immune responses, including cytokine-cytokine receptor interaction, chemokine signaling, TNF signaling, ECM-receptor interaction, and complement and coagulation cascades (Fig. 8D). These findings are consistent with the pronounced immune activation and inflammatory amplification during the early phase of stroke. In contrast, DEGs in SCU@AMExo-treated mice were significantly enriched in pathways closely associated with oxidative stress and inflammation, such as the AGE-RAGE signaling, NOD-like receptor signaling and PI3K-Akt signaling pathways (Fig. 8E). These transcriptomic changes are generally consistent with our experimental observations that SCU@AMExo exerts neuroprotective effects through multiple pathways. Moreover, Gene Ontology (GO) enrichment analysis revealed that significantly enriched terms between the compared groups primarily involved biological processes such as immune system process, inflammatory response, response to stress, cytokine regulation, extracellular matrix reorganization, injury repair, and defense response (Fig. 8F), underscoring the broad regulatory role of SCU@AMExo on the post-stroke inflammatory cascade. Hierarchical clustering analysis further illustrated trends in gene expression patterns (Fig. 8G). A clear separation in expression profiles was observed between the Saline and Sham groups. Notably, samples from the SCU@AMExo group clustered relatively closer to those from the Sham group, suggesting that SCU@AMExo treatment may be associated with a partial reversal of stroke-induced transcriptional alterations. However, because the validation was restricted to a single protein target and no quantitative polymerase chain reaction(qPCR) or additional Western blot validation of representative DEGs was performed, the validation of RNA-seq findings in the present study is limited, so these transcriptomic observations should be interpreted cautiously and regarded as exploratory.

Preliminary biosafety assessment

To systematically evaluate the biosafety of SCU@AMExo, its effects on cell viability were first examined in bEnd.3 cells. As shown in Fig. 9A, SCU@AMExo did not exhibit any obvious cytotoxicity over a concentration range of 10–320 µg/mL. Even at 320 µg/mL, the highest concentration tested, cell viability remained relatively high, indicating the favorable cytocompatibility of the formulation. Subsequently, an in vivo safety assessment was conducted in healthy mice following administration every other day for 14 days. During the treatment period, mice in the SCU@AMExo group exhibited body weight gain curves comparable to those of the control group, with no evidence of weight loss or growth retardation (Fig. 9B), suggesting that SCU@AMExo does not adversely affect overall physiological status. Histopathological examination further confirmed the favorable biosafety profile of SCU@AMExo. Hematoxylin and eosin (H&E) staining revealed no observable abnormalities, such as inflammatory cell infiltration, tissue damage, or necrosis, in major organs, including the brain, heart, liver, spleen, lung, and kidney (Fig. 9C). The cellular arrangement, tissue morphology, and structural integrity in the SCU@AMExo-treated group were comparable to those in the saline group, indicating minimal organ toxicity. In addition, blood biochemical analyses showed no significant differences between the two groups in renal function-related parameters, including creatinine (CREA) and blood urea nitrogen (BUN) (Fig. 9D, E), or in liver function-related markers, including aspartate aminotransferase (AST) and alanine aminotransferase (ALT) (Fig. 9F, G), suggesting that SCU@AMExo does not impose a burden on hepatic or renal function. Moreover, the levels of immunoglobulin G (IgG) and inflammatory cytokines, including TNF-α and IL-10, remained unchanged (Fig. 9H-J), indicating that SCU@AMExo does not induce systemic inflammatory responses or disrupt basal immune homeostasis.

Fig. 9.

Biosafety evaluation of different nanoparticles. (A)Cell viability of bEnd.3 after incubation with different concentrations of SCU@AMExo (n = 6). (B) Evaluation of body weight of mice after treatment with saline or SCU@AMExo (n = 5). (C)Typical images of H&E staining of main organs in SCU@AMExo group and saline group mice (Scale bars: 50 μm). Evaluation of CREA (D), BUN (E), AST (F), ALT (G), IgG (H), TNF-α (I), IL-10 (J) after treatment with saline or SCU@AMExo (n = 5). Data are presented as mean ± SD; ns, not significant

Taken together, SCU@AMExo exhibits excellent biosafety both in vitro and in vivo, characterized by good cytocompatibility, absence of major organ toxicity, minimal impact on liver and kidney function, and no excessive activation of the immune system. Combined with its pronounced neuroprotective effects in ischemic stroke models, SCU@AMExo shows strong potential for further development in the treatment as a therapeutic platform for neurological disorders and provides a solid safety foundation for its long-term in vivo application as a nanocarrier-based therapeutic system.

Conclusion

In this study, an engineered exosome-based SCU loaded drug delivery system capable of crossing the BBB was developed for ischemic stroke therapy. Encapsulation of SCU improved its stability and enabled effective delivery to the brain parenchyma. In vitro results showed that SCU@AMExo was readily internalized by bEnd.3 cells and efficiently traversed an in vitro BBB model, thereby enhancing viability of injured PC12 cells.

Following intravenous administration in animal models, SCU@AMExo crossed the BBB and accumulated in ischemic brain regions, leading to an attenuation of neuronal ischemia-reperfusion injury. This protective effect was associated with reduced ROS accumulation, improved mitochondrial membrane potential, attenuation of NLRP3 inflammasome-related signaling, and a shift in microglial polarization from the M1 to the M2 phenotype, accompanied by decreased secretion of pro-inflammatory cytokines and partial improvement of the post-ischemic microenvironment.

These effects were associated with improved neurological outcomes in tMCAO mice, including increased spontaneous activity, reduced anxiety-like behaviors after stroke, and improved motor coordination and endurance during long-term recovery. Meanwhile, SCU@AMExo exhibited favorable biosafety in both in vitro and in vivo evaluations. Taken together, these findings indicate that the exosomal carrier successfully enhanced the anti-stroke efficacy of SCU by improving its stability, BBB penetration, and accumulation in ischemic brain tissue, highlighting exosome-mediated delivery as a promising strategy for the brain-targeted application of natural compounds in ischemic stroke therapy. However, validation of the RNA-seq findings was limited in the present study, and therefore these transcriptome-level observations should be interpreted cautiously and regarded as exploratory rather than definitive mechanistic evidence. Moreover, the absence of a direct quantitative comparison with other reported BBB-targeted exosomes or brain-targeted nanoplatforms under identical experimental conditions remains another limitation and warrants further investigation.

Materials and methods

Materials

Color Prestained Protein Marker, SCU, and 8% Paraformaldehyde Fix Solution were purchased from Adamas (Shanghai, China). Fresh milk was obtained from Junlebao Dairy Group (Hebei, China). ANG-2 was sourced from Qiangyao Biotechnology (Shanghai, China). Rhodamine B and PVDF membrane (0.22 μm) were obtained from Merck KGaA (Darmstadt, Germany). DAPI staining solution (C0066) was obtained from Solarbio Science & Technology (Beijing, China). The Calcein/PI Cell Viability/Cytotoxicity Assay Kit, Reactive Oxygen Species Assay Kit, Mitochondrial Membrane Potential Assay Kit with JC-1 and The One-Step TUNEL Apoptosis Assay Kit were purchased from Beyotime Biotechnology (Shanghai, China). ELISA kits for TNF-α (RK00027), and IL-10 (RK00016) were obtained from ABclonal Technology (WuHan, China).

Antibodies against CD206, CD81, TSG101, HSP70, Calnexin, and β-actin were purchased from Proteintech Group (Wuhan, China). Antibodies against NLRP3, CD16/32, and Cleaved-Caspase-1 were from Cell Signaling Technology (Danvers, MA, USA). The IL-1β antibody was obtained from Immunoway Biotechnology Company (Plano, TX, USA). HRP-conjugated secondary antibodies, 5× SDS-PAGE Loading Buffer, and RIPA Lysis Buffer were sourced from Beyotime Biotechnology (Shanghai, China). SDS-PAGE Precast Gels were purchased from ACE Biotechnology (Changzhou, China).

DMEM culture medium and fetal bovine serum (FBS) were purchased from Biochannel (Nanjing, China). The bEnd.3, PC12, and BV2 cell lines were acquired from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China).

Isolation and characterization of MExo

Exosomes were isolated by differential centrifugation of fresh milk. Briefly, fresh milk was centrifuged sequentially at 2,000 × g for 20 min 5,000 × g for 30 min, and 10,000 × g for 90 min to remove casein and lipids. The resulting supernatant was filtered through a 0.45 μm membrane and ultracentrifuged at 100,000 × g for 1 h. The pellet was collected and re-suspended in PBS. For further purification, the suspension was layered onto a sucrose density gradient (30%, 45%, 60%) and ultracentrifuged at 100,000 × g for 2 h. The interface between the 30% and 45% layers was collected, diluted with PBS, and washed by ultracentrifugation at 100,000 × g for 1 h. The final pellet was resuspended in PBS to obtain the MExo stock solution, which was stored at − 80℃ for use.

The hydrodynamic size distribution and zeta potential of MExo were measured by dynamic light scattering (DLS) with a Zetasizer (Nano-ZS90; Malvern Instruments, Malvern, UK). The particle size distribution and concentration were determined by nanoparticle tracking analysis (NTA) using a NanoSight (NS300; Malvern Instruments). To confirm MExo morphology, MExo were negatively stained with uranyl acetate and deposited onto copper grids. Their morphology was then observed using a transmission electron microscope (TEM; JEM 2100 F, JEOL Ltd., Tokyo, Japan).

Encapsulation and loading efficiency of SCU in AMExo

SCU was loaded into exosomes using three distinct physical and chemical methods. For each protocol, 10 mg SCU was incubated with 10 mg exosomes (by protein content) in 20 mL PBS. Sonication: The mixture was subjected to water-bath sonication for 30 s using a cycle of 2 s on /2 s off to prevent overheating. Incubation: The mixture was incubated at room temperature for 30 min under continuous gentle agitation using a magnetic stirrer. Saponin-assisted: A trace amount of saponin was added to the mixture, followed by co-incubation for 30 min at room temperature with gentle stirring. Following the loading process, unencapsulated SCU was removed by ultrafiltration using a 100-kDa molecular weight cut-off (MWCO) centrifugal filter. The amount of SCU encapsulated within the exosomes was quantified to calculate the drug loading capacity (DLC) and encapsulation efficiency (EE). A standard HPLC curve of pure SCU was used for quantification. The DLC and EE were calculated according to the following equations:

Preparation and characterization of SCU@MExo and SCU@AMExo

Preparation: MExo was conjugated with ANG-2 using Mal-PEG3500-SCM. Briefly, Mal-PEG3500-SCM (840 µg) was dissolved in 200 µL of DMF and added to MExo containing 10 mg of total protein. The mixture was stirred in PBS buffer (pH 7.4) for 18 h. ANG-2 (500 µg) was subsequently introduced, and the reaction was allowed to proceed for an additional 4 h. Nitrogen protection was applied throughout the process. Upon completion, Ang-II-grafted MExo (AMExo) was obtained. SCU (10 mg) was then added to the AMExo suspension and incubated under gentle stirring for 30 min to enable passive diffusion of the drug into the vesicles. The reaction mixture was transferred to a 100-kDa MWCO ultrafiltration device and centrifuged at 5,000 × g for 10 min to remove unencapsulated SCU, residual Mal-PEG3500-NHS, and free ANG-2, then SCU@AMExo was obtained.

Characterization: Size and zeta potential of SCU@AMExo and SCU@MExo were analyzed using dynamic light scattering (DLS) with a Zetasizer (Nano-ZS90; Malvern Instruments, Malvern, UK).

Western blot analysis

To verify the exosomal protein markers of MExo and to assess potential changes after drug loading, western blot analysis was performed for the negative exosomal marker calnexin and the positive markers HSP70, TSG101, and CD81. MExo, SCU@MExo, and SCU@AMExo samples were lysed using RIPA buffer and centrifuged at 12,000 rpm for 5 min to obtain protein extracts. Protein concentrations were determined using a BCA assay kit.

Protein samples were mixed with SDS-PAGE loading buffer at a 1:5 dilution and heated at 100 °C for 5 min to ensure complete denaturation. Equal amounts of protein (40 µg per lane) were separated by SDS-PAGE at 120 V for 100 min and subsequently transferred onto PVDF membranes at 250 mA. The membranes were blocked with rapid blocking buffer and incubated with primary antibodies at 4 °C overnight. Afterward, the membranes were incubated with HRP-conjugated secondary antibodies at room temperature for 2 h. Protein bands were visualized using a gel imaging system (Tanno).

In vitro drug release of SCU@AMExo

The in vitro release profile of SCU from SCU@Exos was assessed using a dialysis method. Briefly, a sample containing 1 mg of encapsulated SCU was placed into a dialysis bag (MWCO: 3,000 Da) and immersed in 200 mL of artificial cerebrospinal fluid (aCSF, pH 7.4). The release system was maintained at 37 °C with gentle agitation. At predetermined time intervals, samples of the external medium were withdrawn, and the released SCU concentration was quantified by HPLC. The cumulative release from formulations prepared by the three loading protocols was plotted and compared.

In vitro uptake by bEnd.3 cells

The bEnd.3 cells were seeded in 24-well plates (2 × 10⁴ cells per well) and cultured for 24 h. C6@ AMExo and C6@ MExo were diluted with serum-free DMEM medium into solutions with coumarin-6 concentrations of 1, 2 and 4 µg/mL, and co-incubated with the cells at 37 °C for 2 h. After incubation, the cells were washed with PBS, fixed with 4% paraformaldehyde solution, and the cell nuclei were stained with DAPI. Cellular uptake was observed using an inverted fluorescence microscope. For the quantitative investigation of uptake, the cells were lysed with RIPA after incubation, and the supernatant was centrifuged at 12,000 rpm for 20 min. The fluorescence intensity was detected at 504 nm with a microplate reader, and the protein concentration in the supernatant was measured with a BCA kit.

Assessment of BBB permeability

The bEnd.3 cells were plated in Transwell inserts at a density of 1 × 10⁴ cells per insert and maintained in 24-well plates under standard culture conditions for 7 days. The culture medium in both the apical and basolateral compartments was then replaced with DMEM supplemented with hydrocortisone (1 µM). After additional 3 days of culture, the endothelial cells formed a tightly connected confluent monolayer. Monolayer cell integrity was assessed by measuring transepithelial/transendothelial electrical resistance (TEER) using a Millicell@-ERS voltmeter. Monolayer cells with TEER values higher than 200 Ω·cm² were used as blood-brain barrier models for subsequent studies. Then, C6@ MExo or C6@ AMExo (10 µg/mL) prepared by DMEM medium were added into upper chambers and DMEM medium was added to lower chambers. Then it was transferred to a 37 °C incubator for incubation. The lower chamber culture medium was taken at 2 h, 4 h, and 8 h, and the absorbance was measured at the emission wavelength of 504 nm using a microplate reader to investigate the C6@AMExo in vitro BBB permeability.

OGD/R model establishment

PC12 cells were seeded in 96-well plates at 8 × 10³ cells/well and cultured overnight. For OGD/R induction, the culture medium was replaced with glucose-free medium, and the cells were maintained under hypoxic conditions (1% O₂, 5% CO₂, and 94% N₂) for 6 h. Subsequently, the cells were returned to normal culture conditions with complete medium for 24 h of reoxygenation.

In vitro neuroprotective effect

Cell Viability Assay: The PC12 cells were seeded in 96-well plates (5 × 10³ cells per well) and cultured for 24 h. Subsequently, the cells were OGD for 6 h, and then were returned to reoxygenation and normal glucose conditions, followed by treatment with different concentrations of SCU@AMExo diluted in DMEM for 24 h. Then Cell viability was then assessed using the Cell Counting Kit-8(CCK-8) assay.

Live/Dead Cell Staining Assay: The PC12 cells were seeded in 24-well plates (5 × 10⁴ cells per well) and cultured for 24 h. The cells were OGD for 6 h, then returned to reoxygenation with one of the following (1) DMEM (2) AMExo (3) free SCU or (4) SCU@AMExo. Free SCU and SCU@AMExo were administered at the same dose, corresponding to an equivalent free SCU concentration of 5 µg/mL. In all in vitro experiments in this study involving both free SCU and SCU@AMExo, this concentration was used. A Live/Dead staining solution (containing calcein-AM and propidium iodide (PI)) was then added at a dilution ratio of 1:1000, and the cells were incubated in the dark at room temperature for 30 min. After removal of the staining solution, live (green fluorescence) and dead (red fluorescence) cells were observed and photographed under a fluorescence microscope.

In vitro ROS scavenging activity of SCU@AMExo

PC12 cells were seeded into 24-well plates at a density of 3 × 10⁴ cells per well and cultured for 24 h. The cells were then treated for an additional 24 h with one of the following regimens: (1) 1 µg/mL LPS solution (2) LPS plus free SCU (3) LPS plus AMExo (4) LPS plus SCU@AMExo. Then, the medium was aspirated and the cells were incubated with DCFH-DA staining solution (diluted 1:1000) at 37 °C in the dark for 30 min. Subsequently, the staining solution was removed, and the cells were washed with PBS. Intracellular green fluorescence was observed under a fluorescence microscope. Bright-field images of the same fields were also captured.

Mitochondrial membrane potential detection

PC12 cells were seeded into 24-well plates at a density of 3 × 10⁴ cells per well and cultured for 24 h. The cells were then treated with the various formulations for an additional 24 h. Then, the medium was aspirated and the cells were incubated with JC-1 staining solution (diluted 1:1000) at 37 °C in the dark for 30 min. Subsequently, the staining solution was removed, and the cells were washed with PBS. Green or Red fluorescence was observed under a fluorescence microscope.

In vitro regulation of microglia polarization

The BV2 cells were seeded onto 18-mm diameter glass coverslips placed in 12-well plates(2 × 10⁴ cells per well)and cultured for 24 h. The cells were then treated with the various formulations for an additional 24 h. Subsequently, immunofluorescence staining was performed to detect the expression of M1 phenotype marker CD16/32 and M2 phenotype marker CD206. The procedure was as follows: Cells were fixed with 4% paraformaldehyde (PFA) at room temperature for 10 min, followed by blocking with a solution containing 5% bovine serum albumin (BSA) and 10% goat serum for 1 h at room temperature. The cells were then incubated overnight at 4 °C with primary antibodies against CD16/32 and CD206 diluted in the blocking solution. On the following day, fluorophore-conjugated secondary antibodies were added and incubated for 1 h at room temperature in the dark. After incubation, the cells were mounted with an anti-fluorescence quenching mounting medium containing DAPI. Fluorescence images were captured using an inverted fluorescence microscope.

Establishment of tMCAO mouse models

Animal-related research adhered strictly to the protocol (approval number: 2406047) permitted by the Institutional Animal Care and Use Committee of Nanjing Medical University. All animals were specific pathogen-free (SPF) grade and maintained in SPF animal areas. The transient middle cerebral artery occlusion/reperfusion (tMCAO/R) mouse model was generated using a standard intraluminal filament technique. Briefly, mice (about 20 g) were anesthetized and kept under stable anesthesia throughout the procedure. A midline neck incision was made, and the ipsilateral common carotid artery (CCA), external carotid artery (ECA), and internal carotid artery (ICA) were exposed and gently isolated. The CCA was temporarily secured with a loose slipknot. The distal segment of the ECA was permanently ligated, while a second loose ligature was placed on the proximal ECA, followed by a small arteriotomy between the two knots for filament insertion. Then, silicone-coated threads at the tip (Pingdingshan Yushun Biotechnology Co., Ltd., China) were inserted from ECA into ICA to block the middle cerebral artery (MCA) blood flow. One hour later, the threads were removed for blood reperfusion. The sham group performed the same anesthesia and artery exposure procedure without suture insertion.

In vivo targeting and distribution assay

The tMCAO model mice were administrated with Dir-labeled SCU@AMExo and SCU@MExo through the tail vein. At different time points post administration, the targeting and biodistribution of nanoparticles in mice were evaluated using an IVIS. After 8 h of injection, the mice were sacrificed, and main organs were extracted and imaged with IVIS. The quantification was conducted using Living Image software.

Behavioral tests

Animal grouping and treatment: Mice were randomly assigned to five groups: sham, saline, free SCU, SCU@MExo, and SCU@AMExo. All SCU-related treatments were administered at doses equivalent to 25 mg/kg of free SCU.

Open-field tests: the mice were placed in a monitoring chamber (50 cm ×50 cm ×40 cm) and their movements were recorded over a 5-min period. The total distance traveled, average speed, center entries, and center time were analyzed using Any maze software.

Rotarod test: For the rotating bar test, the mice were trained 3 times a day for 3 days before the formal assessment. The mice were placed in a separate compartment on the rod and the rotor bar was adjusted to accelerate uniformly from 10 rpm to 40 rpm over a 5-min period. The latency to fall were recorded.

Infarct volume assessment

The mice in each group were sacrificed 7 days after administration. The brain tissues were harvested, frozen in liquid nitrogen for 10s, and then cut into 1-mm-wide coronal sections. The slices were stained with 1%(w/v) TTC solution and left at 37℃ for about 20 min until the infarcted region and normal region showed a clear color difference. Images are captured by the camera, and the infarct area was quantified with Image J (v1.54p).

In vivo ROS scavenging activity of SCU@AMExo

The ROS levels in the ischemic brain hemisphere were measured using the fluorescent probe DCFH-DA. Briefly, brain tissues were homogenized in cold PBS after euthanasia. The supernatant was incubated with 10 µM DCFH-DA at 37 °C for 30 min in the dark. Fluorescence intensity (Ex/Em = 488/525 nm) was measured by a microplate reader and normalized to protein concentration.

TUNEL staining

For TUNEL staining, brain paraffin sections were dewaxed and hydrated, followed by antigen retrieval with proteinase K and permeabilization with 0.3% Triton X-100. Sections were then incubated in the dark with a TUNEL reaction kit (Beyotime, Shanghai, China) at 37 °C for 1 h. Finally, nuclei were counterstained with DAPI and sections were mounted for imaging under a fluorescence microscope.

Immunofluorescence staining

Immunofluorescence staining was performed on paraffin-embedded brain sections. Briefly, sections were first dewaxed and hydrated, followed by antigen retrieval using proteinase K. After blocking with 5% BSA for 1 h at room temperature, they were incubated overnight at 4 °C with primary antibodies. To assess macrophage polarization, sections were co-incubated with anti-Iba1 antibody in combination with either anti-CD206 (for M2 phenotype) or anti-CD16/32 (for M1 phenotype). The following day, sections were incubated with appropriate fluorescent secondary antibodies for 1 h at room temperature. Finally, nuclei were counterstained with DAPI, and sections were mounted with an anti-fade mounting medium. Images were captured using a fluorescence microscope.

Detection of inflammatory cytokines in the ischaemic region

After the mice were euthanized, the infarcted brain tissue was rapidly dissected. PBS was then added at a tissue weight (mg) to PBS volume (µL) ratio of 1:10. The mixture was homogenized using a tissue homogenizer to obtain a brain tissue homogenate. The homogenate was centrifuged at 12,000 × g for 20 min at 4 °C to collect the supernatant. The levels of TNF-α and IL-10 in the supernatant were quantified using ELISA kits.

RNA sequencing

Sham-operated mice were treated with SCU@AMExo, whereas tMCAO mice received either normal saline or SCU@AMExo twice before being euthanized. Approximately 100 mg of brain tissue from the infarct region was rapidly dissected, placed into RNase-free cryogenic tubes, and immediately snap-frozen in liquid nitrogen for one hour. RNA extraction and transcriptome sequencing were performed by LC-Bio, and all data analyses were conducted on the LC-Bio cloud platform.

Cytotoxicity of SCU@AMExo

The bEnd.3 cells were seeded into 96-well plates at a density of 5 × 10³ cells per well and routinely cultured for 24 h. Subsequently, the cells were treated with SCU@AMExo at various concentrations (0, 20, 40, 80, 160, and 320 µg/mL) for another 24 h. Cell viability was then assessed using the Cell Counting Kit-8(CCK-8) assay.

Safety evaluation in vivo

Healthy C57BL/6J mice were administered SCU@AMExo or saline via tail vein injections every other day for a total of seven treatments, while their body weight was monitored and recorded. Whole blood was collected 24 h after the final injection and centrifuged at 3000 × g for 20 min to separate the serum. Serum levels of aspartate aminotransferase (AST), alanine aminotransferase (ALT), urea nitrogen (BUN), and creatinine (CREA) were then analyzed using a fully automated biochemical analyzer. Meanwhile, serum levels of IgG, TNF-α, IL-10 were detected via ELISA kits. Furthermore, samples of the heart, liver, spleen, lung, and kidney were processed for staining with H&E to obtain histopathological sections. Data from the experiments were analyzed to investigate the in vivo biosafety of SCU@AMExo.

Statistical analysis

Statistical analyses were performed using GraphPad Prism software. All data are expressed as mean ± standard deviation (SD). Comparisons between groups were conducted using Student’s t-test or one-way ANOVA. The thresholds for statistical significance were defined as follows: ns, not significant, *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

J.N.X. and X.C. designed and carried out the experiments, acquired and analysed the data, and wrote the manuscript. H.L.L., Z.W.D., W.Z., K.F.S., Y.G. and N.S. carried out the experiments. S.N.L., T.Y.Z., Y.Y.X., S.J.H. established the tMCAO/R model. Z.W. and F.F.L. supervised the entire project. H.L.X. and W.L revised the manuscript and supervised the entire project.

Funding

This study was supported by the National Natural Science Foundation of China (82304414, China), Jiangsu Young Science and Technology Talent Support Program (JSTJ-2025-107), Top Talent Support Program for young and middle-aged people of Wuxi Health Committee (HB2023103, China), Wuxi Young Science and Technology Talent Support Program (TJXD-2024-107, China), WJPMF Research Special Fund (320.6750.2025-26-6).

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

All animal procedures conformed to the protocol (approval number: 2406047) permitted by the Institutional Animal Care and Use Committee of Nanjing Medical University.

Consent for publication

All authors consent for publication.

Competing interests

The authors declare no competing interests.