Introduction

Ischemic stroke is a common cerebrovascular disease associated with significant morbidity, disability, recurrence, and mortality rates, affecting 80% of stroke patients in China1. Current clinical management options for ischemic stroke are limited. Intravenous thrombolytic therapy, effective only when administered within 4.5 h of symptom onset, carries the risk of complications like cerebral hemorrhage2. Due to the low rate of intravenous thrombolytic therapy success (less than 3%)3, there is an urgent need for novel treatment strategies. Houshiheisan (HSHS) is a formula derived from the classic Traditional Chinese Medicine text, “The Essentials of the Golden Chamber”. Clinical studies have demonstrated that HSHS can enhance limb motor function, reduce neurological deficit scores in patients with cerebral ischemia, and address vascular risk factors such as hypertension and dyslipidemia4,5,6,7.

Ischemia and hypoxia are now understood to initiate angiogenesis. Angiogenesis primarily involves the proliferation, migration, and sprouting of vascular endothelial cells8,9,10. This process augments local blood flow, improves nutrient delivery to the ischemic site, and may also be associated with neurogenesis, a vital component of brain tissue repair11. However, endogenous angiogenesis induced by cerebral ischemia is a gradual process, and neovascular structures are often immature12. Consequently, stimulating angiogenesis following ischemic stroke can aid in restoring cerebral blood perfusion and thereby mitigating brain tissue injury. Endothelial progenitor cells (EPCs) exhibit characteristics of both endothelial cells and stem cells13. Extracellular vesicles derived from EPCs can promote angiogenesis and have been explored for the treatment of ischemic stroke, myocardial infarction, diabetes, and other conditions14,15.

Extracellular vesicles (EVs) are nanoscale lipid bilayer vesicles actively secreted by diverse cell types16,17. EVs encapsulate a variety of bioactive molecules, including nucleic acids, proteins, lipids, and carbohydrates. These vesicles can traverse the blood-brain barrier and serve as key mediators of intercellular communication18,19,20. Both naturally occurring and engineered EVs have demonstrated anti-inflammatory, anti-apoptotic, neuroprotective, and vasoprotective properties in the treatment of ischemic stroke, which are linked to the microRNAs they contain21,22,23,24,25,26,27. Among these microRNAs, miR-126 is particularly abundant in endothelial cells. miR-126 regulates endothelial cell proliferation, migration, and other functions by targeting genes such as PIK3R2, SPRED1, and SDF-1. This microRNA is intimately involved in regulating vascular inflammation, vascular integrity, and angiogenesis28,29,30,31.

Our previous research demonstrated that HSHS mitigated cerebral ischemic injury in stroke model rats and protected endothelial cells32,33. To explore the potential effects of EVsHSHS on angiogenesis following cerebral ischemia, we utilized an oxygen-glucose deprivation (OGD) model with primary rat microvascular endothelial cells (BMECs) to examine the impact of EVsHSHS on endothelial cell function. Additionally, we validated the effects of EVsHSHS on cerebral ischemic injury and angiogenesis using a permanent middle cerebral artery occlusion (pMCAO) rat model. The results confirmed that EVsHSHS attenuated cerebral ischemic injury and promoted cortical angiogenesis in rats via the miR-126/PIK3R2 pathway. These findings may offer novel insights into the mechanisms underlying Traditional Chinese Medicine formulas.

Results

EVsHSHS promote angiogenesis in BMECs after OGD injury

Isolated EVs were characterized using transmission electron microscopy (TEM), western blot (WB) analysis, and nanoparticle tracking analysis (NTA) (Fig. 1A, B). The concentration of EVsHSHS (2.61 × 1010 particles/mL) was approximately twofold higher than that of EVs secreted by fetal bovine serum (FBS)-cultured EPCs (EVsNC, 1.06 × 1010 particles/mL). We observed that BMECs readily internalized EVs in vitro (Supplementary Fig. 1). After assessing the effects of various concentrations of EVsHSHS on BMEC viability, we selected a concentration of 40 µg/mL, which demonstrated statistically significant differences, for subsequent experiments (Supplementary Fig. 2).

The OGD model was established as previously described32. Cell counting kit-8 (CCK-8) was employed to assess BMEC proliferative capacity. Cell viability was significantly elevated in the EVsHSHS group compared to the EVsNC group (Fig. 1C). Wound healing assay (Fig. 1D,E) and tube formation assay (Fig. 1F,G) revealed that EVsHSHS accelerated BMEC migration and promoted tube formation following OGD. However, EVs secreted by blank-serum-cultured EPCs (EVsBlank) did not exhibit statistically significant differences compared to the EVsNC group, suggesting that components of HSHS, rather than rat serum, are responsible for the observed effects on BMEC function. These findings indicate that EVsHSHS can repair BMEC function after OGD injury and promote angiogenesis.

Fig. 1
figure 1

EVsHSHS enhance the proliferation, migration, and tube formation of BMECs following OGD injury. (A) TEM was used to observe the morphology of EVs, and WB detected the marker proteins CD9, CD63, and TSG101 in EVs and the remaining culture supernatant. Original blots are shown in Supplementary Figures. (B) NTA measured the particle number and concentration of EVsNC and EVsHSHS. The effects of EVsHSHS on BMEC proliferation (C, n = 6), migration (D,E, n = 3), and tube formation (F,G, n = 3) after OGD are presented. Scale bar = 100 μm. CON + PBS, normoxia + PBS; OGD + PBS, hypoxia + PBS; OGD + EVsNC, hypoxia + EVs cultured with FBS; OGD + EVsBlank, hypoxia + EVs cultured with blank-serum; OGD + EVsHSHS, hypoxia + EVs cultured with HSHS-serum. Data are presented as mean ± SD. ###P < 0.001, vs. the CON + PBS group. ***P < 0.001, vs. the OGD + PBS group. ∆P < 0.05, ∆∆P < 0.01, ∆∆∆P < 0.001, vs. the OGD + EVsHSHS group. n.s., no significance.

Full size image

EVsHSHS modulate the miR-126 and PIK3R2/PI3K/AKT pathway to protect BMECs

It is well established that miR-126 plays a key role in angiogenesis28,30. We evaluated the interaction between miR-126 and PIK3R2 using a dual-luciferase reporter assay (Fig. 2A). Following OGD injury, EVsHSHS upregulated miR-126 gene expression in BMECs (Fig. 2B). Western blot analysis demonstrated that EVsHSHS downregulated PIK3R2 protein expression and activated the PI3K/AKT pathway (Fig. 2C–F). No significant differences were observed in the expression of these molecules between the EVsBlank and EVsNC groups. These findings prompted further investigation into the effects of EVsHSHS on cerebral ischemia and the underlying mechanisms in vivo.

Fig. 2
figure 2

EVsHSHS modulate the miR-126 and PIK3R2/PI3K/AKT pathway in BMECs. (A) Dual-luciferase reporter assay assessing the interaction between miR-126 and PIK3R2. (B) Quantitative real-time PCR analysis of miR-126 expression levels in BMECs. (C–F) Representative bands and histograms of key molecules in the PIK3R2/PI3K/AKT pathway. Original blots are shown in Supplementary Figures. CON + PBS, normoxia + PBS; OGD + PBS, hypoxia + PBS; OGD + EVsNC, hypoxia + EVs cultured with FBS; OGD + EVsBlank, hypoxia + EVs cultured with blank-serum; OGD + EVsHSHS, hypoxia + EVs cultured with HSHS-serum. Data are presented as mean ± SD, n = 3–4. •••P < 0.001, vs. the NC mimics + r-PIK3R2-3UTR-wt group. #P < 0.05, ###P < 0.001, vs. the Sham group. *P < 0.05, **P < 0.01, ***P < 0.001, vs. the Model group. ∆P < 0.05, ∆∆P < 0.01, ∆∆∆P < 0.001, vs. the EVsHSHS-H group. n.s., no significance.

Full size image

EVsHSHS attenuate cerebral ischemic injury in pMCAO rats

Following surgery, rats in the Model group, EVsNC-L group, and EVsHSHS-L group displayed a consistent pattern of weight loss. Conversely, rats treated with EVsHSHS-H or EVsNC-H demonstrated a slower rate of weight reduction on days 3 and 4. By day 7, the EVsHSHS-H and EVsNC-H groups exhibited significantly less weight loss compared to the pMCAO model rats (Fig. 3A). The modified neurological severity score (mNSS) results indicated that both the EVsHSHS-H and EVsNC-H groups had lower scores, primarily reflecting improvements in motor function (Fig. 3B).

MRI revealed large cerebral infarcts on the ischemic side in the Model group. Infarct volume was significantly reduced in both the EVsNC-H and EVsHSHS-H groups, with EVsHSHS-H demonstrating superior therapeutic efficacy compared to EVsNC-H (Fig. 3C,D).

Fig. 3
figure 3

EVsHSHS reduce cerebral infarct volume and decrease mNSS in cerebral ischemic rats on day 7 after pMCAO. (A) Changes in body weight of pMCAO rats. Data are presented as mean ± SEM, n = 7. (B) mNSS results in pMCAO rats (normal score, 0; maximal deficit score, 18). 13–18 indicates severe injury; 7–12, moderate injury; 1–6, mild injury. Data are presented as Min to Max, n = 7. (C,D) MRI detection of infarct volume. Infarcted areas are indicated by high density in the cortex and striatum, and outlined in each slice with dotted lines. EVsNC-L, 50 µg/100 g weight of EVsNC; EVsNC-H, 100 µg/100 g weight of EVsNC; EVsHSHS-L, 50 µg/100 g weight of EVsHSHS; EVsHSHS-H, 100 µg/100 g weight of EVsHSHS. Data are presented as mean ± SD, n = 3. *P < 0.05, ***P < 0.001, vs. the Model group. ∆∆P < 0.01, ∆∆∆P < 0.001, vs. the EVsHSHS-H group.

Full size image

EVsHSHS improve cortical angiogenesis in pMCAO rats

Relative cerebral blood flow (rCBF) was assessed using MRI perfusion-weighted imaging with arterial spin labeling. Blood flow profiles in the right and left hemispheres of rats in the Sham group were comparable, whereas cerebral perfusion was markedly reduced on the ischemic side in the Model group, particularly within the cingulate cortex, motor cortex, sensory cortex, piriform cortex, and corpus striatum (Fig. 4A). rCBF in the cingulate cortex and piriform cortex was significantly elevated in the EVs groups compared to the Model group. Additionally, EVsHSHS-H also increased rCBF in the motor cortex, sensory cortex, and corpus striatum in pMCAO rats (Fig. 4B).

Fluorescence microscopy revealed that red fluorescent-labeled EVs were present in brain tissue, primarily localized to the ischemic penumbra, supporting their therapeutic potential (Supplementary Fig. 3). Immunofluorescence analysis demonstrated a significant reduction in vascular density and discontinuous vessels in the cortex of the Model group. Conversely, vascular density increased following EC injection. Among the EVs groups, the EVsHSHS-H group exhibited significantly greater cortical vascular density and more continuous vessels than the other groups (Fig. 4C, D).

Additionally, hematoxylin and eosin (H&E) staining revealed that in the Model group, the rat cortical ischemic penumbra exhibited a disorganized tissue structure with decreased neuronal cell density, infiltrated inflammatory cells, irregular vascular lumens, and stripped intima. In contrast, the EVsNC-H and EVsHSHS-H groups showed histological improvements (Fig. 4E). These findings suggest that EVsHSHS-H attenuated cortical pathological injury in the ischemic penumbra of pMCAO rats, promoted angiogenesis, and consequently improved blood perfusion.

Fig. 4
figure 4

EVsHSHS enhance cortical angiogenesis and improve perfusion in cerebral ischemic rats on day 7 after pMCAO. (A) Cerebral blood flow across groups. (B) rCBF in different brain regions. (C,D) Vascular density in the cortical tissue. Scale bar = 100 μm. (E) HE staining of rat brain sections. Scale bar = 50 μm (top), scale bar = 1 mm (bottom). EVsNC-L, 50 µg/100 g weight of EVsNC; EVsNC-H, 100 µg/100 g weight of EVsNC; EVsHSHS-L, 50 µg/100 g weight of EVsHSHS; EVsHSHS-H, 100 µg/100 g weight of EVsHSHS. Data are presented as mean ± SD, n = 3. ###P < 0.001, vs. the Sham group. *P < 0.05, **P < 0.01, ***P < 0.001, vs. the Model group. ∆P < 0.05, vs. the EVsHSHS-H group.

Full size image

EVsHSHS regulate miR-126 and PIK3R2/PI3K/AKT pathway in pMCAO rats

Quantitative real-time PCR results indicated a significant reduction in miR-126 expression in the cortex of rats in the Model group. Compared to the Model group, miR-126 expression levels were significantly elevated in both the EVsNC-H and EVsHSHS-H groups (Fig. 5A). Western blot analysis revealed a significant upregulation of PIK3R2 expression in the cortex of the Model group, accompanied by a significant downregulation of PI3K and p-AKT levels. Conversely, PIK3R2 protein levels were significantly lower in both the EVsHSHS-L and EVsHSHS-H groups. Furthermore, the levels of PI3K, p-AKT, and the p-AKT/AKT ratio were significantly higher in the EVsHSHS-H group compared to the EVsNC-H group (Fig. 5B–E).

Fig. 5
figure 5

EVsHSHS regulate miR-126 expression and key molecules of the PIK3R2/PI3K/AKT pathway in cerebral ischemic rats on day 7 after pMCAO. (A) miR-126 expression levels in the cortex. (B–E) Protein levels of PIK3R2, PI3K, and p-AKT in the cortex. Original blots are shown in Supplementary Figures. EVsNC-L, 50 µg/100 g weight of EVsNC; EVsNC-H, 100 µg/100 g weight of EVsNC; EVsHSHS-L, 50 µg/100 g weight of EVsHSHS; EVsHSHS-H, 100 µg/100 g weight of EVsHSHS. Data are presented as mean ± SD, n = 3. #P < 0.05, ###P < 0.001, vs. the Sham group. *P < 0.05, **P < 0.01, ***P < 0.001, vs. the Model group. ∆P < 0.05, ∆∆P < 0.01, vs. the EVsHSHS-H group.

Full size image

Discussion

The traditional Chinese medicine formula Houshiheisan (HSHS) has been demonstrated to markedly improve limb motor and language functions in patients with cerebral stroke and mitigate vascular injury factors such as hypertension and hyperlipidemia4,34,35. Previous research has shown that HSHS significantly reduces cerebral infarction volume, enhances intracranial hemodynamics, and promotes nerve regeneration in rats with cerebral ischemia33,36,37,38,39. In vitro, HSHS-serum attenuated injury in umbilical vein endothelial cells following OGD, and active ingredients of HSHS reduced inflammation in BV2 cells32,40. Given that Chinese medicine requires proper decoction to fully exert its therapeutic effects, incorrect preparation methods may compromise its efficacy. Furthermore, patients with cerebral ischemia often experience swallowing difficulties, which limits the oral administration of medication.

Currently, there are two primary types of biological therapies for stroke: cell therapies and EVs. Transplantation of stem cell-like cells, including mesenchymal stem cells (MSCs), induced pluripotent stem cells (iPSCs), or EPCs, into animals with cerebral ischemia has been demonstrated to reduce infarct volume and attenuate cerebral edema41,42,43. However, cell transplantation is constrained by low cellular activity and the risk of vascular obstruction44. The utilization of EVs offers a promising approach to circumvent these limitations.

EVs are small, lipid bilayer-enclosed structures released by cells. These vesicles contain a diverse array of nucleic acids, proteins, and other biomolecules, functioning as carriers of cellular information45,46. EVs exhibit lower immunogenicity and toxicity than traditional therapies and can traverse the blood-brain barrier, making them promising candidates for treating neurological disorders like ischemic stroke47,48. The biological content of EVs can be manipulated by pretreating donor cells or directly modifying EVs to optimize therapeutic outcomes49,50,51. EPCs possess endothelial and stem cell13 characteristics, facilitating the repair of damaged endothelial cells and promoting neovascularization in healthy endothelium52,53,54. Notably, EPCs secrete more EVs than ordinary endothelial cells, and EPC-derived EVs have been demonstrated to enhance endothelial cell proliferation and migration, elevate pro-angiogenic factor levels, and stimulate vascular repair55,56,57.

Chinese medical formulas and individual monomers have also been demonstrated to influence various diseases through EVs. For example, the Jian-Pi-Yi-Shen formula has been shown to regulate microRNAs in serum-derived EVs in chronic kidney disease rats58. Additionally, ginsenoside-Rg1 and ferulic acid have been reported to promote angiogenesis in EPC-derived EVs by modulating microRNA expression59,60. In our study, the uptake of EVsHSHS by BMECs significantly improved cell viability, accelerated migration, and enhanced lumen formation following OGD injury. Compared to EVsNC, EVsHSHS exhibited a more pronounced effect on enhancing BMEC function. No statistical differences between EVsBlank and EVsNC were observed in vitro, suggesting that specific components within HSHS, rather than the rat serum itself, were responsible for the observed pro-angiogenic effects.

In the in vivo experiment, EVsNC and EVsHSHS were labeled with DiR and injected into cerebral ischemic rats via tail vein injection on day 3. On day 7, red fluorescence indicative of injected EVs was observed in brain tissue sections, confirming their entry into the brain. This finding provided the material basis for the therapeutic effects of EVs. Indeed, the traditional oral administration route requires drug ingredients to be absorbed by the digestive system before entering the bloodstream. Conversely, EVs can directly enter the bloodstream through intravenous injection, leading to a faster onset of action. This study observed that after injection of EVsHSHS, rats in the high-dose EVsHSHS group exhibited improved overall conditions, fewer limb motor deficits, enhanced feeding and water intake, and consequently reduced weight loss. Further testing revealed improvements in neurological deficits, decreased pathological damage, reduced cerebral infarction volume, and increased cortical vascular density in rats. However, further investigation is needed to elucidate the precise mechanisms underlying these effects. Notably, EVsHSHS-H significantly outperformed EVsNC-H in reducing cerebral infarction volume (Fig. 3C,D). EVsHSHS-H also demonstrated a favorable trend in reducing the mNSS, increasing rCBF in the sensory cortex, motor cortex, and striatum, and increasing cortical vascular density. These findings collectively suggest that HSHS enhances the protective effects of EPCs-derived EVs against cerebral ischemic injury.

miR-126 is significantly downregulated in the circulation of patients with cerebral ischemia and can serve as a potential biomarker61,62,63. Studies have demonstrated that miR-126-3p/5p inhibits microglial activation and reduces neuronal death by suppressing TNF expression. Additionally, it attenuates inflammation and reduces blood-brain barrier permeability by regulating the levels of vascular endothelial adhesion molecule-1 and E-selectin64. Under hypoxic conditions, miR-126 has been implicated in regulating angiogenesis by inhibiting PIK3R265–68. This study confirmed the binding of miR-126 to PIK3R2 through dual-luciferase reporter assays. In OGD-injured BMECs and pMCAO rats, EVsHSHS upregulated the expression of miR-126, which may be attributed to the higher miR-126 content in EVsHSHS compared to EVsNC or the ability of EVsHSHS contents to promote miR-126 expression in cells and rats. However, the miR-126 content within the EVs themselves was not quantified. Future experiments will focus on detecting microRNA content in various EVs and using inhibitors/mimics to elucidate the key mechanisms involved in EVsHSHS-mediated effects. In non-cancer cells, increased PIK3R2 expression inhibits PI3K and AKT activation. For instance, in a rat model of ischemia-reperfusion-induced pain hypersensitivity, reduced PIK3R2 expression was found to promote protein expression of PI3K and p-AKT69. In umbilical vein endothelial cells, PIK3R2 knockdown resulted in elevated levels of p-AKT expression70. Similarly, knocking down PIK3R2 in ovarian granulosa cells elevated the protein expression levels of PI3K and AKT. Conversely, increased PIK3R2 expression decreased the levels of these proteins71. During ischemia and hypoxia, the PI3K/AKT pathway maintains endothelial cell activity, accelerates migration, and promotes angiogenesis by regulating cytokines such as HIF-1α, VEGFA, and eNOS72,73,74,75. Consistent with previous literature, EVsHSHS in this study increased the protein expression levels of PI3K and p-AKT, which is opposite to their regulatory effect on PIK3R2. The regulation of miR-126 and the PIK3R2/PI3K/AKT pathway may represent the molecular mechanisms underlying the pro-angiogenic effects of EVsHSHS after cerebral ischemia.

A limitation of this study is the lack of exploration into the specific mechanisms through which HSHS modifies EPCs and EVs. Based on existing research and our findings, we hypothesize that these changes involve two primary pathways: (1) regulation of EV generation or secretion, as evidenced by the increased number and altered size distribution of EVs released from EPCs after HSHS incubation (Fig. 1B). HSHS may influence EV production or secretion by modulating ESCRT complexes, the syndecan-syntenin-ALIX pathway, tetraspanins, or small GTPases76. (2) modification of EPCs, resulting in altered EV contents. Previous experiments demonstrated an increase in EPC numbers in the cortex of pMCAO rats following HSHS gavage administration (Supplementary Fig. 4), along with enhanced EPC viability and migration in vitro (Supplementary Fig. 5). Therefore, we hypothesized that HSHS may influence the information carried by EVs, such as miRNAs, by modifying EPCs. Moreover, aging is associated with an increased risk of stroke, potentially reduced angiogenesis, and aggravated inflammation, leading to long-term functional impairment77. Investigating the efficacy of HSHS in aged animals would be more clinically relevant.

The clinical translation of EV therapy faces significant challenges. Quality control of EVs remains a critical issue, with no standardized dosage guidelines for laboratory or clinical use. Intravenous injection, the most common administration route, leads to substantial EV loss during delivery to target areas78. Preventing premature clearance of EVs from the bloodstream is a major hurdle. While EVs possess inherent targeting capabilities through membrane receptors or ligands that interact with target cells, lipid composition and protein content can influence their tropism for specific organs79. Unfortunately, many natural EVs are phagocytosed by macrophages, leading to off-target effects80. To enhance targeting, strategies such as modifying EV surface receptors or selecting EVs with specific components as carriers can be employed. For instance, EVs modified with GE11 peptide can target EGFR by binding specifically to tumor cell membrane proteins81. Different integrins can alter the pharmacokinetics of EVs and increase their accumulation in specific organs82. Comprehensive analysis of EV components from various sources is essential for addressing these challenges. Despite technical limitations related to quality control, dosage standards, and utilization/targeting, EV therapy continues to demonstrate promising potential.

Conclusion

This study confirms that EVsHSHS promote angiogenesis after ischemic stroke, potentially through the regulation of the miR-126 and PIK3R2/PI3K/AKT pathway (Fig. 6). These findings offer novel insights into the application of Traditional Chinese Medicine for cerebral ischemia and could contribute to advancements in modern research in this field.

Fig. 6
figure 6

Graphical abstract.

Full size image

Materials and methods

The animal study was approved by the Institution Animal Care and Use Committee of Capital Medical University (No. AEEI-2019-001). The study complied with ARRIVE guidelines. Rats were anesthetized and euthanized according to the American Veterinary Medical Association (AVMA) Guidelines for the Euthanasia of Animals (2020). Rats used for primary cell isolation, weighing less than 100 g, were sacrificed by cervical dislocation. Adult rats were anesthetized with isoflurane during surgery and sample collection, and euthanized at the end of the experiment.

Primary EPC isolation and culture

SPF-grade, 3 to 4-week-old male Sprague-Dawley (SD) rats (70–100 g) were obtained from Beijing Vital River Laboratories (license number: SYXK (Beijing) 2021-0006). Rats were euthanized by cervical dislocation under sterile conditions. The femurs and tibiae were isolated bilaterally, and the bone marrow cavity was flushed with phosphate-buffered saline (PBS) using a syringe. The resulting suspension was filtered through a 100 μm mesh to remove large debris. After washing with PBS and centrifugation (1000 rpm, 5 min), the pellet was resuspended in EGM-2 complete medium (Lonza, CC-3162). Culture flasks were pre-coated with type I rat tail collagen (Corning, 354236). The medium was changed for the first time after 72 h and then every other day. Collected EPCs were identified by their characteristic morphology and positive staining for CD34 and CD133 using immunofluorescent labeling in Supplementary Figs. 583,84.

Primary BMECs isolation and culture

SPF-grade, 2 to 3-week-old male SD rats (40–60 g) were procured from Beijing Vital River Laboratories (license number: SYXK (Beijing) 2021-0006). Cervical dislocation was used to euthanize the rats, and the isolation and culture procedures were adapted from the literature85. Under sterile conditions, the brain was isolated, and only the cortex was retained after removing the meninges and large blood vessels. The cortex was minced, and a mixture of 0.1% type II collagenase (Gibco, 17101-015) and DNase I (Roche, 10104159001, 30 U/mL) was added for digestion in a 37 °C, 5% CO2 incubator (Thermofisher, 150i) for 90 min. Following centrifugation, the digested product was purified with 20% BSA (prepared in PBS). Further purification of the microvascular segments was achieved by adding 0.1% collagenase/dispase (Roche, 10269638001) and DNase I (20 U/mL) to the precipitate. After digestion for 60 min, the above operation was repeated to collect the microvascular segments. Culture bottles were pre-coated with 0.5% gelatin (Sigma, V900863). The microvascular segments were cultured in an ECM complete medium (Sciencell, 1001, containing 4 µg/mL puromycin dihydrochloride). The medium was replaced after 48 h and then replaced every other day. 2 µg/mL puromycin dihydrochloride (Lablead, P067) was added during each cell passage, and the medium was replaced after 24 h.

EVs isolation and characterization

Following the methods established in our previous study32,36, HSHS extract was prepared. After 7 days administration to rats, blank-serum and HSHS-serum were collected. All three serums (fetal bovine serum, blank-serum, and HSHS-serum) underwent ultracentrifugation at 100,000×g for 12 h at 4 °C to deplete native EVs. The resulting supernatant from each serum was then used for separate EPC cultures. Conditioned medium from these cultures was subsequently processed using a commercially available EV isolation kit (Umibio, UR52121) according to the manufacturer’s instructions. The extracted EVs were quantified using the BCA method and stored at − 80 °C. Particle size and concentration analysis of the EVs was performed by NTA using a NanoSight NS300 (Malvern Panalytical, Britain) and NTA 3.4 Build 3.4.4 software at Umibio (Shanghai) Co.Ltd. Transmission electron microscopy (TEM) was employed to visualize the morphology of the EVs. Finally, WB analysis was conducted to detect the presence of surface marker proteins CD9 (ab92726, 1:2000 dilution), CD63 (ab68418, 1:1000 dilution), and TSG101 (ab133586, 1:5000 dilution) on the EVs. All primary antibodies were purchased from Abcam.

Establishment of OGD model and treatment

The third-generation BMECs were separated into five groups: CON + PBS (normoxia + PBS), OGD + PBS (hypoxia + PBS), OGD + EVsNC (hypoxia + EVs cultured with FBS), OGD + EVsBlank (hypoxia + EVs cultured with blank-serum), and OGD + EVsHSHS (hypoxia + EVs cultured with HSHS-serum). After culturing EPCs in various sera for 48 h, EVs within the conditioned medium were collected and diluted to an identical concentration. As detailed in our earlier studies32, the culture medium was substituted with glucose-free DMEM (Gibco, 11966025), and BMECs were subjected to hypoxia (37°C, 93% N2, 5% CO2, 2% O2) for 6 h to induce OGD damage. Upon establishing the OGD model, 40 µg/mL of EVs were introduced to the respective treatment groups, whereas an equivalent volume of PBS was added to the control groups. Subsequently, all groups were cultured for an additional 24 h.

Cell viability assay

Cell viability was assessed using the CCK-8 assay (Vazyme, A311). A working solution of CCK-8 was prepared by diluting the CCK-8 reagent in the DMEM basal medium (Corning, 10-013-CMR). After treatment, the culture medium was replaced with the CCK-8 working solution and incubated for one hour. Three to five blank wells were included for background correction. The optical density (OD) at 450 nm was measured using a microplate reader (Molecular Devices, SpectraMax Plus384). Cell viability was calculated as follows:

$${\text{Cell viability }}\left( \% \right)=\left( {\text{OD}}_{\text{test}} - {\text{OD}}_{\text{blank}} \right)/\left( {\text{OD}}_{\text{control}} - {\text{ OD}}_{\text{blank}}\right) \times 100\% .$$

Wound healing test

Cells were subjected to OGD treatment and then subjected to a scratch wound assay. Briefly, 6-well plates were pre-coated with 0.5% gelatin and horizontal lines were drawn on the back of the plates at 0.5–1 cm intervals. After OGD treatment, a 200 µL pipette tip was used to create a scratch perpendicular to the bottom surface of the well. Cell debris was removed by rinsing with PBS. 40 µg/mL of EVs were added to each treatment group, while an equal volume of PBS was added to the control groups. Cell migration was recorded at 0 and 24 h post-treatment, and the migration rate was analyzed using ImageJ software.

Tube formation test

Pre-cooled reagents were used for this experiment. Matrigel (Corning, 354234) was pipetted into a 96-well plate at a volume of 70 µL per well. After incubating at 4°C for 30 min, the plate was transferred to a 37 °C incubator for solidification. BMECs were harvested following oxygen-glucose deprivation treatment and administration and seeded into the Matrigel-coated wells at a density of 3 × 104 cells per well. The plate was then incubated for 6 to 10 h. The number of tube-like structures formed by the BMECs was quantified using ImageJ software.

Transwell migration assay

EPCs were seeded into the upper chamber of a transwell plate and incubated for 24 h with EGM-2 medium containing either blank serum or HSHS serum. After 24 h, the upper chamber was replaced with a serum-free medium, while the lower chamber was filled with medium containing 20% FBS. Following 2 h of incubation, cells were washed with PBS three times, fixed in 4% paraformaldehyde for 15 min, washed again, and stained with crystal violet dye for 20 min. Cells on the upper side of the membrane were removed with a cotton swab. The number of cells in the field of view was counted using ImageJ software.

Dual-luciferase reporter assay

The r-PIK3R2-3UTR target plasmid was constructed and produced by Hanbio Technology Co., Ltd. 293T cells were transfected with the target plasmid and rno-miR-126a-3p or a Negative Control (N.C). Fluorescence within the cells was observed 48 h post-transfection using a microplate reader, adhering to the kit protocol.

Quantitative real-time PCR (qPCR)

RNA was isolated from BMECs or the cerebral cortex using TRIzol Reagent. The concentration and purity of RNA samples were assessed using an ultra-micro spectrophotometer (Molecular Devices, SpectraMax QuickDrop), with an acceptable OD260/OD280 ratio ranging from 1.8 to 2.2. Quantitative real-time PCR was performed using a 20 µL reaction volume according to the manufacturer’s instructions (Toyobo, QRT-201). Relative gene expression levels were calculated using the 2−ΔΔCt method. Primers targeting miR-126 and the internal reference U6 were synthesized by Servicebio Technology Co., Ltd., and their sequences are listed in Supplementary Table 1.

Western blot (WB)

Proteins were extracted from BMECs or the cerebral cortex using an extraction solution containing radioimmunoprecipitation assay (RIPA) buffer, phenylmethylsulfonyl fluoride (PMSF), and protease/phosphatase inhibitors. Total protein concentration was determined using the bicinchoninic acid (BCA) method. WB was performed to analyze the protein expression levels of PIK3R2 (1:2000), PI3K (1:1000), p-AKT (1:2000), AKT (1:5000), β-actin (1:10,000), and GAPDH (1:5000) were analyzed by WB. PIK3R2 (ab180967) was purchased from Abcam, while PI3K (#4257), AKT (#4691), and p-AKT (#4060) were obtained from Cell Signaling Technology. GAPDH (GTX100118) and β-actin (GTX109639) were purchased from GeneTex. Bands were visualized and photographed using a chemiluminescent gel imager (BIO-RAD, ChemiDoc XRS+), and band intensities were quantified using ImageJ.

Construction of permanent middle cerebral artery occlusion (pMCAO) model

Male SD rats, SPF grade, weighing 300–350 g, were used in this experiment. Obtained from Beijing Vital River Laboratories (license number: SYXK (Beijing) 2021-0006), the rats were housed at the Laboratory Animal Center of Capital Medical University (license number: SYXK (Beijing) 2021-0030). Following a one-week acclimatization period, the rats were randomly assigned to either a Sham or a pMCAO group. The pMCAO surgery was performed according to established protocols33,86. Throughout the experiment, animal welfare and ethical regulations were strictly followed. Isoflurane anesthesia (4–5% for induction, 2-2.5% for maintenance) in a 2:1 N2O/O2 atmosphere was used. Supine-positioned rats underwent blunt dissection of the right common carotid artery, followed by its proximal ligation and ligation of the right external carotid artery. A 4 − 0 monofilament nylon suture (Beijing Cinotech Co.Ltd., Beijing, China) inserted into the right internal carotid artery blocked blood flow to the right middle cerebral artery. The Sham group underwent the identical surgical procedure except for embolisation. Post-surgical pain management consisted of meloxicam (1 mg/kg) administered subcutaneously once daily. To facilitate feeding, food was provided within the cage. Additionally, a sugar saline solution (0.9% NaCl and 5% glucose) was administered daily to ensure the rats’ basic energy needs.

At 6 h postoperatively, the rats were assessed using the Zea Longa 5-grade scoring method86. Scoring criteria: 0–no obvious neurological symptoms; 1–inability to fully extend the left front paw; 2–rotating to the left side; 3–tilting to the left side when walking; 4–walking only when stimulated and loss of consciousness. Rats with scores of 1–3 were considered successful modeled, while those with scores of 0 or 4 were excluded. One rat was excluded from the experiment, and six rats died within 7 days after pMCAO (two from the Model group and one from each of the remaining groups).

Animals administration

Forty-two rats were randomly assigned to each of the following five groups: Model, EVsNC-L, EVsHSHS-L, EVsNC-H, and EVsHSHS-H. Additionally, a Sham group of six rats was included. EVs were administered intravenously via the tail vein at 0, 24, and 72 h post-modeling. The injection dose was adjusted based on previous studies64,65,66 and was 50 or 100 micrograms per 100 g of body weight, depending on the group. The EVsNC-L and EVsNC-H groups received EVs cultured with FBS, while the EVsHSHS-L and EVsHSHS-H groups received EVs cultured with HSHS-serum. The Model and Sham groups were administered equivalent volumes of saline. Prior to the 72 h injection, EVs were labeled with DiR dye (Umibio, UR21017). Rats were euthanized on day 7 for further analysis.

Modified neurological severity score (mNSS)

Rats were weighed daily, and the resulting weight changes were recorded relative to their initial body weights. The mNSS was assessed on day 7, using the scoring criteria outlined in a previous study87. Higher scores indicate more severe behavioral deficits. The specific scoring criteria are provided in Supplementary Table 2.

Magnetic resonance imaging (MRI)39

On day 7 post-operation, rat brains were scanned using a PharmaScan® 7.0T Small Animal Magnetic Resonance Imaging System. Cerebral infarct volume was assessed with T2-weighted imaging (T2WI). T2WI images of the coronal section of the rat brain were acquired using a fast spin echo (FSE) sequence with the following parameters: repetition time = 4400 ms, echo time = 45 ms, inversion angle = 180°, field of view = 33 mm × 33 mm, matrix = 256 × 256, number of slices = 38, and slice thickness = 0.7 mm. High-signal areas in each section were analyzed and processed with ImageJ. The ischemic infarct area of each section was manually outlined, and cerebral infarct volume was calculated as the sum of the high-signal areas per slice multiplied by the slice thickness.

Arterial spin labeling (ASL) perfusion-weighted imaging was used to measure changes in cerebral perfusion in pMCAO model rats. Statistical rCBF was calculated to assess cerebral perfusion impairment. A fluid-attenuated inversion recovery (FLAIR) sequence with the following parameters was used to acquire images: repetition time = 18,000 ms, echo time = 25 ms, inversion angle = 90°, field of view = 3 cm × 3 cm, inversion time = 300 ms, matrix = 128 × 128, slice thickness = 2 mm, and number of excitations = 1. PharmaScan® Paravision version 5.1 was used to analyze the cerebral blood flow images and calculate rCBF in the cingulate cortex, motor cortex, sensory cortex, piriform cortex, and striatum of rats. rCBF was calculated as the ratio of cerebral blood flow on the ischemic side to that on the contralateral side.

Immunofluorescence staining

Following perfusion with cold saline, the entire brain was removed and immediately frozen at − 80 °C. The brain was subsequently embedded in an OCT compound and sectioned into 20 μm thick slices using a cryostat (Leica, CM1950). EVs labeled with red fluorescence within the cortex were observed directly after sectioning under a fluorescence microscope (Nikon, TI2-U). Sections were then fixed, subjected to antigen retrieval, and blocked. A lectin (1:1500) conjugated with a green fluorescent dye was added and incubated for 24 h. After washing, sections were mounted with a DAPI-containing mounting medium. Vessels were visualized and quantified for fluorescence intensity using a fluorescence microscope (Leica, TCS SP8) and ImageJ software.

Hematoxylin and eosin (H&E) staining

Following cardiac perfusion and fixation, the entire brain was excised. Subsequently, the brain tissue was dehydrated, embedded in paraffin, and sectioned into 4-micrometer-thick slices. After dewaxing, the sections were stained with H&E dye, followed by a series of ethanol-mediated dehydrations and mounting. Pathological damage to the cortical ischemic penumbra was evaluated using a fully automated slide scanning system (Pannoramic scan, 3D HISTECH).

Statistical analysis

Statistical analysis were conducted using GraphPad Prism 9.4.1 software. Changes in body weight were presented as mean ± standard error (SEM), while other measurement data were expressed as mean ± standard deviation (SD). Initially, a normality test was performed. For normally distributed data, ANOVA was used to compare multiple groups, followed by Bonferroni or Student-Newman-Keuls (SNK) tests for pairwise comparisons. Non-normally distributed data were presented as medians with interquartile ranges. The Kruskal-Wallis test was employed to compare mNSS results among multiple groups, with Bonferroni-corrected P values used for pairwise comparisons. A P-value of < 0.05 was considered statistically significant.