Paraventricular oxytocin neurons attenuate post-ischemic brain injury by suppressing microglia-mediated neuroinflammation

Abstract

Background

Neuroinflammation is caused by the overactivation of microglia, contributing to secondary brain injury in ischemic stroke. Oxytocin (OXT), a neuropeptide synthesized by neurons in the paraventricular nucleus (PVN) of the hypothalamus, has demonstrated potential in mitigating inflammatory responses across various pathological conditions. However, research on the role of PVN^OXT neurons in ischemic stroke is limited, and the modulatory effect of these neurons on neuroinflammation remains unclear.

Methods

Transient middle cerebral artery occlusion (tMCAO) was performed in mice. OXT levels were measured in the peri-infarct cortex, serum, and cerebrospinal fluid (CSF), and OXTR expression was assessed in the peri-infarct cortex after tMCAO. Chemogenetic approaches were used to selectively activate PVN^OXT neurons in OXT-Cre mice, after which neurological function, infarct volume, and blood–brain barrier integrity were evaluated. To investigate the mechanisms underlying the regulation of ischemic injury by PVN^OXT neurons, RNA sequencing of the ipsilateral ischemic hemisphere was performed. Additional analyses, including flow cytometry, molecular assays, and Transwell migration experiments, were conducted to validate the downstream signaling pathways and cellular responses.

Results

OXT levels significantly reduced in the peri-infarct cortex, serum, and CSF, whereas OXTR expression increased in the peri-infarct cortex after tMCAO. Chemogenetic activation of PVN^OXT neurons increased OXT levels in both the brain and circulation, reduced infarct volume, and improved neurological outcomes. In addition, transcriptomic analysis identified CXCL3 as one of the most significantly downregulated chemokines after the activation of PVN^OXT neurons, and CXCL3 downregulation was associated with reduced neutrophil chemotaxis. Further in vivo and in vitro investigations demonstrated that PVN^OXT neurons inhibit microglial CXCL3 expression via the OXTR–ERK signaling pathway, thereby restricting the infiltration of neutrophils. In contrast, the administration of recombinant CXCL3 promoted the recruitment of neutrophils and exacerbated ischemic injury.

Conclusions

PVN^OXT neurons alleviate post-ischemic brain injury by inhibiting the secretion of CXCL3 from microglia, consequently reducing neutrophil chemotaxis. These results underscore the therapeutic potential of targeting PVN^OXT neurons and their downstream signaling pathways to mitigate immune-mediated damage in ischemic stroke.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12964-026-03007-7.

Introduction

Ischemic stroke remains a leading cause of mortality and long-term disability worldwide and accounts for the majority of stroke burden [1]. Although current reperfusion strategies, including intravenous thrombolysis and mechanical thrombectomy, have significantly improved clinical outcomes, their benefits are limited by narrow therapeutic windows and the risk of ischemia–reperfusion (I/R) injury [2]. Importantly, even after successful recanalization, secondary neuroinflammatory responses triggered by I/R critically contribute to neuronal damage and neurological deterioration [3]. Therefore, targeting post-ischemic neuroinflammation represents a major unmet challenge in the treatment of stroke.

Under physiological conditions, microglia function as resident immune sentinels of the central nervous system, maintaining tissue homeostasis through continuous surveillance [4, 5]. In response to brain ischemia, microglia are activated rapidly and adopt a pro-inflammatory phenotype that is characterized by the release of cytokines, chemokines, and reactive oxygen species, which promote neuronal death and exacerbate infarct progression [6, 7]. Beyond these direct effects, activated microglia coordinates the recruitment of peripheral immune cells, particularly neutrophils, via chemokine signaling [8, 9].

Neutrophils, which are among the earliest infiltrating leukocytes after stroke and further aggravate neuroinflammation by adhering to the endothelium, impairing microcirculation, and releasing cytotoxic mediators [10, 11]. Chemokines are integral to the regulation of neutrophil recruitment to inflamed brain tissue. Notably, C-X-C motif chemokine ligand 3 (CXCL3) binds selectively to C-X-C chemokine receptor 2 (CXCR2) and acts as a highly effective chemoattractant for neutrophils [12]. Prior research has demonstrated that inhibition of the CXCL3–CXCR2 signaling pathway diminishes neutrophil infiltration and mitigates pathological inflammation across various disease models [13, 14]. However, whether microglia regulate neutrophil chemotaxis after stroke through CXCL3 signaling remains unclear.

Oxytocin (OXT), a neuropeptide synthesized by neurons in the hypothalamic paraventricular nucleus (PVN), is well known for its roles in social behavior, reproduction, and stress regulation [15–17]. Emerging evidence has revealed the role of OXT as a neuroprotective and immunomodulatory factor in neurological disorders. In particular, OXT has shown to suppress microglial activation and the production of pro-inflammatory mediators, including NLRP3, IL-1β, and IL-18 [18, 19]. Consistently, the activation of PVN^OXT neurons has been shown to have anti-inflammatory and cardioprotective effects [20]. Moreover, reduced endogenous OXT levels have been associated with poor clinical outcomes in patients with disorders such as autism spectrum disorder and aneurysmal subarachnoid hemorrhage [21, 22]. Despite these findings, whether PVN^OXT neurons regulate post-stroke neuroinflammation through microglia-dependent chemokine signaling remains unclear.

In this study, we demonstrate that the activation of PVN^OXT neurons increases endogenous OXT levels, reduces infarct volume, and improves neurological function after ischemic stroke. Mechanistically, this neuroprotective effect is mediated by the suppression of microglial CXCL3 expression via the OXTR–ERK signaling pathway, thereby limiting neutrophil infiltration. Collectively, these findings reveal a previously unrecognized neural–immune regulatory axis and highlight a potential therapeutic strategy for ischemic stroke.

Materials and methods

Animals

Both male and female C57BL/6 wild-type mice and OXT-Cre mice (Catalog: C001666; Cyagen, Suzhou, China) aged 8–10 weeks were used. The animals were maintained under standard laboratory conditions (22 °C, 12-h light–dark cycle) with ad libitum access to food and water. All experimental procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of the Chinese Institute for Brain Research (CIBR, Approval #: CIBR-IACUC-121) and conducted in accordance with institutional guidelines. Surgeries and outcome assessments were performed by investigators who were blinded to group allocation.

Stereotaxic injection of viral vectors

Both male and female OXT-Cre mice were anesthetized with an intraperitoneal injection of 2.5% tribromoethanol (250 mg/kg) and placed on a stereotaxic apparatus (RWD, China). Body temperature was maintained at 37 ± 0.5 °C using a heating pad throughout the procedure. The skull was adjusted to ensure that the bregma and lambda were level in the same horizontal plane.

For chemogenetic manipulation of PVN^OXT neurons, adeno-associated viral (AAV) vectors were bilaterally injected into the PVN using a microsyringe pump (RWD, China) at a rate of 20 nl/min. The stereotaxic coordinates relative to the bregma were AP − 0.6 mm, ML ± 0.25 mm, and DV − 5.0 mm. The injection needle was left in place for an additional 10 min after completion of the infusion to allow full diffusion of the viral particles before being withdrawal slowly. The viral vectors (brainVTA, China) used included AAV2/9-EF1α-DIO-hM3Dq-mCherry (> 5.0 × 10¹² vector genomes/ml, 120 nl per side) for neuronal activation and AAV2/9-EF1α-DIO-mCherry (> 5.0 × 10¹² vector genomes/ml, 120 nl per side) as a control. After surgery, the mice were allowed to recover for at least three weeks before the subsequent experiments to ensure sufficient viral expression.

Transient middle cerebral artery occlusion (tMCAO) model

Transient focal cerebral ischemia was induced in adult mice using the intraluminal filament method as previously described [7]. Briefly, mice were anesthetized with an intraperitoneal injection of 2.5% tribromoethanol (250 mg/kg) and placed on a heating pad to maintain body temperature at 37 ± 0.5 °C throughout the procedure.

A midline neck incision was made to expose the right common carotid artery (CCA), external carotid artery (ECA), and internal carotid artery (ICA). The CCA was temporarily occluded using a microvascular clip to prevent backflow during filament insertion. The ECA was ligated, and a silicone-coated monofilament (Doccol, USA) was inserted through the ECA stump into the ICA to occlude the origin of the middle cerebral artery (MCA). The filament was advanced approximately 9–10 mm from the carotid bifurcation until slight resistance was felt, indicating successful occlusion. Once the filament was positioned, the CCA clip was removed to restore blood flow through the CCA during the ischemic period. Cerebral blood flow (CBF) was continuously monitored using a laser Doppler flowmeter (SIM Opto-Technology, China), and only mice showing a 75% or greater reduction in CBF were included in subsequent analyses. After 60 min of occlusion, the filament was carefully withdrawn to allow reperfusion, and the incision was sutured. The mice in the sham-operated group underwent the same surgical procedure except for the filament insertion step. Following surgery, all the mice were returned to a temperature-controlled recovery chamber and given free access to food and water. Postoperative care included monitoring for neurological signs and maintaining hydration.

Drug administration

To selectively activate PVN^OXT neurons, clozapine-N-oxide (CNO; 1 mg/kg; brainVTA, China) was administered intraperitoneally in sterile saline once daily for three consecutive days, beginning 12 h after reperfusion following tMCAO. The dose was selected according to previous reports [23].

To assess the role of CXCL3 in post-stroke inflammation, recombinant mouse CXCL3 (150 µg/kg; R&D Systems) was diluted in sterile phosphate-buffered saline (PBS) and administered via tail vein injection once daily for three consecutive days after reperfusion, beginning 12 h after reperfusion. Control mice received equivalent volumes of PBS. The dose and regimen were selected according to previous reports [12].

Primary microglial culture

Primary microglia were isolated from the cerebral cortices of postnatal day 1 C57BL/6 mice. The cortices were dissected, the meninges were removed, and the tissues were digested with 0.25% trypsin/EDTA (Gibco, USA) and 0.025% DNase I (Biofroxx, Germany) at 37 °C for 10 min. Digestion was terminated by the addition of Dulbecco’s modified Eagle’s medium (DMEM; Gibco, USA) supplemented with 10% fetal bovine serum (FBS; Gibco, USA) and 50 U/mL penicillin–streptomycin (Gibco, USA). The cell suspension was centrifuged at 1000 rpm for 5 min, resuspended in DMEM, and filtered through an 80-µm cell strainer. Cells were seeded into poly-D-lysine-coated (0.01 mg/mL; Sigma, USA) 25 cm² flasks and maintained in DMEM supplemented with 10% FBS at 37 °C in a humidified 5% CO₂ incubator. After 7 days in vitro, the microglia were detached from the mixed glial cultures by mechanical shaking at 100 rpm for 5 h, after which they were collected and used for subsequent experiments.

Small interfering RNA (siRNA) transfection

CXCL3-specific siRNAs (sense: 5′-CACCCAGACAGAAGTCATAGCC-3′; antisense: 5′-CCGTTGGGATGGATCGCTTT-3′) were synthesized by Sangon Biotech (China). A scrambled siRNA, which produced results similar to those observed in the control group, was used as a negative control. siRNAs were complexed with Lipofectamine 2000 (Invitrogen, USA) in OptiMEM (Gibco, USA) and added to primary microglia at a final concentration of 160 nM. The cells were incubated for 72 h, after which the medium was replaced with fresh DMEM containing 10% FBS.

Oxygen-glucose deprivation/reoxygenation (OGD/R) and drug treatments

Primary microglia were subjected to OGD/R to simulate ischemic conditions. The cells were rinsed twice with glucose-free DMEM and incubated in glucose-free DMEM under hypoxia (1% O₂, 5% CO₂, and 94% N₂) at 37 °C for 4 h. During reoxygenation, 1% pyruvate/glucose and 10% FBS were added, and the cells were returned to normoxic conditions. Control cells were maintained in standard glucose-containing DMEM under normoxia.

Following OGD/R, the cells were treated with OXT (10 µM; MCE, China), the OXTR antagonist L-368,899 (10 µM; MCE, China), or the ERK inhibitor U0126 (10 µM; MCE, China). After 24 h of treatment, the cells or culture supernatants were collected for subsequent analyses. Drug concentrations were selected on the basis of previously published studies [24–26].

Bone marrow neutrophil isolation

Neutrophils were isolated from the bone marrow of C57BL/6 mice. Femurs and tibias were harvested, and bone marrow cells were flushed out with ice-cold PBS. Neutrophils were purified using density gradient centrifugation with a commercial neutrophil isolation kit (TBD, China) according to the manufacturer’s instructions. The purity of the neutrophils was routinely assessed by flow cytometry using the makers Ly6G and CD11b.

Neutrophil migration assay

A total of 1 × 10⁶ neutrophils were seeded onto a 6.5-mm, 3-µm pore size, polyester membrane Transwell inserts (Corning, USA), and the inserts were then placed into conditioned medium collected from the OGD/R microglia. Transwell systems were incubated at 37 °C for 12 h, and the migration rate was quantified by counting the neutrophils that migrated to the lower compartment.

Magnetic resonance imaging (MRI)

MRI was performed to evaluate the infarct volume on day 3 after tMCAO. All scans were conducted using a 9.4 Tesla small-animal MRI scanner (BioSpec 94/30 USR; Bruker BioSpin, Germany) equipped with a 154 mm volume transmitter coil and a dedicated mouse brain surface receiver coil.

The mice were anesthetized with 3.5% isoflurane for induction and maintained at 1.5% isoflurane in a medical gas mixture (70% N₂O and 30% O₂) throughout the scanning procedure. The core body temperature was maintained at 37.0 ± 0.5 °C using a feedback-controlled circulating warm-water system. T₂-weighted axial brain images were acquired using a fat-suppressed RARE sequence with the following parameters: repetition time = 2200 ms, echo time = 33 ms, matrix size = 240 × 200, and field of view = 24 × 20 mm. T₂ mapping was performed using a multislice gradient-echo sequence to enhance lesion contrast and structural delineation. MRI data were analyzed with ITK-SNAP (v.4.2.2; http://www.itksnap.org), with infarct volume defined as hyperintense regions on T2-weighted images across the entire brain. To account for brain atrophy, the final lesion volume was calculated as follows: lesion volume = contralateral parenchyma volume − (ipsilateral parenchyma volume − infarct volume).

RNA sequencing (RNA-seq)

RNA-seq was performed on ipsilateral ischemic hemisphere tissues collected from male mice on day 3 after tMCAO from both the tMCAO and tMCAO + OXT groups, with four mice included per group. Sequencing and analysis were conducted by OE Biotech, China. Differentially expressed genes (DEGs) were identified using DESeq2 with thresholds of |log2FC| > 1.5 and adjusted p < 0.05. Gene Ontology (GO) analysis and gene set enrichment analysis (GSEA) were performed using the ClusterProfiler R package. Protein–protein interaction (PPI) networks were analyzed through the STRING database and visualized in Cytoscape. Heatmaps and volcano plots were generated using R (v4.2.2).

ELISA

Peripheral serum was extracted from the mice and stored at 4 °C. Cerebrospinal fluid (CSF) was collected from the cisterna magna. The peri-infarct cortex was homogenized in RIPA buffer, and protein concentrations were determined using a BCA protein assay kit (Beyotime, China). OXT and CXCL3 levels were quantified using commercial ELISA kits (FineTest, CUSABIO, China) according to the manufacturer’s instructions. OXT and CXCL3 levels were measured in sham mice and on days 1, 3, and 7 after tMCAO. In separate experiments, after chemogenetic activation of PVN^OXT neurons, OXT and CXCL3 levels were assessed on day 3 post‑tMCAO.

Immunofluorescence assay

On day 3 after tMCAO, the mice were deeply anesthetized with tribromoethanol and transcardially perfused with ice-cold PBS, followed by 4% paraformaldehyde (PFA). The brains were then carefully removed and post-fixed in 4% PFA overnight at 4 °C. The tissues were subsequently cryoprotected by sequential immersion in 20% and 30% sucrose solutions until they sank. The brains were sectioned into 30-µm thick slices using a cryostat (Leica, Germany).

For immunofluorescence staining, free-floating sections were first permeabilized with 1% Triton X-100 for 15 min and then blocked with 5% bovine serum albumin (BSA) for 30 min. The sections were incubated with primary antibodies at 4 °C overnight. The following primary antibodies were used: goat anti-Iba1 (1:500; novus, NB100-1028), rabbit anti-CXCL3 (1:500; Abcam, ab220431), rat anti-Ly6g (1:200; Proteintech, 65078-1), and goat anti-CD31 (1:500; R&D Systems, AF3628). Following incubation with the primary antibodies, the sections were washed and incubated with secondary antibodies, including donkey anti-goat (1:500; Abcam, Alexa Fluor^® 594), donkey anti-rabbit (1:500; Abcam, Alexa Fluor^® 488), and donkey anti-rat (1:500; Abcam, Alexa Fluor^® 488), for 1 h at room temperature. Finally, the sections were mounted with anti-fade mounting medium containing DAPI (Solarbio, China). Images were acquired using a confocal laser scanning microscope (Nikon AX R MP, Japan) and analyzed with ImageJ software 1.80 (National Institutes of Health, USA).

Flow cytometry

Peripheral blood and ipsilateral ischemic hemisphere tissues were collected from mice on day 3 post-ischemic stroke and processed into single-cell suspensions. Red blood cells were lysed using RBC lysis buffer, and the brain tissues were enzymatically digested with collagenase IV, DNase I, and neutral protease at 37 °C, followed by centrifugation and PBS washes.

The cells were divided into unstained, single-stained, and multicolor groups. Each sample was adjusted to 1 mL in PBS, stained with 0.3 µL of live/dead dye for at least 20 min at 4 °C in the dark, washed, and incubated with the surface antibodies BV650-CD45, PerCP-Cy5.5 CD11b, APC-Ly6C, APC-CXCR2 and PE-Ly6G (BioLegend, USA) for another 20 min under the same conditions. Immune populations were gated on live singlet cells as follows: microglia (CD45^int CD11b⁺), infiltrating monocytes/macrophages (CD45^hi CD11b⁺ Ly6C⁺), and neutrophils (CD45^hi CD11b⁺ Ly6G⁺).

For intracellular staining, the cells were fixed for 30 min, permeabilized with 1 mL of permeabilization buffer, and incubated with FITC-CXCL3 (Aviva Systems Biology, USA) for 3 h at room temperature in the dark. After being washed with PBS, the samples were analyzed on an Attune flow cytometer (Thermo Fisher, USA), and the data were processed using FlowJo v10 software. A total of 5,000 CD45⁺ events were collected per sample. Single-color controls were used for compensation, and fluorescence-minus-one (FMO) controls were used to determine negative gating thresholds for CXCL3 and CXCR2. Although no myelin/debris removal was performed, the use of live/dead staining, FSC/SSC gating, and FMO controls allowed reliable identification of target populations.

Western blotting

For tissue samples, peri‑infarct cortex tissues were collected from sham‑operated mice and from tMCAO mice on days 1, 3, and 7 after tMCAO for OXTR detection, and on day 3 after tMCAO for BBB‑related proteins (Occludin, VE‑cadherin, ZO‑1). For cell samples, cultured cells were harvested at the indicated time points after OGD/R for detection of CXCL3, ERK, p‑ERK, p38, and p‑p38.

Total proteins were extracted using cold lysis buffer, and the protein concentrations were determined with a BCA kit (Beyotime, China). Samples containing equal amounts of protein (30 µg per lane) were separated by 10% SDS-PAGE and transferred to PVDF membranes. After being blocked with 5% milk for 2 h at room temperature, the membranes were incubated overnight at 4 °C with primary antibodies and then incubated with HRP-conjugated secondary antibodies (1:2000; Beyotime, China) for 2 h at room temperature. The primary antibodies used for Western blotting included rabbit anti-OXTR (1:1000, AF4667, Affinity, China), rabbit anti-Occludin (1:1000, A24601, ABclonal, China), rabbit anti-VE-cadherin (1:1000, ab33168, Abcam, USA), rabbit anti-ZO-1 (1:1000, A0659, ABclonal, China), rabbit anti-CXCL3 (1:1000, 86261, Proteintech, China), rabbit anti-ERK (1:1000, A4782, ABclonal, China), rabbit anti-p-ERK (1:1000, AF1015, Affinity, China), rabbit anti-p38 (1:1000, A4771, ABclonal, China), rabbit anti-p-p38 (1:1000, #9211, Cell Signaling Technology, USA), and mouse anti-β-actin (1:20000, 66009, Proteintech, China). ImageJ software 1.80 was used for the analyses.

Evans blue leakage

Blood–brain barrier (BBB) permeability was evaluated using an Evans blue extravasation assay. Evans blue dye (2%, 4 mL/kg; Sigma-Aldrich, USA) was injected into the mice via the tail vein at 3 days after tMCAO. After 4 h of circulation, the mice were deeply anesthetized and transcardially perfused with PBS to remove intravascular dye. The brains were then carefully removed and weighed, followed by fixation in 4% PFA at 4 °C for 24 h. The brains were subsequently cut into six coronal Sect.  (2 mm per slice) and photographed for macroscopic evaluation of Evans blue leakage. For quantitative analysis, the brain tissues were homogenized and incubated in 1 mL of formamide (F103362, Aladdin, China) at 65 °C for 48 h in the dark to extract the dye. After centrifugation at 4000 rpm for 5 min, the supernatant was collected, and the absorbance at 610 nm was measured using a microplate reader. The evans blue content was quantified on the basis of absorbance values and normalized to the tissue weight.

Quantitative real-time PCR (qRT‒PCR)

Peri‑infarct cortex tissues were collected on day 3 after tMCAO. Total RNA was extracted from the peri-infarct cortex using a chloroform substitute reagent (Leagene, China) according to the manufacturer’s instructions. RNA was subsequently reverse transcribed into complementary DNA (cDNA) using the FastPure Cell/Tissue Total RNA Isolation Kit V2 (RC112-01, Vazyme Biotech Co., Ltd., China) with gDNA Eraser to remove genomic DNA contamination. qRT‒PCR was performed using a QuantStudio™ 1 Plus Real-Time PCR System (Applied Biosystems, Thermo Fisher Scientific, USA). The relative mRNA expression levels of target genes were normalized to those of β-actin as the internal control and calculated using the 2^-ΔΔCt method. All reactions were performed in triplicate. The sequences of the primers used were as follows: Cxcl1: forward 5′-GCTGGGATTCACCTCAAGAA-3′, reverse 5′-TGGGGACACCTTTTAGCATC-3′; Cxcl2: forward 5′-GTTTCTGGGGAGAGGGTGAG-3′, reverse 5′-TGTTCTACTCTCCTCGGTGC-3′; Cxcl3: forward 5′-AGACCATCCAGAGCTTGACG-3′, reverse 5′-GGACTTGCCGCTCTTCAGTA-3′; Cxcl5: forward 5′-TTCCTCAGTCATAGCCGCAA-3′, reverse 5′-TGGATCCAGACAGACCTCCT-3′; and β-actin: forward 5′-GGCTGTATTCCCCTCCATCG-3′, reverse 5′-CCAGTTGGTAACAATGCCATGT-3′.

Neurological and behavioral assessments

Neurological and behavioral functions were evaluated before surgery and at days 1, 3, and 7 after tMCAO to assess post-ischemic functional recovery. All the behavioral tests were performed in a quiet environment, and the experimenters were blinded to the group assignments throughout the procedures.

Modified neurological severity score (mNSS)

The mNSS reflects a comprehensive evaluation of the motor function, sensory, balance function and reflexes of experimental animals to assess neurological deficits and the degree of neurological impairment in animals with brain ischemia. The scoring system ranges from 0 to 18 points, with higher scores indicating more severe neurological deficits.

Rotarod test

Motor coordination and balance were assessed using an accelerating rotarod apparatus with a rod diameter of 5 cm. The mice were individually placed on the rotating rod, and after a brief acclimatization period, the rotation speed was gradually increased from 5 rpm to 40 rpm over 120 s. The rotation was then maintained at a constant speed of 40 rpm for an additional 180 s, for a total testing duration of 300 s. The latency to fall (the time each mouse remained on the rod before falling) was automatically recorded.

Corner test

The corner test was performed to evaluate sensorimotor and postural asymmetry following ischemic injury. The mice were allowed to walk freely between two connected boards (30 cm × 20 cm × 1 cm each) positioned at an angle of 30°. When entering the corner, the animal typically rears and turns either to the left or the right to exit. Each mouse was tested 20 times per session, and the number of turns on each side was recorded. To prevent bias, consecutive turns to the same side were not counted. The asymmetry score was calculated using the following formula: right turn ratio= number of right turns/total number of turns.

Statistical analyses

All the statistical analyses were performed using GraphPad Prism 9 (La Jolla, CA, USA). The data are presented as the mean ± SEM. Comparisons between two groups were conducted using two-tailed unpaired Student’s t-tests. For comparisons among multiple groups, one-way or two-way ANOVA was performed as appropriate, followed by Tukey’s post hoc test. Statistical significance was defined as P < 0.05.

Results

Temporal changes in OXT and OXTR levels following tMCAO

We first characterized the temporal profile of OXT levels following tMCAO using ELISA. OXT concentrations in the peri-infarct cortex, serum, and CSF were markedly reduced on day 1 post-injury, followed by a progressive recovery toward baseline by day 7 (Fig. 1A–B, Supplementary Fig. 1A). Given that OXT signaling is mediated primarily through the cognate receptor for OXT, OXTR, we next assessed OXTR protein expression in the peri-infarct cortex. Western blotting analysis revealed the robust upregulation of OXTR on days 1–3 after tMCAO, and OXTR expression subsequently decreased by day 7 but remained elevated compared with that in the sham controls (Fig. 1C–D). Analyses in female mice also showed an early reduction in OXT levels across these compartments, accompanied by a transient upregulation of OXTR expression (Fig. 1E–H, Supplementary Fig. 1B). Therefore, OXT–OXTR signaling was temporally dysregulated following brain I/R, suggesting a potential role in the neurobiological response to ischemic brain injury.

Fig. 1.

Dynamic changes in OXT and OXTR expression following tMCAO. A–B Quantification of OXT levels in the peri-infarct cortex A and serum B of male mice at the indicated time points after tMCAO (n = 6/group; one-way ANOVA). C–D Representative Western blotting images and corresponding quantification of OXTR protein expression in the peri-infarct cortex of male mice following tMCAO (n = 6/group; one-way ANOVA). E–F Quantification of OXT levels in the peri-infarct cortex E and serum F of female mice at the indicated time points after tMCAO (n = 6/group; one-way ANOVA). G–H Representative Western blotting images and corresponding quantification of OXTR protein expression in the peri-infarct cortex of female mice following tMCAO (n = 6/group; one-way ANOVA). I Schematic illustration of stereotaxic viral injection targeting the PVN. J Representative immunofluorescence images showing viral expression at the PVN injection site. K–L Quantification of OXT levels in the peri-infarct cortex K and serum L from male and female mice on day 3 after tMCAO across different groups (n = 6/group; unpaired t-test). All the data are presented as the mean ± SEM. ns, not significant; * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001

Activation of PVN^OXT neurons alleviates I/R injury

The PVN is considered a central hub of endogenous OXT production. We first injected an anterograde tracing virus into the PVN of OXT-Cre mice and observed PVN^OXT-derived fibers in the peri-infarct cortex (Supplementary Fig. 1D). Furthermore, chemogenetic activation was performed by injecting excitatory or control viral vectors into the PVN of OXT-Cre mice, followed by CNO administration after tMCAO (Fig. 1I–J). Notably, OXT levels in both the peri-infarct cortex and serum were significantly increased in the tMCAO + OXT group compared with the tMCAO group on day 3 after tMCAO, in both male and female mice (Fig. 1K–L). OXT concentrations in the CSF were also increased following the activation of PVN^OXT (Supplementary Fig. 1C), which may indicate a potential contribution of centrally derived OXT to the observed effects.

We next assessed whether the activation of PVN^OXT neurons confers neuroprotection following I/R injury (Fig. 2A). MRI analyses revealed a marked reduction in infarct volume in the tMCAO + OXT group on day 3 after tMCAO (Fig. 2B–C). Consistent with this finding, Western blotting and Evans blue assays indicated attenuated BBB disruption (Fig. 2D–I). Behavioral assessments revealed improved neurological outcomes in the tMCAO + OXT mice, including a reduced mNSS and enhanced performance in the rotarod and corner tests (Fig. 2J–L). Consistently, activation of PVN^OXT neurons also exerted a neuroprotective effect in female mice (Supplementary Fig. 2). Collectively, these data demonstrate that the activation of PVN^OXT neurons enhances OXT signaling and is associated with neuroprotective effects against I/R injury.

Fig. 2.

Chemogenetic activation of PVN^OXT neurons attenuates I/R injury in male mice. A Schematic timeline of viral injection, CNO administration, and tMCAO procedures. B–C Representative T2-weighted MR images and quantification of infarct volume at 3 days post-stroke (n = 6/group; unpaired t-test). D–G Representative Western blotting images and densitometric analyses of Occludin, VE-cadherin, and ZO-1 in the peri-infarct cortex (n = 6/group; one-way ANOVA). H–I Representative Evans blue images and quantification of Evans blue extravasation (n = 6/group; one-way ANOVA). J–L Behavioral performance was assessed by the mNSS J, corner test K, and rotarod test L at days 1, 3, and 7 after tMCAO (n = 8/group; two-way ANOVA). All the data are presented as the mean ± SEM. ns, not significant; * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001

Activation of PVN^OXT neurons suppresses CXCL3 expression and neutrophil chemotaxis

To elucidate the molecular mechanisms underlying the neuroprotective effects of PVN^OXT neurons following I/R injury, RNA-seq was performed on ipsilateral ischemic hemisphere tissues from male mice collected on day 3 after tMCAO. Sample-to-sample distance analysis revealed the clear segregation of transcriptional profiles between the two groups (Fig. 3A). Differential expression analysis, as visualized by volcano plots and heatmaps, revealed CXCL3 as one of the most prominently downregulated genes in the tMCAO + OXT group compared with the tMCAO control group (Fig. 3B–C).

Fig. 3.

Transcriptomic profiling reveals the suppression of CXCL3-mediated neutrophil chemotaxis after PVN^OXT neuron activation. A RNA-seq was performed on ipsilateral ischemic hemisphere tissues from male mice on day 3 after tMCAO. Sample-to-sample distances indicated clear separation between the tMCAO and tMCAO + OXT groups. B–C Volcano plot and heatmap of genes that were differentially expressed between the tMCAO and tMCAO + OXT groups. D GO enrichment analysis indicating the downregulation of pathways related to chemokine activity, CXCR chemokine receptor binding, and neutrophil chemotaxis. E Chord diagram illustrating the association of CXCL3 with gene networks involved in neutrophil migration and chemotaxis. F GSEA showing significant suppression of the neutrophil migration pathway in the tMCAO + OXT group

CXCL3 is a key chemokine that signals through CXCR2 to promote neutrophil recruitment. Consistent with this role, chord diagram analysis linked CXCL3 to gene networks enriched in neutrophil migration and chemotaxis (Fig. 3E). GO enrichment analysis further demonstrated the significant downregulation of terms related to “chemokine activity”, “CXCR chemokine receptor binding”, and “neutrophil chemotaxis” (Fig. 3D). In parallel, GSEA revealed the marked suppression of “neutrophil migration” pathways in the tMCAO + OXT group (Fig. 3F).

To determine whether these transcriptomic changes translated into cellular effects, flow cytometric analysis of cells isolated from the ipsilateral ischemic hemisphere revealed that the activation of PVN^OXT neurons significantly reduced the proportion of infiltrating neutrophils, whereas the percentages of microglia and monocytes/macrophages remained largely unaffected (Supplementary Fig. 3B–E). This effect was further corroborated at the tissue level by immunofluorescence analysis, which revealed decreased neutrophil accumulation in the peri-infarct cortex of tMCAO + OXT mice (Supplementary Fig. 3F–G). Collectively, these findings indicate that the activation of PVN^OXT neurons may suppress CXCL3 expression and limit neutrophil recruitment, thereby contributing to the attenuation of I/R injury.

Activation of PVN^OXT neurons suppresses microglial CXCL3 expression via inhibition of ERK phosphorylation

To validate our previous transcriptomic findings, we first examined the mRNA expression and temporal dynamics of CXCL family members in the peri-infarct cortex. Following PVN^OXT neuron activation, the expression of CXCL3 was among the most strongly downregulated at the mRNA level, whereas the expression of CXCL2 decreased modestly, and the expression of CXCL1 and CXCL5 was unaffected (Supplementary Fig. 4A–D). Consistent with these changes in mRNA expression, ELISA revealed that in the peri-infarct cortex, CXCL3 levels peaked on day 3 after tMCAO and had returned to baseline by day 7, indicating a temporal pattern inverse to that of OXT (Fig. 4A). On day 3, CXCL3 levels were significantly elevated in tMCAO mice compared with those in sham mice and were markedly reduced by PVN^OXT neuron activation (Fig. 4B). A similar temporal pattern was also observed in serum (Supplementary Fig. 4E–G), suggesting that CXCL3 is a key downstream effector of PVN^OXT signaling.

Fig. 4.

PVN^OXT neuron activation reduces microglial CXCL3 expression after tMCAO. A ELISA quantification of CXCL3 and OXT levels in the peri-infarct cortex of mice at different time points after tMCAO. B CXCL3 levels on day 3 after tMCAO in different experimental groups (n = 6/group; one-way ANOVA). C Representative flow cytometry plots of cells isolated from the ipsilateral ischemic hemisphere showing CXCL3 expression in microglia from different experimental groups. D Proportion of CXCL3⁺ microglia among total microglia (n = 6/group; one-way ANOVA). E MFI of CXCL3 in microglia from different experimental groups (n = 6/group; one-way ANOVA). F–G Representative immunofluorescence images of CXCL3/Iba-1 double staining in the peri-infarct cortex and corresponding quantification of CXCL3⁺ microglia. Scale bar = 100 μm (n = 6/group; one-way ANOVA). All the data are presented as the mean ± SEM. * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001

Previous studies have shown that OXT suppresses LPS-induced microglial inflammation [27]. On this basis, we next investigated whether PVN^OXT neurons regulate microglial CXCL3 expression following tMCAO. Flow cytometry analysis of cells isolated from the ipsilateral ischemic hemisphere revealed that compared with the sham treatment, tMCAO significantly increased both the proportion of CXCL3⁺ microglia and their mean fluorescence intensity (MFI). Notably, PVN^OXT neuron activation markedly reduced both the proportion and the MFI of CXCL3⁺ microglia (Fig. 4C–E). A similar effect was observed in female mice (Supplementary Fig. 5). Consistent with these findings, immunofluorescence staining further confirmed that the number of CXCL3⁺ microglia in the peri-infarct cortex was substantially lower in the tMCAO + OXT group than in the tMCAO group (Fig. 4F–G). Collectively, these results demonstrate that PVN^OXT neuron activation is associated with suppression of microglial CXCL3 expression following tMCAO.

To further elucidate the underlying mechanism, we performed in vitro experiments using primary microglia subjected to OGD/R (Fig. 5A). Flow cytometry and Western blotting analyses revealed that compared with OGD/R alone, OXT treatment significantly reduced CXCL3 expression. This inhibitory effect was largely abolished by co-treatment with the OXTR antagonist L-368,899 (Fig. 5B–E), suggesting that OXT primarily acts through OXTR to regulate CXCL3 expression. Given previous evidence implicating ERK signaling in CXCL3 transcription [13], we next examined ERK activation. Western blotting analysis demonstrated that OXT markedly reduced ERK phosphorylation (p-ERK) in microglia following OGD/R (Fig. 5D, F). Pharmacological inhibition of ERK with U0126 did not further decrease CXCL3 levels beyond the decrease caused by OXT treatment alone (Fig. 5G–I), indicating that ERK signaling is involved in the regulation of CXCL3 expression. We also assessed the activity of the p38 pathway, another regulator of chemokine transcription. The results suggested that OXT may also modestly affect p38 signaling, although its potential role in regulating CXCL3 requires further investigation (Supplementary Fig. 6). Collectively, these findings indicate that activation of PVN^OXT neurons may attenuate I/R injury through OXT–OXTR signaling in microglia, in which ERK-associated regulation of CXCL3 expression appears to play an important role.

Fig. 5.

OXT suppresses microglial CXCL3 expression via the OXTR–ERK signaling pathway in vitro. A Schematic illustration of the OGD/R model and treatment conditions (OXT, L-368,899). B Representative flow cytometry fluorescence intensity distribution plots showing CXCL3 expression in primary microglia under different treatment conditions. C Quantification of the CXCL3 MFI (n = 6/group; one-way ANOVA). D–F Representative Western blotting images and densitometric analyses of CXCL3, ERK, and p-ERK in primary microglia under the indicated conditions (n = 6/group; one-way ANOVA). G Schematic illustration of the OGD/R model and treatment conditions (OXT, U0126). H Representative flow cytometry fluorescence intensity distribution plots showing CXCL3 expression under ERK inhibition conditions. I Quantification of the CXCL3 MFI (n = 6/group; one-way ANOVA). All the data are presented as the mean ± SEM. ns, not significant;** p < 0.01; *** p < 0.001; **** p < 0.0001

OXT reduces neutrophil chemotaxis by suppressing the CXCL3–CXCR2 axis

To further determine whether OXT regulates neutrophil chemotaxis through CXCL3 signaling, we investigated the transcriptomic dataset. PPI analysis revealed a strong association between CXCL3 and its cognate receptor CXCR2 (Fig. 6A). Consistent with these findings, flow cytometry analysis of cells isolated from the ipsilateral ischemic hemisphere revealed that both the number and percentage of CXCR2⁺ neutrophils were significantly greater in the tMCAO group than in the sham group, and these increases were accompanied by an increase in the CXCR2 MFI (Fig. 6B–F). The activation of PVN^OXT neurons markedly reduced the overall number of CXCR2⁺ neutrophils, without significantly affecting their MFI or relative proportion, suggesting that OXT primarily limits neutrophil infiltration rather than directly modulating CXCR2 expression.

Fig. 6.

OXT inhibits neutrophil chemotaxis via the CXCL3–CXCR2 axis. A PPI network illustrating the association between CXCL3 and CXCR2. B Representative flow cytometry plots showing CXCR2 expression in neutrophils from different experimental groups. C Quantification of CXCR2⁺ neutrophils per 1 million total cells (n = 6/group, one-way ANOVA). D Quantification of the proportion of CXCR2⁺ cells within the neutrophil population (n = 6/group; one-way ANOVA). E Representative flow cytometry fluorescence intensity distribution plots showing CXCR2 expression in neutrophils under different treatment conditions. F Quantification of the CXCR2 MFI in neutrophils (n = 6/group; one-way ANOVA). G Schematic illustration of the Transwell migration assay. H Quantification of neutrophil migration under the indicated conditions (n = 6/group; one-way ANOVA). All the data are presented as the mean ± SEM. ns, not significant; * p < 0.05; ** p < 0.01; **** p < 0.0001

To functionally validate these findings, a Transwell migration assay was performed. Primary microglia were subjected to OGD/R and treated with OXT in the presence or absence of the OXTR antagonist L-368,899. Conditioned media were then added to the lower chamber of a Transwell system containing purified neutrophils (Fig. 6G). Neutrophil migration toward OXT-treated microglial supernatants was significantly reduced, whereas OXTR blockade restored migratory capacity (Fig. 6H), indicating that OXT regulates microglia-dependent neutrophil chemotaxis via OXTR signaling. To directly assess the role of CXCL3, microglial CXCL3 expression was silenced using siRNA (si_CXCL3). Compared with OGD/R, CXCL3 knockdown significantly attenuated neutrophil migration (Fig. 6H). Notably, combined CXCL3 silencing and OXT treatment did not result in an additive inhibitory effect, further suggesting that OXT primarily exerts its effect by regulating CXCL3. Collectively, these findings demonstrate that OXT attenuates neutrophil chemotaxis primarily through suppression of the CXCL3–CXCR2 signaling axis.

Exogenous CXCL3 administration exacerbates neutrophil infiltration and I/R injury

To further validate the functional role of CXCL3 in neutrophil recruitment and I/R injury in vivo, recombinant CXCL3 protein was administered via tail vein injection for three consecutive days following tMCAO (Fig. 7A). Flow cytometry analysis of cells isolated from the ipsilateral ischemic hemisphere revealed that exogenous CXCL3 significantly increased the number and percentage of CXCR2⁺ neutrophils compared with those in the tMCAO group, and the MFI of CXCR2 was also increased (Supplementary Fig. 7), indicating enhanced neutrophil recruitment.

Fig. 7.

Exogenous CXCL3 administration exacerbates I/R injury in male mice. A Schematic timeline of viral injection, CXCL3 administration, and tMCAO procedures. B–C Representative T2-weighted MR images and quantification of infarct volume at 3 days poststroke (n = 6/group; unpaired t test). D–G Representative Western blot images and the results of densitometric analyses of Occludin, VE-cadherin, and ZO-1 in the peri-infarct cortex (n = 6/group; one-way ANOVA). H-I Representative Evans blue images and quantification of Evans blue extravasation (n = 6/group; one-way ANOVA). J–L Behavioral performance was assessed by the mNSS J, corner test K, and rotarod test L at days 1, 3, and 7 after tMCAO (n = 8/group; two-way ANOVA). All the data are presented as the mean ± SEM. ns, not significant; * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001

We next assessed the effect of CXCL3 on ischemic brain injury. MRI revealed that compared with tMCAO mice, CXCL3-treated mice exhibited significantly greater infarct volumes on day 3 after tMCAO (Fig. 7B–C). Consistent with these findings, western blot analyses revealed exacerbated BBB disruption following CXCL3 administration (Fig. 7D–G), which was further confirmed by increased Evans blue extravasation (Fig. 7H–I). Behavioral assessments revealed that compared with control mice, CXCL3-treated mice displayed impaired neurological recovery, as evidenced by higher mNSS values and poorer performance on the rotarod and corner tests (Fig. 7J–L). Similarly, injection of CXCL3 also exacerbated brain infarction and neurological injury in female mice (Supplementary Fig. 8). Collectively, these results demonstrate that exogenous CXCL3 enhances neutrophil infiltration and exacerbates cerebral I/R injury, further supporting a pathogenic role for CXCL3 in this process.

Discussion

This study provides a comprehensive evaluation of the neuroprotective role of PVN^OXT neurons in cerebral I/R injury and elucidates the mechanisms underlying this role. We observed a pronounced and early reduction in OXT levels in the peri-infarct cortex, serum, and CSF following tMCAO, accompanied by marked upregulation of OXTR. Chemogenetic activation of PVN^OXT neurons restored OXT levels across these compartments, reduced infarct volumes, preserved BBB integrity, and improved neurological outcomes. Mechanistically, transcriptomic and cytological analyses revealed that OXT suppresses microglial ERK phosphorylation, leading to the downregulation of CXCL3 expression. This decrease in CXCL3 expression attenuates neutrophil recruitment, thereby mitigating neuroinflammation and alleviating I/R injury.

Endogenous OXT levels are reduced in several neurological disorders, including aneurysmal subarachnoid hemorrhage, intracerebral hemorrhage, and autism [18, 22, 28]. Consistent with these observations, we found that OXT concentrations markedly decreased after tMCAO, accompanied by the upregulation of OXTR during the early phase (days 1–3). The reduction in OXT may result from impaired PVN neuronal activity, depleted of peptide stores, or disrupted axonal transport, although these mechanisms were not investigated directly in this study. In contrast, increased OXTR expression may be associated with injury-induced neuroinflammatory signaling, which could enhance cellular responsiveness to OXT. In support of this hypothesis, inflammatory stimuli upregulate OXTR via NF-κB–dependent pathways [29], suggesting that OXTR upregulation may represent a compensatory, anti-inflammatory response following brain injury.

OXT is synthesized in the hypothalamic PVN and widely distributed throughout the brain [30, 31]. Exogenous OXT administration has been shown to attenuate neuroinflammation and neuronal injury in models of both ischemic and hemorrhagic stroke, thereby promoting functional recovery [32–34]. However, whether the direct modulation of PVN^OXT neuronal activity confers neuroprotection and the underlying neuroimmune mechanisms remain unclear. In the present study, chemogenetic activation of PVN^OXT neurons significantly increased OXT levels in the peri-infarct cortex, serum, and CSF, accompanied by reduced brain injury and improved neurological outcomes. Given that OXT signaling can be influenced by sex hormones [35, 36], we additionally assessed the effects in female mice. Activation of PVN^OXT neurons was associated with protective effects in female mice, although further studies are needed to clarify potential sex-related influences and differences. Collectively, these findings suggest that PVN^OXT neurons may represent a potential therapeutic target for ischemic stroke.

Recent studies have shown that OXT attenuates microglial activation and neuroinflammatory responses through multiple signaling pathways. For example, the OXTR agonist carbetocin alleviates early microglial activation and confers long-term neuroprotection [37], whereas OXT suppresses LPS-induced microglial activation by inhibiting ERK and p38 phosphorylation, thereby reducing pro-inflammatory cytokine release [27]. In the present study, our data indicate that the activation of PVN^OXT neurons significantly reduced microglial CXCL3 expression. Further analysis revealed a distinct temporal relationship between OXT and CXCL3 following ischemic injury: OXT levels reached their nadir in day 1 after tMCAO, whereas CXCL3 expression peaked on day 3, suggesting a sequential response during the acute phase. The delayed increase in CXCL3 expression likely reflects a secondary inflammatory process requiring transcriptional and translational activation. Mechanistically, these findings are consistent with a role for OXT in modulating ERK phosphorylation and CXCL3 expression in microglia. OXTR antagonism partially reversed these effects without fully restoring ERK activity to baseline levels, suggesting the involvement of additional regulatory mechanisms, including potential receptor cross-talk between OXTR and vasopressin receptors [38]. Notably, pharmacological inhibition of ERK did not further reduce CXCL3 levels beyond the reduction induced by OXT treatment, providing additional support for the involvement of ERK signaling in this process.

CXCL3 exerts its biological effects mainly through the effects of CXCR2 on neutrophils, thereby promoting neutrophil recruitment and activation [16]. Although the CXCL3–CXCR2 axis has been implicated in various inflammatory conditions [39, 40], its role in ischemic stroke remains insufficiently defined. In the present study, exogenous CXCL3 administration significantly enhanced CXCR2⁺ neutrophil infiltration in vivo, whereas OXT treatment or microglial CXCL3 knockdown markedly reduced neutrophil chemotaxis in vitro. These findings identify CXCL3 as a critical downstream mediator linking microglial activation to neutrophil recruitment. Through modulation of the CXCL3–CXCR2 axis, OXT indirectly limits neutrophil infiltration and attenuates neuroinflammation following ischemic injury.

In alignment with the findings of previous studies [41, 42], our results further support the neuroprotective effects of PVN^OXT neuronal activation. Although the deep anatomical location of the PVN has historically limited clinical translation, recent advances in noninvasive neuromodulation techniques, such as precise magnetic stimulation systems (pMSS), have enabled the targeted modulation of deep brain structures [43, 44]. Notably, pMSS has been shown to activate PVN neurons, increase endogenous OXT release, and improve neurological outcomes in disease models [45, 46]. These advances suggest that targeting PVN^OXT neurons may represent a feasible and translatable strategy for modulating neuroinflammation after stroke.

This study has several limitations. First, microglia were identified by flow cytometry primarily as CD45^intCD11b⁺ cells, without the use of more specific markers such as TMEM119 or P2RY12. This conventional gating may not fully capture microglial heterogeneity after ischemic stroke, limiting the specificity of our interpretation. Second, whether OXT regulates CXCL3 expression in other cell types and how intercellular interactions contribute to I/R pathology require further investigation. Third, PVN neurons may be indirectly influenced by stroke-related signals, such as systemic inflammation or activation of the stress axis [47, 48]. Although we demonstrated dynamic changes in OXT levels following stroke and showed that modulation of PVN^OXT neuronal activity improves stroke outcomes, the activity status of PVN^OXT neurons after ischemic injury, as well as the mechanisms underlying the reduction in OXT, remain to be fully elucidated. Future studies should address these questions to further clarify the regulatory network of the OXT system.

In conclusion, the results of this study demonstrate that the activation of PVN^OXT neurons attenuates cerebral I/R injury by suppressing microglial CXCL3 expression via the OXT–OXTR–ERK signaling pathway. This effect reduces neutrophil recruitment, preserves BBB integrity, and improves neurological outcomes. These findings highlight a previously unrecognized neuroendocrine–immune regulatory axis and provide a mechanistic basis for developing neuromodulation-based therapeutic strategies for ischemic stroke.

Abbreviations

OXT

Oxytocin

OXTR

Oxytocin receptor

I/R

Ischemia–reperfusion

PVN

Hypothalamic paraventricular nucleus

BBB

Blood–brain barrier

CXCL3

C-X-C motif chemokine ligand 3

CXCR2

C-X-C chemokine receptor 2

NET

Neutrophil extracellular trap

CIBR

Chinese Institute for Brain Research

IACUC

Institutional Animal Care and Use Committee

AAV

Adeno-associated viral

tMCAO

Transient middle cerebral artery occlusion

CCA

Common carotid artery

ECA

External carotid artery

ICA

Internal carotid artery

MCA

Middle cerebral artery

CBF

Cerebral blood flow

CNO

Clozapine-N-oxide

siRNA

Small interfering RNAs

OGD/R

Oxygen-glucose deprivation/reoxygenation

MRI

Magnetic resonance imaging

RNA-seq

RNA sequencing

DEGs

Differentially expressed genes

GO

Gene Ontology

PPI

Protein–protein interaction

PFA

Paraformaldehyde

BSA

Bovine serum albumin

mNSS

Modified neurological severity score

SON

Supraoptic nucleus

pMSS

Precise magnetic stimulation system

MFI

Mean fluorescence intensity

CSF

Cerebrospinal fluid

BBB

Blood–brain barrier

qRT‒PCR

Quantitative real-time PCR

DMEM

Dulbecco’s modified Eagle’s medium

FBS

Fetal bovine serum

Authors' contributions

RL and Xy-Y contributed to study design, performed the majority of experiments, analyzed data, prepared figures, and drafted the manuscript. Hz-G, Hj-F, Xb-F, Jy-L, and Wt-G assisted with animal surgeries, behavioral testing, molecular experiments, imaging, and data analysis. Xt-L, Yp-S and Yb-W contributed to RNA-seq experiments, bioinformatics analysis, and interpretation of transcriptomic data. WW, Jq-J, and Lq-J conceived and supervised the study, secured funding, and critically revised and approved the final manuscript. All authors read and approved the final manuscript.

Funding

This study was supported by grants from the National Natural Science Foundation of China (8240051347), the Beijing Natural Science Foundation (L252081), and the Capital Medical Science Innovation Center Organized Research Special Program (CX25YZ05).

Data availability

All data analyzed and presented in this study are available from the corresponding author on reasonable request. The raw sequence data reported in this paper have been deposited in the Genome Sequence Archive (Genomics, Proteomics & Bioinformatics 2025) in National Genomics Data Center (Nucleic Acids Res 2025), China National Center for Bioinformation / Beijing Institute of Genomics, Chinese Academy of Sciences (GSA: CRA033671) that are publicly accessible at https://ngdc.cncb.ac.cn/gsa.

Declarations

Ethics approval and consent to participate

All experimental procedures were conducted at the Chinese Institute for Brain Research (CIBR) and approved by the Institutional Animal Care and Use Committee (IACUC) of CIBR (Approval #: CIBR-IACUC-121). Data were acquired and reported in accordance with the ARRIVE 2.0 guidelines.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.