CircRNA circ_0004058 Modulates Early Brain Injury in Subarachnoid Hemorrhage Through miR-221-3p and VE1 Activation Pathway

Abstract

Subarachnoid hemorrhage (SAH) frequently results in early brain injury (EBI), which remains a major barrier to favorable neurological recovery. Understanding the molecular underpinnings of EBI is crucial for developing targeted therapeutics. Circular RNAs (circRNAs) have emerged as influential molecular players in various brain injury contexts. This study focuses on one such molecule, circ_0004058, examining its impact on EBI through interaction with miR-221-3p and the VE1 signaling pathway. Utilizing an established SAH rodent model, our team conducted a detailed investigation of the expression patterns and interactions involving circ_0004058. Our analyses revealed a significant post-SAH upregulation of circ_0004058, which affected miR-221-3p activity and VE1 signaling. Furthermore, functional modulation of circ_0004058 expression altered the severity of EBI, presenting evidence that it serves as a critical determinant in the injury process. The results suggest that circ_0004058 holds promise as a therapeutic target, offering new possibilities for the development of strategies to mitigate SAH-induced brain damage. Through this study, circ_0004058 is highlighted not only as a biomarker but also as a possible avenue for therapeutic modulation in SAH management.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12975-025-01383-9.

Introduction

Subarachnoid hemorrhage (SAH) represents a serious type of hemorrhagic stroke [1–3]. Within the initial 72 h post-SAH, commonly defined as the early brain injury (EBI) period, secondary brain injury occurs along with microcirculatory dysfunction, neuroinflammation, blood–brain barrier (BBB) breakdown, cerebral edema, and neuronal death [4]. There is a need to elucidate the pathological mechanisms involved in EBI and identify potential targets for SAH therapies.

Circular RNAs (circRNAs) are a class of endogenous non-coding RNAs that exert crucial regulatory functions in the development and progression of neurological disorders. Growing evidence suggests that a wide range of circRNAs exhibit altered expression profiles after stroke; these dysregulated circRNAs can influence BBB integrity and serve as potential biomarkers for disease monitoring [5, 6]. Because abnormally expressed circRNAs can also promote vascular smooth muscle cell remodeling, they are potential targets for cerebral hemorrhage therapies [7, 8]. This mechanistic study investigated the roles of a novel circRNA, circ_0004058 in modulating EBI following SAH in a rat model.

Bioinformatic analysis using the starBase database revealed putative binding sites for miR-221-3p on circ_0004058. microRNAs (miRNAs) are small non-coding RNAs with key regulatory roles in human diseases because of their ability to modulate gene expression [9]. miR-221-3p reportedly promotes vascular endothelial cell proliferation and angiogenesis [10]. Furthermore, miR-221-3p is strongly linked to immune cell dynamics, notably promoting the polarization of M2 macrophages toward a pro-inflammatory state by suppressing the JAK3/STAT3 activation [11].

In a clinical study, miR-221-3p expression was upregulated in patients with SAH [12]. This upregulation implies that miR-221-3p may function as a potential biomarker for predicting neurological prognosis following SAH.

Target prediction analysis using the miRWalk database identified the lymphatic vessel endothelial hyaluronan receptor-1 (LYVE1) gene as a potential downstream target of miR-221-3p. LYVE1 is a major receptor for hyaluronan in lymphatic vessel endothelial cells [13]. LYVE1 can mediate dendritic cell transport in vivo [14] and regulate downstream signaling pathways that affect endothelial cell permeability; thus, LYVE1 may be closely associated with angiogenesis [15]. Notably, the recruitment and accumulation of bone marrow-derived LYVE1-positive macrophages have been shown to be critical for establishing a dense and functional vascular network during neovascularization [16]. Furthermore, SAH induces extensive immune cell infiltration into brain tissues [17].

In this study, we hypothesized that LYVE1 is a key mediator in the pathophysiology of SAH. Accordingly, we investigated the regulatory interactions among circ_0004058, miR-221-3p, and LYVE1, which also explored potential molecular mechanisms of circ_0004058/miR-221-3p/LYVE1 regulatory network involvement in SAH-induced EBI.

Methods

Establishment of the SAH Rat Model

A This study was reviewed and approved by the Institutional Animal Ethics Committee of Huzhou Institute for Drug Control (Animal Use License No.: SYXK (Zhe) 2018–0015). A total of 288 male SPF Sprague–Dawley (SD) rats (8 weeks old, 300–350 g) were obtained from Beijing Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China). The animals were housed in groups under standard laboratory conditions (humidity 60%–65%, temperature 22–25 °C, 12-h light/dark cycle) and acclimated for one week prior to experiments.

The SAH model was established using the endovascular perforation technique. Briefly, rats were anesthetized with isoflurane (5% induction, 3% maintenance in 65% air and 35% oxygen), and the right common carotid artery, external carotid artery (ECA), and internal carotid artery (ICA) were surgically exposed. A 4–0 nylon filament was introduced via the ECA into the ICA and advanced 2 mm beyond resistance to induce perforation. In the sham group, the filament was inserted without perforation. Post-surgery, animals were recovered in a Heated cage. SAH induction was confirmed at 24, 48, and 72 h using neurological scoring, Evans blue extravasation assay, and SAH grading. Neurological deficits were assessed using a modified Garcia scoring system, with scores ranging from 3 to 18, where lower scores indicate more severe impairment. For Evans blue quantification, 2% Evans blue dye (4 mL/kg) was injected via the tail vein and allowed to circulate for 2 h before sacrifice. Brains were removed, homogenized in formamide, and incubated at 55 °C for 24 h. The supernatant was collected after centrifugation, and absorbance was measured at 620 nm with a spectrophotometer to quantify blood–brain barrier permeability [18].

Rats were randomly assigned to 16 experimental groups (n = 18/group):Sham, SAH, oe-NC (SAH + lentiviral negative control), oe-LYVE1 (SAH + LYVE1 overexpression), NC inhibitor + sh-NC, miR-221-3p inhibitor + sh-NC, miR-221-3p inhibitor + sh-LYVE1, oe-NC + NC mimic, oe-circ_0004058 + NC mimic, oe-circ_0004058 + miR-221-3p mimic, oe-NC + sh-NC, oe-circ_0004058 + sh-NC, oe-circ_0004058 + sh-LYVE1, oe-NC + anti-IgG (IgG control), oe-LYVE1 + anti-IgG, and oe-LYVE1 + anti-CSF1R. All SAH groups received the indicated lentiviral constructs or antibody treatments. The sh-LYVE1 sequence was 5'-CCGGCCAGGTGTCATGCAGAATTATCTCGAGATAATTCTGCATGACACCTGGTTTTTG-3', and the sh-NC sequence was 5'-UUUGGUGGGUAGUAAUGGGUUCGUA-3'. In each group, 8 rats were allocated for Evans blue staining and 8 for other assays. Lentiviral vectors (1 × 10⁸ TU/mL, 3 μL per side) were stereotactically injected into the bilateral lateral ventricles three days prior to SAH induction. Lentiviral injections were administered using a Hamilton syringe through a burr hole positioned 1.0 mm posterior, 1.5 mm lateral, and 2.5 mm ventral to bregma. The injection rate was maintained at 0.2 µL/min. Following injection, the syringe was left in place for 5 min to prevent cerebrospinal fluid reflux. The burr hole was sealed with bone wax, and the scalp incision was sutured [19, 20]. Lentiviruses were supplied by Shanghai GeneChem Co., Ltd. (Shanghai, China; titer 1 × 10¹¹ PFU, injection volume ~ 50 µL per rat). Antibodies were administered by intracerebroventricular injection two days before SAH induction: IgG (GTX00612-00, MOPC-21, GeneTex) and CSF1R antibody (BE0213, clone AFS98, BioXCell) were administered at 30 mg/kg based on body weight [21, 22]. All surgical and injection procedures were performed under sterile conditions. Exclusion criteria: Rats that did not survive surgery or did not meet the neurological criteria evaluated by the Garcia scoring system (score > 15) were excluded from further analysis (Table S1).

Validation of the Intravascular Perforation SAH Model

To validate the success of the intravascular perforation-induced SAH model, we conducted a series of evaluations, including neurological scoring, cerebral edema measurement, and histopathological examination. The neurological deficits were scored using a Double-blind method based on the Garcia scoring system at 24, 48, and 72 h post-SAH. A lower score indicated more severe neurological impairment, confirming the successful induction of SAH. Cerebral edema was assessed by measuring brain water content via the wet-dry weight method. Histopathological validation was performed using Evans blue staining to assess BBB integrity, Nissl staining to evaluate neuronal injury, and TUNEL staining for the detection of apoptotic cells in hippocampal tissue.

Transcriptome Sequencing and Identification of Differentially Expressed Genes (DEGs) in SAH

Total RNA was isolated from samples of rat brain tissue (SAH: n = 10; sham-operated: n = 10; full description of modeling is provided in the section “Construction of SAH Rat Model”), then quantified using a Qubit® RNA Assay Kit (Shanghai Baoji Biotechnology, Shanghai, China, HKR2106-01) with the Qubit®2.0 Fluorometer (Life Technologies, Q33216), a Nanometer spectrophotometer (IMPLEN), and an RNA Nano 6000 Assay Kit (Agilent, 5067–1511) on the Bioanalyzer 2100 platform. For Library construction, 3 μg of total RNA per sample was processed using the NEBNext® Ultra™ RNA Library Prep Kit for Illumina® (E7435L, New England Biolabs, Beijing, China). Indexed libraries were clustered on the cBot Cluster Generation System using the TruSeq PE Cluster Kit v3-cBot-HS (PE-401–3001, Illumina) and sequenced on the Illumina HiSeq 550 system to produce 125/150 bp paired-end reads. High-quality reads were aligned to the rat reference genome using Hisat2.

DEGs between SAH and sham groups were identified with the Limma package in R, with the following thresholds: |log(fold change)|> 1 and p < 0.05. Volcano plots were constructed using the ggplot2 package in R software.

Microarray-Based Gene Expression Profiling

Microarray-based gene expression profiling was performed using datasets GSE161870 and GSE161913 from the Gene Expression Omnibus (GEO) database. SAH-associated genes were identified via the GeneCards database and intersected with DEGs from RNA-seq analysis, leading to the identification of LYVE1 as a key SAH-related gene. Upstream miRNAs targeting LYVE1 were predicted using the miRWalk database and cross-referenced with miRNA expression data from GSE161870, resulting in the selection of miR-221-3p. Subsequently, circBank was used to predict circRNAs targeting miR-221-3p, and an overlapping analysis with circRNA profiles from GSE161913 identified circ_0004058. Thus, an SAH-specific circRNA–miRNA–mRNA competitive endogenous RNA (ceRNA) network was constructed. The information of database is shown in Table S2.

SAH Grading Assessment

SAH severity was evaluated using a previously established grading system based on the extent of subarachnoid blood in six regions of the basal cistern [23]. Each region was scored from 0 to 3: 0 for no blood, 1 for slight bleeding, 2 for moderate clotting with visible arteries, and 3 for dense clotting obscuring all arteries. The cumulative score was calculated, and animals with a total score > 8 were considered to have moderate to severe SAH and included for further analysis. Grading was performed by investigators blinded to group allocation.

Neurological Function Scoring and Assessment of Cerebral Edema

Neurological Function was assessed at 24, 48, and 72 h post-SAH using the Garcia scoring system, which assesses six parameters: spontaneous activity, symmetry of limb movement, forepaw extension, climbing, body proprioception, and response to vibrissae stimulation [18] (Table S3). Each rat was scored three times, with a maximum total score of 18; lower scores indicated greater neurological impairment. The 72-h timepoint was selected for subsequent animal experiments.

Following behavioral evaluation, rats were euthanized under deep anesthesia, and brains were harvested for edema measurement. Fresh brain tissue was weighed (wet weight), then dried at 120 °C for 48 h to obtain dry weight. Brain water content was calculated using the formula: (wet weight – dry weight)/wet weight × 100% [24, 25].

Evans Blue Staining

At 24 h post-SAH, Evans blue dye (4% in 0.9% saline, 2 mL/kg; Sigma) was administered via the right femoral vein. After 3 h, rats were transcardially perfused with saline to eliminate intravascular dye. The ipsilateral Hemisphere was collected, homogenized in 1mL of 50% trichloroacetic acid, and centrifuged. The resulting supernatant was diluted 1:4 with ethanol. The dye concentration was measured with a fluorescence reader (emission: 680 nm, excitation: 620 nm) [26].

Nissl Staining

Paraffinized brain tissue sections were dewaxed and hydrated. Slides were transferred to a solution of toluidine blue (Sigma, MFCD00011934) at 60 °C; soaked in distilled water; dehydrated in serial solutions of 70% ethanol, 95% ethanol, and 100% ethanol; and permeabilized in xylene. Next, the sections were mounted with DPX medium and examined under a light microscope (Bingyu Optical Instruments Co., Ltd., Shanghai, China, XP-330) [27].

TUNEL Staining

Neuronal apoptosis in brain tissues was detected using a TUNEL Assay Kit (ab66110, Abcam, Cambridge, MA, USA). Brain tissue sections were permeabilized in 0.05% Triton X-100 (P0096, Beyotime, Shanghai, China) for 30 min, then exposed to equilibration buffer and terminal deoxyribonucleotidyl transferase (TdT); the mixture without TdT was used as an NC. Subsequently, 50 μL of TUNEL detection solution and the corresponding secondary antibodies were added; sections were incubated with the mixtures for 60 min at 37 °C in the dark. Neuron nuclei of the hippocampus were stained with 4',6-diamidino-2-phenylindole (DAPI; 5 mg/mL). After sealing with anti-fade mounting medium, slides were visualized using a Zeiss LSM 510 META confocal microscope (Carl Zeiss). The numbers of TUNEL-positive cells were expressed as percentages [28, 29].

Flow Cytometry

Rat brain tissue sections were prepared in a McIlwain tissue slicer and enzymatically digested in serum-free medium containing collagenase type A (3 mg/mL, Roche) and DNase I (25 μg/mL, Sigma) at 37 °C for 1 h in a shaking water bath. Next, sections were incubated with fluorescein-conjugated antibodies to LYVE1⁺ (Alexa Fluor™ 488, 53–0443-82, Invitrogen) and CD68⁺ (Alexa Fluor™ 647, 51–0689-42, Invitrogen) for 1 h in the dark, then resuspended in 0.5 mL of phosphate-buffered saline. Data were acquired on a BD LSRII flow cytometer using Diva Software (BD Biosciences) and analyzed with FlowJo software (version 9.9.6) [30].

Isolation and Culture of Primary Rat Hippocampal Neurons

Primary hippocampal neurons were isolated from embryonic day 17.5 rat embryos. Briefly, the hippocampus was dissected and cut into pieces, which were digested with 0.25% ethylene diamine tetra acetic acid (EDTA)-free trypsin at 37 °C for 20 min and gently shaken every 5 min. After the removal of tissue fragments, cells were suspended in Dulbecco's modified Eagle medium (DMEM) containing 20% fetal bovine serum(FBS) and seeded in poly-L-lysine (Sigma, 50 μg/mL)-precoated dishes at 37 °C for 4 h to allow adhesion. Next, cells were incubated in basal neural medium (1% GlutaMAX and 2% B27, Gibco) in a humidified chamber at 37 °C with 5% CO₂. Cells were examined by immunofluorescence staining using an anti-MAP2 antibody (rabbit, A17409, 1:200, ABclonal) [31–33].

HEK-293 T Cell Culture

HEK-293 T cells (ATCC) were cultured in DMEM (10,569,044, Gibco) containing 10% FBS (10,099,141, Gibco), 2 mM L-glutamine (Sigma), 100 U/mL penicillin, and 100 μg/mL streptomycin at 37 °C with 5% CO₂.

Isolation and Culture of Primary rBMECs

Primary rat brain microvascular endothelial cells (rBMECs) were isolated from cerebral cortices of Sham and SAH group rats (n = 8 per group). After euthanasia, brains were removed and gently rolled on sterile blotting paper to remove the meninges. The cortices were dissected and enzymatically digested in two sequential steps: initially with type II collagenase (1 mg/mL, C6885, Sigma) and DNase I (20 mg/mL, Sigma-Aldrich), followed by treatment with collagenase/dispase (1 mg/mL, #10,269,638,001, Sigma) and DNase I. The resulting cell suspensions were collected for total RNA extraction and subsequent RT-PCR analysis [34].

Cell Culture, Transfection, and Transduction

Primary rat hippocampal neurons (4 × 10⁵ cells/well) were seeded in six-well plates and transfected at 70–80% confluence using Lipofectamine 2000 (11,668–019, Invitrogen). Cells were transduced with lentiviral vectors (RiboBio Co., Ltd., Guangzhou, China) encoding mimic-NC, miR-221-3p mimic, inhibitor-NC, miR-211-5p inhibitor, oe-NC, oe-circ_0004058, sh-NC (5′-AAGACAUUGUGUGUCCGCCTT-3′), or sh-circ_0004058 (5′-GGAUAAAGACACUCCUAGAAU-3′). After 72 h, the medium was replaced with puromycin (4 μg/mL) for initial selection. Surviving cells were expanded in 2 μg/mL puromycin for an additional 7 days, followed by culture in puromycin-free medium to establish stable overexpression or knockdown lines.

Rat macrophages (RMA-BMs) and primary rat hippocampal neurons were co-cultured. RMA-BMs were placed in lower Transwell chambers, whereas transduced primary rat hippocampal neurons were placed in upper Transwell chambers to detect regulatory relationships between factors. Primary rat hippocampal neurons in the lower Transwell chamber were treated for 24 h with the GW4869 inhibitor (HY-19363, MedChemExpress) [35]. Dimethyl sulfoxide treatment was used as a control condition.

Reverse Transcription Quantitative Polymerase Chain Reaction (RT-qPCR)

Total cellular RNA was extracted using TRIzol (Thermo Fisher Scientific, 16,096,020), followed by reverse transcription with the PrimeScript RT Reagent Kit (Takara Biotechnology, Dalian, China) for mRNAs and circRNAs, and the PrimeScript miRNA RT Kit (Takara) for miRNAs. RT-qPCR was performed using an RT-qPCR Kit (Q511-02, Vazyme Biotech, Nanjing, China) on a Bio-Rad CFX96 real-time PCR thermocycler (Bio-Rad, Hercules, CA, USA). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was regarded as an internal reference for circ_0004058 and genes analyzed in this study, while U6 small nuclear RNA served as the reference gene for miR-221-3p quantification. Fold changes were determined by the 2^−△△Ct method. All primer sequences (Sangon Biotechnology Co., Ltd., Shanghai, China) are shown in Table S4 [36].

Western Blotting

Total protein was extracted from brain tissues or hippocampal neurons, then separated and transferred onto membranes. After blocking with 5% bovine serum albumin, membranes were incubated overnight at 4 °C with rabbit monoclonal antibodies against LYVE1 (1:1000, #67,538, Cell Signaling Technology), vascular endothelial growth factor (VEGF)-A (1:1000, ab214424, Abcam), and GAPDH (ab181602, 1:10,000, Abcam). The next day, membranes were exposed to HRP-conjugated secondary antibody (1:5000, ab6721, Abcam) for 2 h. A chemiluminescence reagent was employed to visualize protein bands, and captured images were quantified with ImageJ software (v1.48, NIH). GAPDH served as the internal reference for both LYVE1 and VEGF-A [37].

Immunohistochemistry (IHC)

Tissue sections preserved in formalin and embedded in paraffin were first deparaffinized and rehydrated, followed by subsequent immunohistochemical staining procedures. Sections were incubated with primary antibodies against VEGF-A (ab1316, 1:200, Abcam, UK) and CD31 (ab182981, 1:200, Abcam, UK), followed by an HRP-conjugated anti-rabbit secondary antibody (12–348, 1:1000, Sigma-Aldrich, USA). Immunoreactivity was visualized using a DAB detection kit (ab64238, Abcam, USA). After counterstaining, dehydration, and mounting, sections were examined under a light microscope. Five random fields per section were selected to calculate the percentage of positively stained cells.

Luciferase Assay

The 3′-untranslated region (3'-UTR) of the LYVE1 gene was cloned into the pmirGLO vector (E1330, Promega Corporation, Madison, WI, USA) and named LYVE1-wild type (WT). An LYVE1 construct with mutations at miR-221-3p binding sites (LYVE1-Mutant [MUT]) was generated. Luciferase reporter vectors were co-transfected with miR-221-3p mimic or NC mimic into HEK293 cells; subsequently, luciferase activity was measured in accordance with the method described by Promega and normalized to Renilla luciferase. Interactions between circ_0004058 and miR-221-3p were explored in a similar manner [38, 39].

RNA Immunoprecipitation (RIP) Assay

RIP was performed using the Magna RIP RNA-Binding Protein Immunoprecipitation Kit (#17–700, Millipore) to assess circ_0004058 and miR-221-3p interactions. Cell lysates were incubated overnight at 4 °C with anti-Ago2 (2 μg, ab186733, Abcam) or control IgG (2 μg, ab182931, Abcam). Immunoprecipitates were digested with proteinase K, and the bound RNA was purified for RT-qPCR analysis [40].

RNA Pull-Down Assay

Primary rat hippocampal neurons overexpressing circ_0004058 were transfected with biotinylated WT miR-221-3p and MUT miR-221-3p (50 nM each). After 48 h, cell lysates were incubated with RNase-free streptavidin-coated M-280 magnetic beads (S3762, Sigma) pre-blocked with yeast tRNA. The enrichment of circ_0004058 bound to miR-221-3p was quantified by RT-qPCR [41].

Fluorescence In Situ Hybridization (FISH)

The Ribo™ lncRNA FISH Probe Mix (Red) (RiboBio) was used for this assay. Primary rat hippocampal neurons were seeded on slides in six-well plates (1 × 10⁵ cells/well). After cells had reached approximately 80% confluence, they were treated with proteinase K (2 μg/mL), glycine, and acetamidine reagent, then hybridized with 250 μL of prehybridization solution, followed by 250 μL of hybridization solution containing the probe (300 ng/mL). Nuclei were counterstained with DAPI (1:800), and fluorescence signals were visualized with a Leica confocal microscope by capturing five random fields per sample(Leica, Germany) [42].

Statistical Analysis

Sample sizes were estimated using G*Power with an alpha level of 0.05 and power of 0.80. Final animal numbers were adjusted according to experimental needs. Data were analyzed with SPSS 21.0 and reported as mean ± standard deviation (SD). Normality was tested with the Shapiro–Wilk method. For normally distributed data, unpaired t-tests were used for two-group comparisons, and one-way ANOVA with Tukey’s post hoc test for multiple groups. Repeated measures ANOVA analyzed time-course data. Non-parametric tests were applied to non-normally distributed data. Statistical significance was set at p < 0.05.

Results

LYVE1 May Have a Key Role in SAH Onset

To explore the mechanism underlying SAH pathogenesis, we constructed an SAH rat model. SAH rats had lower neurological function scores and greater cerebral edema, compared with sham-operated rats (Fig. S1A and B). Additionally, pathological scores were higher in SAH rats than in sham-operated rats (Fig. S1C). Based on these results, 72 h post-SAH was selected for all subsequent animal experiments.

Macroscopic examination revealed more severe cerebral hemorrhage in SAH rats than in controls (Fig. S1D). Evans blue staining demonstrated significantly elevated BBB permeability in SAH rats, as indicated by enhanced dye extravasation in brain tissue compared with sham controls (Fig. S1E). Nissl staining showed extensive neuron pyknosis, organelle swelling, loss of plasma membrane integrity, and loss of intracellular content in hippocampal tissue of SAH rats, compared with sham-operated rats (Fig. S1F). TUNEL staining further confirmed increased neuronal apoptosis in the hippocampal region of SAH animals relative to the Sham group (Fig. S1G).

Subsequently, brain tissues from SAH (n = 10) and sham-operated (n = 10) rats were subjected to high-throughput transcriptome sequencing and differential analysis. We identified 24 DEGs (11 upregulated and 13 downregulated) in SAH samples (Fig. 1A). Subsequently, we searched the GeneCards database to identify genes related to "subarachnoid hemorrhage," which were then intersected with the DEGs from SAH samples to reveal two key genes involved in SAH onset: LYVE1 and angiopoietin-1 receptor (TEK) (Fig. 1B). RNA sequencing revealed that both LYVE1 and TEK were significantly downregulated in SAH samples (Fig. 1C and D). However, accumulating evidence highlights a more direct involvement of LYVE1 in pathological processes relevant to SAH. For instance, pathological overexpression of VEGF-A in diabetic retinopathy has been shown to promote lymphangiogenesis and drive substantial infiltration of LYVE1-positive macrophages [43]. Moreover, the VEGFC/VEGFR3 signaling axis plays a pivotal role in fibrosis-associated lymphangiogenesis, in which LYVE1—a receptor expressed on lymphatic endothelial cells (LECs)—serves as a key regulatory component. Notably, modulation of lymphangiogenesis through LYVE1-related pathways has been reported to alleviate fibrosis to some extent [44]. In contrast, TEK has been less directly implicated in these mechanisms. Taken together, and supported by the smaller p-value (p = 0.0031) observed for LYVE1 compared with TEK, we selected LYVE1 for subsequent analyses as the more biologically and statistically relevant candidate.

Fig. 1.

Identification of key genes involved in SAH onset and validation in vivo. A: Volcano plot of differentially expressed genes between samples of brain tissue from SAH (n = 10) and sham (n = 10) rats according to RNA sequencing; red represents upregulated genes and green represents downregulated genes. B: Venn diagram of differentially expressed genes according to RNA sequencing and SAH-related genes analyzed by GeneCards. C–D: Expression levels of LYVE1 (C) and TEK (D) extracted from RNA sequencing data. E–F: mRNA and protein levels of LYVE1 in the hippocampal tissue of sham-operated and SAH rats, as detected by RT-qPCR (E) and Western blotting (F). G: Flow cytometry detection of LYVE1⁺CD68⁺ macrophages in the hippocampal tissue of sham-operated and SAH rats. H–I: mRNA and protein levels of VEGF-A in the hippocampal tissue of sham-operated and SAH rats, as detected by RT-qPCR (H) and Western blotting (I). (J) Immunohistochemistry detection of CD31 and VEGFA protein levels in the hippocampal tissue of each group of rats. n = 8 rats per treatment. *** p < 0.001 vs. sham-operated rats

RT-qPCR and Western blotting demonstrated that LYVE1 expression in hippocampal tissue was lower among SAH rats than among sham-operated rats (Fig. 1E and F). There is evidence that LYVE1-positive macrophages play an important role in angiogenesis. Flow cytometric analysis was employed to assess the population of LYVE1⁺CD68⁺ macrophages in hippocampal tissues of SAH and sham-operated rats. We found that there were fewer LYVE1⁺CD68⁺ macrophages in hippocampal tissue among SAH rats than among sham-operated rats (Fig. 1G). Additionally, both mRNA and protein expression levels of VEGF-A were diminished in the hippocampal tissue of SAH rats, compared with sham-operated rats (Fig. 1H and I). Immunohistochemistry revealed that CD31 and VEGFA levels in hippocampal tissue were markedly lower in the SAH group than in the Sham group(Fig. 1J).

Overexpression of LYVE1 Alleviates Brain Injury, Increases the Number of LYVE1⁺CD68⁺ Macrophages, and Promotes Angiogenesis in SAH Rats

Next, we overexpressed LYVE1 in SAH rats to explore the role of LYVE1 on SAH. RT-qPCR and Western blot analyses confirmed successful upregulation of LYVE1 expression in hippocampal tissue following oe-LYVE1 treatment (Fig. 2A and B). Treatment with oe-LYVE1 increased the neurological function score and decreased cerebral edema (Fig. 2C and D). Furthermore, the pathological score decreased after oe-LYVE1 treatment (Fig. 2E).

Fig. 2.

Effects of LYVE1 overexpression on brain injury in SAH rats. ote: SAH rats were treated with oe-NC or oe-LYVE1. A: RT-qPCR detection of LYVE1 mRNA expression in the hippocampal tissue of SAH rats. B: Western blot of LYVE1 protein expression in the hippocampal tissue of SAH rats. C–D: Neurological function scores (C) and extent of cerebral edema (D) in SAH rats. E: The pathological scores of SAH rats. F: Representative images of brain specimens collected after SAH induction and quantitative assessment of SAH severity; G: Evans blue staining was used to evaluate BBB permeability and quantify Evans blue dye extravasation in each group; scale bar = 0.5 cm; H: Nissl staining of neuronal injury in the hippocampal tissue of SAH rats. I: TUNEL staining of neuronal apoptosis in the hippocampal tissue of SAH rats. J: Flow cytometry analysis of LYVE1⁺CD68⁺ macrophages in the hippocampal tissue of SAH rats. K–L: mRNA and protein levels of VEGF-A in the hippocampal tissue of SAH rats, as detected by RT-qPCR (K) and Western blotting (L); M: Immunohistochemistry was used to detect CD31 and VEGFA protein levels in the hippocampal tissue of each group. n = 8 rats per treatment. ** p < 0.01, *** p < 0.001vs. treatment with oe-N

In addition, representative images of brain specimens collected after SAH induction showed that, compared with the oe-NC group, the oe-LYVE1 group exhibited a reduced amount of cerebral hemorrhage (Fig. 2F). Evans blue staining further showed that, compared with the oe-NC group, the oe-LYVE1 group exhibited reduced Evans blue dye extravasation (Fig. 2G).

Nissl staining indicated alleviated neuronal injury in the hippocampal tissue of oe-LYVE1-treated rats (Fig. 2H). The number of apoptotic neurons was also reduced in the hippocampal tissue of oe-LYVE1-treated rats (Fig. 2I). Flow cytometry demonstrated a higher number of LYVE1⁺CD68⁺ macrophages in the hippocampal tissue of oe-LYVE1-treated rats (Fig. 2J). Finally, the mRNA and protein levels of VEGF-A were increased in the hippocampal tissue of oe-LYVE1-treated rats (Fig. 2K and L). Immunohistochemistry detection of angiogenesis markers showed that, compared with the oe-NC group, the levels of CD31 and VEGFA in the hippocampal tissue were lower in the oe-LYVE1 group (Fig. 2M).

miR-221-3p Targets LYVE1 and Inhibits Its Expression

We predicted miRNAs upstream of LYVE1 using the miRWalk database, identifying 1690 miRNAs. Subsequently, we conducted differential analysis of the SAH-associated miRNA dataset GSE161870, which showed 102 differentially expressed miRNAs (Fig. 3A). By intersection analysis of these two Lists, we generated 39 candidate miRNAs (Fig. 3B and C). Given that miRNAs typically exert negative regulatory effects on their target genes, we concentrated on those that were upregulated in SAH samples. Among the candidates, miR-221-3p, previously implicated in brain injury, was selected for further investigation. RT-qPCR revealed elevated miR-221-3p expression in the hippocampal tissue of SAH rats compared with Sham controls (Fig. 3D).

Fig. 3.

Identification of miRNAs upstream of LYVE1. A: Volcano plot of differentially expressed miRNAs between control (n = 2) and SAH (n = 2) samples in the GSE161870 dataset. B: Venn diagram of regulatory miRNAs upstream of LYVE1, as predicted by miRWalk database, and differentially expressed miRNAs in the GSE161870 dataset. C: Heat map of the expression of candidate miRNAs in the GSE161870 dataset. D: miR-221-3p expression in the hippocampal tissue of sham-operated and SAH rats, as detected by RT-qPCR. E: miR-221-3p binding sites on LYVE1 mRNA, as predicted by miRWalk database. F: miR-221-3p binding to LYVE1, as confirmed by dual-luciferase reporter assay. G: Expression levels of miR-221-3p and LYVE1 in RMA-BMs co-cultured with primary rat hippocampal neurons that had been transfected with miR-221-3p mimic or miR-221-3p inhibitor and then treated with GW4869, as detected by RT-qPCR. H: Western blot of LYVE1 protein expression in RMA-BMs co-cultured with primary rat hippocampal neurons that had been transfected with miR-221-3p mimic or miR-221-3p inhibitor and then treated with GW4869. n = 8 rats per treatment. *** p < 0.001 vs. treatment with NC mimic or NC inhibitor. Cell experiments were repeated three times

The miRWalk database predicted miR-221-3p binding sites in LYVE1 mRNA (Fig. 3E). A dual-luciferase reporter assay demonstrated that transfection with miR-221-3p mimic reduced the luciferase activity of LYVE1-WT without affecting the luciferase activity of LYVE1-MUT (Fig. 3F).

Primary rat hippocampal neurons were isolated and good purity was confirmed by immunofluorescence (Fig. S2A). Next, RMA-BMs were co-cultured with primary rat hippocampal neurons that had been transfected with miR-221-3p mimic or miR-221-3p inhibitor (Fig. S2B). In RMA-BMs co-cultured with primary rat hippocampal neurons that had been transfected with miR-221-3p mimic, miR-221-3p expression was increased and LYVE1 expression was decreased. Transfection with miR-221-3p inhibitor produced the opposite effect (Fig. 3G and H). We speculated that miR-221-3p could be transmitted from primary rat hippocampal neurons to macrophages via exosomes, thereby regulating LYVE1. To test this hypothesis, we treated the co-culture system with the exosome inhibitor GW4869 and found that the expression levels of miR-221-3p and LYVE1 were similar in macrophages treated with GW4869 plus miR-221-3p mimic or miR-221-3p inhibitor (Fig. 3G and H).

miR-221-3p-Mediated Reduction of LYVE1 Exacerbates EBI in SAH Rats

Next, we investigated the specific ways in which miR-221-3p-mediated LYVE1 regulation affects SAH. RT-qPCR analysis revealed that administration of miR-221-3p inhibitor combined with sh-NC resulted in reduced miR-221-3p levels and elevated LYVE1 expression in the hippocampal tissue of SAH rats. However, co-treatment with miR-221-3p inhibitor and sh-LYVE1 led to suppression of LYVE1 expression, while miR-221-3p levels remained unaffected (Fig. 4A). Furthermore, LYVE1 protein expression was increased in the hippocampal tissue of rats treated with miR-221-3p inhibitor + sh-NC; this increase was reversed by sh-LYVE1 (Fig. 4B).

Fig. 4.

Effects of miR-221-3p-mediated regulation of LYVE1 on EBI in SAH rats. Note: SAH rats were treated with miR-221-3p inhibitor alone or in combination with sh-LYVE1. A: RT-qPCR detection of miR-221-3p expression and LYVE1 mRNA expression in the hippocampal tissue of SAH rats. B: Western blot of LYVE1 protein expression in the hippocampal tissue of SAH rats. C–D: Neurological function scores (C) and extent of cerebral edema (D) in SAH rats. E: The pathological scores of SAH rats; F: Representative images of brain specimens collected after SAH induction and quantitative assessment of SAH severity. G: Evans blue staining to assess BBB permeability and quantify Evans blue dye extravasation in each group; scale bar = 0.5 cm; H: Nissl staining of neuronal injury in the hippocampal tissue of SAH rats. I: TUNEL staining of neuronal apoptosis in the hippocampal tissue of SAH rats. J: Flow cytometry analysis of LYVE1⁺CD68.⁺ macrophages in the hippocampal tissue of SAH rats. K–L: mRNA and protein levels of VEGF-A in the hippocampal tissue of SAH rats, as detected by RT-qPCR (K) and Western blotting (L). M: Immunohistochemistry to detect CD31 and VEGFA protein levels in the hippocampal tissue of each group. n = 8 rats per treatment. *** p < 0.001

In the absence of miR-221-3p, the neurological function score was increased, and the extent of cerebral edema was decreased; further LYVE1 silencing had the opposite effects (Fig. 4C and D). miR-221-3p inhibition led to lower pathological scores, whereas further LYVE1 silencing led to higher scores (Fig. 4E). In addition, representative images of brain specimens collected after SAH induction revealed a marked reduction in cerebral hemorrhage in the miR-221-3p inhibitor + sh-NC group relative to the NC inhibitor + sh-NC group (Fig. 4F). However, this protective effect was reversed in the miR-221-3p inhibitor + sh-LYVE1 group, which exhibited greater hemorrhage. Evans blue staining results (Fig. 4G) indicated significantly decreased dye extravasation in the miR-221-3p inhibitor + sh-NC group relative to the NC inhibitor + sh-NC group, whereas additional LYVE1 knockdown reversed this effect, leading to increased leakage. Nissl staining revealed that miR-221-3p inhibitor + sh-NC-treated rats had minimal neuronal injury in hippocampal tissue; this finding was reversed by treatment with sh-LYVE1 (Fig. 4H). TUNEL showed decreased neuronal apoptosis after miR-221-3p inhibition, whereas co-treatment with sh-LYVE1 elevated neuronal cell death (Fig. 4I).

Flow cytometry revealed that miR-221-3p inhibition increased the number of LYVE1⁺CD68⁺ macrophages in the hippocampus, an effect reversed by LYVE1 knockdown (Fig. 4J). Similarly, VEGF-A mRNA and protein levels were elevated following miR-221-3p inhibition but declined upon LYVE1 silencing (Fig. 4K and L). Immunohistochemical analysis further showed that CD31 and VEGFA expression were enhanced in the miR-221-3p inhibitor + sh-NC group compared to controls, while additional LYVE1 suppression significantly reduced their levels (Fig. 4M).

circ_0004058 Competitively Binds to miR-221-3p and Upregulates LYVE1 Expression, Thus Attenuating EBI in SAH Rats

Given that both circRNAs and mRNAs can bind miRNAs via miRNA response elements, we utilized the circBank database to predict circRNAs targeting miR-221-3p and identified 3,956 potential candidates. Subsequent differential analysis of the circRNA expression dataset GSE161913 revealed 28 differentially expressed circRNAs (Fig. 5A). By intersection analysis of these two lists, we obtained two circRNAs: circ_0000826 and circ_0004058 (Fig. 5B). Independent sample t-tests were performed on the sequencing data of two circRNAs. The expression of circ_0000826 showed an upward trend without statistical significance, while circ_0004058 was significantly downregulated in the SAH group (p = 0.03) (Fig. 5C), which is consistent with the typical regulatory pattern between circRNAs and miRNAs. Thus, we hypothesized the involvement of a circ_0004058/miR-221-3p/LYVE1 regulatory axis in the pathogenesis of SAH.

Fig. 5.

Relationship among circ_0004058, miR-221-3p, and LYVE1. A: Volcano plot of differentially expressed circRNAs between control (n = 4) and SAH (n = 5) samples in the GSE161913 dataset. B: Venn diagram of regulatory circRNAs upstream of miR-221-3p, as predicted by circBank database, and differentially expressed circRNAs in the GSE161913 dataset. C: Differential expression analysis of circ_0000826 and circ_0004058 in the GSE161913 dataset (Control, n = 4; SAH, n = 5). D: Stability of circ_0004058, as determined by RNase R digestion. E: circ_0004058 expression in the hippocampal tissue of sham-operated and SAH rats, as detected by RT-qPCR. F: circ_0004058 and miR-221-3p binding sites, as predicted by starBase database. G: Binding of circ_0004058 to miR-221-3p, as confirmed by dual-luciferase reporter assay. H: circ_0004058 binding to miR-221-3p, as detected by RNA pull-down assays. I: circ_0004058 binding to miR-221-3p, assessed by RIP assays. J: Localization of circ_0004058 and miR-221-3p in primary rat hippocampal neurons, as detected by FISH assays. K: Expression levels of miR-221-3p and LYVE1 in RMA-BMs co-cultured with primary rat hippocampal neurons that had been transfected with oe-circ_0004058 or sh-circ_0004058 and then treated with GW4869, as detected by RT-qPCR. L: Western blot of LYVE1 protein expression in RMA-BMs co-cultured with primary rat hippocampal neurons that had been transfected with oe-circ_0004058 or sh-circ_0004058 and then treated with GW4869. n = 8 rats per treatment. ** p < 0.01, *** p < 0.001 vs. treatment with oe-NC or sh-NC. Cell experiments were repeated three times

The circular nature of circ_0004058 was confirmed via RNase R digestion, which demonstrated decreased GAPDH content and unaltered circ_0004058 content after RNase R digestion (Fig. 5D). Additionally, RT-qPCR showed that circ_0004058 expression in brain tissue was lower among SAH rats than among sham-operated rats (Fig. 5E). Moreover, endothelial cells play a key role in maintaining BBB integrity and driving angiogenesis. We further investigated the expression and regulation of LYVE1 and circ_0004058 in endothelial cells (rBMECs). After isolating rBMECs from Sham and SAH rats, RT-qPCR analysis revealed markedly reduced levels of circ_0004058 and LYVE1 in rBMECs from the SAH group compared with the Sham group (Fig. S3A–B), suggesting that these genes may play roles in endothelial function related to BBB protection and angiogenesis.

We used the starBase database to predict circ_0004058 and miR-221-3p binding sites (Fig. 5F). Dual-luciferase reporter assays demonstrated that luciferase activity was decreased in cells co-transfected with miR-221-3p mimic and circ_0004058 WT, whereas no change was observed in cells co-transfected with miR-221-3p mimic and circ_0004058 MUT (Fig. 5G). These results indicated that circ_0004058 could target and regulate miR-221-3p.

Primary rat hippocampal neurons expressing miR-221-3p-WT had a higher level of circ_0004058, compared with neurons expressing miR-221-3p-MUT (Fig. 5H). RIP assays revealed enrichment of both circ_0004058 and miR-221-3p after immunoprecipitation using the anti-Ago2 antibody (Fig. 5I). FISH revealed co-localization of circ_0004058 and miR-221-3p in the cytoplasm of primary hippocampal neurons (Fig. 5J).

Next, RMA-BMs were co-cultured with primary rat hippocampal neurons that had been transfected with oe-NC, oe-circ_0004058, sh-NC, or sh-circ_0004058 (Fig. S2C, D). RT-qPCR demonstrated that LYVE1 expression was increased while miR-221-3p expression was decreased in RMA-BMs co-cultured with primary rat hippocampal neurons that had been transfected with oe-circ_0004058. Where as in the absence of circ_0004058, miR-221-3p expression was increased, and LYVE1 expression was decreased (Fig. 5K). Western blotting results were consistent with RT-qPCR findings in terms of LYVE1 expression (Fig. 5L). Upon further treatment with GW4869, miR-221-3p and LYVE1 had similar expression levels in the co-culture system (Fig. 5K and L).

To investigate the effect of circ_0004058-mediated regulation of miR-221-3p on SAH, SAH rats were treated with oe-NC + NC mimic, oe-circ_0004058 + NC mimic, or oe-circ_0004058 + miR-221-3p mimic. RT-qPCR revealed downregulation of miR-221-3p and upregulation of both circ_0004058 and LYVE1 in the hippocampal tissue of oe-circ_0004058 + NC mimic-treated rats; these effects were partially reversed by oe-circ_0004058 + miR-221-3p mimic treatment—there was no change in circ_0004058 expression (Fig. 6A). Western blotting results were consistent with RT-qPCR findings in terms of LYVE1 expression (Fig. 6B).

Fig. 6.

Effects of circ_0004058-mediated regulation of miR-221-3p and reduced EBI in SAH rats. Note: SAH rats were treated with oe-circ_0004058 alone or in combination with miR-221-3p mimic. A: RT-qPCR detection of circ_0004058, miR-221-3p, and LYVE1 expression in the hippocampal tissue of SAH rats. B: Western blot of LYVE1 protein expression in the hippocampal tissue of SAH rats. C–D: Neurological function scores (C) and extent of cerebral edema (D) in SAH rats. E: The pathological scores of SAH rats. F: Representative images of brain specimens collected after SAH induction and quantitative assessment of SAH severity. G: Evans blue staining to evaluate BBB permeability and quantitative analysis of Evans blue dye extravasation in each group; scale bar = 0.5 cm. H: Nissl staining of neuronal injury in the hippocampal tissue of SAH rats. I: TUNEL staining of neuronal apoptosis in the hippocampal tissue of SAH rats. J: Flow cytometry analysis of LYVE1⁺CD68⁺ macrophages in the hippocampal tissue of SAH rats. K–L: mRNA and protein levels of VEGF-A in the hippocampal tissue of SAH rats, as detected by RT-qPCR (K) and Western blotting (L). M: Immunohistochemistry to detect CD31 and VEGFA protein levels in the hippocampal tissue of each group. n = 8 rats per treatment. * p < 0.05, ** p < 0.01, *** p < 0.001 vs. treatment with oe-NC + NC mimic or oe-circ_0004058 + NC mimic

As shown in Fig. 6C and D, neurological function scores increased, and the extent of cerebral edema decreased. These protective effects were reversed upon co-treatment with oe-circ_0004058 + miR-221-3p mimic. The pathological scores of rats treated with oe-circ_0004058 + NC mimic were reduced, and this change was abolished by treatment with oe-circ_0004058 + miR-221-3p mimic (Fig. 6E). In addition, representative images of brain specimens collected after SAH induction revealed diminished cerebral hemorrhage in the oe-circ_0004058 + NC mimic group relative to the oe-NC + NC mimic group, whereas hemorrhage was exacerbated by miR-221-3p overexpression(Fig. 6F). Evans blue staining results (Fig. 6G) revealed that vascular leakage was reduced in the oe-circ_0004058 + NC mimic group compared with the oe-NC + NC mimic group, whereas co-treatment with miR-221-3p mimic restored dye extravasation to elevated levels.

Nissl staining revealed that circ_0004058 overexpression inhibited neuronal injury, whereas miR-221-3p overexpression promoted injury (Fig. 6H). TUNEL staining revealed reduced neuronal apoptosis in the hippocampal tissue of rats overexpressing circ_0004058; this effect was attenuated by overexpression of miR-221-3p (Fig. 6I). These findings indicate that circ_0004058 can regulate EBI in SAH rats by inhibiting miR-221-3p.

Further flow cytometry analysis indicated an increased number of LYVE1⁺ CD68⁺ macrophages in rats overexpressing circ_0004058, which was reversed upon miR-221-3p overexpression (Fig. 6J). Overexpression of circ_0004058 in the hippocampal tissue of rats led to higher mRNA and protein expression levels of VEGF-A, whereas miR-221-3p overexpression produced the opposite effect (Fig. 6K and L). Immunohistochemical analysis of angiogenic markers CD31 and VEGFA revealed that the oe-circ_0004058 + NC mimic group exhibited significantly increased expression compared to the oe-NC + NC mimic group, whereas these increases were suppressed in the oe-circ_0004058 + miR-221-3p mimic group (Fig. 6M).

The circ_0004058/miR-221-3p/LYVE1 Axis Regulates EBI in SAH Rats

These findings indicate that circ_0004058 exerts protective effects in SAH by modulating the miR-221-3p/LYVE1 axis. RT-qPCR results showed that overexpression of circ_0004058 in the hippocampal tissue of SAH rats led to decreased miR-221-3p expression and increased levels of both circ_0004058 and LYVE1. However, in response to co-treatment with oe-circ_0004058 and sh-LYVE1, LYVE1 expression was reduced while the expression levels of circ_0004058 and miR-221-3p were not significantly altered (Fig. 7A). Western blotting results were similar to RT-qPCR findings in terms of LYVE1 expression (Fig. 7B).

Fig. 7.

Effects of the circ_0004058/miR-221-3p/LYVE1 axis on EBI in SAH rats. Note: SAH rats were treated with oe-circ_0004058 alone or in combination with sh-LYVE1. A: RT-qPCR detection of circ_0004058, miR-221-3p, and LYVE1 expression in the hippocampal tissue of SAH rats. B: Western blot of LYVE1 protein expression in the hippocampal tissue of SAH rats. C–D: Neurological function scores (C) and extent of cerebral edema (D) in SAH rats. E: The pathological scores of SAH rats. F: Representative images of brain specimens collected after SAH induction and quantitative assessment of SAH severity. G: Evans blue staining to evaluate BBB permeability and quantitative analysis of Evans blue dye extravasation in each group; scale bar = 0.5 cm. H: Nissl staining of neuronal injury in the hippocampal tissue of SAH rats. I: TUNEL staining of neuronal apoptosis in the hippocampal tissue of SAH rats. J: Flow cytometry analysis of LYVE1⁺CD68⁺ macrophages in the hippocampal tissue of SAH rats. K–L: mRNA and protein levels of VEGF-A in the hippocampal tissue of SAH rats, as detected by RT-qPCR (K) and Western blotting (L). M: Immunohistochemistry to detect CD31 and VEGFA protein levels in the hippocampal tissue of each group. n = 8 rats per treatment. *** p < 0.001 vs. treatment with oe-NC + sh-NC; # p < 0.05 vs. treatment with oe-circ_0004058 + sh-NC

circ_0004058 overexpression increased neurological function scores and reduced the extent of cerebral edema; these effects were attenuated by LYVE1 silencing (Fig. 7C and D). Pathological scores were decreased upon circ_0004058 overexpression, but the decrease was abolished by LYVE1 silencing (Fig. 7E). Representative brain images after SAH induction showed reduced hemorrhage in the oe-circ_0004058 + sh-NC group compared with oe-NC + sh-NC, while hemorrhage increased upon LYVE1 silencing (oe-circ_0004058 + sh-LYVE1 vs. oe-circ_0004058 + sh-NC) (Fig. 7F). Similarly, Evans blue staining showed reduced dye leakage in the oe-circ_0004058 + sh-NC group, which was reversed in the oe-circ_0004058 + sh-LYVE1 group (Fig. 7G).

Nissl and TUNEL staining demonstrated that the inhibitory effects of circ_0004058 overexpression on neuronal injury and neuronal apoptosis were partially abolished by LYVE1 silencing (Fig. 7H and I). Moreover, LYVE1 silencing reversed the circ_0004058 overexpression-mediated increase in the number of LYVE1⁺CD68⁺ macrophages, as indicated by flow cytometry analysis (Fig. 7J). As shown in Fig. 7K and L, the circ_0004058 overexpression-induced upregulation of VEGF-A mRNA and protein levels was abrogated by LYVE1 silencing. Furthermore, immunohistochemical analysis revealed that CD31 and VEGFA levels were elevated in the hippocampal tissue of the oe-circ_0004058 + sh-NC group compared to the oe-NC + sh-NC group, whereas LYVE1 silencing markedly reduced their expression relative to the oe-circ_0004058 + sh-NC group (Fig. 7M).

LYVE1⁺CD68⁺ Macrophages Alleviate EBI and Promote Angiogenesis in SAH Rats

To explore the effects of LYVE1⁺ CD68⁺ macrophages on SAH, we used an anti-CSF1R antibody to inhibit macrophage activity. RT-qPCR showed that LYVE1 expression was significantly upregulated in the hippocampal tissue of rats treated with oe-LYVE1 + anti-IgG, while miR-221-3p and circ_0004058 levels remained unchanged. Anti-CSF1R administration did not alter the expression of any of these molecules (Fig. 8A). Western blotting confirmed the upregulation of LYVE1 protein in the oe-LYVE1 + anti-IgG group, consistent with the RT-qPCR results (Fig. 8B).

Fig. 8.

Effects of LYVE1⁺CD68⁺ macrophages on EBI in SAH rats. Note: SAH rats were treated with oe-LYVE1 alone or in combination with anti-CSF1R antibody. A: RT-qPCR detection of circ_0004058, miR-221-3p, and LYVE1 expression in the hippocampal tissue of SAH rats. B: Western blot of LYVE1 protein expression in the hippocampal tissue of SAH rats. C–D: Neurological function scores (C) and extent of cerebral edema (D) in SAH rats. E: The pathological scores of SAH rats. F: Representative images of brain specimens collected after SAH induction and quantitative assessment of SAH severity. G: Evans blue staining to assess BBB permeability and quantitative analysis of Evans blue dye extravasation in each group; scale bar = 0.5 cm. H: Nissl staining of neuronal injury in the hippocampal tissue of SAH rats. I: TUNEL staining of neuronal apoptosis in the hippocampal tissue of SAH rats. J: Flow cytometry analysis of LYVE1⁺CD68⁺ macrophages in the hippocampal tissue of SAH rats. K–L: mRNA and protein levels of VEGF-A in the hippocampal tissue of SAH rats, as detected by RT-qPCR (K) and Western blotting (L). n = 8 rats per treatment. M: Immunohistochemistry to detect CD31 and VEGFA protein levels in the hippocampal tissue of each group. ** p < 0.01, *** p < 0.001 vs. treatment with oe-NC + anti-IgG; # p < 0.05 vs. treatment with oe-LYVE1 + anti-IgG

As shown in Fig. 8C and D, the oe-LYVE1-mediated increase in neurological function score and decrease in extent of cerebral edema were reversed by anti-CSF1R antibody treatment (Fig. 8E). In addition, representative images of brain specimens collected after SAH induction revealed decreased hemorrhage in the oe-LYVE1 + anti-IgG group compared to oe-NC + anti-IgG, while anti-CSF1R treatment reversed this effect(Fig. 8F). Evans blue staining results (Fig. 8G) showed that compared with the oe-NC + anti-IgG group, the oe-LYVE1 + anti-IgG group displayed significantly reduced Evans blue dye extravasation, while compared with the oe-LYVE1 + anti-IgG group, the oe-LYVE1 + anti-CSF1R group showed markedly increased Evans blue dye extravasation. Nissl staining and TUNEL staining showed that anti-CSF1R antibody treatment abolished the decreases in neuronal injury and apoptosis in response to LYVE1 overexpression (Fig. 8H and I). Flow cytometry revealed that the LYVE1 overexpression-mediated increase in the number of LYVE1⁺CD68⁺ macrophages was also reversed by anti-CSF1R antibody treatment (Fig. 8J). Finally, oe-LYVE1-mediated upregulation of VEGF-A mRNA and protein levels was reversed by anti-CSF1R antibody treatment (Fig. 8K and L). Immunohistochemical analysis showed that CD31 and VEGFA levels were significantly elevated in the hippocampal tissue of the oe-LYVE1 + anti-IgG group compared with the oe-NC + anti-IgG group, whereas these levels were markedly reduced following anti-CSF1R treatment (oe-LYVE1 + anti-CSF1R) relative to the oe-LYVE1 + anti-IgG group (Fig. 8M).

Discussion

This study is the first to reveal the protective role of circ_0004058 in mitigating SAH-induced EBI by regulating the miR-221-3p/LYVE1 axis. We found that circ_0004058 competes with miR-221-3p, preventing it from downregulating LYVE1 expression, thereby alleviating EBI and promoting angiogenesis in the SAH model. This discovery provides a new potential target for SAH treatment and offers a novel perspective for understanding the pathological mechanisms of EBI.

Our findings highlight the pivotal role of LYVE1 in SAH-induced EBI. Specifically, LYVE1 expression is significantly decreased in the hippocampal tissue of SAH rats, which may impact the number of LYVE1 + CD68 + macrophages, thereby impairing angiogenesis (Fig. 9). LYVE1 overexpression significantly alleviated brain injury, as indicated by enhanced neurological scores, reduced brain edema, and lower levels of neuronal apoptosis. These neuroprotective effects are likely mediated by enhanced recruitment and activity of LYVE1⁺ macrophages, which contribute to angiogenesis and the preservation of BBB integrity.

Fig. 9.

Schematic diagram of the mechanism by which the circ_0004058/miR-221-3p/LYVE1 axis affects SAH-induced EBI

Moreover, miR-221-3p was identified as a negative regulator of LYVE1 by targeting its 3′-UTR. Suppressing miR-221-3p resulted in increased LYVE1 expression, which alleviated neuronal injury and promoted angiogenesis. These results suggest that miR-221-3p exacerbates EBI, while LYVE1⁺CD68⁺ macrophages have a protective role in alleviating EBI and promoting angiogenesis. LYVE1⁺ macrophages directly promote endothelial cell proliferation and neovascularization by secreting pro-angiogenic factors such as VEGF and FGF2. Studies have shown that infiltration of LYVE1⁺ macrophages is positively correlated with increased vascular density in ischemic brain injury models [43]. In addition, LYVE1 may enhance macrophage adhesion to endothelial cells by regulating hyaluronic acid recognition on the macrophage surface, thereby further supporting angiogenesis [45]. LYVE1⁺ macrophages may also stabilize inter-endothelial junctions and maintain BBB integrity through the Angiopoietin-1/Tie2 signaling pathway [46]. Moreover, LYVE1⁺ macrophages may regulate microglial polarization through paracrine effects. For example, macrophage-derived IL-4 and IL-13 can promote M2-type anti-inflammatory polarization of microglia, suppress neuroinflammation, and promote tissue repair [47]. M2-type microglia further secrete BDNF and Arg1, enhancing BBB protection and supporting neuronal survival [48]. LYVE1 + macrophages can also secrete VEGF and Angiopoietin-2 to activate the PI3K/AKT pathway in endothelial cells, thereby promoting neovascularization [49]. In a post-SAH brain edema model, LYVE1 overexpression significantly increased microvascular density and reduced neuronal apoptosis [50]. It has also been proposed that in acute central nervous system injuries, the regulation of circRNAs can improve cognitive function, promote angiogenesis, inhibit apoptosis and inflammation, modulate autophagy, and protect the BBB through various molecules and pathways, including miRNAs, NF-κB, PI3K/AKT, Notch1, and TET [51, 52]. Moreover, application of the XGBoost algorithm revealed that patients with favorable neurological outcomes exhibited significantly lower miR-152-3p, miR-221-3p, and miR-34a-5p levels compared with those with poor outcomes, suggesting their potential as predictive biomarkers for neurological prognosis [12].

circ_0004058 plays a pivotal role in this study by acting as a competing endogenous RNA (ceRNA). It binds to miR-221-3p, thereby preventing the downregulation of LYVE1 and mitigating SAH-induced EBI. This mechanism underscores the broader role of circRNAs as sponges for miRNAs, indirectly regulating mRNA expression. For instance, previous research has shown that circNLRP3 acts as a sponge for miR-221-5p, upregulating its target mRNA A20 [53]. Similarly, circ_0004058 exerts its protective effects against SAH-induced EBI by regulating the miR-221-3p/LYVE1 axis.

In comparison with other studies, while miR-221-3p’s role in brain injury has been reported, these studies mostly focus on miRNA regulation of BBB function without delving into specific downstream targets like LYVE1 [54–56]. Our study not only confirms the role of miR-221-3p in EBI but also further elucidates the specific mechanism by which it exacerbates brain injury through downregulation of LYVE1.

Moreover, unlike other studies, we are the first to identify the significant regulatory role of circ_0004058 in EBI, offering a novel molecular target and potential therapeutic avenue. Although the function of circRNAs as miRNA sponges has been widely reported, this mechanism remains underexplored in the context of SAH-related brain injury. In summary, our results suggest that in the future, synthetic stable circ_0004058 mimics (such as circularized RNAs or circRNA encapsulated in lipid nanoparticles) could be designed to target SAH lesions and enhance LYVE1 expression. Simultaneous modulation of circ_0004058 (upregulation), miR-221-3p (inhibition), and LYVE1 (activation) may produce synergistic effects and avoid the limited efficacy of single-target interventions. The development of multifunctional nanoparticles (e.g., gold nanoshells or exosomes) or novel drugs carrying a circ_0004058 mimic, miR-221-3p inhibitor, and LYVE1 agonist simultaneously could demonstrate significant synergistic therapeutic effects in SAH models by combining vascular repair, anti-inflammation, and enhanced lymphatic drainage mechanisms. Further studies are required to validate the safety and clinical efficacy of such approaches, but the discovery of this regulatory axis undoubtedly opens new avenues for precision medicine in SAH treatment.

Despite the promising findings, this study has several limitations. First, although the sample size in our animal experiments was adequate to detect statistical differences, it may limit the generalizability of the results. Larger studies are required to confirm these results and explore their applicability to diverse genetic backgrounds and clinical settings. Second, the SAH rat model used may not fully recapitulate the complexity of human SAH, especially regarding comorbidities and environmental influences. Third, only male Sprague–Dawley rats were used in this study to avoid the potential confounding effects of estrous cycle–related hormonal fluctuations on pain sensitivity, inflammatory responses, and drug metabolism. While this approach enhanced internal consistency, it also precluded the evaluation of sex-specific differences, and future studies should therefore include both male and female animals. Fourth, while our study focused on the miR-221-3p/LYVE1 axis, it remains unclear whether other miRNA-mediated pathways contribute to EBI in SAH, which warrants further exploration. Finally, the mechanistic insights gained from this study need to be validated in human tissues to confirm their clinical relevance.

Future studies should address these limitations by expanding sample sizes, employing more clinically relevant models, and exploring additional signaling pathways. Furthermore, translating these findings into human studies could enhance our understanding of the role of circ_0004058 and miR-221-3p in SAH and potentially lead to novel therapeutic interventions.

Conclusion

In summary, this study demonstrated that circ_0004058 could alleviate SAH-induced EBI, increase the number of LYVE1⁺CD68⁺ macrophages, and promote angiogenesis by mediating the miR-221-3p/LYVE1 axis. To our knowledge, this is the first mechanistic study to reveal the role of circ_0004058 in SAH-induced EBI. The inspiring results of our study provide a potential biomarker and therapeutic target for SAH.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Hua Gu conducted the experiments, analyzed data, and contributed to manuscript drafting. Yong Cai designed and supervised the study, interpreted results, and finalized the manuscript. All authors read and approved the final version of the manuscript.

Funding

This study was supported by Zhejiang Province Basic Public Welfare Research Program (LGF21H090001).

Data Availability

Data is provided within the manuscript or supplementary information files.

Declarations

Ethics Approval

This study was reviewed and approved by the Institutional Animal Ethics Committee of Huzhou Institute for Drug Control (Animal Use License No.: SYXK (Zhe) 2018–0015).

Consent to Participate

Not applicable.

Consent for Publication

Not applicable.

Competing interests

The authors declare no competing interests.