MicroRNA-29b-3p promotes endothelial cell inflammation via NTRK2/PI3K/NLRP3 pathway in Moyamoya disease

Abstract

Moyamoya disease (MMD) is a rare chronic progressive vascular anomaly of the skull base whose molecular regulatory mechanisms remain poorly understood. We therefore aimed to analyze the molecular mechanisms involved in the development of MMD from the perspective of miRNA regulation of mRNA. Raw gene expression profiles (GSE178501, GSE157628 and GSE189993) were downloaded from the Gene Expression Omnibus database and used to identify differentially expressed genes and perform functional enrichment analysis. Differentially expressed miRNAs and their predicted target genes were validated by RT-qPCR. Oxygen glucose deprivation was applied to induce an inflammatory injury cell model in human brain microvascular endothelial cells (BMEC). 973 differentially expressed mRNAs and 3 differentially expressed miRNAs were identified in three sets of gene expression profiles. RT-qPCR confirmed that the miR-29b-3p was upregulated in leukocytes of MMD and that the expression of NTRK2 was downregulated. Dual-luciferase reporter assay indicated that NTRK2 was the direct target of miR-29b-3p. Overexpression of NTRK2 improved the viability of BMEC and increased the protein levels of NTRK2 and pPI3K, while suppressed the expression of NLRP3, IL1β, and TNF-α. miR-29b-3p treatment partially abolished the protective effect of NTRK2 and diminished the effect of NTRK2 on PI3K/NLRP3 pathway. In conclusion, this study provided a novel insight into the pathophysiological mechanisms of MMD and demonstrated that the miR-29b-3p/NTRK2/PI3K/NLRP3 axis plays a pivotal role in the progression of MMD.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40001-025-03521-3.

Introduction

Moyamoya disease (MMD) is an unusual chronic obstructive vascular illness [1, 2] of the brain. Moyamoya indicates “puff of smoke” in Japanese and represents the angiographic outlook of collateral vessels that develop around obstructed cerebral vessels in MMD patients [3]. These obstructions are described by a hyperplastic narrow or narrowing of the end of the inner carotid artery and its surrounding branches, including the proximal anterior cerebral artery and middle cerebral artery (MCA) [4]. The main clinical manifestations of MMD are ischemia or hemorrhage, or it may be asymptomatic [4] and MMD usually has very serious sequelae, such as stroke, epilepsy, or neuropsychological disorders, which are related to high states of disability [1, 5].

MMD has a high incidence rate in East Asian population and shows obvious familial clustering, which strongly suggests that genetic factors play an important role in its pathogenesis [1, 6–9]. The onset of this condition is associated with gender, age, and certain immune-inflammatory states [10, 11]. The pathological changes in MMD stenosis vessels involved intima hyperplasia and luminal stenosis [3, 12]. Previous studies reveal that intima cell hyperplasia leads to lumen stenosis of MMD and endothelial cells participate in the neointima of MMD [13, 14].

The maturation of sequencing technologies is allowing the influence of the transcriptome on illness to be gradually described. microRNAs (miRNAs) are brief pieces of endogenous non-coding RNAs that manage the performance of numerous genes by promoting messenger RNA (mRNA) degradation or inhibiting its translation [15–19]. Recently, there have been some survey of miRNAs and mRNAs contained in MMD, and a few surveys of miRNA–mRNA regulatory relationships in MMD [9, 15, 20, 21]. For example, it was discovered that let-7c targets homo sapiens ring finger protein 213 (RNF213) and that this process is contained in the pathogenesis of MMD [22]. The outcomes of a two vascular occlusion + brain–muscle–joint angiogenesis experiment in rats indicated that miR-126-5p accelerated endothelial cell multiplication and angiogenesis by mediating risen performance of eNOS, CD31, and VEGF by the PI3K/Akt approach, which accelerated the recovery of rats’ cognitive function [23]. Nevertheless, although these surveys have provided new insights into MMD, its regulatory systems are complex and require farther detailed investigation.

It is worth noting that among numerous miRNAs related to MMD, the role of miR-29b-3p has not been fully explored. MiR-29b-3p is closely related to endothelial cell proliferation and angiogenesis [24]. Although a previous bioinformatics study suggested that it may be dysregulated in MMD, the finding lacks rigorous experimental validation [25]. More importantly, the specific function of miR-29b-3p in the inflammatory response of MMD endothelial cells, its clear target genes, and downstream signaling pathways are currently completely unknown. Elucidating the precise regulatory mechanism of miR-29b-3p is expected to reveal new pathways for the pathogenesis of MMD and may provide new targets for diagnosis and treatment.

Accordingly, we established the miRNA–mRNA regulative network in MMD by acting bioinformatics analysis of miRNA and mRNA expression profiles in MMD patients. We also applied real-time quantitative reverse-transcription polymerase chain reaction (RT-qPCR) to identify a series of remarkably altered miRNAs and their predicted target genes. Then, we investigated the potential miRNA–mRNA regulative axis in the progression of MMD in human brain microvascular endothelial cells (BMEC). The aforementioned investigations aimed to indicate the molecular regulatory systems of MMD and provide clues for clinical treatment.

Materials and methods

Data collection

MMD-related mRNA/miRNA microarray datasets (GSE178501, GSE157628, and GSE189993) were gained from the Gene Expression Omnibus (GEO) database and used to identify differentially expressed genes and perform functional enrichment analysis (Fig. 1). After selecting, three sets of microarray data (two sets of mRNA microarray data of MCA wall tissue and one set of miRNA microarray data of circulating leukocytes) remained (Table S1). A total of 40 samples were applied for mRNA and miRNA analysis.

Fig. 1.

The workflow of analysis

Data processing

The raw Agilent microarray data were processed using R software (v4.1.1). First, the oligo package (v1.56.0) is utilized for data reading, background correction (using RMA algorithm), and quantile normalization to eliminate technical differences between different microarrays. Subsequently, the probe was annotated with gene symbols using its corresponding GPL platform annotation file. For mRNA expression matrices containing multiple datasets, we merged the normalized data and used the remove Batch Effect function in the limma package (v3.48.3) to perform batch effect correction on the integrated expression matrix using dataset sources (GSE189993 and GSE157628) as batch variables. The miRNA dataset (GSE178501), as an independent set, has also undergone the same background correction and normalization process. Differential expression analysis was performed using the limma (v3.48.3) package. For both mRNA and miRNA, the significance threshold was set to | log2 Fold Change (FC) |> 1 and the false discovery rate (FDR) corrected by the Benjamin–Hochberg (BH) method was P.adjust < 0.05. Differential genes were visualized applying the ggplot2 (v3.3.5) package and the pheatmap (v1.0.12) package.

WGCNA analysis

The WGCNA (v1.70.3) package was applied to gain candidate module genes related to MMD, in the base of previously identified differential mRNAs. WGCNA identifies gene co-expression networks from topological overlap, identifies co-expressed gene groups of models, and correlates these with phenotypes. We first systematically evaluated a series of soft threshold powers (β, ranging from 1 to 20). We determine the optimal soft threshold by analyzing the trend of scale-free topological fitting index (R²) and average connectivity with increasing β value. We chose β = 10 as the final parameter, because under this condition, the network achieved a satisfactory scale-free topology (R² = 0.908) while maintaining a relatively high average connectivity, which ensures that the network has a robust topology while retaining a large amount of biological information To verify the stability of the constructed gene module, we tested neighboring thresholds, such as β = 8 and β = 12, and confirmed that under these parameters, the gene composition of the core module most significantly correlated with MMD phenotype remained highly stable (with over 90% overlap of core genes), indicating that our findings were insensitive to threshold selection and the results were reliable. Then, the correlation matrix was converted into an adjacency matrix, which was then converted into a topological overlap matrix (TOM) to gain various groups of models. These groups of models were then related to the phenotypes, and those with a high correlation with MMD were selected for subsequent analysis.

Prediction of target genes of miRNAs

Target gene prediction for differential miRNAs was conducted applying the multiMIR (v1.14.0) package. A threshold of 30% was set to predict differential miRNA and candidate mRNA-binding positions.

GO and KEGG enrichment

Functional analysis of mRNA was acted through GO and the KEGG. KEGG and GO enrichment analysis was acted applying the org.Hs.eg.db (v3.13.0) package and cluster Profiler (v4.0.5) package, and data analysis for visualization was acted applying the ggplot2 (v3.3.5) package and GOplot (v1.0.2) package. The P values of enrichment analysis were corrected using the BH method to control for FDR. We use P.adjust < 0.01 as the screening criterion for functional enrichment.

Study population

Twenty patients were diagnosed with MMD in the First Affiliated Hospital of Ningbo University (China), and these participants were contained in this research. The inclusion criteria were (1) an age 18–65 years; (2) having a diagnosis of MMD in the base of the 2021 Japanese guidelines for the treatment and diagnosis of MMD [7]; and (3) having provided consent that is informed to receive the research program. Exclusion criteria were (1) smog syndrome suggested by clinical examination or laboratory examination; (2) patients with history of stroke; (3) having undergone hemodialysis; and (4) having current pregnancy, tumors, or immune diseases. 20 age- and sex-matched healthy adults were identified and served as the normal control group. There were 10 females and 10 males in the MMD group, aged 32–60 years, and 10 females and 10 males in the normal control group, aged 32–58 years. There were no statistically remarkable differences between the two groups with regard to age, sex, and underlying disease related to atherosclerosis (Table 1). All of the participants or their legally authorized representatives marked a consent that is informed form, which was in approval of the ethics committee in the First Affiliated Hospital of Ningbo University.

Table 1.

The clinical characteristics for all participants

Character	Control (n = 20)	MMD (n = 20)	P
Age (year)	48.00 ± 8.34	47.95 ± 9.62	0.99
Man (n)	10	10	0.99
TG (mmol/L)	1.38 ± 0.67	1.84 ± 1.00	0.095
TC (mmol/L)	4.76 ± 1.02	4.58 ± 1.49	0.66
HDL (mmol/L)	1.29 ± 0.38	1.09 ± 0.26	0.071
LDL (mmol/L)	3.01 ± 0.71	2.88 ± 1.03	0.664
ApoA (mg/dL)	1.34 ± 0.32	1.10 ± 0.25	0.017
ApoB (mg/dL)	0.91 ± 0.21	0.81 ± 0.32	0.245
ApoE (mg/L)	44.39 ± 17.87	54.76 ± 19.67	0.103

Sample collection and total RNA extraction

Under fasting situations, a peripheral venous blood sample was collected from the anterior elbow vein into an ethylene diamine tetra acetic acid anticoagulation tube. Within 1 h, 5 ml of the whole blood sample was mixed with 5 ml of physiological saline, and the resulting solution was slowly dropped into a 15 ml centrifuge tube including 5 ml of lymphocyte isolation solution (Beijing Solarbio Science & Technology Co., Ltd., China). The sample was centrifuged at 400 g for 15 min. Next, leukocytes were isolated from the middle section, and the mixed erythrocytes were then withdrawn with erythrocyte lysis buffer and stored at −80 °C. Total leukocyte RNA was extracted applying TRIzol reagent (Invitrogen, USA), and the concentration and purity of the extracted total RNA were tested using NanoDrop 2000 (Thermo Scientific, USA), with an A260/A280 ratio between 1.8 and 2.1.

RT-qPCR

Several differentially expressed miRNAs were reverse-transcribed applying the stem-loop method and then subjected to RT-qPCR. The stem-loop Reverse Transcription primers (Guangzhou RiboBio Co., Ltd., China) and a TransScript One-Step gDNA Removal and cDNA Synthesis SuperMix reverse-transcription kit (Transgen Biotech, China) were applied to reverse transcribe 1000 ng of RNA into complementary DNA (cDNA). RT-qPCR was acted using an IQ SYBR Green Supermix kit (Bio-Rad, USA) with cel-miRNA-39-3p (Guangzhou RiboBio Co., Ltd., China) as the external reference and Homo sapiens actin beta (ACTB) as the internal reference (Table S2). The relative expression of miRNAs or mRNAs was calculated through the 2-ΔΔCt method..

Cell culture

Human embryonic kidney 293 T (HEK293T) cells and human brain microvascular endothelial cells (BMEC) (ctcc-003–0113) were used in this research. Cells were cultured in 96-well plates or 6-well plates applying Dulbecco’s modified eagle’s medium with 1% penicillin/streptomycin (ScienCell, USA) and 10% fetal bovine serum at 37 °C.

Luciferase reporter assay

The 3’ UTR gene fragments of NTRK2 were cloned and amplified as the NTRK2-wild type (WT). The mutant type (MUT) was constructed by targeted mutagenesis of miR-29b-3p-binding site in the 3’ UTR of the target gene NTRK2, as predicted by TargetScan database analysis. The target sequences of WT or MUT were cloned into the pSI-Check2 reporter vector (C8021, Promega Corporation, WI, USA), and then, the vector and miR-29b-3p mimic or negative control (NC) mimic were co-transfected into HEK293T cells. Cells were collected for fluorescence assay at 48 h after transfection. The luciferase activity was testified in the base of the standard means of the guide manual (Dual-Luciferase^® Reporter Assay Systems, Promega).

Cell mechanistic experiments

We applied oxygen–glucose deprivation (OGD) to induce BMEC to construct an inflammatory cell model. The BMEC were cultured in glucose-free DMEM medium and hypoxic conditions (1% O², 94% N₂, 5% CO₂, 37 °C) for 6 h to construct a cell model of inflammatory injury induced by OGD [26]. To ensure transfection efficiency, we conducted a pre experiment to optimize the conditions before the formal experiment. Transfect BMEC with FAM fluorescently labeled negative control mimic at the same concentration as miR-29b-3p mimic, and quantify by fluorescence microscopy and flow cytometry to confirm stable transfection efficiency of over 75% in BMEC. Similarly, the pcDNA3.1 empty vector expressing green fluorescent protein (GFP) was used for parallel transfection, and its efficient expression (efficiency > 80%) was confirmed by fluorescence microscopy in both BMEC and HEK293T cells before conducting formal plasmid transfection experiments. The pcDNA3.1-NTRK2 (NTRK2) or pcDNA3.1- empty vector (Vehicle), miR-29b-3p mimic, or NC mimic were constructed by HANBIO (Shanghai, China). The vectors were transfected into the BMEC applying Lipofectamine 3000 (Invitrogen, Carlsbad, CA, USA) in the base of the producer’s guide.

Cell viability assays

The cells with a density of 3000 cells/well were plated in 96-well plates. Three parallel wells were set up for each group. After treatment, 10 μl Cell Counting Kit-8 solution was added to each well and incubated for 1 h at 37 °C under 5% CO2. At last, the absorbance was testified applying a Microplate Reader (Bio-Rad, Hercules, CA, USA) at 450 nm.

Western blot analysis

Protein was extracted from BMEC cells applying RIPA lysis solution (Beyotime, Shanghai, China) with protease and phosphate inhibitors. After protein concentration quantification, 40 μg of protein was taken for SDS-PAGE and subsequently transferred to a 0.45 μm nitrocellulose filter membrane, followed by incubation with primary antibody diluted in 5% skim milk for 8–12 h at 4 °C and secondary antibody for 2 h at room temperature. The protein bands were developed applying enhanced chemiluminescence reagents (Genesee Scientific, CA, USA), and analyzed applying Image J software (NIH, USA). The primary antibodies were as follows: rabbit anti-PI3K (1:1000, Cat#4228, Cell Signaling Technology, MA, USA), anti-NLRP3 (1:1000, Cat#ab214185, Abcam, MA, USA), anti-NTRK2 (1:1000, Cat# 29,961–1-AP, Proteintech, Wuhan, China), and anti-β-actin (1: 20,000, Cat# 20,536–1-AP, Proteintech), and the second antibodies was purchased from Biyuntian Company.

Enzyme-linked immunosorbent assay (ELISA)

The contents of IL-1β (Elabscience Biotechnology, Wuhan, China) and TNF-α (Elabscience Biotechnology) were testified applying ELISA kits in the base of the guides.

Statistical analysis

GraphPad Prism 8 software (San Diego, CA, USA) was applied for statistical analysis and receiver-operating characteristic (ROC) curve analysis. The Student’s t test was used for comparisons between MMD and control groups. For experiments involving multiple comparisons, we used one-way analysis of variance (ANOVA) combined with Tukey’s post hoc test, which inherently corrected for multiple comparisons. G*Power 3.1 software was used for post hoc statistical power analysis, with a 75% efficacy threshold as the confidence level. All P values in the report are P.adjust. The data were indicated as mean ± standard deviation (SD). The two-tailed value P < 0.05 was deemed statistically significant.

Results

Differential gene expression analysis and functional enrichment of differential mRNAs

The 973 differentially expressed mRNAs were selected between MMD patients and controls which indicated that 297 were downregulated and 676 were upregulated in MMD patients (Fig. 2A, B, Table S3). The 973 differentially expressed mRNAs were also subjected to GO functional enrichment analysis with P.adjust < 0.01 as the selecting situation and applying 121 functions (Table S4). The Go bar chart shows the top 10 most enriched functions in each of the Molecular Function (MF), Biological Process (BP), and Cellular Component (CC) in the groups of GO (Fig. 2C). KEGG analysis of the 973 differentially expressed mRNAs applying P < 0.01 as the filtering condition indicated that they were enriched in nine pathways (Fig. 2D, Table S5), which were the pathways for chronic myeloid leukemia, thyroid hormone signaling pathway, Ras signaling, MAPK signaling, insulin signaling, human cytomegalovirus infection, and PI3K–Akt signaling. These enriched functions and pathways indicated that the 973 differentially expressed mRNAs may be contained in the complex biological processes underpinning MMD.

Fig. 2.

Differences mRNA and miRNA visualization and enrichment analysis. A Hot map. B Volcanic map. C Go bar chart, with the P value as the sorting indicator, and the top10 is selected in BP, CC and MF to visualize it. D KEGG bubble diagram. E Hot map. F Volcanic map

In MMD patients, three miRNAs with differential expression were chosen compared to controls, revealing that one miRNA (miR-29b-3p) was increased, while two miRNAs (miR-486-5p, miR-451a) were decreased (Fig. 2E, F, Table S6).

WGCNA

Stratified clustering was acted on 40 samples with 40 normal samples (Fig. 3A). In contrast with other settings, setting the soft threshold power to 10 and the scale-free topological fit index R² = 0.908, and the average connectivity of the mRNA groups improved compared to other settings (Fig. 3B). An adjacency matrix was constructed in the base of the optimal soft threshold, followed by topological overlap matrix transformation and divided into different gene modules. A total of six gene high co-expression modules with common expression patterns were identified (Fig. 3C). The heatmap indicated the TOM (Fig. 3D). The Pearson correlations of each of the mRNAs in the six modules with MMD were trialed, and the blue part of the co-expression was strongly positively correlated with MMD prevalence (R = 0.74, P = 6 × 10^–8) (Fig. 3E). Next, 137 candidate mRNAs in the blue module were sought for subsequent analysis (Table S7).

Fig. 3.

Different mRNA performs the relevant diagram of WGCNA analysis. A Cluster diagram of the different mRNA sample. B Difference in the soft threshold selection process when the mRNA performs WGCNA analysis. C Identification of the difference mRNAs to express the module; the correlation between modules and MMD disease. D Heatmap of the TOM; E the correlation between modules and MMD disease

Construction of MMD-related miRNA–mRNA regulatory relationships and functional enrichment of targeted mRNA

A miRNA–mRNA molecular regulatory network was constructed from 137 candidate mRNAs and three differentially expressed miRNAs. By applying the MultiMIR package [27] for predicting the binding sites of differentially expressed miRNAs and candidate mRNAs, 48 miRNA–mRNA targeting relationships were gained, involving three miRNAs and 41 mRNAs. The corresponding relationship network was delineated for the 48 regulatory relationships (Fig. 4A). The red arrows indicated highly expressed miRNAs corresponding to lowly expressed mRNAs, and the blue arrows indicate lowly expressed miRNAs corresponding to highly expressed mRNAs. GO function enrichment analysis was acted on lowly expressed mRNAs applying 66 functions (Table S8). These mRNAs were enriched in 48 BP group functions, 4 CC group functions, and 14 MF group functions (Fig. 4B, C). KEGG pathway analysis was performed on lowly expressed mRNAs using P < 0.05 as a selecting condition and indicated that they were enriched in six pathways (Fig. 4D, E, Table S9), such as the Parkinson’s disease and tryptophan metabolism pathways. GO functional enrichment analysis was also acted on the highly expressed mRNAs applying P < 0.01 as the selecting condition. This indicated that they were enriched in 63 functions (Fig. 4F, Table S10), with 49 enriched in BP group functions, two enriched in CC group functions, and 12 enriched in MF group functions. KEGG pathway analysis was acted on the highly expressed mRNAs applying a selecting condition of P < 0.05 and indicated that they were enriched in three pathways (Fig. 4G, H, Table S11), such as the neuroactive ligand–receptor interaction pathway.

Fig. 4.

Differential miRNA targeted candidate mRNA regulatory network map and enrichment analysis. A Different miRNA targeted candidate mRNA regulation network. Red arrows point to mRNA corresponding to high-expression miRNA corresponding to low expression, and blue arrows point to low-expression mRNA corresponding to high-expression mRNA. The purple round is mRNA and the yellow triangle is miRNA. B GO air bubble map. C GO bar chart, with P value as the sorting indicator, and TOP10 is selected in BP, CC and MF for visualization. D Rich in the KEGG pathway. E Enriched geneticization in the KEGG pathway. F GO bar chart, which uses the P value as the sorting indicator. Top10 is selected in BP, CC, and MF to visualize it. G KEGG bubble diagram. H Enriched geneticization in the KEGG pathway

RT-qPCR validation

Three differential miRNAs (miR-29b-3p, miR-486-5p, andmiR-451a) and seven differential mRNAs (NTRK2, CYP1B1, NKX2-5, CRHR1, MARK4, PDLIM2, and SLC11A1) screened by bioinformatics were validated by RT-qPCR among 20 MMD volunteers and 20 healthy controls. Compared with the leukocytes of the 20 healthy controls, miR-29b-3p was upregulated (P = 0.040, Power = 79.2%, Fig. 5A) and NTRK2 was downregulated (P = 0.024, Power = 79.1%, Fig. 5B) in the leukocytes of the 20 MMD patients. Nevertheless, we did not detect any remarkable varies in the expression of miR-486-5p, miR-451a, CYP1B1, NKX2-5, CRHR1, MARK4, PDLIM2, and SLC11A1 (P < 0.05, Power < 75.0%, Fig. 5C-J). We also discovered that the inflammatory factors IL-1β and TNF-α in the plasma of MMD patients were higher than those in the control group (Fig. 5K, L). NTRK2 expression was negatively correlated with miR-29b-3p and IL-1β expression in participants according to correlation analysis (P < 0.05, Fig. 5M). There was a positive correlation between miR-29b-3 and the expression levels of IL-1β and TNF-α (P < 0.05, Fig. 5M). ROC curves analysis indicated that miR-29b-3p (area under curve [AUC] = 0.719, 95% confidence interval [CI]: 0.531 − 0.907), NTRK2 (AUC = 0.768, 95% CI: 0.605 − 0.931), and miR-29b-3p + NTRK2 (AUC = 0.867, 95% CI: 0.729 − 1.000, Fig. 5N) can be applied as predictors of MMD.

Fig. 5.

Clinical sample validation analysis. A Expression difference of miR-29b-3p between MMD and control group. B Expression difference of NTRK2 between MMD and control group. C-J Expression of CYP1B1, miR-451a, NKX2-5, miR-486-5p, CRHR1, MARK4, PDLIM2, and SLC11A1 between MMD and controls. K and L IL-1β and TNF-α were significantly increased in MMD compared with controls. M Correlation analysis of miRNA and NTRK2 with clinical data. N ROC analysis of miRNA and NTRK2 with MMD. Data in A–L were analyzed using Student’s t test, and the results were presented as mean ± SD. n = 20 per group

Cell mechanistic experiments

To testify the possible mechanism of miR-29b-3p and NTRK2. The TargetScan database was applied to predict the binding sites of miR-29b-3p and the outcomes indicated that miR-29b-3p had 2 binding sites to the 3’UTR of NTRK2 gene (Fig. 6A). The miR-29b-3p performance was remarkably risen in the miR-29b-3p mimic treated group in contrast with the NC mimic group (Fig. 6B). Dual-luciferase reporter assay indicated that miR-29b-3p transfection could reduce the luciferase activity of NTRK2-WT group in contrast with that in the NC mimic + NTRK2-WT group (P < 0.05), but did not have any impact NTRK2-MUT group (P > 0.05, Fig. 6C), which indicated that NTRK2 was the direct target of miR-29b-3p.

Fig. 6.

Functional experimental analysis of miRNA and NTRK2 in inflammatory cell model. A Putative miR-29b-3p-binding sequence and mutation sequence of NTRK2 mRNA. B The miR-29b-3p expression was increased after miR-29b-3p mimic transfection. n = 3. C Binding of miR-29b-3p to NTRK2 was conformed using luciferase reporter assays. n = 3. D Changes in cell viability after different treatments. n = 5. E Representative western blot bands of NTRK2, pPI3K, NLPR3, and β-Actin. F–H Densitometric quantification of NTRK2, pPI3K, NLPR3 in different cell treatment groups. n = 3. I, J Changes of IL-1β and TNF-α in different cell treatment groups detected by ELISA. n = 3. Data were presented as mean ± SD. Data in B and C were analyzed using Student’s t test, while data in D–J were analyzed using one-way ANOVA–Tukey test

OGD was applied to induce an inflammatory injury cell model in BMEC. The viability of BMEC was reduced after OGD treatment when in contrast with the Blank group (P < 0.001, Fig. 6D). The viability of BMEC was improved when cells were overexpressed NTRK2, but then lowered by miR-29b-3p treatment (P < 0.001, Fig. 6D).

The WB results indicated that expressions of NTRK2 and p-PI3K were lowered (P < 0.001, Fig. 6E-H), while the NLRP3 was increased at 6 h after OGD treatment in contrast with the blank group (P < 0.01, Fig. 6I). ELISA data indicated that the expression levels of IL-1β and TNF-α were consistent with NLRP3 at 6 h after OGD treatment (P < 0.01, Fig. 6J, K). NTRK2 overexpression treatment farther risen the levels of NTRK2 and p-PI3K, while lowered the expression of NLRP3, IL-1β, and TNF-α in the OGD + NTRK2 group when in contrast with the OGD + Vehicle group (P < 0.05, Fig. 6E-K). Consistently, miR-29b-3p mimic also lowered the levels of NTRK2 and p-PI3K, while risen the expression of NLRP3, IL-1β, and TNF-α in the OGD + NTRK2 + miR-29b-3p group when in contrast with the OGD + NTRK2 + NC mimic group (P < 0.01, Fig. 6E-K). From a mechanistic perspective, we discovered that inhibition of miR-29b-3p risen the expression of NTKP2 and prevented oxygen–glucose deprivation-induced endothelial cell inflammation across the NTRK2/PI3K/NLRP3 pathway (Fig. 7).

Fig. 7.

miR-29b-3p aggravates neuroinflammation through NTRK2/PI3K/NLRP3 pathway in MMD. When stimulated by external adverse factors, miR-29b-3p may target NTRK2, thereby reducing its expression and affecting endothelial cell proliferation and inflammation through the PI3K/NLRP3 signaling pathway, ultimately affecting the occurrence of MMD

Discussion

MMD is a cerebrovascular disease of unknown etiology, which mainly refers to a pathological state of chronic occlusion of cerebral blood vessels [28]. The pathogenesis of MMD is very complicated, and it is usually diagnosed clinically by radiological methods [29]. Nevertheless, no researches have examined miRNA–mRNA co-expression in MMD, and little is currently known about the molecular mechanisms underlying the pathogenesis of MMD. Therefore, a comprehensive analysis of differentially expressed miRNAs and mRNAs could help to unravel the pathogenesis of MMD. According to the results of this study, there was a differential expression of 973 mRNAs and three miRNAs between MMD patients and healthy controls based on microarray data analysis. A software analysis indicated that novel miRNA–mRNA molecular regulatory mechanisms are contained in the cerebrovascular systems of MMD. The core regulatory molecule of MMD, miR-29b-3p, and its target gene, NTRK2, were identified by RT-qPCR validation. And in vitro experiments indicated that miR-29b-3p promotes BMEC inflammation by targeting NTRK2. That is, miR-29b-3p targets NTRK2 through the PI3K/NLRP3 pathway, thereby affecting the development of MMD.

miRNA-106b, miRNA130a, miRNA-126, miRNA-196a, miRNB-125-3p, and let-7c are aberrantly expressed in the serum of patients with MMD [22, 30, 31]. Additionally, miRNA-6165, miRNA-3679-5p, miRNA6760-5p, miRNA574-5p, miR-421, miR-361-5p, miR-320a, and miR-29b-3 are abnormally expressed in the cerebrospinal fluid of MMD patients [32, 33]. However, circulating serum miRNAs only reveal extracellular miRNA secreted by multiple cells through exosomes; they do not reveal the true expression of specific intracellular miRNAs. This hinders farther studies of miRNA-related target genes and cellular functions. Although an RNF213 variant [34], the HLA family [35]; TGF-β1 and VEGF [36]; inflammatory factors [37]; and miRNA-29b-3p, miRNA-451a, and miRNA-486-5p [15] play important parts in MMD, the pathogenesis of this disease remains unknown. Accordingly, in this research, we analyzed microarray data and derived intracellular miRNA–mRNA molecular regulatory mechanisms to help unravel the pathogenesis of MMD.

The pathology of MMD is characterized by abnormal proliferation of vascular smooth muscle cells (VSMC), which causes thinning and eccentric fibromuscular thickening of the intima, leading to progressive narrowing and occlusion of vessel vessels and collateral development [29, 38, 39]. The current researches have indicated that the pathophysiological process of MMD may include multiple biological processes or signaling pathways, such as neovascularization, immune and inflammatory response, VSMC proliferation, apoptosis, metastasis, and thrombosis. Both KEGG and GO analysis indicated that differentially expressed mRNAs were enriched in the MAPK, Ras, and PI3K signaling pathway, which is related to the outcomes of former research [23, 40, 41]. Our in vitro experiments confirmed that miR-29b-3p inhibits PI3K signaling and activates NLRP3 inflammasome by targeting NTRK2, thereby exacerbating the inflammatory response and cellular damage of brain microvascular endothelial cells (BMEC). However, further elaboration is needed on how this molecular pathway contributes to the overall vascular pathological landscape of MMD. Based on our findings and existing literature, we propose that this pathway may play a central role in driving several key clinical features of MMD. First, endothelial instability is one of the pathological cornerstones of MMD [42]. In this study, overexpression of miR-29b-3p under OGD conditions exacerbated the loss of BMEC activity and the release of inflammatory factors, while overexpression of NTRK2 showed a protective effect. This indicates that in MMD patients, elevated miR-29b-3p may directly lead to endothelial dysfunction, apoptosis, and barrier integrity damage by inhibiting NTRK2 and its downstream PI3K survival signaling pathway. This endothelial instability makes the vascular endothelium more susceptible to damage, creating conditions for plasma protein leakage, inflammatory cell infiltration, and subsequent vascular events. Second, thickening of the neointimal layer is the direct cause of MMD vascular stenosis and occlusion [3]. Our data show that the activation of NLRP3 inflammasome regulated by miR-29b-3p/NRRK2 axis and the subsequent production of IL-1β and TNF-α are not only markers of endothelial inflammation, but may also serve as key paracrine signals. These inflammatory factors can act on smooth muscle cells in the intima of blood vessels, inducing their transition from contractile to synthetic form and promoting their abnormal proliferation and migration to the endometrium, thereby accelerating the formation of new endometrium and narrowing of the lumen [43]. Therefore, we speculate that the miR-29b-3p/NTRK2/PI3K/NLRP3 pathway in endothelial cells is one of the early events that initiate and maintain this pathological vascular remodeling. Regarding the formation of "smoke like" blood vessels. On the one hand, chronic inflammation and abnormal activation of endothelium driven by this pathway may promote a disordered, fragile, and dysfunctional pathological angiogenesis, which is a typical feature of "smoke like" blood vessels. They have abnormal structures and are prone to rupture and bleeding [44]. On the other hand, the PI3K signaling pathway itself plays an important role in angiogenesis [45]. In the chronic ischemic environment of MMD, local and complex regulation may attempt to compensate through various mechanisms such as this pathway, but the ultimate result is a distorted and inefficient collateral circulation network due to the overall dysfunction of the pathway. RT-qPCR validation outcomes indicated that, in contrast with controls, the expression of miR-29b-3p was remarkably higher and that of NTRK2 was remarkably lower in MMD patients. This is consistent with the regulation of target genes by miRNAs, and the outcomes of miR-29b-3p expression are consistent with the experimental outcomes of Kang et al. [15]. And dual-luciferase reporter assay showed that NTRK2 was the direct target of miR-29b-3p. This indicated a molecular regulatory mechanism: miR-29b-3p–NTRK2. Previous studies have shown that knocking down miR-29b-3p can inhibit the expression of PI3K [46]. Based on this, this study did not conduct functional impairment experiments. The dual-luciferase reporter gene and NTRK2 overexpression strongly support the model of miR-29b-3p promoting endothelial cell inflammation by targeting NTRK2 and regulating the PI3K/NLRP3 axis. As the pathophysiological changes of an arterial wall differ from those of leukocytes, and the mRNA microarray data were derived from samples gained from the middle arterial walls of MMD patients, the RT-qPCR validation of the differential mRNAs was suboptimal. There are few diagnostic biomarkers for MMD. The results of this study showed that the combination of circulating miR-29b-3p and NTRK2 exhibited encouraging diagnostic efficacy in distinguishing MMD patients from healthy controls (AUC = 0.867), indicating their enormous potential as novel diagnostic biomarkers for MMD. Compared to traditional imaging diagnosis, a stable plasma biomarker can provide a fast, low-cost, and non-invasive diagnostic tool, which may play an important role in the initial screening of high-risk populations.

The findings of this study, particularly the upregulation of circulating miR-29b-3p, have opened up two promising pathways for its clinical translation. In terms of early detection, miR-29b-3p, as a stable circulating RNA, has the characteristic of becoming an ideal liquid biopsy marker. Meanwhile, the use of miR-29b-3p inhibitors can serve as an innovative therapy. By systemic or local administration, these inhibitors can precisely reduce excessive miR-29b-3p, thereby relieving its inhibition of NTRK2, restoring PI3K mediated survival signaling, and ultimately inhibiting NLRP3 inflammasome driven destructive inflammation. This oligonucleotide-based targeted therapy is expected to directly correct the core molecular defects of MMD, representing a potential shift in the future MMD treatment model. In addition, gene therapy utilizes viral vectors to deliver functional NTRK2 genes to vascular wall cells, achieving their long-term expression. However, we must be aware that an essential step in advancing the findings of this study to clinical applications is prospective validation in larger, multicenter independent cohorts. This is exactly our core plan for the future. It is crucial to compare miR-29b-3p with other emerging biomarkers to clarify its potential localization in the field of MMD diagnosis. At present, the RNF213 gene is the strongest genetic susceptibility marker for MMD in East Asian populations [47]. However, as a diagnostic tool, it has limitations as not all patients carry this variant, and it is a static genetic information that cannot dynamically reflect disease activity or treatment response. In contrast, miR-29b-3p, as a regulated molecule, may fluctuate in circulating levels with disease progression or therapeutic intervention. Therefore, miR-29b-3p is not intended to replace the genetic testing of RNF213, but may serve as a functional epigenetic complement, combined with genetic markers, to more accurately identify high-risk individuals and potentially be used for monitoring disease progression. Similarly, compared to the conventional systemic inflammatory markers, such as C-reactive protein, erythrocyte sedimentation rate, or cytokines, miR-29b-3p may provide different values. Although MMD is associated with systemic inflammation, traditional inflammatory markers lack specificity and are also common in other infectious or autoimmune diseases. The in vitro experiments of this study suggest that the upregulation of miR-29b-3p is directly related to inflammatory damage to brain endothelial cells, which may be closer to the local pathological process of the vascular wall. Moyamoya patients are highly unstable in their cerebrovascular system due to ischemic stimulation, and are prone to rupture due to pathological neovascularization. Brain endothelial cells as a potential source of vascular instability factors induce high plasticity and disintegration of MMD [48]. Accumulating evidence illustrates that the MMD involves allergic angiitis [49]. MiR-29b-3p is contained in the regulation of various vascular-like illnesses [50–53]. Previous studies reveal that miR-29b-3p negatively regulates retinal microvascular endothelial cells’ proliferation and angiogenesis by targeting VEGFA and PDGFB [54]. Data analysis indicated that NTRK2 was remarkably enriched in the PI3K–Akt signaling pathway. Chen et al. reported [23] that improved revascularization in MMD patients was related to the upregulation of miR-126-5p expression in the temporalis muscle and dura mater. Additionally, miRNA-126-5p promotes endothelial cell VEGF, CD31, and eNOS expression through the PI3K–Akt pathway and thus promotes angiogenesis in rats with chronic ischemia due to double vessel occlusion and brain-myo-angiopathy. PI3K–Akt signaling pathway plays a key role in the proliferation, apoptosis, migration, and differentiation of VSMCs in several cardiovascular diseases, which is also one of the pathogenesis mechanisms contained in MMD [55–57]. Our research showed that in oxygen-glucose deficiency, miR-29b-3p may target NTRK2, thereby reducing its expression and affecting endothelial cell proliferation and inflammation through the PI3K/NLRP3 signaling pathway, which ultimately affects the development of MMD.

Previous studies have confirmed a close correlation between peripheral immune cells and MMD [58]. This study chose to validate the expression of miR-29b-3p in circulating white blood cells and explore its role in MMD endothelial inflammation based on this. More and more evidence suggests that MMD is not an isolated cerebrovascular disease, but a systemic disease associated with systemic immune-inflammation disorders [59]. In this context, circulating white blood cells serve as key effector cells in the systemic immune-inflammatory state, and changes in their molecular expression profiles may to some extent reflect the overall pathophysiological environment of the disease, including sustained attacks or impacts on vascular walls [60]. Second, miRNAs are known to act as messengers for intercellular communication, transported through vesicles such as exosomes in the bloodstream, and taken up by distant cells, thereby coordinating pathological processes [61]. Therefore, we speculate that miR-29b-3p derived from white blood cells may participate in dialog with vascular endothelium through some mechanism that has not yet been elucidated. From a clinical application perspective, obtaining biomarkers from peripheral blood has significant advantages in non-invasive and reproducible procedures. Exploring the correlation between circulating miRNAs and diseases in the absence of routine access to intracranial vascular tissue from patients, our data show a positive correlation between miR-29b-3p and plasma levels of inflammatory factors (IL-1β, TNF-α), providing indirect evidence for the link between circulating miRNAs and systemic and vascular inflammation. However, the level of miR-29b-3p in circulating white blood cells may not be completely equivalent to the expression level of diseased cerebral vascular endothelial cells. Although our in vitro experiments validated the function of the miR-29b-3p/NRRK2 axis in brain microvascular endothelial cells, this does not directly prove that the changes observed in patients’ white blood cells are driving endothelial pathology in vivo. Therefore, our research findings should be regarded as an important starting point and hypothesis generation process. It revealed the value of circulating miR-29b-3p as a potential biomarker for MMD and suggested its possible involvement in endothelial inflammation mechanisms through cell experiments. In the future, it is necessary to directly detect the expression of miR-29b-3p and its target molecules in MMD vascular specimens obtained during surgery, and conduct experiments in suitable animal models to ultimately confirm the direct role and cellular origin of this pathway in the vascular pathology of MMD in vivo.

Previous studies have shown that BMEC is a widely used and mature cell model for studying MMD in vitro [62]. The BMEC monolayer culture system used in this study can effectively simulate the endothelial components of the blood–brain barrier and respond to OGD stress, but it cannot fully reproduce the complex vascular microenvironment in MMD patients. Similarly, the use of HEK293T cells for luciferase reporter gene detection is mainly aimed at efficiently and specifically verifying the direct binding of miRNA to the target gene 3’UTR, as confirmed in Takeda et al.’s study [63]. However, its non-vascular background means that it cannot reflect the unique background that this regulatory event may exist in cerebral vascular cells. Therefore, our in vitro findings should be regarded as a prospective proof of principle for the function of this regulatory axis. These results strongly guide future research directions, but ultimately must be validated in animal models that better simulate the complex pathophysiology of MMD, as well as in patient derived vascular tissue, to confirm its in vivo relevance and importance.

This study has the following limitations. First, our bioinformatics analysis integrates databases from different tissue sources. Although this strategy is a commonly used method to obtain sufficient statistical power in rare disease research, it may introduce tissue-specific bias. The miRNA–mRNA regulatory network in white blood cells may not fully represent the complex local interactions in the cerebral vascular wall, which may affect the reproducibility of the regulatory network we constructed in target tissues. Second, another major limitation of this study is its relatively small clinical sample size. Although this is understandable for the study of a rare disease, and post-efficacy analysis shows that it has sufficient statistical power for major positive findings (Power > 75.0%), it does limit the generalizability of the research and its ability to detect more subtle expression changes. Some of the negative results we observed may be due to insufficient statistical power (Power < 75.0%). In addition, the small sample size may lead to overestimation of diagnostic efficacy indicators. In the next step of our work, we will collaborate with multiple medical centers to establish a large-scale, prospective MMD cohort. This will significantly increase the sample size and enable subgroup analysis of patients based on disease staging, clinical presentation, etc., thereby verifying the reproducibility of the findings of this study and evaluating its broader clinical application value. It is crucial to intervene in feasible MMD animal models for causal validation and testing of treatment strategies in a complete biological system, as this can provide the strongest evidence for this pathway as a therapeutic target.

Conclusion

Overall, in this research, a series of differentially expressed miRNAs and mRNAs were identified from microarrays of samples from healthy controls and MMD patients, and a miRNA–mRNA molecular regulatory network was built. The main regulatory molecule, miR-29b-3p, and its target gene, NTRK2, were identified by RT-qPCR validation. Meanwhile, in vitro experiments showed that miR-29b-3p could increase OGD-induced BMEC inflammation through NTRK2/PI3K/NLRP3 pathway. Our findings provide a new molecular perspective for addressing clinical challenges in the field of MMD. More importantly, it proposes circulating miR-29b-3p as a highly promising liquid biopsy biomarker for early detection and dynamic monitoring of diseases.

Author contributions

Conceptualization: Yi Huang, Xiang Gao, and Jie Sun; methodology: Liangzhe Wei, He Ren, Yuanwei Lin, Xinpeng Deng, Yuchun Liu, Jingjing Zeng, and Jinghui Lin; data curation: Jingjing Zeng and Yuchun Liu; writing—original draft preparation: Liangzhe Wei and Jingjing Zeng; writing—review & editing: Xiang Gao, Jie Sun, and Yi Huang. All authors read and approved the final manuscript.

Funding

This work was funded by The Key Research and Development Program of Zhejiang Province, under Grant No. 2024C03281(SD2), Ningbo Top Medical and Health Research Program, under Grant No. 2022020304, and Ningbo Clinical Research Center for Emergency and Critical Diseases, under Grant No. 2024L003.

Data availability

MMD-related mRNA/miRNA microarray datasets were gained from the Gene Expression Omnibus (GEO) database (http:/www.ncbi.nlm.nih.gov/geo).

Declarations

Ethics approval and consent to participate

The Ethics Committee of First Affiliated Hospital of Ningbo University reviewed and approved this study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.