IGF2BP2 and LGALS8 were stably overexpressed in Human umbilical vein endothelial cells (HUVECs) using Lenti-oeIGF2BP2 and Lenti-oeLGALS8, respectively, constructed by Shanghai GeneChem Co., Ltd.; and knockdown of IGF2BP2 was achieved using Lv-shIGF2BP2 constructed by Shanghai GeneChem Co., Ltd. (Supplementary Table S1). Transient knockdown of LGALS8 in HUVECs was performed with siRNAs (Supplementary Table S1), synthesized by Shanghai GenePharma Co., Ltd., following the manufacturer’s protocol for Lipofectamine 3000 (Invitrogen). After 72 h of infection or 48 h of transfection, cells were harvested for subsequent mRNA or protein expression analysis.

Total RNA was extracted from cells using the Total RNA Extraction Kit (Solarbio). RNA was reverse transcribed into cDNA using the NovoScript^® Plus All-in-One 1st Strand cDNA Synthesis SuperMix (gDNA Purge) (Novoprotein E047-01B) according to the manufacturer’s instructions. Quantitative PCR amplification was then performed using the NovoStart^® SYBR qPCR SuperMix Plus (Novoprotein E096-01A). Relative quantification was conducted using the 2^−ΔΔCT method. All primer sequences are compiled in Supplementary Table S2.

Proteins were extracted from cells using lysis buffers containing detergents and protease inhibitors to prevent degradation. The protein concentration was determined using a BCA assay to ensure equal sample loading. Proteins were separated by molecular weight using SDS-PAGE (sodium dodecyl sulfate-polyacrylamide gel electrophoresis). Following electrophoresis, the separated proteins were transferred to a polyvinylidene difluoride (PVDF) membrane. The membrane was incubated in a blocking solution to prevent non-specific antibody binding. It was then incubated with a primary antibody specific to the target protein (IGF2BP2, abcam, ab124930, 1: 2000; LGALS8, abcam, ab109519, 1:1000; β-Actin, proteintech, 66009-1-Ig, 1:5000). After washing to remove unbound primary antibody, the membrane was incubated with a horseradish peroxidase (HRP)-conjugated secondary antibody (proteintech, SA00001-1, 1:5000; proteintech, SA00001-2, 1:5000). The chemiluminescence signal was visualized using the digital imaging systems.

RNA sequencing (RNA-seq) was conducted by Novogene. Briefly, RNA integrity was assessed using the RNA Nano 6000 Assay Kit of the Bioanalyzer 2100 system (Agilent Technologies, CA, United States). mRNA was purified using poly-T oligo-attached magnetic beads. Fragmentation was carried out using divalent cations under elevated temperature in First Strand Synthesis Reaction Buffer. First strand cDNA was synthesized using random hexamer primers and M-MuLV Reverse Transcriptase. Second strand cDNA synthesis was subsequently performed using DNA Polymerase I and RNase H. Sequencing adapters were then ligated to the ends of the cDNA fragments. PCR was performed with Phusion High-Fidelity DNA Polymerase, and the products were purified and assessed for quality on the Agilent Bioanalyzer 2100 system. Finally, the library preparations were sequenced on an Illumina Novaseq platform.

Raw sequencing data were subjected to quality control checks to remove low-quality reads and adapter sequences. High-quality reads were aligned to the reference genome using HISAT2. The abundance of transcripts was quantified by counting the number of reads mapped to each gene or transcript using featureCounts. FPKM (Fragments Per Kilobase of transcript per Million mapped reads) of each gene was calculated based on the length of the gene and the read count mapped to it. Differential expression analysis was performed using DESeq2 to compare gene expression levels between different conditions or treatments. The resulting p-values were adjusted using the Benjamini and Hochberg’s approach for controlling the false discovery rate. Genes with an adjusted p-value ≤ 0.05 identified by DESeq2 were considered differentially expressed. Gene enrichment analysis was conducted using Metascape (21).

Endothelial cells were cultured under appropriate conditions until they reached the desired confluency. Transcription was inhibited using actinomycin D, which prevents new mRNA synthesis. Endothelial cells were harvested at various time points after transcription inhibition (e.g., 0, 3, 6, and 9 h) to assess mRNA decay over time. Total RNA was extracted from each sample using the Total RNA Extraction Kit (Solarbio) to ensure high RNA integrity and purity. The extracted RNA was reverse transcribed into cDNA and quantified using qRT-PCR with gene-specific primers. The decay rate and half-life of the target mRNA were calculated based on the decrease in mRNA levels over time.

Matrigel was pipetted into pre-chilled μ-Slide 15 Well 3D plates and allowed to polymerize at 37°C for 30 min. HUVECs were harvested, counted, and resuspended in an appropriate culture medium. The cells were then seeded onto the polymerized Matrigel at a density of 1–2 × 10 ⁴ cells per well. The plates were incubated at 37°C in a humidified atmosphere with 5% CO₂. Tube formation was monitored at 24 h using an inverted microscope. Images of the formed tubular structures were captured for subsequent analysis. ImageJ software with the Angiogenesis Analyzer plugin was used to facilitate the analysis.

AB wild-type and Tg (fli1:EGFP) strain zebrafish were all raised in fish farming water at 28°C (water quality: 200 mg instant sea salt was added to every 1 L of reverse osmosis water; conductivity was 450 ~ 550 μS/cm; pH was 6.5 ~ 8.5; hardness was 50 ~ 100 mg/L CaCO₃). The feeding and management comply with the requirements of the international AAALAC certification (certification number: 001458). Zebrafish at the single-cell stage were used for gene target knockdown or rescue experiments. The Medical Research Ethics Committee of Zhengzhou Central Hospital approved all animal experiments (ZXYY2024129).

AB wild-type strain zebrafish at the single-cell stage were randomly selected. A 1 nL mixture of gRNA (320 ng/μL, Supplementary Table S3) and Cas9 protein (800 ng/μL) was injected into each embryo. After 3 days of treatment at 28°C, 20 juvenile fish from each target site were randomly selected for genome extraction, PCR amplification, and sequencing. The PCR products were used to create TA clones, and 15 clones per target site were sequenced to calculate the target cleavage efficiency. Based on the cleavage efficiency and the presence of frameshift mutations, specific gRNAs were selected for subsequent experiments.

Tg (fli1:EGFP) strain zebrafish at the single-cell stage were randomly selected. The knockdown group was injected with a 1 nL mixture of gRNA (320 ng/μL) and Cas9 protein (800 ng/μL) per embryo, while the control group was injected with the same concentration of an ineffective target. For the rescue experiment, a 1 nL mixture of gRNA (320 ng/μL), Cas9 protein (800 ng/μL), overexpression plasmid (25 ng/μL, with DsRed fluorescent protein as a control), and TP mRNA (25 ng/μL) was injected per embryo. After the embryos were cultured at 28°C for 24 h, PTU was added to inhibit pigment growth. The 4 dpf zebrafish were fixed using 1.5% low-melting point agarose. Confocal imaging was performed using a 10X air microscope with a 2X zoom and an excitation wavelength of 488 nm.

RNA-seq data for control tissues and cerebral arteriovenous malformations were obtained from our previous study (16). MeRIP-seq and RNA-seq data for normal and METTL3 knockdown or WTAP knockdown HUVECs were obtained from the GEO database (GSE142386) (15, 16). RNA-seq data for normal and IGF2BP2 knockdown HUVECs generated in this study are available in the Supplementary Table S4.

T-tests were used to analyze the significance of differences in gene expression, cell proliferation, and tube formation. All statistical analyses were performed using GraphPad Prism 9 and R software, with p-values less than 0.05 considered statistically significant.

Previously, we analyzed transcriptional differences between cerebral arteriovenous malformations (AVMs) and normal tissues (Supplementary Figure S1A) (16). Enrichment analysis revealed that downregulated differentially expressed genes (DEGs) were significantly associated with pathways such as metabolism of RNA, RNA biosynthetic processes, epithelial cell differentiation, tube morphogenesis, and cellular responses to stress (Figure 1A; Supplementary Figure S1B). Given the regulatory role of RNA metabolism in vascular development, we further examined DEGs involved in RNA metabolic pathways and identified IGF2BP2 as the gene with the most pronounced expression difference (Figures 1B,C). However, the role of IGF2BP2 in AVM formation and development remains unclear. To determine its role in vascular development, we knocked down IGF2BP2 in endothelial cells (Figures 1D,E). Knockdown of IGF2BP2 significantly reduced the tube-forming ability of endothelial cells (Figures 1F,G), while overexpression of IGF2BP2 in endothelial cells promoted tube formation compared to controls (Figures 1H–K).

Next, we performed transcriptome sequencing of IGF2BP2 knockdown and control endothelial cells to determine how IGF2BP2 regulates vascular development. RNA-seq analysis identified 5,802 DEGs, including 2,870 upregulated and 2,932 downregulated DEGs (Figure 2A; Supplementary Figure S2). As IGF2BP2 acts as a reader for m⁶A modification, enhancing the stability of m⁶A-modified transcripts and preventing their degradation, we analyzed whether the m⁶A modification of downregulated genes in IGF2BP2-deficient endothelial cells was regulated by the N⁶-methyladenosine methyltransferase complex. By integrating MeRIP-seq data of control and METTL3-deficient or WTAP-deficient endothelial cells, we identified 188 target genes (Figure 2B). Enrichment analysis showed that these genes were mainly involved in the vascular endothelial growth factor receptor signaling pathway and vasculature development (Figure 2C), including SULF1, DAB2IP, RNF213, ADGRG1, NPR3, NAGLU, LOX, LGALS8, HOXA3, VAV3, PSEN1, FLT4, and COL1A2 (Figure 2D). We further analyzed whether the expression of the 13 genes enriched in vascular-related pathways was influenced by METTL3 and WTAP. RNA-seq results indicated that RNF213 and LGALS8 were downregulated in both METTL3-deficient and WTAP-deficient endothelial cells, while GPR56 and SULF1 were downregulated in METTL3-deficient and WTAP-deficient endothelial cells, respectively (Figure 2E). In addition, IGF2BP2 expression was significantly correlated with the expression of RNF213 and LGALS8 in cerebral AVMs (Figures 2F,G).

A comprehensive analysis of expression levels and m⁶A modifications suggested that IGF2BP2 may regulate LGALS8 in an m⁶A-dependent manner, thereby influencing vascular development (Figures 3A,B). To investigate this hypothesis, we examined whether IGF2BP2 regulates LGALS8 expression in endothelial cells. Compared to the control, LGALS8 expression was significantly reduced in IGF2BP2 knockdown endothelial cells (Figures 3C,D). Conversely, LGALS8 expression was significantly increased in IGF2BP2-overexpressing endothelial cells (Figures 3E,F). Since, IGF2BP2 enhances the stability of m⁶A-modified transcripts, we analyzed the mRNA half-life of the LGALS8 transcript in both control and IGF2BP2 knockdown endothelial cells. The results indicated that LGALS8 mRNA levels consistently decreased in IGF2BP2 knockdown endothelial cells compared to control endothelial cells at various time points following actinomycin D treatment, and the mRNA half-life of the LGALS8 transcript was shorter in IGF2BP2 knockdown endothelial cells compared to control endothelial cells (Figures 3G,H).

To determine the role of LGALS8 in angiogenesis, we knocked down its expression in endothelial cells (Figures 4A,B). The results of the tube formation assay showed that LGALS8 knockdown significantly reduced the tube-forming ability of endothelial cells (Figures 4C,D). Additionally, we also overexpressed LGALS8 in endothelial cells (Figures 4E,F). The results of the tube formation assay indicated that overexpression of LGALS8 promoted tube formation compared to control endothelial cells (Figures 4G,H).

To examine the role of IGF2BP2 in vascular development, we generated igf2bp2a/b crispant zebrafish using CRISPR-Cas9 (Figures 5A,B). Compared to control Tg (fli1:EGFP) zebrafish, the density of brain vessels was reduced in the igf2bp2a/b crispant zebrafish (Figures 5C,D). Additionally, we constructed rescue zebrafish by injecting an overexpression plasmid together with gRNA and Cas9 protein into embryos. Imaging results showed that the density of brain vessels in the IGF2BP2 rescue zebrafish was significantly increased compared to control zebrafish injected with DsRed fluorescent protein (Figures 5E,F). Similarly, we investigated the role of LGALS8 in vascular development (Figures 5G,H). We found that the density of brain vessels was dramatically reduced in the lgals8a/b crispant zebrafish (Figures 5I,J), and partially restored in the LGALS8 rescue zebrafish (Figures 5K,L).