The gene expression profiling datasets GSE47472 and GSE98278 were acquired from the GEO database (https://www.ncbi.nlm.nih.gov/geo/) for analysis. Among them, GSE47472 included mRNA expression data from 14 AAA neck specimens and 8 normal aortic specimens of organ donors, while GSE98278 contained 48 AAA samples but no normal aortic samples. All data from these two chips were free to use and both of the chips were annotated with the same platform named Illumina HumanHT-12 V4.0 Expression Beadchip (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GPL10558). Thereafter, the raw data from these two datasets underwent data combination, background correction and quantile normalization by “limma” package in R. After preprocessing, the combined dataset consisted of 8 normal aortic samples and 62 AAA samples altogether, which was prepared for further analysis.

The overall differential gene expression analysis based on the combined dataset was performed with the statistical threshold of |log₂Fold Change (FC)|≥1 and adjusted P value <0.05 also by the “limma” R package. In this study, fourteen genes extracted from formal relevant reviews and research articles, including YTHDF1, YTHDF2, YTHDF3, RRP8, ALKBH1, ALKBH2, ALKBH3, YTHDC1, TRMT6, TRMT61A, TRMT61B, TRMT10C, FTO and BMT2, were regarded as common m1A RNA methylation regulators. Then, the differentially expressed m1A regulatory genes (DEMRGs) were identified via the overlapping of differentially expressed genes (DEGs) and 14 candidate m1A regulators. Due to their significantly differential expression level in AAAs, DEMRGs were also defined as AAA-Related m1A regulators.

Based on the DEMRGs and our combined dataset, co-expression analysis aiming to dig the co-expressed genes of DEMRGs in AAA tissues was carried out with “limma” and “Hmisc” packages within R. To be specific, the co-expression analysis was conducted using Pearson correlation as well as the cut-off criteria of |R| ≥ 0.8 and P < 0.05. After that, among the union set of co-expressed genes of all DEMRGs, we performed Gene Ontology (GO) enrichment analysis, which included terms of biological processes (BP), cellular components (CC), molecular functions (MF), and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis utilizing “clusterProfiler” R package. The adjusted P < 0.05 together with q value < 0.05 was considered statistically significant and 15 terms with the top gene counts of each enrichment category were displayed.

A LASSO model was established to primarily screen the qualified biomarkers from AAA-Related m1A regulators in the diagnosis of AAA. The “glmnet” package in R software was used to establish the LASSO regression model (family = “binomial”) based on the gene expression profiles of DEMRGs and the status (AAA or normal) of all the samples. In the current study, the minimum lambda value (λ.min) judged by the minimum binomial deviance was used as a reference to identify the essential genes to be included in the AAA diagnostic model. Additionally, key genes obtained from the LASSO model were then used to conduct Receiver Operating Characteristic (ROC) curve analysis in order to quantitatively evaluate the diagnostic value of DEMRGs in AAA. The ROC curve model was established with HIPLOT online tools (https://hiplot.com.cn/basic/roc) and genes with Area Under Curve (AUC) more than 90% were reckoned as ideal diagnostic biomarkers in identifying AAA.

Because inflammatory immune cell infiltration was a substantial characteristic of AAA pathology, we performed analysis of immune cell infiltration in AAA with CIBERSORT tool in R. CIBERSORT algorithm is an analytical tool from the Alizadeh Lab developed by Newman et al. to provide an estimation of the abundances of member cell types in a mixed cell population, using gene expression data (27). All the relevant R packages and files could be acquired from the CIBERSORT web portal (http://cibersort.stanford.edu/). The gene expression matrix of our combined dataset was input to R and the leukocyte signature matrix (LM22), which contains 547 genes that distinguish 22 human hematopoietic cell phenotypes, was selected as the signature gene file. Afterwards, the result of immune cell infiltration landscape in AAA and normal aortic tissue was output. Then, immune cell types with zero abundance value in more than half of the samples were regarded as extremely low proportion and were thus excluded from our research. The adjusted P < 0.05 was set as the threshold for statistical significance. The relative contents of diverse immune cell types in each sample were visualized as heat maps according to the research outcomes. Moreover, specific immune cell types with significant differences between AAA and normal samples were individually visualized as box plots.

Pearson correlation analysis was also performed to further explore the role of AAA-related m1A regulators during the development of AAA, participating in immune cell infiltration. The degree of correlation among immune cells was analyzed according to the relative proportion of immune cells in each sample. Besides, the relative content of each immune cell type and the expression data of DEMRGs were used to explore the relationship between AAA-Related m1A regulators and immune cells. The Pearson correlation coefficient (R) ≥ 0.45 together with adjusted P < 0.05 was considered as a significantly strong correlation.

Our study obtained human AAA tissue from the clinical cases at the Department of Vascular Surgery, The First Hospital of China Medical University (CMU), undergoing the elective open surgical repair for aneurysm as previously described (14, 28). The AAA population selection and biopsy tissue collection were conducted by China Medical University Aneurysm Biobank (CMU-aB). However, patients with Marfan syndrome, Ehlers-Danlos syndrome, and other identified vascular or connective tissue disorders were excluded from the study. For the inclusive 141 AAA patients in CMU-aB, AAA was diagnosed by computed tomography angiography (CTA). All human AAA samples collection was conducted according to the guidelines from the World Medical Association Declaration of Helsinki. Meanwhile, normal abdominal aorta samples from 18 organ donors were used as the control tissue samples and this was conducted in accordance with the Declaration of Istanbul. Exclusion criteria for the control group included drug history, cancer, infection, and any additional immune-related diseases that might have negatively influenced the study. In addition, each case provided the informed consent for participation. Afterwards, we selected a total of 63 fresh AAA samples and 18 control samples to form our aneurysm biobank applied in this study. The Ethics Committees of the First Hospital of China Medical University approved our study protocol (ethical approval number: 2019-97-2).

All tissue samples were divided into two parts according to different experimental purposes and stored at −80°C condition. The first part consisted of 35 AAA samples and 8 control samples, which was used for total RNA extraction and subsequent reverse transcription PCR as well as RT-qPCR analysis. The second part contained 28 AAA samples and 10 control samples, prepared for protein extraction and quantitative Western blot (WB) analysis.

For the first part of tissue samples, total mRNA was extracted from aortic tissues utilizing TRIzol reagent (93289, Sigma-Aldrich, USA) according to the standard protocol. To ensure the accuracy of the subsequent experiments, only high-quality RNAs with their A260/A280 ratio >1.8 and <2.2 were qualified for use. Thereafter, PrimeScript™ RT Master Mix (RR036A, TaKaRa Bio, Shiga, Japan) was used to synthesize cDNA from total mRNA with the procedure of 15 min of reverse transcription reaction under 37°C followed by 5 s of inactivation of reverse transcriptase under 85°C. For the other part of tissue samples, proteins were extracted from fresh frozen tissues through RIPA lysis buffer (P0013B, Beyotime, Shanghai, China) and later the concentration of each sample was quantified by BCA Protein Assay kit (P0012, Beyotime, Shanghai, China). At last, all the measured protein samples were mixed with 5 ×SDS-PAGE Sample Loading Buffer (P0015, Beyotime, Shanghai, China) in a 4:1 volume ratio. All the cDNA and protein samples undergone preprocessing were then stored at −20°C.

Moreover, the typical 2–3 μm AAA tissue sections utilized to carry out Immunofluorescence (IF) staining analysis were acquired from FangDa Mass Hospital in Yingkou City, Liaoning Province.

RT-qPCR method was applied in the detection of gene relative expression on mRNA level. For all of the tissue-derived samples and cell-derived samples included in the RT-qPCR analysis, the TB Green^® Premix Ex Taq ™ II plus Tli RNaseH (RR820A, TaKaRa Bio, Shiga, Japan) was utilized to conduct RT-qPCR in Applied Biosystems 7500 Real-Time PCR System (Thermo Fisher Scientific, USA) in accordance with the modified two-step amplification protocol. The RT-qPCR conditions were as follows: 30 s of initial denaturation under 95°C; followed by 5 s of denaturation under 95°C, and 34 s of annealing combined with extension under 60°C for 50 cycles. Housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (GAPDH) served as the internal control in every RT-qPCR test for normalization. Each pair of forward and reverse primers was provided by Sangon Biotech (Shanghai, China) and their sequences are shown in Table 1. The relative expression levels of mRNAs in samples were calculated with the 2^−ΔΔCT method.

Western Blot (WB) staining has been widely used in the analysis of gene relative expression on protein level. Protein samples (20 μg of each lane) were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) using 5% spacer gel and 12% separation gel, then transferred onto polyvinylidene difluoride (PVDF) membrane. The membranes were blocked (5% skimmed milk powder in 1 ×TBS with 0.05% Tween 20) for 2 h, followed by incubation with primary antibody at a dilution of 1:1,000 for rabbit anti-YTHDF3 (K005805P, Solarbio, Beijing, China), anti-iNOS (BA0362, BOSTER, Wuhan, China) and 1:10,000 for rabbit anti-GAPDH (D110016, Sangon Biotech, Shanghai, China), anti-β-Actin (abs132001, absin, Shanghai, China), overnight at 4°C. The blots were then incubated with Horseradish Peroxidase (HRP)-labeled Goat Anti-Rabbit IgG (A0208, Beyotime, Shanghai, China) at a dilution of 1:1,000 for 1–2 h at room temperature (RT). After washing the membranes 3 times with Tris-buffered saline with Tween (TBST), immunoreactive bands were developed and detected using hypersensitive Electro Chemical Luminescence (ECL) kit (P1000-100, Applygen, Beijing, China) and the chemiluminescence detection system (FluorChem HD2, ProteinSimple, USA). The densitometry was performed with Image J software 1.53 (W. Rasband, Research Services Branch, NIMH, National Institutes of Health, Bethesda, MD, USA) and normalized to the signal intensity of GAPDH or β-Actin for equal protein loading control of each sample in each experiment.

Since our bioinformatics analysis of immune cell infiltration in AAA revealed a strong correlation between macrophages and YTHDF3, IF double staining analysis of CD68 (surface marker of macrophages) and YTHDF3 was carried out for cellular localization in AAA tissue.

Firstly, typical 2–3 μm paraffin-embedded AAA sections were dewaxed and rehydrated in line with xylene, isopropanol and descending gradient ethanol. Then, the process of antigen epitopes retrieving was performed by boiling the sections within 0.01M citrate buffer (pH = 6.0). After rinsing by Tris-buffered saline (TBS), the exposed tissues were blocked by Immuno Staining Blocking Buffer (P0102, Beyotime, Shanghai, China) for 2 h, followed by incubation overnight using the rabbit anti-YTHDF3 antibody (dilution 1:100; K005805P, Solarbio, Beijing, China) at 4°C in the humidified chamber. After washing by TBS thrice, the Cy3 (red) conjugated Goat Anti-Rabbit IgG (dilution 1:400; GB21303, Servicebio, Wuhan, China) was utilized to incubate the sections for 2 h under RT, light-free condition. Afterwards, sections were washed 3 times with fresh TBS, and incubated with rabbit anti-CD68 antibody (dilution 1:200; abs120102, absin, Shanghai, China) overnight under 4°C in the humid chamber. Being rinsed thrice with TBS, sections were incubated with Goat anti-Rabbit IgG-AlexaFluor 488 (green, dilution 1:400; abs20025, absin, Shanghai, China) at RT condition away from light for another 2 h. The sections were finally washed three times with TBS, and stained by Antifade Mounting Medium with DAPI (P0131, Beyotime, Shanghai, China) for 10 min in a dark chamber. At last, the fluorescence microscope (Nikon, Tokyo, Japan) was used to observe sections and acquire photos of different visual fields.

The murine macrophage cell line RAW264.7 was obtained from Fudancell (FuHeng BioLogy, FH0328, Shanghai, China) and cultured in 10 cm plates with complete growth medium DMEM High Glucose (MA0545, meilunbio, Dalian, China) containing 10% fetal bovine serum (10100-147, Gibco, Thermo Fisher Scientific, USA) and 1% Penicillin/Streptomycin (10,000 U/mL; SV30010, HyClone, USA) in humidified air with 5% CO₂ at 37°C. To inhibit the differentiation of RAW264.7 cell line, cell scrapers were used for cell subculturing instead of 0.25% Trypsin-EDTA.

Untreated RAWs were defined as M0 macrophages. 1 × 10⁶ M0 macrophages were seeded into each chamber of the 6-well plates for 12–24 h. When the cells were adherent, the previous complete growth medium was removed and treatment for M0 macrophages was placed as follows: control group, 2 mL of complete growth medium for each well only; M1 polarization group was induced by 100 ng/mL lipopolysaccharide (LPS; SMB00610, Sigma-Aldrich, USA) and 50 ng/mL recombinant murine interferon-gamma (IFN-γ; C600059, Sangon Biotech, Shanghai, China) with a total medium mixture volume of 2 mL. After applying the above intervention conditions, the cells were further cultured at 37°C and 5% CO₂ for 24 h, followed by total RNA extraction, protein extraction and cell culture supernatant collection according to the protocols mentioned in parts 2.4 and 2.10.

Small interfering RNA (siRNA) against mouse YTHDF3 as well as Negative Control (NC) siRNA was designed and synthesized by RiboBio (Guangzhou, China). The three siRNA sequences arranged for YTHDF3 mRNA silencing were ythdf3-si-1(5′-3′): ACUCAUUGGUUCCUUUAAGGG; (3′-5′): CUUAAAGGAACCAAUGAGUCC; ythdf3-si-2 (5′-3′): UGGAUUUGUACCAUUCUUCUG; (3′-5′): GAAGAAUGGUACAAAUCCAAG; ythdf3-si-3 (5′-3′): UCAUAUUCUGAAUCUCAUCCU; (3′-5′): GAUGAGAUUCAGAAUAUGAAG.

RAW264.7 cells were transfected with 50 nM YTHDF3 siRNAs or NC siRNAs for 48 h, using riboFECT^TM CP Transfection Kit (C10511-05, RiboBio, Guangzhou, China) in line with manufacturer's instructions. M0 macrophages were inoculated into 6-well plates at 5 × 10⁵ cells/well, and cultured overnight at 37°C in 5% CO₂ incubator. Total RNA, protein and cell culture supernatant samples were all extracted or collected after siRNA transfection, prepared for use in the further analysis. The RT-qPCR analysis was applied in the validation of YTHDF3 mRNA silencing efficiency. Additionally, positive control siRNA conjugated with Cy3 fluorophore (siT0000002-1-5, RiboBio, Guangzhou, China) was also used as the indication of successful transfection.

To further explore the role of YTHDF3 in macrophage M1 polarization, induction of M1 polarization (fresh medium with 100 ng/mL LPS and 50 ng/mL IFN-γ, incubating for 24 h) was performed again after 48 h of YTHDF3 siRNA transfection. Relevant samples were collected and stored for subsequent experiments.

The expression levels of macrophage M1-phenotype marker IL-12 in cell culture supernatant were tested through ELISA. To begin with, cell culture media was centrifuged at 2,000 rpm for 20 min under 4°C, and then supernatant was obtained and ready for detection after the removal of impurities and cell debris. We used Mouse IL-12 ELISA Kit (D721111, Sangon Biotech, Shanghai, China) which employs the sandwich enzyme immunoassay technique for ELISA, following the manufacturer's protocol. After the formation of immune complex and its chromogenic reaction with TMB, the absorbance (OD) value of each sample was measured at 450 nm. The concentration of IL-12 in the sample was proportional to the OD value and could be calculated by quantitative method of standard curve.

RIP experiments were performed using a Magna RIP RNA-binding Protein Immunoprecipitation Kit (Millipore, Billerica, MA, USA) in accordance with the manufacturer's instructions. First of all, nuclear protein was extracted from both AAA tissue and healthy aorta tissue for subsequent RIP-Seq. In brief, after coating with anti-YTHDF3 antibody (sc-377119, Santa Cruz Biotechnology, Dallas, USA) or corresponding NC IgG antibody (Millipore, Billerica, MA, USA), cell lysates from AAA group and normal aorta group were independently used to incubate the coated magnetic beads overnight under 4°C. Later, the specific binding of RNA-protein complexes was treated in accordance with specific protocol for the two types of purified co-precipitated RNAs, and the RNA content was then analyzed by adopting NanoDrop 2000c. Furthermore, we eliminated rRNAs from the immune-precipitated RNAs. Later, we commissioned SEQHEALTH Ltd. Co. (Wuhan, China; Contract No. KC2019-D146) to conduct RIP-transcriptome sequencing analysis of the two groups of our immune-precipitated RNAs and the subsequent bioinformatics analysis. The RIP-Seq results were then uploaded to NCBI Sequence Read Archive (SRA) database. At last, the YTHDF3-binding genes in AAA tissue and healthy aorta tissue were compared and the AAA-specific genes bound to YTHDF3 were identified.

The objective of RIP-Seq analysis of YTHDF3 was to explore the potential function of YTHDF3 by identifying the RNAs binding to YTHDF3 protein in human AAA. Finally, the result of RIP-Seq analysis of YTHDF3 was output and used for PPI network construction and the prediction of YTHDF3 downstream target genes in AAA progression.

2298 AAA-specific genes binding to YTHDF3 were identified via RIP-Seq analysis. In order to dig the downstream target genes (DTGs) of YTHDF3 that also play substantial roles in AAA progression, key intersection genes were yielded through the overlapping of these 2,298 genes, upregulated DEGs in AAA and the positive co-expression genes of YTHDF3 in AAA.

The Search Tool for the Retrieval of Interacting Genes (STRING) database (https://www.string-db.org) was employed to construct a PPI network for key intersection genes. Thereafter, the PPI pairs were extracted using a combined score of >0.55. Subsequently, the YTHDF3-centric PPI network was visualized by Cytoscape 3.9.0 software (https://cytoscape.org/). According to the Maximal Clique Centrality (MCC) method and Degree method (Deg), CytoHubba was employed to calculate the MCC value as well as the degree of each protein node in the whole PPI network. Furthermore, MCODE, a plugin in Cytoscape, was utilized to analyze and visualize the sub-networks (highly interconnected clusters) in the PPI network. Lastly, nodes with the top 20 degrees or top 20 MCC values in the whole PPI network were defined as hub genes. These hub genes were also considered as DTGs of YTHDF3 that might participate in the progression of AAA.

Statistical analysis was performed using SPSS 25.0 software (IBM Corp, Chicago, IL, USA) together with GraphPad Prism 8.0.1 software (GraphPad Software, San Diego, CA, USA) to evaluate differences among groups. Either the parametric Student's t-test for unpaired samples or the non-parametric Mann-Whitney U test was applied to analyze based on the distribution of variables. Comparisons among ≥3 groups were conducted using one-way ANOVA (with Brown-Forsythe test). For correlation analysis in our bioinformatics analysis section, correlations between different variables were analyzed through Pearson correlation analysis. A difference of P value <0.05 suggested statistical significance. All experiments were repeated in triplicates.

Based on the combined dataset and the statistical threshold of |log₂FC| ≥ 1 and adjusted P < 0.05, the differential expression analysis of genes between AAA samples and controls was conducted, as shown in the heat map (Figure 1A) and volcano plot of DEGs (Figure 1C). Later, the overlapping between DEGs and 14 common m1A regulators was carried out and 8 DEMRGs were identified through the analysis. The expression pattern of DEMRGs was shown in the mosaic plot (Figure 1B) and volcano plot (Figure 1D). 8 DEMRGs, containing YTHDC1, YTHDF1-3, RRP8, TRMT61A, FTO, and ALKBH1, were distinctly displayed in Table 2. The significantly differential expression of DEMRGs between AAA groups and control groups was illustrated in box plots (Figures 2A–H).

The expression profile of the 8 DEMRGs was extracted and used to establish the LASSO model (Figures 2I,J). Then, 6 genes, including YTHDC1, YTHDF2-3, RRP8, FTO, and ALKBH1 were identified via the LASSO binomial regression analysis based on the value of λ.min = 0.001374812 and therefore were qualified for the further ROC curve analysis. Thereafter, ROC curve analysis demonstrated the AUC value of ALKBH1 (96.37), YTHDF3 (95.77%), YTHDC1 (91.94%), YTHDF2 (87.10%), FTO (81.05%), and RRP8 (80.04%), indicating ALKBH1, YTHDF3 and YTHDC1 may serve as potential biomarkers of AAA with high diagnostic values (Figure 2K). To be specific, YTHDF3 has the highest diagnosis value (with AUC > 95%) among all the up-regulated DEMRGs, which implies the significance of studying the potential function of YTHDF3 in our subsequent analysis.

After the co-expression analysis of all the DEMRGs, a union set of 7,237 co-expressed genes were yielded and then input into the gene pool which was used for GO and KEGG enrichment analysis (Supplementary Material 1). As indicated by GO analysis, these co-expressed genes were mainly involved in BPs, including RNA catabolic process, neutrophil-mediated immunity, protein targeting, regulation of cellular amide metabolic process and regulation of cell cycle phase transition. In terms of CCs, these genes were mostly enriched in the mitochondrial inner membrane, mitochondrial matrix, cell-substrate junction and focal adhesion. With regard to MFs, these genes were mainly enriched in transcription co-regulator activity, GTPase regulator activity, cadherin binding and ubiquitin protein ligase binding, etc. Additionally, the KEGG pathway enrichment results showed that these genes were significantly enriched in pathways participated in chemical carcinogenesis-reactive oxygen species (ROS), endocytosis and protein processing in endoplasmic reticulum (ER). The outcome of functional enrichment and pathway analysis of co-expressed genes was presented in Supplementary Material 2 and Table 3.

Through the immune cell infiltration analysis, the relative contents of 22 types of immune cells in 62 AAA samples and 8 normal samples were calculated by the CIBERSORT algorithm (Figure 3A). The landscape of immune infiltration in AAA tissues indicated the significantly higher relative proportions of M0 macrophages, M1 macrophages, plasma cells and activated mast cells in AAA than normal tissues (Figures 3C–F). Moreover, after filtering out 8 immune cell types with extremely low proportions, the Pearson correlation analysis was performed to evaluate the correlation among immune cells as well as the correlation between DEMRGs and immune cells in all samples (Figures 3B, 4I). For example, activated dendritic cells were markedly positively correlated with naive B cells (R = 0.53) and monocytes (R = 0.49); M0 macrophages were significantly correlated with activated mast cells (R = 0.47).

As for the relationship between DEMRGs and immune cells, we found that YTHDF3 had the most significantly positive correlation with various immune cells, such as activated mast cells (R = 0.73), Treg cells (R = 0.59), plasma cells (R = 0.52), M1 macrophages (R = 0.53), and M2 macrophages (R = 0.51). Among these immune cells, macrophages served as the appropriate research subject for the further analysis of YTHDF3's regulatory effect, due to its known important role in the formation and progression of AAA. Besides, some of the typical correlations between DEMRGs and immune cells were presented in scatter diagrams (Figures 4A–H).

Thereafter, 6 up-regulated DEMRGs, including YTHDC1, YTHDF1-3, RRP8, and TRMT61A, were validated by RT-qPCR analysis for their relative expression at mRNA level (Figure 5). In our current study, the mRNA expressions of YTHDC1, YTHDF1, and YTHDF3 were observed up-regulating in the AAA tissue samples relative to normal tissues (FC = 4.606, P = 0.043; FC = 26.927, P value = 0.031; FC = 5.426, P = 0.043). Differences in YTHDF2, RRP8, and TRMT61A at mRNA levels were not significant (P = 0.364, P = 0.567, and P = 0.775).

Hence, 3 AAA-Related m1A regulators named YTHDC1, YTHDF1, and YTHDF3 were validated as up-regulated DEMRGs in our AAA samples.

We selectively performed Western Blot analysis for YTHDF3 in AAA and normal aorta tissues, because of its potential high diagnostic value, its strong positive correlation with immune infiltration and its up-regulating expression at mRNA level in human AAAs. Figures 6A,B displayed the blot images of YTHDF3 and its relative expression levels measured through Western blotting in all samples. The protein expression level of YTHDF3 in AAA tissue samples was significantly increased compared with healthy control aortic tissue samples (FC = 3.475, P = 0.012), in accordance with the RT-qPCR result.

Later, to further validate the relationship between YTHDF3 and macrophages, the AAA wall co-localization of YTHDF3 with macrophage surface marker CD68 was examined using IF double staining analysis (Figure 6C). The IF result showed a significant infiltration of macrophages in AAA adventitia. Here, we could also clearly observe that YTHDF3 and CD68 were co-expressed in the one cell in AAA adventitia, which indicated the potential significance of studying the role of YTHDF3 in macrophage functions.

Since the immune cell infiltration analysis together with the IF experiment illustrated the strong correlation between YTHDF3 and M1/M2 macrophages, we started to focus on the association between YTHDF3 and macrophage pro-inflammatory M1 polarization, which was involved in the immune-driven destruction of the aortic wall. Firstly, after the LPS/IFN-γ stimulus, the expression of M1 phenotype marker CD86 and IL12 significantly increased, indicating an ideal M1 macrophages experimental model (Figures 7A,B). Thereafter, the relative expression of YTHDF3 in M0 macrophage and M1 macrophage was determined by Western Blot analysis and the result showed an up-regulating tendency of YTHDF3 in M1 macrophage than M0 macrophage (FC = 1.412) but with the P = 0.149 (Figures 7C,D).

For the aim of further exploring the mechanism underlying the regulation of YTHDF3 in macrophage M1/M2 polarization, siRNA transfection experiment was performed to specifically knock down the expression of YTHDF3 in RAW264.7 cells. The RT-qPCR result of YTHDF3 indicated an ideal ythdf3 knockdown efficiency (FC = 0.199, P < 0.0001, Supplementary Material 3). Later, the RT-qPCR analysis was conducted to analyze the relative expression of M1 phenotype markers (CD86, iNOS, TNFα) and M2 phenotype markers (CD206, Arg-1, TGFβ). The results demonstrated that siRNA knockdown of YTHDF3 inhibited macrophages M1 polarization and at the same time promoted macrophages M2 polarization (Figures 8A,B).

For the YTHDF3 knockdown group, subsequent induction of macrophage M1 polarization was carried out and the expression of Inducible Nitric Oxide Synthase (iNOS, NOS2) and Interleukin 12 (IL12), two of the biomarkers and functional products of M1 macrophages, was tested through Western Blot and ELISA. The comparisons were performed among si-ythdf3+M1 polarization macrophage group, M1 macrophage group and M0 macrophage group. The result showed that YTHDF3 knockdown significantly impaired LPS/IFN-γ induced macrophage M1 polarization, and meanwhile attenuated the secretion of inflammatory factor IL12, significantly reversing the M0 to M1 polarization of macrophages (Figures 8C,E).

First of all, the RIP-Seq result of YTHDF3 in AAA tissue and healthy control sample was independently uploaded to the dataset SRX9734810 and SRX9734811, respectively, in NCBI SRA database. Afterwards, 2,298 AAA-specific YTHDF3-binding genes were yielded via the comparison with control group (Supplementary Material 4). Then, the overlapping of AAA-specific YTHDF3-binding genesYTHDF3, up-regulated DEGs in AAA and positively co-expressed genes of YTHDF3 in AAA was conducted to obtain an intersection of 681 genes (Figure 9A). In addition, these 681 key intersection genes and YTHDF3 were input into the STRING database to construct a PPI network (Supplementary Material 5). The two-level PPI network centered on YTHDF3 was visualized by Cytoscape (Figure 9B). Thereafter, hub genes with top 20 MCC values or node degrees were identified and visualized in circular networks by CytoHubba (Figures 9C,D and Tables 4, 5). After the combination of two hub gene lists, the final 30 hub genes were predicted as YTHDF3 downstream target genes (DTGs) participating in AAA progression. At last, MCODE plugin in Cytoscape was used to perform sub-network analysis of the whole PPI network. The sub-network with the highest cluster score was displayed in Figures 9E,F and hub genes identified by MCC method and Degree method were marked yellow in the diagram, respectively.