Abstract
Background
Endothelial to mesenchymal transition (EndMT) is a key driver of atherosclerosis (AS). This study integrated multiple RNA transcriptomic and single-cell transcriptomic datasets to identify shared gene expression signatures between EndMT and AS.
Methods
GEO expression profiles were analyzed via differential expression, weighted gene co-expression network analysis (WGCNA), and functional enrichment analyses. A Random Forest model evaluated key genes, validating their expression and diagnostic performance. Single-cell RNA sequencing (scRNA-seq) further assessed their expression patterns, and a Transcription Factor (TF)-mRNA-miRNA regulatory network.
Results
Fifteen overlapping genes were identified, from which six candidates were screened through a Protein-Protein Interaction (PPI) network and validated in vitro. The application of a Random Forest model identified two key hub genes, CCL5 and IL1B, which were significantly upregulated. Their diagnostic Area Under the Curve (AUC) was >0.7 when assessed separately and increased to >0.85 in a combined model. Three shared TFs (NFKBIA, NFKB1, RELA) of the Nuclear Factor-kappa B (NF-κB) pathway were identified, highlighting its central role in disease progression.
Conclusions
CCL5 and IL1B are key genes potentially regulating EndMT-mediated AS via inflammatory pathways including Nucleotide-binding oligomerization domain-like (NOD-like) receptor, Tumor Necrosis Factor (TNF), and NF-κB signaling, offering promising diagnostic and therapeutic targets for AS.
Keywords: Endothelial to mesenchymal transition (EndMT), Atherosclerosis, Bioinformatics, Machine learning
Introduction
Atherosclerosis (AS) is a chronic, progressive disease that serves as the primary pathological basis for various cardiovascular and cerebrovascular diseases, and is clinically characterized by ischemic heart disease, ischemic stroke, and peripheral artery disease (Lusis, 2000). Official statistics indicate AS is responsible for approximately 20 million deaths globally each year, making it a cardiovascular disease that seriously threatens human health and life (Martin et al., 2025). However, the molecular mechanisms underlying AS progression remain incompletely understood, highlighting the need for further investigation into AS pathogenesis and the identification of potential therapeutic candidates.
Endothelial to mesenchymal transition (EndMT) is widely recognized as a critical pathological process in the initiation and progression of AS (Xue et al., 2025). EndMT is regulated by multiple signaling pathways, including transforming growth factor-beta (TGF-β) (Huang et al., 2021; Dong et al., 2023), Notch and Wnt/β-catenin pathways (You et al., 2025; Feng et al., 2025), and is also significantly influenced by the local inflammatory microenvironment (Jia et al., 2022). During the pathological process, endothelial cells (ECs) lose their endothelial markers, such as platelet endothelial cell adhesion molecule-1 (PECAM-1), von willebrand factor (vWF), and acquire a mesenchymal or smooth muscle-like phenotype, including α-smooth muscle actin (α-SMA) and SM22 alpha (SM22α) (Jiang et al., 2025; Singh et al., 2024; Gole et al., 2022). Previous studies have demonstrated that EndMT is involved in multiple stages of AS development, including lesion formation and plaque instability (Liang et al., 2022; Chen, Schwartz & Simons, 2020). The transdifferentiated cells acquire migratory and invasive capabilities, excessively synthesize and deposit extracellular matrix (ECM), thereby disrupting the vascular endothelial layer structure, compromising barrier function, and increasing local vascular permeability (Libby & Soehnlein, 2025). These changes not only promote the infiltration and accumulation of lipids and inflammatory cells but also continuously enlarge the inflammatory response in the vascular wall, collectively leading to the formation, progression, and eventual instability of atherosclerotic plaques (Ajoolabady et al., 2024). Although the core role of EndMT in AS has been confirmed, its precise molecular regulatory network, the specific EC subtypes that initiate the process, and the complex interplay with numerous factors in the local lesion microenvironment-particularly the dynamic changes and functional heterogeneity across different stages of AS progression-remain incompletely elucidated.
Machine learning has emerged as an essential methodology in fundamental research by effectively handling rapidly expanding biological data, constructing predictive models of potential biological processes, and presenting complex bioinformation in a simplified and intuitive manner for enhanced comprehension (Chen et al., 2025). This study integrated machine learning with bioinformatics to identify key genes involved in EndMT-mediated AS. Differentially expressed genes associated with both AS samples and the EndMT process were identified, and the diagnostic value of the key genes was evaluated by analyzing multiple RNA transcriptome datasets. The expression patterns of the key genes were further characterized through analysis of a single-cell sequencing dataset and external experimental validation. Furthermore, construction of a Transcription Factor (TF)–mRNA–miRNA regulatory network predicted potential regulatory relationships involving of key genes at both transcriptional and post-transcriptional levels (Wang et al., 2024). This study elucidated the pathogenesis of AS from the perspective of EndMT at its early stages, aiming to identify potential therapeutic targets and thereby provide novel insights into the disease mechanisms, early diagnosis, and treatment of AS.
Methods
Data acquisition and processing
Gene expression profiling data for AS (GSE100927, GSE28829, and GSE97210) and EndMT (GSE236730, GSE281995, and GSE118446) were obtained from the NCBI GEO (http://www.ncbi.nlm.nih.gov/geo) database. In the GSE236730 dataset, acetate-treated experimental groups were excluded from subsequent analyses. GSE100927, GSE236730, and GSE281995 were designated as the training set to identify overlapping genes between AS and EndMT. The datasets GSE28829, GSE97210, and GSE118446 were utilized as external validation sets to confirm the expression levels and diagnostic performance of the key genes. A Protein-Protein Interaction (PPI) network was constructed using the common differential expression genes (DEGs), and the random forest (RF) algorithm was subsequently applied, which led to the identification of the key genes. The expression levels of candidate genes were validated through in vitro experiments. The expression patterns of the key genes were further characterized through single-cell sequencing analysis. Finally, a TF-mRNA-miRNA regulatory network was constructed based on the identified key genes. Figure 1 illustrated the overall workflow of this study.
Figure 1. Study flowchart.
EndMT, endothelial to mesenchymal transition; AS, Atherosclerosis; HC, Heathy Control; DGEs, Differentially Expressed Genes; WGCNA, Weighted Gene Coexpression Network Analysis; ME, Module Eigengene; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes.
Identification of DEGs and functional enrichment analysis
DEGs in AS (GSE100927) and EndMT (GSE236730, GSE281995) were identified using the R package “limma” (https://bioconductor.org/packages/limma), with selection criteria set at adjusted p-value ≤ 0.05 and |log2(FC)| ≥ 1. Heatmaps and volcano plots were visualized using the “ggplot2” and “ggrepel” packages. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses of the DEGs were performed using the R package “clusterProfiler”. The “enrichplot” package was employed to explore biological functions and signaling pathways, with results visualized using the “ggplot2” package.
Weighted gene co-expression network analysis
Weighted gene co-expression network analysis (WGCNA) is a bioinformatics method that clusters functionally related genes into modules based on gene expression correlation and explores the associations between these modules and traits such as diseases. The “WGCNA” package in R was employed to analyze the dataset GSE100927 (top 50% of genes with high standard deviation), aiming to identify gene modules with significant biological relevance. An initial assessment of various soft-thresholding powers (β) was conducted to construct a scale-free network, and a β value of 10 was ultimately selected as the optimal soft-thresholding parameter to optimize network construction. The network was partitioned into modules using the dynamic tree cut algorithm with a minimum module size of 40 genes. Similar modules were subsequently consolidated via a dynamic hybrid merging method with a merge threshold set at 0.25. Pearson correlation coefficients between module eigengenes (MEs) and sample traits were subsequently calculated to identify modules significantly associated with specific traits.
Protein-protein interaction network construction
The overlapping DEGs were uploaded to the STRING database to construct a PPI network, wherein interactions with a confidence score greater than 0.9 were identified as high-confidence markers and visualized using Cytoscape software.
Quantitative polymerase chain reaction
TGF-β1 is a canonical cytokine for inducing EndMT. In this experiment, human umbilical vein endothelial cells (HUVECs) were treated with 10 ng/mL TGF-β1 for 3 days as the experimental group. HUVECs were purchased from Wuhan Jinkairui Biotechnology Co., Ltd (Wuhan, China). Cells were subjected to total RNA extraction utilizing Trizol reagent (Invitrogen; Thermo Fisher Scientific, Waltham, MA, USA). Subsequently, cDNA was synthesized employing the PrimeScriptTM RT reagent kit (TaKaRa Biotechnology, Shiga, Japan). The quantitative polymerase chain reaction (qPCR) process was executed using a SYBR Green PCR kit (TaKaRa Biotechnology) on a DNA Engine Opticon 2 System (Bio-Rad laboratories, Hercules, CA, USA). The primer sequences, designed utilizing NCBI’s Primer-BLAST function, are presented in Table S1 Normalization of mRNA expression was conducted relative to GAPDH expression. All aforementioned experiments were replicated a minimum of three times to ensure reliability.
Machine learning
Based on the GSE100927 dataset, five machine learning algorithms, including Lasso regression, support vector machine (SVM), RF, XGBoost, and k-nearest neighbor (KNN), were used to construct classification models using default hyperparameter settings. Samples were randomly divided into training and validation sets at a 7:3 ratio. Model performance was evaluated by 5-fold cross-validation in the training set and further assessed in the validation set using the area under the receiver operating characteristic curve (AUC), accuracy, and F1-score. The model with the best overall performance was selected for subsequent analyses.
Expression analysis and diagnostic evaluation of key hub genes
To validate the above findings, the expression levels of key genes in control and disease groups were assessed across six datasets (GSE100927, GSE28829, GSE97210, GSE236730, GSE281995, and GSE118446). Statistical differences between groups were analyzed using Student’s t-test (p < 0.05). Receiver operating characteristic (ROC) curves were generated using the “pROC” package, and an AUC > 0.7 was considered indicative of satisfactory diagnostic performance. Furthermore, a multigene diagnostic model was developed by combining these two key genes using logistic regression. ROC curves were plotted to compare the diagnostic performance between individual genes and the combined model.
Single-cell dataset processing and gene expression analysis
Single-cell transcriptomic datasets GSE159843 and GSE159677 were processed and analyzed using the Seurat package. After quality control and normalization, the datasets were integrated using Canonical Correlation Analysis (CCA) to identify shared cellular structures and reduce dataset-related variations. Endothelial cell subtypes in the GSE159843 dataset were annotated according to the method described by Lebas et al. (2024) Proliferative cells were identified by MKI67 and PTTG1 expression, tip cells by PGF and CXCR4 expression, and MSCs by the expression of mesenchymal markers, including SERPINE1 and FN1. Cell types in the GSE159677 dataset were annotated using the SingleR algorithm. The expression patterns of key genes across cell populations were visualized and compared using the “FeaturePlot” and “DimPlot” functions. The expression of key genes across different groups was examined using the “geom_violin” function.
TF-mRNA-miRNA regulatory network
TF-mRNA and mRNA-miRNA interaction data were retrieved from the TRRUST database (https://www.grnpedia.org/trrust/) and the miRTarBase database (https://mirtarbase.cuhk.edu.cn/) within the NetworkAnalyst platform. Following a screening process of TFs, mRNAs, and miRNAs, a comprehensive TF-mRNA-miRNA regulatory network was constructed and visualized using Cytoscape.
Statistical analysis
Transcriptional data and scRNA-seq analysis was performed in R software (version 4.5.0; R Core Team, 2025). Statistical analyses were performed using R software along with packages such as rstatix and ggpubr. Unpaired Student’s t-test was used for comparisons between the two groups. The false discovery rate (FDR) method was applied to adjust p-values for multiple testing. An adjusted p-value (p.adj) < 0.05 was considered significant, and p.adj < 0.01 was regarded as highly significant. The significance levels were set as follows: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.
Results
Differential expression analysis
To identify differentially expressed genes associated with AS, differential expression analysis was performed between the disease and control groups. Gene expression differences between AS and control groups were analyzed using the “limma” package in R, and DEGs were identified. Hierarchical clustering analysis was performed to assess the expression patterns of DEGs and visualize their distribution across samples based on expression profiles. In the AS dataset GSE100927, 501 DEGs were identified between the AS and control groups, including 370 upregulated and 131 downregulated genes (Figs. 2A, 2B). In the EndMT-related dataset GSE281995, 605 DEGs were identified compared with the control group, comprising 295 upregulated and 310 downregulated genes (Figs. 2C, 2D). Similarly, GSE236730 identified 657 DEGs, including 334 upregulated and 323 downregulated genes (Figs. 2E, 2F).
Figure 2. Identification of DEGs in AS (GSE100927) and EndMT (GSE236730, GSE281995).
(A, C, E) Volcano plots of all DEGs in GSE100927, GSE236730 and GSE281995, bule indicates downregulated DEGs, and red indicates upregulated DEGs. (B, D, F) A heatmap of the top DEGs in GSE100927, GSE236730 and GSE281995.
Functional enrichment analysis
To explore the potential biological functions of the identified DEGs, GO and KEGG enrichment analyses were performed based on the identified DEGs. The enriched terms included biological processes, cellular components, molecular functions, and signaling pathways related to AS. A total of 19 overlapping DEGs identified by Venn diagram analysis (Fig. 3A) were subjected to GO and KEGG enrichment analyses to characterize their potential biological functions. GO enrichment analysis showed that these DEGs were mainly enriched in cytokine receptor binding, cytokine activity, and fibronectin binding in the molecular function category. In the biological process category, significant enrichment was observed in inflammatory response-related processes, including the positive regulation of inflammatory response, cellular response to lipopolysaccharide, and cellular response to molecules of bacterial origin. KEGG pathway analysis revealed that these DEGs were significantly enriched in several inflammation-related pathways, including rheumatoid arthritis, cytokine-cytokine receptor interaction, Nucleotide-binding oligomerization domain-like (NOD-like) receptor signaling pathway, and IL-17 signaling pathway (Fig. 3C). These results indicate that the overlapping DEGs are associated with inflammatory processes potentially involved in EndMT and AS.
Figure 3. Overlapping gene functional enrichment analysis.
(A) Venn diagram identifies overlapping DEGs. (B, C) The enrichment analysis results of GO and KEGG pathway.
WGCNA identifies key gene modules associated with AS
WGCNA is widely used to identify co-expression modules associated with disease traits (Langfelder & Horvath, 2008). WGCNA was performed on the GSE100927 dataset to identify co-expression modules associated with AS. The optimal soft-thresholding power β was determined as 10 (Fig. 4A). Using a module merging threshold of 0.25, eight co-expression modules were identified. Pearson correlation analysis was performed to evaluate the relationships between MEs, among modules, and with clinical traits (Fig. 4B). The turquoise module showed a strong positive correlation with AS (r = 0.73), while the black (r = −0.45) and blue (r = −0.56) modules showed negative correlations with AS (Fig. 4C). By intersecting genes from the AS-associated turquoise module with DEGs identified by differential expression analysis, 15 overlapping genes were obtained (Fig. 4D).
Figure 4. WGCNA analysis of the significant modules in atherosclerosis.
(A) The selection of soft threshold in atherosclerosis (GSE100927). (B) Gene cluster dendrogram for atherosclerosis by dynamic tree cut algorithm. (C) Heatmap of the association between modules and clinical traits in atherosclerosis. (D) Venn diagram demonstrates the intersection of common genes obtained by WGCNA and DEGs. (E, F) The chord diagram showing significantly enriched signaling pathways in GO and KEGG and distribution of DEGs in each pathway.
GO and KEGG enrichment analyses were performed on the 15 overlapping DEGs to characterize their potential biological functions. GO enrichment analysis showed that these genes were mainly enriched in biological processes including cellular response to lipopolysaccharide and leukocyte chemotaxis, cellular components such as lysosomal lumen, and molecular functions including chemokine binding (Fig. 4E). KEGG pathway analysis revealed that these genes were mainly enriched in lipid and atherosclerosis, NOD-like receptor signaling pathway, Tumor Necrosis Factor (TNF) signaling pathway, and other inflammation-related pathways (Fig. 4F). These results suggest that the overlapping DEGs are associated with inflammatory and immune-related processes in AS.
PPI network analysis and candidate gene identification
The 15 overlapping genes were submitted to the STRING database to construct a PPI network, and interaction information was obtained (Fig. 5A). The network data were exported as a TSV file and imported into Cytoscape for visualization and analysis. After removing disconnected nodes, six candidate genes, including CCL5, CXCL1, CSF3, TLR2, GZMA, and IL1B, were selected (Fig. 5B).
Figure 5. Identification and validation of candidate genes.
(A, B) Identification of candidate genes based on the constructed PPI. (C–H) The relative expression levels of CCL5, CXCL1, CSF3, TLR2, GZMA, and IL1B genes in normal HUVECs and HUVECs with EndMT by qPCR. n = 3. *p < 0.05; **p < 0.01.
In this study, in vitro experiments utilizing qPCR were performed to measure the expression levels of six candidate genes during TGF-β1-induced EndMT in HUVECs. To confirm the successful induction of EndMT in HUVECs, we first evaluated the expression of the EndMT-associated markers vWF and SM22α, together with the six candidate genes identified following TGF-β1 treatment, as shown in Fig. S1. As shown in Figs. 5C–5H, HUVECs undergoing TGF-β1-induced EndMT exhibited increased expression of all six candidate genes compared with controls.
Determination of machine learning and identification of key hub genes
To compare the classification performance of different machine learning models, Lasso regression, SVM, RF, XGBoost, and KNN were evaluated using the GSE100927 training dataset. As shown in Figs. 6A–6C, the RF model achieved the highest accuracy (0.867) and F1-score (0.900), with an AUC value of 0.865. The out-of-bag (OOB) error rate was further used to assess RF model stability and determine optimal parameters. Based on the overall evaluation results, RF was selected for subsequent analyses.
Figure 6. Identification of key genes through machine learning.
(A–C) Evaluation of five machine learning algorithms using GSE100927. (D, E) Based on the random forest algorithm, the correlation between the total number of trees and the error rate in the AS as well as the ranking of gene relative importance scores. (F, G) Based on the random forest algorithm, the correlation between the total number of trees and the error rate in the EndMT (GSE236730) as well as the ranking ofgene relative importance scores.
The six candidate genes were used as input features for the RF model, and genes with a relative importance score > 0 were retained as informative features. In the AS dataset, the RF model selected three genes, including CCL5, CSF3, and IL1B (Figs. 6D, 6E). Similarly, RF analysis of the EndMT dataset (GSE236730) identified three genes, including CCL5, CXCL1, and IL1B (Figs. 6F, 6G). Two genes (CCL5 and IL1B) shared between the two models were selected for subsequent analyses.
The expression patterns of the two key genes were evaluated in the EndMT training datasets (GSE236730 and GSE281995), the validation dataset (GSE118446), the AS training dataset (GSE100927), and the validation datasets (GSE28829 and GSE97210). Both genes were significantly upregulated in disease groups across multiple datasets (Figs. 7A, 7B). These findings suggest that CCL5 and IL1B may serve as key regulatory genes in EndMT-mediated AS.
Figure 7. Validation of key genes expression levels.
(A, B) Expression levels of CCL5 and IL1B in GSE236730, GSE281995, GSE118446, GSE100927, GSE28829, and GSE1097210. * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001.
Diagnostic potential of key hub genes in EndMT and AS
ROC analyses were performed in both EndMT and AS datasets to evaluate the diagnostic performance of CCL5 and IL1B. The individual diagnostic value of CCL5 and IL1B was first assessed separately. A logistic regression-based biomarker model integrating the expression levels of these two genes was constructed (Figs. 8A–8D). The combined model achieved AUC values of 0.914 (95% CI [0.857–0.971]) in GSE100927, 1.000 (95% CI [1.000–1.000]) in GSE236730, 0.870 (95% CI [0.706–1.000]) in GSE28829, and 1.000 (95% CI [1.000–1.000]) in GSE118446. The CCL5/IL1B model showed good discrimination between disease and control groups across both AS and EndMT datasets.
Figure 8. Diagnostic potential of IL1B and CCL5 in patients with AS and EndMT.
(A–D) ROC curve of the two key genes and multi-marker diagnostic model in test and validation datasets for AS and EndMT.
scRNA-seq validation of key hub gene expression
The expression patterns of IL1B and CCL5 were validated by analyzing single-cell transcriptomic data from EndMT (GSE159843: 3 HC, 2 EndMT) and AS patients (GSE159677: 3 HC, 3 AS). Analysis of the single-cell datasets using the R package “Seurat” yielded 16,624 cells from EndMT patients and 15,850 cells from AS patients. Based on gene expression profiles, we identified eight cell populations in AS patients (Fig. 9A) and three EC subpopulations in EndMT (Fig. 9B). We then assessed the expression levels of two key genes in different cell populations. In AS patients, the expression levels of the key genes were significantly elevated in immune-related cells, including monocytes, NK cells, and T cells (Fig. 9C). In EndMT, ECs with mesenchymal characteristics (labeled as MSCs) and tip cells (Goveia et al., 2020) showed significantly elevated expression levels of key genes (Fig. 9D). As shown in Figs. 9E–9J, single-cell datasets were analyzed to validate the expression patterns of the key genes between disease and control groups. Both CCL5 and IL1B were significantly upregulated in AS and EndMT disease groups. These findings suggest that these two genes may represent potential molecular markers associated with AS and EndMT.
Figure 9. Single-cell transcriptomic characterization of gastric tissues from AS and EndMT patients.
(A) UMAP visualization of 15,850 cells in the AS scRNA-seq (GSE159677). (B) UMAP visualization of 16,624 cells in the EndMT scRNA-seq (GSE159843). (C, D) The dot plot illustrates the expression levels of key genes across different cell populations in AS and EndMT. (E, G) UMAP visualization of different cell groups in AS and EndMT. (F, H) UMAP visualization of expression levels of key genes in AS and EndMT. (I, J) Violin plot showing expression of key genes in HC vs. AS and HC vs. EndMT.
Construction of the TF-mRNA-miRNA regulatory network
TF-mRNA-miRNA regulatory networks have been widely used to explore potential regulatory interactions and identify molecular mechanisms associated with disease progression (Xu et al., 2022). Therefore, a TF-mRNA-miRNA regulatory network was constructed to identify transcriptional and post-transcriptional regulators of the key gene. Initially, mRNA-miRNA interaction pairs were predicted, resulting in the identification of 64 human miRNAs targeting the two key genes. Subsequently, TF-mRNA interaction pairs were predicted, leading to the identification of 12 TFs. Ultimately, a comprehensive TF-mRNA-miRNA regulatory network was constructed based on the two key genes, 64 miRNAs, and 12 TFs (Fig. 10). Notably, three transcription factors (NFKBIA, NFKB1, and RELA) were shared by CCL5 and IL1B, all of which are components of the NF-κB signaling pathway. These findings suggest that the NF-κB pathway may be involved in inflammatory regulation associated with EndMT and AS.
Figure 10. The miRNA-mRNA-TF regulatory network was constructed based on key genes (IL1B, CCL5).
Bule arrows represent miRNAs, red diamonds denote mRNAs, and green circles signify TFs.
Previous studies have demonstrated that miRNAs participate in the regulation of AS progression through post-transcriptional mechanisms, suggesting their potential roles in disease-associated regulatory networks (Schober, Nazarijahantigh & Weber, 2015; Su et al., 2019). Among the identified miRNAs, miR-149-5p, miR-21-5p, miR-34a-5p, miR-146a-5p, and miR-221-5p have been documented to participate in AS-related processes. However, their specific roles in EndMT-mediated AS require further elucidation (Kim, 2018).
Discussion
Accumulating evidence suggests that EndMT is associated with AS progression. Previous studies demonstrated that high glucose (Lu et al., 2021) and oxLDL (Wang et al., 2022) promote EndMT in human aortic endothelial cells, thereby impairing endothelial function. However, the detailed regulatory networks, specific drivers, stage-specific associations, and precise interventional strategies targeting these processes remain incompletely understood. Integrating transcriptomic analysis, bioinformatics, and machine learning approaches may facilitate the identification of molecular signatures and potential regulatory relationships associated with EndMT and AS. In this study, transcriptomic and single-cell transcriptomic datasets were integrated to characterize the molecular signatures associated with EndMT and AS. The identified genes may represent candidate biomarkers associated with AS.
The study initially identified 19 common DEGs through intersection analysis of EndMT and AS datasets. Subsequently, WGCNA of AS samples revealed the most significantly positively correlated module, whose intersection with the common DEGs yielded 15 overlapping DEGs. These DEGs were predominantly enriched in inflammation-related pathways, such as the NOD-like receptor signaling pathway and TNF signaling pathway, thereby indicating a strong association between EndMT-mediated AS and inflammatory responses. Through integration of PPI network analysis and RF machine learning, CCL5 and IL1B were identified as key hub genes and diagnostic biomarkers for AS. The elevated expression of CCL5 and IL1B under disease conditions was consistently observed in in vitro experiments, multiple independent GEO datasets, and single-cell RNA sequencing analyses, suggesting their potential value as AS-associated biomarker candidates and warranting further investigation into their biological roles and clinical relevance.
Interleukin-1β (IL1B) is a crucial mediator of inflammatory responses and a member of the interleukin-1 (IL-1) cytokine family. This cytokine is primarily produced by activated macrophages (Cavalli et al., 2021). IL1B has been reported to be associated with inflammatory responses and EndMT-related changes during atherosclerosis development (Blirando et al., 2015). The underlying mechanism may involve the activation of classical inflammatory signaling pathways such as NF-κB, thereby upregulating the expression of key EndMT-related transcription factors including Snail and Twist (Cho et al., 2025). Furthermore, IL1B interacts with vascular smooth muscle cells and fibroblasts to establish a pro-inflammatory microenvironment that amplifies the local inflammatory response (Hettwer et al., 2022); stimulates ECM remodeling, and promotes plaque destabilization (Libby, 2017; Shen et al., 2021). These findings collectively highlight IL1B as a promising potential biomarker and therapeutic target for EndMT-mediated AS (Ridker et al., 2017).
C-C motif chemokine ligand 5 (CCL5), a secreted cytokine belonging to the CC chemokine family (Zlotnik & Yoshie, 2012), is primarily produced by T lymphocytes, macrophages, endothelial cells, and certain tumor cells. It plays a central role in recruiting leukocytes, such as T cells, monocytes, and eosinophils, to inflammatory sites (Jongstra-Bilen et al., 2021). In this study, CCL5 expression was demonstrated to be significantly upregulated during the AS process mediated by EndMT. The underlying mechanism may involve CCL5 binding to its specific receptors, thereby activating downstream inflammatory signaling pathways such as ERK1/2 and NF-κB (Zeng et al., 2022; Gao et al., 2025). Targeted inhibition of CCL5 or its receptor signaling pathway may emerge as a novel therapeutic strategy to simultaneously mitigate inflammation and suppress EndMT, thereby stabilizing atherosclerotic plaques (Márquez, van der Vorst & Maas, 2021).
Based on these findings, CCL5 and IL1B may represent potential candidates for therapeutic exploration in EndMT-associated AS; however, further functional studies are required to determine whether they play causal roles in disease progression and can serve as effective therapeutic targets. Clinical studies have demonstrated that elevated IL1B levels are significantly associated with the severity of atherosclerotic disease and adverse cardiovascular outcomes. Consequently, targeting IL1B or its downstream signaling pathways may offer a novel therapeutic strategy for intervening in EndMT-mediated AS (Lavillegrand et al., 2024). Previous studies have shown that genetic deletion of CCR5, the principal receptor for CCL5, was found to significantly reduce atherosclerotic plaque formation and size. The finding functionally demonstrates the crucial role of the CCL5/CCR5 signaling axis in atherosclerotic progression and its association with plaque instability, thereby identifying CCL5 as a pivotal chemokine within the inflammatory milieu of AS (Zernecke et al., 2006).
MiRNAs primarily regulate post-transcriptional gene expression, while transcription factors control the transcription of target genes, with both playing critical roles in biological processes and disease pathogenesis (Zhang et al., 2015). A TF–mRNA–miRNA regulatory network was constructed based on the two key genes. Within this network, we identified several key regulators—including transcription factors NFKBIA (Ghosh & Hayden, 2012), NFKB1 (Luo et al., 2017), and RELA (Beinke & Ley, 2004), along with miRNAs miR-204-5p (Tian et al., 2024), miR-146a-5p (Raitoharju et al., 2011), miR-221-5p (Bazan et al., 2015), miR-34a-5p (Li, Wang & Li, 2015), miR-21-5p (Fan et al., 2014), and hsa-miR-149-5p (Shu et al., 2024)—all of which have been reported to be associated with inflammatory response pathways and atherosclerosis. Under basal conditions, NFKBIA is known to inhibit NF-κB activation, and its inactivation can lead to sustained NF-κB signaling. Downregulation of NFKB1 leads to endothelial dysfunction, exacerbated apoptosis, and enhanced inflammatory responses. Furthermore, miRNAs including miR-204-5p, miR-146a-5p, miR-221-5p, miR-34a-5p, miR-21-5p, and hsa-miR-149-5p have been implicated in exacerbating AS progression and contributing to plaque instability through their modulation of distinct signaling pathways. This suggests that the two key genes may influence the development of EndMT-mediated AS through their regulation of TFs and miRNAs. The underlying molecular mechanisms require further validation through subsequent research.
In summary, this study suggests that IL1B and CCL5 exhibit potential value as candidate biomarkers associated with AS, particularly in the context of EndMT-related pathological processes. However, the limited sample sizes of the analyzed datasets, especially the scarcity of EndMT-related samples, may have influenced the observed diagnostic performance and resulted in potentially overestimated AUC values. Moreover, the available EndMT datasets were generated using different induction strategies, and this heterogeneity may introduce additional variability into gene expression patterns. Although the integration of multiple EndMT datasets may partially reduce the influence of model-specific effects, further studies using larger and more standardized EndMT cohorts are required. In addition, no independent cohort specifically representing EndMT-associated AS was available for validation. Future studies involving larger clinical cohorts and comprehensive in vivo and in vitro functional experiments are needed to further validate these findings and clarify the biological roles of IL1B and CCL5 in AS development.
Conclusion
In conclusion, this study integrated multiple transcriptomic datasets and single-cell RNA sequencing data to characterize gene expression patterns associated with EndMT and AS, and identified CCL5 and IL1B as genes with potential biomarker value. Consistent expression patterns of these genes were observed across bioinformatic analyses, independent datasets, and in vitro cellular models. These findings suggest that CCL5 and IL1B may be involved in EndMT-associated AS pathology and provide a basis for further investigation into their biological functions and clinical relevance.
Supplemental Information
Acknowledgments
The authors used Gemini (Model: Gemini 3 Pro), developed by Google LLC, to assist with the grammatical refinement and polishing of the manuscript. The authors reviewed and edited the output to ensure accuracy and take full responsibility for the content of the publication.
Funding Statement
This work was supported by the Grants from the Natural Science Foundation of Chongqing (CSTB2025NSCQ-GPX1257), Scientific Research Project of Chongqing Municipal Education Commission (Youth Program) [KJQN202600614], Postdoctoral Science Foundation of Chongqing (cstc2021jcyj-bshX0181) and the Doctoral Foundation of Chongqing University of Posts and Telecommunications (A2022-11). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Contributor Information
Zhengjun Hou, Email: houzj@cqupt.edu.cn.
Kunxian Shu, Email: shukx@cqupt.edu.cn.
Additional Information and Declarations
Competing Interests
The authors declare that they have no competing interests.
Author Contributions
Zhengjun Hou conceived and designed the experiments, performed the experiments, analyzed the data, prepared figures and/or tables, authored or reviewed drafts of the article, and approved the final draft.
Sihan Chen conceived and designed the experiments, performed the experiments, analyzed the data, prepared figures and/or tables, authored or reviewed drafts of the article, and approved the final draft.
Yuliang Cui performed the experiments, prepared figures and/or tables, and approved the final draft.
Qinfeng Ma analyzed the data, prepared figures and/or tables, and approved the final draft.
Jiandong Chen analyzed the data, prepared figures and/or tables, and approved the final draft.
Kunxian Shu conceived and designed the experiments, authored or reviewed drafts of the article, and approved the final draft.
Data Availability
The following information was supplied regarding data availability:
The data and code are available in the Supplemental Files.
References
- Ajoolabady et al. (2024).Ajoolabady A, Pratico D, Lin L, Mantzoros CS, Bahijri S, Tuomilehto J, Ren J. Inflammation in atherosclerosis: pathophysiology and mechanisms. Cell Death & Disease. 2024;15(11):817. doi: 10.1038/s41419-024-07166-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bazan et al. (2015).Bazan HA, Hatfield SA, OMalley CB, Brooks AJ, Lightell DJ, Woods TC. Acute loss of miR-221 and miR-222 in the atherosclerotic plaque shoulder accompanies plaque rupture. Stroke. 2015;46(11):3285–3287. doi: 10.1161/STROKEAHA.115.010567. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Beinke & Ley (2004).Beinke S, Ley SC. Functions of NF-kappaB1 and NF-kappaB2 in immune cell biology. Biochemical Journal. 2004;382(2):393–409. doi: 10.1042/BJ20040544. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Blirando et al. (2015).Blirando K, Blaise R, Gorodnaya N, Rouxel C, Meilhac O, Vincent P, Limon I. The stellate vascular smooth muscle cell phenotype is induced by IL-1β via the secretion of PGE2 and subsequent cAMP-dependent protein kinase A activation. Biochimica et Biophysica Acta—Molecular Cell Research. 2015;1853(12):3235–3247. doi: 10.1016/j.bbamcr.2015.09.019. [DOI] [PubMed] [Google Scholar]
- Cavalli et al. (2021).Cavalli G, Colafrancesco S, Emmi G, Imazio M, Lopalco G, Maggio MC, Sota J, Dinarello CA. Interleukin 1α: a comprehensive review on the role of IL-1α in the pathogenesis and treatment of autoimmune and inflammatory diseases. Autoimmunity Reviews. 2021;20(3):102763. doi: 10.1016/j.autrev.2021.102763. [DOI] [PubMed] [Google Scholar]
- Chen, Schwartz & Simons (2020).Chen PY, Schwartz MA, Simons M. Endothelial-to-mesenchymal transition, vascular inflammation, and atherosclerosis. Frontiers in Cardiovascular Medicine. 2020;7:53. doi: 10.3389/fcvm.2020.00053. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Chen et al. (2025).Chen Y, Tang Z, Tang Z, Fu L, Liang G, Zhang Y, Tao C, Wang B. Identification of core immune-related genes CTSK, C3, and IFITM1 for diagnosing Helicobacter pylori infection-associated gastric cancer through transcriptomic analysis. International Journal of Biological Macromolecules. 2025;287(1):138645. doi: 10.1016/j.ijbiomac.2024.138645. [DOI] [PubMed] [Google Scholar]
- Cho et al. (2025).Cho DI, Ahn JH, Kang BG, Hwang I, Cho HH, Jun JH, Yoo J, Cho M, Yoo SJ, Kim HS, Kim YS, Ahn Y. ANGPTL4 prevents atherosclerosis by preserving KLF2 to suppress EndMT and mitigates endothelial dysfunction. Arteriosclerosis, Thrombosis, and Vascular Biology. 2025;45(10):1742–1761. doi: 10.1161/ATVBAHA.125.322700. [DOI] [PubMed] [Google Scholar]
- Dong et al. (2023).Dong Y, Wang B, Du M, Zhu B, Cui K, Li K, Yuan K, Cowan DB, Bhattacharjee S, Wong S, Shi J, Wang DZ, Chen K, Bischoff J, Linton MF, Chen H. Targeting epsins to inhibit fibroblast growth factor signaling while potentiating transforming growth factor-β signaling constrains endothelial-to-mesenchymal transition in atherosclerosis. Circulation. 2023;147(8):669–685. doi: 10.1161/CIRCULATIONAHA.122.063075. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Fan et al. (2014).Fan X, Wang E, Wang X, Cong X, Chen X. MicroRNA-21 is a unique signature associated with coronary plaque instability in humans by regulating matrix metalloproteinase-9 via reversion-inducing cysteine-rich protein with Kazal motifs. Experimental and Molecular Pathology. 2014;96(2):242–249. doi: 10.1016/j.yexmp.2014.02.009. [DOI] [PubMed] [Google Scholar]
- Feng et al. (2025).Feng Y, Li C, Chen J, Xiao X, Mao Q, Zhao H, Wang J, Liu B. Endothelial dysfunction in atherosclerosis: from classical pathways to emerging mechanisms. Vessel Plus. 2025;9:8. doi: 10.20517/2574-1209.2025.39. [DOI] [Google Scholar]
- Gao et al. (2025).Gao R, Shi J, Lyu Y, Ren B, Wei W, Cheng J, Chen J, Zhou Y, Chen J, Sun X, Jiang J, Li B, Yang K. ALKBH5 regulates macrophage senescence and accelerates atherosclerosis by promoting CCL5 m6A modification. Arteriosclerosis, Thrombosis, and Vascular Biology. 2025;45(6):928–944. doi: 10.1161/ATVBAHA.125.322508. [DOI] [PubMed] [Google Scholar]
- Ghosh & Hayden (2012).Ghosh S, Hayden MS. Celebrating 25 years of NF-κB research. Immunological Reviews. 2012;246(1):5–13. doi: 10.1111/j.1600-065X.2012.01111.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Gole et al. (2022).Gole S, Tkachenko S, Masannat T, Baylis RA, Cherepanova OA. Endothelial-to-mesenchymal transition in atherosclerosis: friend or foe? Cells. 2022;11(19):2946. doi: 10.3390/cells11192946. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Goveia et al. (2020).Goveia J, Rohlenova K, Taverna F, Treps L, Conradi L-C, Pircher A, Geldhof V, de Rooij LPMH, Kalucka J, Sterpin L, Vinckier S, Schoonjans L, Kernstock J, Windross SJ, Brüning U, García-Caballero M, Zheng Y, Qian J, Teuwen L-A, Khan S, Boeckx B, Wauters E, Decaluwé H, De Leyn P, Vansteenkiste J, Weynand B, Sagaert X, Verbeken E, Wolthuis A, Topal B, Everaerts W, Bohnenberger H, Emmert A, Panovska D, De Smet F, Staal FJT, Mclaughlin RJ, Impens F, Lagani V, Vinckier S, Mazzone M, Schoonjans L, Dewerchin M, Eelen G, Karakach TK, Yang H, Wang J, Bolund L, Lin L, Thienpont B, Li X, Lambrechts D, Luo Y, Carmeliet P. An integrated gene expression landscape profiling approach to identify lung tumor endothelial cell heterogeneity and angiogenic candidates. Cancer Cell. 2020;37(1):21–36. doi: 10.1016/j.ccell.2019.12.001. [DOI] [PubMed] [Google Scholar]
- Hettwer et al. (2022).Hettwer J, Hinterdobler J, Miritsch B, Deutsch MA, Li X, Mauersberger C, Moggio A, Braster Q, Gram H, Robertson AAB, Cooper MA, Groß O, Krane M, Weber C, Koenig W, Soehnlein O, Adamstein NH, Ridker P, Schunkert H, Libby P, Kessler T, Sager HB. Interleukin-1β suppression dampens inflammatory leucocyte production and uptake in atherosclerosis. Cardiovascular Research. 2022;118(13):2778–2791. doi: 10.1093/cvr/cvab337. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Huang et al. (2021).Huang Q, Gan Y, Yu Z, Wu H, Zhong Z. Endothelial to mesenchymal transition: an insight in atherosclerosis. Frontiers in Cardiovascular Medicine. 2021;8:734550. doi: 10.3389/fcvm.2021.734550. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Jia et al. (2022).Jia M, Li Q, Guo J, Shi W, Zhu L, Huang Y, Li Y, Wang L, Ma S, Zhuang T, Wang X, Pan Q, Wei X, Qin Y, Li X, Jin J, Zhi X, Tang J, Jing Q, Li S, Jiang L, Qu L, Osto E, Zhang J, Wang X, Yu B, Meng D. Deletion of BACH1 attenuates atherosclerosis by reducing endothelial inflammation. Circulation Research. 2022;130(7):1038–1055. doi: 10.1161/CIRCRESAHA.121.319540. [DOI] [PubMed] [Google Scholar]
- Jiang et al. (2025).Jiang H, Zhou Y, Zhang W, Wang B, Wu J, Zhang G. Molecular mechanisms of endothelial–mesenchymal transition and its pathophysiological feature in cerebrovascular disease. Cell & Bioscience. 2025;15(1):49. doi: 10.1186/s13578-025-01393-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Jongstra-Bilen et al. (2021).Jongstra-Bilen J, Tai K, Althagafi MG, Siu A, Scipione CA, Karim S, Polenz CK, Ikeda J, Hyduk SJ, Cybulsky MI. Role of myeloid-derived chemokine CCL5/RANTES at an early stage of atherosclerosis. Journal of Molecular and Cellular Cardiology. 2021;156(10):69–78. doi: 10.1016/j.yjmcc.2021.03.010. [DOI] [PubMed] [Google Scholar]
- Kim (2018).Kim J. MicroRNAs as critical regulators of the endothelial to mesenchymal transition in vascular biology. BMB Reports. 2018;51(2):65–72. doi: 10.5483/bmbrep.2018.51.2.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Langfelder & Horvath (2008).Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics. 2008;9(1):559. doi: 10.1186/1471-2105-9-559. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lavillegrand et al. (2024).Lavillegrand JR, Al-Rifai R, Thietart S, Guyon T, Vandestienne M, Cohen R, Duval V, Zhong X, Yen D, Ozturk M, Negishi Y, Konkel J, Pinteaux E, Lenoir O, Vilar J, Laurans L, Esposito B, Bredon M, Sokol H, Diedisheim M, Saliba AE, Zernecke A, Cochain C, Haub J, Tedgui A, Speck NA, Taleb S, Mhlanga MM, Schlitzer A, Riksen NP, Ait-Oufella H. Alternating high-fat diet enhances atherosclerosis by neutrophil reprogramming. Nature. 2024;634(8033):447–456. doi: 10.1038/s41586-024-07693-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lebas et al. (2024).Lebas M, Chinigò G, Courmont E, Bettaieb L, Machmouchi A, Goveia J, Beatovic A, Van Kerckhove J, Robil C, Angulo FS, Vedelago M, Errerd A, Treps L, Gao V, Delgado De la Herrán HC, Mayeuf-Louchart A, L’homme L, Chamlali M, Dejos C, Gouyer V, Garikipati VNS, Tomar D, Yin H, Fukui H, Vinckier S, Stolte A, Conradi LC, Infanti F, Lemonnier L, Zeisberg E, Luo Y, Lin L, Desseyn JL, Pickering JG, Kishore R, Madesh M, Dombrowicz D, Perocchi F, Staels B, Pla AF, Gkika D, Cantelmo AR. Integrated single-cell RNA-seq analysis reveals mitochondrial calcium signaling as a modulator of endothelial-to-mesenchymal transition. Science Advances. 2024;10(32):eadp6182. doi: 10.1126/sciadv.adp6182. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Li, Wang & Li (2015).Li N, Wang K, Li PF. MicroRNA-34 family and its role in cardiovascular disease. Critical Reviews in Eukaryotic Gene Expression. 2015;25(4):293–297. doi: 10.1615/critreveukaryotgeneexpr.2015015396. [DOI] [PubMed] [Google Scholar]
- Liang et al. (2022).Liang G, Wang S, Shao J, Jin YJ, Xu L, Yan Y, Günther S, Wang L, Offermanns S. Tenascin-X mediates flow-induced suppression of EndMT and atherosclerosis. Circulation Research. 2022;130(11):1647–1659. doi: 10.1161/CIRCRESAHA.121.320694. [DOI] [PubMed] [Google Scholar]
- Libby (2017).Libby P. Interleukin-1 beta as a target for atherosclerosis therapy: biological basis of CANTOS and beyond. Journal of the American College of Cardiology. 2017;70(18):2278–2289. doi: 10.1016/j.jacc.2017.09.028. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Libby & Soehnlein (2025).Libby P, Soehnlein O. Inflammation in atherosclerosis: lessons and therapeutic implications. Immunity. 2025;58(10):2383–2401. doi: 10.1016/j.immuni.2025.09.012. [DOI] [PubMed] [Google Scholar]
- Lu et al. (2021).Lu L, Zhong Z, Gu J, Nan K, Zhu M, Miao C. ETS1 associates with KMT5A to participate in high glucose-mediated EndMT via upregulation of PFN2 expression in diabetic nephropathy. Molecular Medicine. 2021;27(1):74. doi: 10.1186/s10020-021-00339-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Luo et al. (2017).Luo JY, Li XM, Zhou Y, Zhao Q, Chen BD, Liu F, Chen XC, Zheng H, Ma YT, Gao XM, Yang YN. Mutant DD genotype of NFKB1 gene is associated with the susceptibility and severity of coronary artery disease. Journal of Molecular and Cellular Cardiology. 2017;103(4):56–64. doi: 10.1016/j.yjmcc.2017.01.005. [DOI] [PubMed] [Google Scholar]
- Lusis (2000).Lusis AJ. Atherosclerosis. Nature. 2000;407(6801):233–241. doi: 10.1038/35025203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Martin et al. (2025).Martin SS, Aday AW, Almarzooq ZI, Anderson CA, Arora P, Avery CL, Baker-Smith CM, Beaton AZ, Boehme AK, Buxton AE, Carson AP, Commodore-Mensah Y, Evenson KR, Freeman-Hildreth Y, Gidding SS, Godoy-Sanz B, Glynn RJ, Hall ME, Heitner SB, Howard VJ, Huffman MD, Joseph JJ, Kanaya AM, Khan SS, Kitsios GD, Koton S, Lemaitre RN, Lewis TT, Lichtman JH, Loaten G, Luepker RV, Magnani JW, Majumdar U, Makanji SY, Manuel DG, Mozaffarian D, Mussolino ME, Navaneethan SD, Ning H, Parikh NI, Poudel R, Rebholz CM, Rexrode KM, Roth GA, Sampson UKA, Satou GM, Sharma G, Shihaneh AM, Sims M, Smilowitz NR, Spartano NL, St-Onge MP, Thiruvathukal RL, Turakhia MP, Tsao CW, Valero-Elizondo J, van der Ende MY, Virani SS, Vora N, Widmer RJ, Wong ND, Yaghi S, Yancy CW, Yeboah J, Yu B, Zilberstein J. 2025 Heart disease and stroke statistics: a report of US and global data from the American Heart Association. Circulation. 2025;151(8):e41–e660. doi: 10.1161/CIR.0000000000001303. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Márquez, van der Vorst & Maas (2021).Márquez AB, van der Vorst EPC, Maas SL. Key chemokine pathways in atherosclerosis and their therapeutic potential. Journal of Clinical Medicine. 2021;10(17):3825. doi: 10.3390/jcm10173825. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Raitoharju et al. (2011).Raitoharju E, Lyytikäinen LP, Levula M, Oksala N, Mennander A, Tarkka M, Klopp N, Illig T, Kähönen M, Karhunen PJ, Laaksonen R, Lehtimäki T. miR-21, miR-210, miR-34a, and miR-146a/b are up-regulated in human atherosclerotic plaques in the tampere vascular study. Atherosclerosis. 2011;219(1):211–217. doi: 10.1016/j.atherosclerosis.2011.07.020. [DOI] [PubMed] [Google Scholar]
- R Core Team (2025).R Core Team . R: a language and environment for statistical computing. Vienna: R Foundation for Statistical Computing; 2025. Version 4.5.0. [Google Scholar]
- Ridker et al. (2017).Ridker PM, Everett BM, Thuren T, MacFadyen JG, Chang WH, Ballantyne C, Fonseca F, Nicolau J, Koenig W, Anker SD, Kastelein JJP, Cornel JH, Pais P, Pella D, Genest J, Cifkova R, Lorenzatti A, Forster T, Kobalava Z, Vida-Simiti L, Flather M, Shimokawa H, Ogawa H, Dellborg M, Rossi PRF, Troquay RPT, Libby P, Glynn RJ. Antiinflammatory therapy with canakinumab for atherosclerotic disease. The New England Journal of Medicine. 2017;377(12):1119–1131. doi: 10.1056/NEJMoa1707914. [DOI] [PubMed] [Google Scholar]
- Schober, Nazarijahantigh & Weber (2015).Schober A, Nazarijahantigh M, Weber C. MiRNA-mediated mechanisms of the cellular stress response in atherosclerosis. Nature Reviews Cardiology. 2015;12(6):361–374. doi: 10.1038/nrcardio.2015.38. [DOI] [PubMed] [Google Scholar]
- Shen et al. (2021).Shen J, Zhao M, Zhang C, Sun X. IL-1β in atherosclerotic vascular calcification: from bench to bedside. International Journal of Biological Sciences. 2021;17(15):4353–4364. doi: 10.7150/ijbs.66537. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Shu et al. (2024).Shu LX, Cao LL, Guo X, Wang ZB, Wang SZ. Mechanism of efferocytosis in atherosclerosis. Journal of Molecular Medicine. 2024;102(7):831–840. doi: 10.1007/s00109-024-02439-3. [DOI] [PubMed] [Google Scholar]
- Singh et al. (2024).Singh A, Bhatt KS, Nguyen HC, Frisbee JC, Singh KK. Endothelial-to-mesenchymal transition in cardiovascular pathophysiology. International Journal of Molecular Sciences. 2024;25(11):6180. doi: 10.3390/ijms25116180. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Su et al. (2019).Su Y, Yuan J, Zhang F, Lei Q, Zhang T, Li K, Guo J, Hong Y, Bu G, Lv X, Liang S, Ou J, Zhou J, Luo B, Shang J. MicroRNA-181a-5p and microRNA-181a-3p cooperatively restrict vascular inflammation and atherosclerosis. Cell Death & Disease. 2019;10(5):365. doi: 10.1038/s41419-019-1599-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Tian et al. (2024).Tian Z, Ning H, Wang X, Wang Y, Han T, Sun C. Endothelial autophagy promotes atheroprotective communication between endothelial and smooth muscle cells via exosome-mediated delivery of miR-204-5p. Arteriosclerosis, Thrombosis, and Vascular Biology. 2024;44(8):1813–1832. doi: 10.1161/ATVBAHA.123.319993. [DOI] [PubMed] [Google Scholar]
- Wang et al. (2022).Wang ZC, Niu KM, Wu YJ, Liang CY, Yan XY, Gao S, Zhao TC, Gao J, Geng DF, Hou JS. A dual Keap1 and p47phox inhibitor ginsenoside Rb1 ameliorates high glucose/ox-LDL-induced endothelial cell injury and atherosclerosis. Cell Death & Disease. 2022;13(9):824. doi: 10.1038/s41419-022-05274-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wang et al. (2024).Wang S, Wang Q, Zhao K, Zhang S, Chen Z. Exploration of the shared diagnostic genes and mechanisms between periodontitis and primary Sjögren’s syndrome by integrated comprehensive bioinformatics analysis and machine learning. International Immunopharmacology. 2024;141:112899. doi: 10.1016/j.intimp.2024.112899. [DOI] [PubMed] [Google Scholar]
- Xu et al. (2022).Xu Y, Tan Y, Zhang X, Cheng M, Hu J, Liu J, Chen X, Zhu J. Comprehensive identification of immuno-related transcriptional signature for active pulmonary tuberculosis by integrated analysis of array and single cell RNA-seq. Journal of Infectious Diseases. 2022;85(5):534–544. doi: 10.1016/j.jinf.2022.08.017. [DOI] [PubMed] [Google Scholar]
- Xue et al. (2025).Xue Z, Han M, Sun T, Wu Y, Xing W, Mu F, Zhang Z, Liu J, Liang X, Ling L, Wang J, Wang J, Li X, Huang B, Wang D. PIM1 instigates endothelial-to-mesenchymal transition to aggravate atherosclerosis. Theranostics. 2025;15(2):745–765. doi: 10.7150/thno.102597. [DOI] [PMC free article] [PubMed] [Google Scholar]
- You et al. (2025).You N, Liu G, Yu M, Chen W, Fei X, Sun T, Han M, Qin Z, Wei Z, Wang D. Reconceptualizing endothelial-to-mesenchymal transition in atherosclerosis: signaling pathways and prospective targeting strategies. Journal of Advanced Research. 2025;77(Suppl 1):419–441. doi: 10.1016/j.jare.2024.12.049. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zeng et al. (2022).Zeng Z, Lan T, Wei Y, Wei X. CCL5/CCR5 axis in human diseases and related treatments. Genes & Diseases. 2022;9(1):12–27. doi: 10.1016/j.gendis.2021.08.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zernecke et al. (2006).Zernecke A, Liehn EA, Gao JL, Kuziel WA, Murphy PM, Weber C. Deficiency in CCR5 but not CCR1 protects against neointima formation in atherosclerosis-prone mice: involvement of IL-10. Blood. 2006;107(11):4240–4243. doi: 10.1182/blood-2005-09-3922. [DOI] [PubMed] [Google Scholar]
- Zhang et al. (2015).Zhang HM, Kuang S, Xiong X, Gao T, Liu C, Guo AY. Transcription factor and microRNA co-regulatory loops: important regulatory motifs in biological processes and diseases. Briefings in Bioinformatics. 2015;16(1):45–58. doi: 10.1093/bib/bbt085. [DOI] [PubMed] [Google Scholar]
- Zlotnik & Yoshie (2012).Zlotnik A, Yoshie O. The chemokine superfamily revisited. Immunity. 2012;36(5):705–716. doi: 10.1016/j.immuni.2012.05.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The following information was supplied regarding data availability:
The data and code are available in the Supplemental Files.