Elucidating the role of SF3B3 in coronary atherosclerosis: integrating bioinformatics and machine learning for advanced insights

Abstract

Background

Atherosclerosis (AS) is a chronic inflammatory disease that compromises vascular health and underlies major cardiovascular events. SF3B3, a core spliceosome component, mediates exon–intron processing, yet its role in AS remains unclear.

Methods

We performed integrative analyses of differentially expressed genes across GEO datasets (GSE43292 and GSE9820) and immune-related gene sets from ImmPort. Feature selection was refined using LASSO regression and SVM-RFE. Functional enrichment was assessed via GSEA and GSVA, while immune associations were evaluated with CIBERSORT and ESTIMATE. Key findings were validated in an independent dataset (GSE9820).

Results

Thirteen hub genes associated with SF3B3 were identified. SF3B3 expression correlated with enhanced antiviral defense, cytokine production, and immune signaling pathways. Higher SF3B3 levels were positively associated with adaptive immune populations—including memory and naïve B cells, CD4⁺ and CD8⁺ T cells, follicular helper T cells, and regulatory T cells—while inversely correlated with activated memory CD4⁺ T cells, monocytes, macrophages, eosinophils, and activated dendritic cells.

Conclusions

SF3B3 is closely linked to immune infiltration patterns in AS, highlighting its potential as a biomarker and a candidate target for therapeutic intervention. These findings provide a framework for future mechanistic studies and clinical applications.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13019-026-03979-0.

Introduction

Coronary artery disease (CAD), a leading cause of global mortality, accounted for nearly 18 million deaths in 2017, with projections suggesting an increase to 23.4 million by 2030 [1]. The development of CAD is closely associated with the formation of atherosclerotic plaques, leading to coronary obstruction, myocardial ischemia, and tissue necrosis, thereby heightening the risk of acute cardiovascular syndromes (ACS) [2]. Recent years have witnessed a notable rise in AS prevalence, often linked to diets high in fats and sugars but low in fiber, presenting significant public health challenges [3]. AS, a chronic inflammatory arterial disease, initiates with inflammatory processes in the arterial intima, characterized by the infiltration and retention of low-density lipoprotein (LDL) in the arterial walls. This process leads to the formation of atherosclerotic plaques composed of inflammatory cells and LDL deposits. Despite extensive research, the genomic regulatory mechanisms of AS remain incompletely understood [4]. Metabolism, encompassing essential biochemical reactions, plays a critical role in the onset and progression of various diseases, including metabolic, oncological, and cardiovascular disorders [5]. Disruption of metabolic pathways is a common pathological aspect of these diseases. Recent developments in metabolic research have illuminated the intricate links between metabolic processes and disease progression, revealing new therapeutic targets and biomarkers [6]. Moreover, the critical role of metabolism in modulating immune responses and inflammation is gaining recognition. Metabolic pathways significantly influence immune cell function, affecting their differentiation, activation, and efficacy. Alterations in immune cell metabolism are increasingly identified as contributors to autoimmune diseases, chronic inflammatory conditions, and compromised defenses against pathogens [7]. Understanding these intricate relationships between metabolism, immune responses, and inflammation is crucial for advancing therapeutic strategies. As metabolic pathways become increasingly recognized for their influence on immune cell function and disease progression, they present promising targets for novel interventions aimed at mitigating the impact of CAD and AS [8].

SF3B3 (Splicing Factor 3B Subunit 3) is an integral component of the spliceosome, a complex multiprotein machinery responsible for the excision and ligation of exons and introns from precursor messenger RNA (pre-mRNA) within the cell [9]. SF3B3 plays a pivotal role in this process, participating in the assembly and stabilization of small nuclear ribonucleoprotein (snRNP) complexes. Specifically, SF3B3 is a constituent of the U2 snRNP, which recognizes and binds to the branch point sequence on pre-mRNA, ensuring the precision and efficiency of splicing events [10]. Studies have demonstrated that aberrations in SF3B3 function are associated with various diseases, particularly cancer. Mutations or altered expression levels of the SF3B3 gene can lead to anomalous RNA splicing, thus driving the onset and progression of tumors [11]. For instance, SF3B3 mutations have been widely reported in certain types of leukemia and solid tumors. These mutations often result in dysfunctional splicing factors, leading to extensive alterations in gene expression profiles that ultimately promote cancer cell growth and survival [12]. Moreover, SF3B3 has emerged as a potential therapeutic target. Given its central role in RNA splicing, the development of drugs targeting SF3B3 or its associated complexes shows promising potential [13]. Several small molecule inhibitors have been developed to disrupt SF3B3 function, thereby inhibiting cancer cell proliferation and inducing apoptosis [14]. In summary, SF3B3, as a critical component of the RNA splicing machinery, is under extensive investigation for its functions and roles in disease. By elucidating the molecular mechanisms of SF3B3 and its involvement in pathological processes, scientists hope to develop new diagnostic and therapeutic strategies to address a variety of diseases linked to SF3B3 dysfunction.

Recent studies have highlighted the pivotal role of the immune microenvironment in shaping disease progression, not only in cancer but also in cardiovascular disorders such as AS [15]. In AS, chronic vascular inflammation drives endothelial dysfunction, plaque formation, and immune cell infiltration, creating a heterogeneous tissue milieu that mirrors aspects of the metabolic and immune dysregulation observed in tumors. Immune cells—including T cells, B cells, macrophages, and dendritic cells—interact with vascular and stromal components in a nutrient- and oxygen-limited environment, with their functional states profoundly influenced by local signaling cascades and metabolic stress. These observations underscore the importance of dissecting molecular regulators that coordinate immune activity within diseased vascular tissue. Spliceosome components, such as SF3B3, are central to pre-mRNA processing and have been implicated in modulating immune responses in diverse pathological contexts. Despite extensive characterization in cancer, the expression patterns and functional roles of SF3B3 in AS remain largely unexplored. High-throughput transcriptomic profiling, including historical microarray datasets archived in GEO, provides a valuable resource for systematically investigating dysregulated gene networks and immune-associated pathways in well-phenotyped AS cohorts. While next-generation sequencing offers unprecedented resolution, microarray-based studies continue to offer complementary insights, particularly when analyzed using rigorous integrative bioinformatics frameworks [16, 17]. In this study, we reanalyzed AS-associated expression microarray datasets, employing rigorous preprocessing, batch-effect correction, and multi-algorithm feature selection to uncover reproducible transcriptional signatures across independent cohorts. Plaque-derived transcriptomic data (GSE43292) were used for initial feature discovery, capturing the local vascular microenvironment and disease-relevant molecular perturbations. To evaluate the robustness and systemic relevance of these signatures, validation was performed using circulating immune-cell RNA-seq data (GSE9820). While these datasets originate from distinct compartments, cross-context validation enables identification of transcriptional patterns and immune-associated regulatory networks that are consistent across both vascular tissue and peripheral immune cells, thereby prioritizing biomarkers with broad translational potential. By integrating differential expression analysis with immune infiltration profiling, we specifically examined the role of SF3B3 and its associated molecular networks in shaping the vascular immune landscape. This framework complements sequencing-based studies by emphasizing cross-cohort reproducibility and highlighting candidate targets for mechanistic investigation and therapeutic exploration (Fig. 1).

Fig. 1.

Framework

Materials and methods

We used the approaches proposed by Zi-Xuan Wu, et al. [18].

Source of transcriptional profiling data

The Gene Expression Omnibus (GEO), curated by the National Center for Biotechnology Information (NCBI), is a publicly accessible repository for high-throughput gene expression datasets submitted worldwide, encompassing microarray and next-generation sequencing platforms. In this study, relevant datasets were retrieved from GEO using the following query: (“Atherosclerosis” [MeSH] AND miRNA [All Fields] AND normal) AND (“Homo sapiens” [Organism] AND “Non-coding RNA profiling by array” [Filter]). Included studies reported plasma miRNA profiles from patients with AS and non-atherosclerotic controls. For genes represented by multiple probes, probe-level values were averaged to obtain a single expression value per gene. To integrate data across distinct microarray platforms, expression matrices were normalized and batch effects adjusted using the Limma [19] and sva packages in R4.1.0 [20]. Surrogate variable analysis was applied post-standardization to account for latent technical variation, and principal component analysis (PCA) was used to assess the effectiveness of batch correction. Rather than minimizing cross-platform heterogeneity solely as a confounder, this variation was leveraged as a robustness test for biomarker stability (Table 1).

Table 1.

The clinical characteristics of patients

GSE43292		GSE9820
Variables	Number of samples	Variables	Number of samples
Gender		Gender
Male/female	unknown	Male/Female	64/89
Diagnosis		Diagnosis
AS/Not-AS	32/32	AS/Not-AS	87/66
Tissue		Tissue
Macroscopically intact tissue/ Atheroma plaque	32/32	Macroscopically intact tissue/ Atheroma plaque	unknown

Differential expression analysis between AS and control samples was conducted using the limma framework within each dataset. Multiple testing correction was applied using the Benjamini–Hochberg false discovery rate (FDR) method, and differentially expressed genes (DEGs) were defined by |log₂ fold change|> 1 and FDR-adjusted P < 0.05. These DEGs formed the basis for downstream immune infiltration analyses. Two independent GEO datasets were analyzed: GSE43292 (GPL6244) as the training cohort for feature discovery, and GSE9820 (GPL6255) as an external validation cohort, minimizing circular inference and cross-platform bias (Table 1). Feature selection was further refined through multi-algorithm consensus strategies (LASSO, SVM-RFE, Random Forest), ensuring reproducibility and robustness of candidate biomarkers. Additionally, genes involved in fatty acid metabolism were curated from established resources (Table S1).

Immune landscape characterization and predictive modeling using machine learning approaches

To delineate the association between SF3B3 expression and immune-related biological processes, the immune microenvironment was quantitatively characterized at the individual sample level using the CIBERSORT and ESTIMATE algorithms. The relative proportions of 22 infiltrating immune cell subsets were inferred for each sample, and differences between AS and healthy controlss were evaluated using the Wilcoxon rank-sum test. To establish a predictive model with high accuracy and robustness, LASSO regression was implemented using the glmnet package, incorporating rigorous cross-validation to minimize overfitting and enhance model generalizability across high-dimensional biological data. Feature selection was further optimized through SVM-RFE, as implemented in the e1071 package, enabling the systematic identification of features that contributed most substantially to classification performance. Cross-validation was applied throughout to assess model error rates and predictive precision, thereby strengthening model reliability. In parallel, a Random Forest classifier was constructed to refine feature selection and further mitigate overfitting through ensemble learning. By aggregating predictions from multiple decision trees generated via bootstrap sampling and randomized feature selection, this approach enhanced both model stability and classification accuracy. DEGs and candidate biomarkers were identified using the randomForest and ggplot2 packages, facilitating the prioritization of features with high discriminatory power. An integrative feature importance analysis was subsequently performed by synthesizing outputs from LASSO regression, Random Forest, and SVM-based models, providing a comprehensive and robust evaluation of the mechanistic relevance of candidate genes in AS pathogenesis. The resulting biomarkers were thus refined for downstream functional and translational analyses. Model performance was assessed using receiver operating characteristic (ROC) curve analysis, with the area under the curve (AUC) serving as the primary metric of diagnostic accuracy. The ROC curve depicts the trade-off between sensitivity and specificity across varying thresholds, where an AUC of 1.0 indicates perfect discrimination and values approaching 0.5 reflect no predictive capability. To validate model performance, ROC analyses were conducted using the pROC package in R, integrating AS-associated clinical outcomes with key DEGs across multiple datasets. External validation was performed using the independent GSE9820 cohort, further confirming the robustness and generalizability of the predictive model. Collectively, this ROC-based evaluation framework provides a rigorous assessment of biomarker diagnostic potential and supports their prospective clinical applicability.

Functional annotation and pathway enrichment analyses using GO and KEGG

To elucidate the biological functions and signaling pathways associated with the differentially expressed genes (DEGs), Gene Ontology (GO) enrichment analysis was performed. The biological processes (BP), molecular functions (MF), and cellular components (CC) regulated by the differentially expressed autophagy-related genes were systematically annotated. Pathway enrichment analysis was further conducted based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. All enrichment analyses were implemented in R using the clusterProfiler, org.Hs.eg.db, enrichplot, and ggplot2 packages, enabling comprehensive functional annotation and high-quality visualization of enriched pathways.

Integrated gene set enrichment analyses using GSEA and GSVA

To capture coordinated biological activities at the pathway level, global gene set enrichment analyses were conducted using Gene Set Enrichment Analysis (GSEA). This approach enabled the identification of functionally coherent gene sets and signaling pathways exhibiting differential activation across the study cohorts. Enrichment scores and corresponding visualizations were generated to delineate dynamic pathway alterations across distinct risk stratifications. In parallel, R-based analyses were performed to investigate the impact of differential SF3B3 expression on GO categories (BP, MF, and CC) as well as on associated signaling pathways, drawing from curated gene sets available in the GSEA Molecular Signatures Database (MSigDB; https://www.gsea-msigdb.org/gsea/index.jsp). Collectively, these analyses provided a systems-level perspective on the biological processes and molecular pathways modulated by SF3B3 in AS.

Correlation analyses between diagnostic biomarkers and immune cell infiltration

Immune infiltration analysis is predicated on the quantitative assessment of immune cell composition and functional states within the disease microenvironment, including key immune populations such as T lymphocytes, macrophages, and dendritic cells. Integrating immune cell profiling with biomarker expression enables the elucidation of immunoregulatory mechanisms underlying disease initiation and progression. In the present study, immune infiltration analyses centered on SF3B3 were conducted using a suite of bioinformatics tools. Differential gene expression was assessed using the limma package, while GSVA and GSEABase were applied to evaluate pathway-level immune activity. Data visualization and restructuring were performed using the ggpubr and reshape2 packages. Spearman’s rank correlation analysis was employed to quantify associations between diagnostic biomarkers and immune cell infiltration patterns, thereby providing insights into their coordinated roles within the disease microenvironment.

Integrative analysis of miRNA–lncRNA regulatory networks in AS

Non-coding RNAs, particularly microRNAs (miRNAs) and long non-coding RNAs (lncRNAs), serve as pivotal regulators of gene expression. miRNAs predominantly exert post-transcriptional control by modulating mRNA stability and translation, whereas lncRNAs regulate gene expression through diverse mechanisms, including chromatin remodeling, transcriptional activation, and transcriptional interference. Increasing evidence underscores the complex cross-talk between miRNAs and lncRNAs, giving rise to competing endogenous RNA (ceRNA) networks that fine-tune gene regulatory landscapes. In this study, we sought to delineate shared regulatory axes and developmental trajectories involving miRNAs and lncRNAs in AS. Experimentally validated target genes of the identified miRNAs and lncRNAs were retrieved from the miRTarBase and PrognoScan databases. An integrated regulatory network encompassing mRNA–miRNA–lncRNA interactions and their common targets in AS was subsequently constructed and visualized using Cytoscape.

Results

DEG identification and principal component analysis

We integrated the GSE43292 and GSE9820 datasets and applied batch-effect correction using the SVA package to mitigate platform-specific differences. To investigate patient stratification, principal component analysis (PCA) was performed on the 2,798 DEGs identified between AS and control samples. The first two principal components captured the majority of transcriptional variance, revealing a separation of patients into clusters corresponding to higher- and lower-risk profiles, denoted as “treat” and “control” cohorts (Fig. 2a–b). Specific DEGs contributed disproportionately to this separation. The genes displayed in the heatmap (Fig. 2c) were selected from the 2,798 differentially expressed genes based on a combination of statistical significance and effect size, prioritizing those with the most pronounced absolute log₂ fold changes and adjusted P values after multiple-testing correction. Accordingly, DPP3, ADCY4, VAV1, CNDP2, RIPK3, ITGAL, PLCB2, PSME1, TAP1, PARP12, and STAT1 were preferentially enriched in the coronary artery disease (CAD) group, whereas CALD1, MYLK, MYL9, WASF3, PDE5A, SPARC, PGRMC1, PROS1, MAOB, and CDC14B predominated in the healthy controls (Fig. 2c). These representative genes were chosen to illustrate the most robust and biologically interpretable transcriptional differences underlying group separation, rather than to imply exclusivity or exhaustiveness. Notably, C1QA and C1QB were significantly upregulated, whereas SPARC, PROS1, CTNNAL1, GNG11, MEIS1, FSTL1, SH3BGRL2, and JAM3 were downregulated in AS samples (Fig. 2d; Table S1). Among the 23 hub genes identified via integrative feature selection approaches (LASSO, SVM-RFE, and Random Forest), SF3B3 was selected for further investigation due to its consistent and robust association with immune-related pathways, its high connectivity within the DEG-derived network, and its moderate diagnostic performance (AUC ≈ 0.665) across independent datasets. Functional enrichment and immune-infiltration analyses further highlighted SF3B3 as a central node linking transcriptional dysregulation to alterations in the immune microenvironment, justifying its prioritization for in-depth characterization. Collectively, these results demonstrate that SF3B3 captures critical transcriptional and immunological heterogeneity in AS, providing a rationale for its focused analysis in subsequent sections.

Fig. 2.

Principal component analysis. (a-b) Analysis of PCA. (c) Heatmap. (d) Volcano map. Some of these DEGs were significantly up-regulated (C1QB and C1QA). However, some genes were significantly down-regulated (SPARC, PROS1, CTNNAL1, GNG11, MEIS1, FSTL1, SH3BGRL2, C1QB, JAM3)

Construction of the model

To construct a robust gene signature for AS, we employed LASSO and Cox regression analyses to optimize gene selection, as demonstrated in Fig. 3a and b. Subsequently, the SVM-RFE technique was utilized to develop a machine learning model, confirming the model’s high accuracy and reliability with an accuracy rate of 0.696 and an error rate of 0.304 (Fig. 3c and d). Further investigation using Random Forest analysis identified several key genes (Fig. 3e). These DEGs were then analyzed through a comprehensive approach using Lasso regression, SVM-RFE, and Random Forest algorithms. This integrated analytical strategy successfully pinpointed 23 critical hub genes,, confirmed by consensus among the outputs from the three methodologies (Fig. 3f; Table S2). Based on these results, SF3B3 was selected for in-depth analysis, underlining its potential significance in the pathophysiology of AS.

Fig. 3.

The development of the signature. (a): To construct a robust gene signature for AS, we employed LASSO and Cox regression analyses to optimize gene selection. Regression of the AS-related genes using LASSO. (b): Cross-validation is used in the LASSO regression to fine-tune parameter selection. (c-d): Accuracy and error of this model. (e): Random forest analysis. (f): Venn. Based on these results, SF3B3 was selected for in-depth analysis, underlining its potential significance in the pathophysiology of AS

DEG identification and visualization

We visualized the expression patterns of the 23 hub genes across the AS and normal sample groups (Figure S1). To evaluate their discriminative potential, we calculated receiver operating characteristic (ROC) curves for each gene (Figure S2). The area under the curve (AUC) values were as follows: SF3B3 (0.665), KLRD1 (0.639), ASPA (0.650), LGALS3BP (0.638), RABEP1 (0.614), STYK1 (0.621), SDK2 (0.622), NDFIP1 (0.613), IMPA1 (0.609), PKD1L3 (0.622), OR2T8 (0.608), DAOA (0.596), C10orf111 (0.612), WDR77 (0.612), UBE2L3 (0.613), KCNA10 (0.600), KLHL10 (0.504), KCTD10 (0.586), TAS2R38 (0.589), SPANXB1 (0.591), OR56B4 (0.569), LRRC19 (0.583), and MT1H (0.545).

While these analyses provide an initial assessment of predictive performance, we recognize that characterizing AUC values in the range of ~ 0.55–0.67 as “high” is overly optimistic. Furthermore, the observation that AUCs are occasionally higher in the validation cohort than in the training cohort suggests that aspects of the preprocessing pipeline—such as batch correction or cross-platform integration—may have introduced artifacts that influence model performance. Accordingly, the reported AUCs should be interpreted as preliminary indicators of discriminative potential rather than definitive measures of predictive accuracy. Future analyses incorporating rigorously harmonized datasets, additional independent cohorts, and alternative normalization strategies will be necessary to more robustly evaluate model quality.

Validation of hub genes

GSE9820 was used for validation to boost our model’s confidence and prediction accuracy of these hub genes. What’s interesting is that these DEGs are showed significant differences in GSE9820 analysis (Fig. 4a). In the GSE9820 analysis of hub genes, we analyzed the ROC of these genes, showing that the accuracy of these genes is high. SF3B3 (AUC:0.831), KLRD1 (AUC:0.741), ASPA (AUC:0.774), LGALS3BP (AUC:0.750), STYK1 (AUC:0.752), SDK2 (AUC:0.755), NDFIP1 (AUC:0.794), PKD1L3 (AUC:0.714), WDR77 (AUC:0.748), KCTD10 (AUC:0.820), TAS2R38 (AUC:0.723. These results also confirmed the high reliability and accuracy of our model (Fig. 4b).

Fig. 4.

Validation of Hub Genes. (a) Expression of 23 hub genes in GSE9820 analysis. (b) ROC of 23 hub genes

Differential expression analysis centered on SF3B3

SF3B3 was prioritized as a key gene to delineate its contributions to AS pathophysiology. To identify SF3B3-associated transcriptional changes, Spearman correlation analysis was performed between SF3B3 expression and all DEGs in the training cohort (GSE43292), and genes exhibiting a correlation coefficient |r|> 0.4 with an adjusted P < 0.05 were designated as SF3B3-linked DEGs. This approach yielded 10 genes demonstrating robust co-expression patterns with SF3B3 (Fig. 5). Patients were stratified into high and low SF3B3 expression clusters based on the median SF3B3 expression value across the cohort. Differential expression analysis between these clusters revealed distinct expression patterns: the high-expression cluster encompassed SFXN2, CDCA7, NSUN5, TMC6, ARHGEF1, TBC1D10C, P2RY8, NCR3, and PRKD2, whereas the low-expression cluster included PDGFC, ATP10D, SORT1, CTNNA1, PDLIM7, LATS2, ZAK, KCNMB1, and TRIM36 (Fig. 5a–b). Correlation matrices further elucidated the relationships between SF3B3 and these DEGs, providing a comprehensive visualization of potential regulatory interactions (Fig. 5c; Table S3). Collectively, these results highlight SF3B3 as a central node within its associated transcriptional network, revealing its potential regulatory and pathological roles in AS.

Fig. 5.

DEG Identification of SF3B3. (a): Heatmap. (b): Volcano map. (c): Correlation matrix diagram

Enrichment analysis of DEGs of SF3B3

Figure 6 presents the differential expression analysis of SF3B3, a precursor to functional enrichment analyses aimed at elucidating its biological roles. GO enrichment identified 65 principal targets categorized under MF and BP. The MF category was predominantly associated with enzyme inhibitor activity, carbohydrate binding, immune receptor activity. BPs included response to response to mononuclear cell differentiation, activation of immune response, lymphocyte differentiation. CC was predominantly associated with external side of plasma membrane, secretory granule membrane, cytoplasmic vesicle lumen. Additionally, KEGG pathway analysis highlighted significant involvement of overexpressed genes in pathways such as Cytokine-cytokine receptor interaction (hsa04060), PI3K-Akt signaling pathway (hsa04151), Osteoclast differentiation (hsa04380). These findings, illustrated in Fig. 6 and Table S4a-b, provide insights into the molecular mechanisms through which SF3B3 influences immune and inflammatory responses, potentially impacting AS pathology.

Fig. 6.

For PMGs, GO, and KEGG analyses were performed. (a): The GO circle illustrates the barplot, chord, circos, and cluster of the selected gene’s logFC. (b): The KEGG barplot, chord, circos, and cluster illustrates the scatter map of the logFC of the indicated gene

GSEA of analysis

To elucidate the biological functions impacted by differential expression of SF3B3, we applied GSEA using an array of computational tools including limma for differential expression analysis, org.Hs.eg.db for gene annotation, clusterProfiler and enrichplot for visualization of enrichment results. GSEA facilitated the identification of significant functional alterations among the DEGs linked to SF3B3. In the analysis of BP ribosome biogenesis, CC ribosomal subunit, MF catalytic activity acting on dna. Conversely, in the low expression group, functional enrichments were noted in BP monocyte chemotaxis, BP neuroinflammatory response, CC tertiary granule (Fig. 7a). KEGG pathway analysis revealed that high SF3B3 expression groups showed significant enrichment in pathways like cell cycle, dna replication, primary immunodeficiency. Low expression groups exhibited enrichment in pathways associated with cytokine cytokine receptor interaction, leishmania infection, lysosome (Fig. 7b, Table S4).

Fig. 7.

GSEA of Analysis in SF3B3. (a): GO. (b): KEGG

GSVA of analysis

For the GSVA, we utilized tools such as reshape2 for data restructuring, ggpubr for publication-quality visualizations, along with limma, GSEABase, and GSVA packages for robust analytical assessments. This analysis aimed to pinpoint functional alterations in the SF3B3 DEGs. In GO terms, the high expression group showed enrichment in processes such as BP regulation of t helper 1 cell differentiation, BP chondrocyte hypertrophy, BP regulation of t helper cell differentiation, BP endothelial cell apoptotic process, BP positive regulation of t helper cell differentiation, BP limb joint morphogenesis, BP regulation of extrinsic apoptotic signaling pathway via death domain receptors, BP regulation of extrinsic apoptotic signaling pathway (Fig. 8a). KEGG analysis highlighted enrichment in pathways including regulation of autophagy, vascular smooth muscle contraction, nod like receptor signaling pathway, cytosolic dna sensing pathway, tgf beta signaling pathway, circadian rhythm mammal, complement and coagulation cascades (Fig. 8b).

Fig. 8.

GSVA of Analysis in SF3B3. (a): GO. (b): KEGG

Immune landscape characterization

Figure 9 delineate the immune landscape of AS, with SF3B3 identified as a central regulator of immune contexture. Immune cell infiltration was quantified using two complementary computational algorithms: CIBERSORT, which deconvolutes bulk expression data to estimate relative abundances of 22 immune cell subsets, and ESTIMATE, which provides stromal and immune scores reflecting the overall infiltration level. The dual-algorithm approach was employed to enhance robustness and cross-validate inferred immune profiles. For PCA, genes contributing to the 2,798 DEGs were included, capturing the principal axes of transcriptional variance relevant to patient stratification. Patients were separated into high- and low-SF3B3 expression groups based on the median SF3B3 expression (Fig. 9a). This stratification revealed pronounced differences in adaptive and innate immune populations: SF3B3 expression correlated positively with naive and memory B cells, plasma cells, resting and naive CD4⁺ T cells, CD8⁺ T cells, follicular helper T cells, and regulatory T cells (Tregs), and negatively with activated CD4⁺ memory T cells, monocytes, M0/M1 macrophages, eosinophils, and activated dendritic cells (Fig. 9b, 12; Table S5). Correlation matrices and heatmaps were generated to provide a comprehensive overview of co-variation among immune cell types and SF3B3-linked transcriptional profiles. Lollipop plots further visualized the strength and direction of these associations (Fig. 9c–f), facilitating intuitive interpretation of immune interactions within SF3B3-defined risk clusters. PCA analysis confirmed that SF3B3 expression underlies a consistent immune-based patient stratification (Fig. 9e). Other hub genes with lower predictive performance are reported in the Supplementary Materials; their limited discriminatory capacity underscores SF3B3 as the most robust immune-associated biomarker in this study. Through PCA analysis, immune-based patient categorization was again successfully executed (Fig. 9e). A Lollipop was created to display the expression patterns of Correlation Coefficient. T cells CD8, NK cells activated, T cells follicular helper, T cells CD4 naive, T cells gamma delta, T cells regulatory (Tregs), T cells CD4 memory resting (Fig. 9f). B cells memory, B cells naive, T cells CD4 memory resting, T cells CD4 naive, T cells CD8, T cells follicular helper, T cells regulatory (Tregs) were shown to be positively associated with SF3B3, While, T cells CD4 memory activated, Monocytes, Macrophages M1, Macrophages M0, Eosinophils, Dendritic cells activated were shown to be negatively associated with SF3B3 (Figure S5). (Table. S5). Collectively, these findings position SF3B3 as a key modulator of the immune landscape in AS and prioritize it for further functional and mechanistic exploration, providing a focused framework for investigating its role in vascular inflammation.

Fig. 9.

Immune landscape characterization. (a) Expression of immune function. (b) Expression of immune cells (c) Correlation rectangle plot. (d) Heatmap. (e) PCA analysis. (f) The expression patterns of correlation coefficient

Identification of common RNAs and construction of miRNAs-LncRNAs shared genes network

Figure 10 positions SF3B3 as a central hub within a regulatory network encompassing AS-associated miRNAs and lncRNAs, providing insight into its potential post-transcriptional modulation. Through systematic curation from three public databases, 53 miRNAs and 115 lncRNAs associated with AS were identified. Integrating these with SF3B3-linked DEGs obtained via LASSO regression and SVM-RFE, we constructed a high-confidence miRNA–lncRNA–gene network. This network comprises 10 lncRNAs (including RP11-326C3.10, CDR1-AS, RP11-830F9.6, RP5-894D12.5, RP1-34P24.3, C10orf91, LINC01043, and RP11-326C3.14) and 19 miRNAs (including hsa-miR-590-3p, hsa-miR-377-3p, hsa-miR-365a-3p, hsa-miR-877-3p, hsa-miR-665, and hsa-miR-92a-2-5p) (Fig. 10; Table S5). Functional annotation of these noncoding RNAs suggests their involvement in pathways central to vascular inflammation, immune regulation, and endothelial homeostasis, implicating them as modulators of AS pathophysiology. Notably, several miRNAs and lncRNAs within this network have been previously linked to immune cell differentiation, cytokine signaling, or lipid metabolism, indicating that SF3B3 may influence atherosclerotic progression not only through direct transcriptional control but also via intricate post-transcriptional regulatory circuits. These findings highlight the potential of SF3B3 to act as a key integrator of coding and noncoding RNA interactions, providing a mechanistic framework for future experimental validation in AS.

Fig. 10.

miRNAs-LncRNAs shared genes network. Note: Red circles are mrnas, blue quadrangles are miRNAs, and green triangles are lncRNAs

Discussions

AS, a covert yet pivotal contributor to vascular pathology, remains a foremost etiological factor in cardiovascular disease-associated morbidity and mortality. In light of this, the imperative pursuit of refined diagnostic modalities capable of accurately delineating atherosclerotic risk assumes paramount significance. The progression of atherosclerotic plaques unfolds through a labyrinthine trajectory, bespeaking immunological intricacies reminiscent of oncogenic processes. Consequently, a profound comprehension of the intricate orchestration of immune-associated genes in plaque evolution emerges as an imperative. SF3B3, as a pivotal component of the RNA spliceosome, has garnered significant attention in recent years for its role in cardiovascular diseases. Emerging evidence indicates that aberrant SF3B3 function is intricately linked to the onset and progression of cardiovascular conditions [21]. For instance, alterations in SF3B3 expression can affect the gene expression profiles of cardiomyocytes and vascular smooth muscle cells, leading to abnormalities in cardiac and vascular function [22]. In diseases such as AS, abnormal SF3B3 expression may result in the mis-splicing of key genes, thereby influencing processes like cell proliferation, migration, and apoptosis [23]. Furthermore, the role of SF3B3 in cardiovascular diseases is also evident in its association with inflammatory responses. Chronic inflammation is a common feature of cardiovascular diseases, and SF3B3 can modulate the splicing of inflammation-related genes, impacting the intensity and duration of inflammatory reactions [24]. For example, following myocardial infarction, changes in SF3B3 expression may affect the production and release of inflammatory mediators, subsequently influencing myocardial repair and remodeling processes [25]. The present investigation leverages the potent synergy of bioinformatics methodologies, harmonizing model construction with computational interrogation, thereby laying a robust foundation for subsequent basic and clinical inquiries. This methodological paradigm aims to unravel the intricate tapestry of AS and translate these insights into efficacious therapeutic strategies and clinically relevant paradigms.

In our comprehensive study of AS, we identified 10 DEGs intricately linked with SF3B3. By employing a synergistic approach that combines Lasso regression with SVM-RFE, we refined this extensive list to isolate a critical subset of genes directly relevant to the pathogenesis of AS. A subsequent crossover analysis highlighted 23 central hub genes, including CYBB, SRSF5, CDC7, ARPC5L, RUNX3, SMARCD1, FBXW4, ZNF212, SLC39A7, GPRC5D, DNAJC24, and notably, SF3B3 itself. Validation using external datasets confirmed the diagnostic significance of these genes, firmly integrating them into the complex molecular landscape that orchestrates AS. While our findings are substantial, they do not establish direct causative relationships between these identified genes and specific transcriptional regulators essential for AS modulation. Nevertheless, within this constellation of hub genes, SF3B3 stands out due to its established role in modulating inflammatory processes and immune responses. The identification and characterization of these hub genes not only pave the way for further studies but also underscore the need for a deeper understanding of their regulatory mechanisms. This enhanced understanding is crucial for decoding the intricate molecular basis of AS and could lead to the development of more effective therapeutic interventions, ultimately aiming to mitigate the impact of this debilitating condition.

SF3B3, as a core component of the RNA spliceosome, has garnered increasing recognition for its significant role in AS. AS, a chronic arterial disease characterized by lipid deposition, inflammation, and fibrotic tissue proliferation, forms the primary pathological basis for myocardial infarction and other cardiovascular events [26]. By regulating RNA splicing, SF3B3 influences gene expression, thereby playing a pivotal role in this complex pathological process. SF3B3 is crucial in regulating the functions of vascular endothelial cells and vascular smooth muscle cells [27]. Alterations in SF3B3 expression can lead to the mis-splicing of key genes, affecting cell proliferation, migration, and apoptosis. For instance, aberrant SF3B3 expression may cause dysfunction in endothelial and smooth muscle cells, thereby promoting the formation and progression of atherosclerotic lesions [28]. Furthermore, SF3B3 is closely associated with inflammatory responses, which play a critical role in the initiation and progression of AS. SF3B3 can significantly influence the intensity and duration of inflammatory reactions by regulating the splicing of inflammation-related genes [29]. Specifically, SF3B3 may affect the production of cytokines, chemokines, and adhesion molecules, thus modulating the recruitment and activation of inflammatory cells and ultimately impacting the stability of atherosclerotic plaques. Additionally, SF3B3 dysfunction is linked to lipid metabolism disorders, a hallmark of AS characterized by lipid accumulation in the arterial wall [30]. SF3B3 can regulate lipid synthesis, transport, and storage by influencing the splicing of lipid metabolism-related genes. For example, abnormal SF3B3 expression may reduce the expression of low-density lipoprotein (LDL) receptors, leading to elevated LDL levels in the blood and exacerbating AS [31]. Given SF3B3’s multifaceted role in AS, its potential as a therapeutic target is under active exploration. Drugs targeting SF3B3 or its associated splicing mechanisms may correct abnormal gene splicing, improve cell function and inflammatory responses, and thereby slow or reverse the progression of AS [32]. Our research underscores the pivotal role of SF3B3, alongside other differentially expressed genes, within the pathophysiological landscape of AS. Data from the GSE9820 study highlight the prognostic utility of SF3B3-associated traits. However, exploration into the genomic alterations related to SF3B3 remains in its early stages, revealing a significant gap in our understanding. Addressing this gap is crucial for unraveling the intricate genomic interactions involving SF3B3, which could revolutionize therapeutic approaches for AS. This focused exploration holds the potential to advance our comprehension of AS pathogenesis and foster the development of innovative, targeted treatments.

AS epitomizes chronic inflammatory diseases intimately linked with immune mechanisms, where the gradual development of arterial intimal plaques heralds cardiovascular diseases. This pathophysiological progression begins with early inflammation, advancing through the complex evolution and eventual rupture of atherosclerotic plaques [33]. A myriad of immune cells, including monocytes/macrophages, T lymphocytes, dendritic cells, and mast cells, play critical roles in the development of AS through dynamic and intricate interactions. In the arterial environment, monocytes differentiate into macrophages, initiating lipid uptake and foam cell formation—hallmarks of atherosclerotic lesions [34]. CD4 + T cells amplify the inflammatory milieu, contributing to the proliferation of pro-inflammatory cytokines. Dendritic cells, positioned at the interface of innate and adaptive immunity, regulate the immune response within atherosclerotic lesions [34]. Mast cells exacerbate this process by releasing inflammatory mediators, intensifying inflammation and promoting plaque instability. The interplay between immune and vascular cells in AS is orchestrated by a complex network of cytokines, chemokines, and inflammatory mediators. Crucial pro-inflammatory cytokines such as IL-1β, IL-6, and TNF-α play essential roles in endothelial activation and plaque progression [35]. Concurrently, chemokines direct the migration of immune cells to sites of inflammation, augmenting the inflammatory response. Toll-like receptors (TLRs) and NOD-like receptors (NLRs), recognizing pathogen-associated molecular patterns (PAMPs) and damage-associated molecular patterns (DAMPs), initiate the inflammatory cascade characteristic of AS [36]. Deciphering these intricate immunological interactions has significantly advanced our understanding of AS, uncovering novel therapeutic avenues. Immunomodulatory strategies, targeting specific immune cells or inflammatory pathways, offer groundbreaking approaches to preventing and stabilizing atherosclerotic plaques, potentially reducing cardiovascular events [36]. In our research, we observed that certain immune cells, including memory B cells, naive B cells, resting memory CD4 + T cells, naive CD4 + T cells, CD8 + T cells, follicular helper T cells, and regulatory T cells (Tregs), were positively associated with SF3B3. Conversely, activated memory CD4 + T cells, monocytes, M1 macrophages, M0 macrophages, eosinophils, and activated dendritic cells were negatively associated with SF3B3. By examining the immunological intricacies of AS, our research establishes a foundation for innovative treatments aimed at significantly improving clinical outcomes for this challenging condition. These findings underscore the potential for targeted immunotherapies to modulate specific aspects of the immune response, thereby offering new hope in the fight against AS and its associated cardiovascular risks. The complex interplay between AS and metabolic dynamics represents a rapidly advancing frontier in biomedical research. Recent bioinformatics approaches have begun to dissect these multifaceted relationships [37–39]. Notably, Zemin Tian demonstrated that immunogenic cell death in endothelial cells contributes to the chronic inflammatory milieu characteristic of AS. Complementing this, Shuangyang Mo highlighted molecular and immunological parallels between non-alcoholic fatty liver disease and AS, identifying PLCXD3, CCL19, and PKD2 as key biomarkers bridging these conditions. Concurrently, Chi Ma developed an autophagy-related gene model, revealing 19 significant genes that advance diagnostic and therapeutic exploration in AS. The present study identifies SF3B3 as a potential splicing-related regulator in AS, highlighting its novel association with immune landscape remodeling. By integrating multi-cohort microarray datasets and applying multi-algorithm feature selection, we demonstrate that SF3B3 expression correlates with both adaptive and innate immune cell populations, suggesting its involvement in modulating pro- and anti-inflammatory responses within the vascular microenvironment. These findings provide a new perspective on the transcriptional regulation underlying AS, complementing prior work on immunogenic cell death, autophagy-related gene networks, and metabolic reprogramming in endothelial and immune cells. Mechanistically, SF3B3 may influence AS progression through alternative splicing of immune- or metabolism-related transcripts, thereby affecting cytokine production, immune cell differentiation, and inflammatory signaling cascades. Such a role aligns with the broader functions of spliceosome components in regulating gene expression programs across diverse pathological contexts. We acknowledge several limitations inherent to this study. First, the analysis relies on publicly available microarray datasets, which may introduce platform-specific biases and constrain the generalizability of our findings. Second, sex- and cell-type-specific metadata are incomplete, limiting the ability to fully resolve confounding variables. Third, experimental validation, including in vitro or in vivo functional assays, is lacking; however, the exploratory bioinformatics focus of this study is intended to generate hypotheses and prioritize candidate biomarkers, rather than establish causality. Future studies leveraging next-generation sequencing, single-cell resolution, and mechanistic experiments will be essential to corroborate SF3B3’s functional role and to delineate its impact on immune modulation in AS. Collectively, this work underscores SF3B3 as a novel candidate regulator linking RNA splicing to immune remodeling in AS, providing a foundation for future mechanistic and therapeutic investigations aimed at improving disease understanding and intervention strategies.

Conclusions

This investigation delves deeply into the intricate mechanisms underlying disease pathogenesis, emphasizing the pivotal role of SF3B3 in tissue repair and inflammation and its substantial implications for the prognosis of AS. Utilizing advanced predictive modeling techniques, we meticulously map the transcriptional landscape of SF3B3, uncovering significant expression differences between AS-affected tissues and their healthy counterparts.

Abbreviations

AS

Coronary atherosclerosis

GO

Gene ontology

TCM

Traditional chinese medicine

MF

Molecular functions

KEGG

Kyoto encyclopedia of genes and genomes

GEO

Gene expression omnibus

SF3B3

Splicing factor 3B subunit 3

BP

Biological processes

CC

Cellular components

DEGs

Differentially expressed genes

Author contributions

Nan Yang, Jifang Liu and Qianqian Zhang drafted and revised the manuscript. Nan Yang and Dongli Jiang were in charge of data collection. Nan Yang and Man Zheng were in charge of design of frame. Zongtao Wang and Zhongxiu Guo conceived and designed this article, in charge of syntax modification and revised of the manuscript. All the authors have read and agreed to the final version manuscript.

Funding

No Funding.

Data availability

The datasets generated during and/or analyzed during the current study are available in the appendix.

Declarations

Ethics approval and consent to participation

This manuscript is not a clinical trial, hence the ethics approval and consent to participation are not applicable.

Consent for publication

All authors have read and approved this manuscript to be considered for publication.

Competing interests

The authors declare no competing interests.

Acknowledgement

We thank Zhenquan Fu for his technical support in data analysis and figure generation.