Integration of transcriptome and mendelian randomization analyses to explore the molecular mechanisms of atrial fibrillation

Abstract

Background

Atrial fibrillation (AF) is a prevalent arrhythmia associated with significant adverse outcomes and elevated mortality rates, though its underlying molecular mechanisms remain poorly understood. This study sought to identify key genes linked to AF through the integration of transcriptome analysis and Mendelian randomization (MR), supplemented by bioinformatics.

Methods

Differentially expressed genes (DEGs) were identified by comparing patients with AF to those without the condition. MR analysis was used to assess causal relationships between key genes and expression quantitative trait loci-associated outcomes. Single-cell sequencing, combined with Area Under the Curve for Single Cell analysis, was used to assess the activity of immune and metabolic pathways. This was further complemented by immunoinfiltration analysis.

Results

Among the 503 DEGs identified—312 upregulated and 191 downregulated—MR analysis highlighted ST8SIA4 and SLPI as potential protective genes in AF pathogenesis. ST8SIA4 was predominantly enriched in immune and inflammatory pathways and demonstrated associations with γδ T cells, M2 macrophages, and CD8⁺T cells. Conversely, SLPI was implicated in coagulation pathways, with expression observed in endothelial cells, macrophages, and T cells, and exhibited a negative correlation with eosinophil levels.

Conclusions

ST8SIA4 and SLPI are associated with a reduced AF risk highlighting their potential involvement in the pathogenesis of the disease and their promise as therapeutic targets.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13019-026-03845-z.

Introduction

Atrial fibrillation (AF) is one of the most prevalent types of cardiac arrhythmia [1–3], characterized by high mortality and disability rates [4–6]. Despite recent advances in interventional therapeutic techniques, their long-term efficacy in disease management remains suboptimal. Current antiarrhythmic drugs are often limited by suboptimal efficacy, poor tolerability, and significant adverse effects [7, 8]. Notably, emerging research has elucidated the critical role of immune-coagulation crosstalk in AF pathogenesis: inflammatory cytokines promote atrial fibrosis by activating the coagulation cascade, while thrombin-mediated immune cell infiltration establishes a vicious cycle [9, 10].

At the genetic level, genome-wide association studies (GWAS) have demonstrated that approximately 22% of AF risk originates from common genetic variants [11]. To date, over 100 risk loci have been identified, the majority residing within non-coding regions of the genome [12, 13]. This suggests that these variants likely influence disease development primarily through the regulation of transcriptional activity.

Given the limited current therapeutic options and persistently high AF mortality, this study aims to systematically identify differentially expressed genes (DEGs) and their regulatory networks within the atrial tissue-specific expression profile by implementing an integrated multi-omics strategy. Methodologically, this involves combining Mendelian randomization (MR) analysis, expression quantitative trait loci (eQTL) analysis, and gene set enrichment analysis (GSEA/GSVA) techniques. The ultimate objective of this research is to provide a new scientific basis for personalized anti-coagulation and anti-fibrotic therapeutic strategies.

Materials and methods

Data acquisition

The Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/info/datasets.html), a resource within the National Center for Biotechnology Information (NCBI), provides extensive datasets on gene expression. For this study, the GSE79768 series matrix file was retrieved from the GEO database, annotated using the GPL570 platform. This dataset included expression profiles from a total of 13 samples, comprising of 6 control group samples and 7 disease group samples. Additionally, single-cell expression data from the GSE224995 dataset were obtained, encompassing 12 samples with complete single-cell expression profiles for single-cell analysis.

Exposure Data: eQTL data were sourced from the eQTLGen Consortium database (https://www.eqtlgen.org), which is dedicated to exploring gene expression patterns in blood and elucidating the genetic basis of complex traits [14]. The eQTLGen project, currently in its second phase, focuses on conducting large-scale genome-wide meta-analyses of blood-derived samples.

Outcome Data: The outcome-related GWAS included in this study were predominantly conducted among participants of European ancestry. Aggregated outcome data were obtained from the EBI database (EBI-A-GCST90038689), which provides access to publications, top associations, and complete summary statistics [15]. The GWAS Catalog [16] data are mapped to the Genome Assembly and the database of Single Nucleotide Polymorphisms (dbSNP) Build. This study analyzed a total of 481,061 individuals, including 3,537 diagnosed with AF.

Differential expression analysis

The limma package [17], an R software tool, is widely used for differential expression analysis to identify genes that exhibit significant differences in expression between comparative groups. In this study, the limma package was used to analyze molecular differences within the dataset and to identify DEGs between control and disease samples. DEGs were screened based on the criteria of a p-value < 0.05 and an absolute log fold change (|logFC|) > 0.585. Additionally, a volcano plot and heatmap were generated.

GO and KEGG functional profiling

The R package ClusterProfiler [18] was used for the functional annotation of DEGs to systematically examine their functional associations. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses were conducted to categorize the relevant functional domains. Enrichment pathways in GO and KEGG with both p-values and q-values less than 0.05 were considered statistically significant.

Mendelian randomization analysis

The EBI database (https://www.ebi.ac.uk/gwas/) contains data from hundreds of GWAS of many summary statistics. By filtering the EBI database to the end of the ID in the GWAS summary data (https://gwas.mrcieu.ac.uk/) to extract the relevant causal relationships in eQTLs, single nucleotide polymorphisms (SNPs) related to the significance threshold of each gene in the whole locus (p < 10^–8) were selected as potential instrumental variables (IVs). Linkage disequilibrium (LD) between SNPs was calculated, and only SNPs with p < 5 × 10^–8 were retained among SNPs with R² < 0.001 (clumping window size = 10,000 kb).

Inverse variance-weighted (IVW) meta-analysis was used to combine Wald estimates for each SNP, along with three additional statistical methods—MR Egger (which assumes that instrument strength is independent of direct effects, per the InSIDE assumption), weighted median (allowing robust causal estimation when up to 50% of IVs are invalid), and weighted mode (providing greater sensitivity for detecting causal effects compared to MR-Egger regression, with reduced bias and lower type I error rates). When only a single SNP met the criteria for causal analysis, the Wald ratio method was applied.

These methods collectively assessed the reliability of causal relationships, providing an overall estimate of the effect of cis-expression and certain trans-regional gene expressions in whole blood from patients with AF. Finally, the residual method was used to validate the causality of the selected results.

We conducted colocalization analysis using the coloc method along with eQTL summary data and atrial fibrillation GWAS. A 100-kilobase region around the index SNP was used to calculate the posterior probability. In the coloc results, H3 represents the posterior probability that the two traits (gene expression and atrial fibrillation) are related but have different causal variants; H4 represents the posterior probability that the two traits are associated and share a single causal variant. A SNP.PP.H4 > 0.75 was used as the colocalization threshold.

GSEA enrichment analysis

GSEA [18] was conducted to further investigate differences in signaling pathways between high- and low-expression groups. The background gene set used for this analysis was the annotated gene set from version 7.0 of the MsigDB database, serving as the reference for subtype pathway annotation. Differential expression analysis was conducted on pathway-related genes across the subtypes. Gene sets with significant enrichment, defined by an adjusted p-value of less than 0.05, were ranked based on their consistency scores. GSEA is frequently used to examine the associations between disease phenotypes and their underlying biological significance.

GSVA

GSVA [19] is a nonparametric, unsupervised approach used to assess gene set enrichment at the transcriptomic level. GSVA assesses the biological functions of samples by comprehensively scoring gene sets of interest, translating gene-level alterations into pathway-level changes. In this study, gene sets were obtained from the Molecular Signatures Database (MsigDB), and the GSVA algorithm was applied to assign comprehensive scores to each gene set. The goal of this analysis was to assess potential biological functional changes across different samples.

Single-cell sequencing analysis

The expression profiles were processed using the Seurat package, and genes with low expression levels were filtered out. The data were standardized and normalized for uniformity. Principal component analysis and t-distributed Stochastic Neighbor Embedding (t-SNE) analyses were then conducted to reduce dimensionality. The optimal number of principal components was determined using an elbow plot, while t-SNE analysis was used to depict the spatial relationship between clusters. Cell cluster annotation was conducted using the "celldex" package [20], focusing on cells significantly associated with disease pathogenesis. Finally, marker genes for each cell subtype were identified from the single-cell expression profiles by setting the FindAllMarkers function to a log-fold change threshold of 1.

Transcriptional regulation analysis of key genes

In this study, transcription factors were predicted using the R package "RcisTarget." [21] The calculations conducted by RcisTarget were based on motif analysis. The normalized enrichment score for each motif was derived from the total number of motifs in the database. Beyond the motifs annotated in the source data, additional annotations were inferred based on motif similarity and gene sequences. To estimate the overrepresentation of each motif within the gene set, the AUC was calculated for each motif-gene set pair by constructing recovery curves that ranked motifs according to the gene set. The NES for each motif was subsequently determined using the AUC distribution of all motifs within the gene set.

Immune cell infiltration analysis

The CIBERSORT [22] method, widely used for the evaluation of immune cell composition in the microenvironment, operates on the principle of support vector regression. This method identifies immune cell subtypes using 547 biomarkers that distinguish 22 human immune cell phenotypes, including T cells, B cells, plasma cells, and various myeloid subsets, through deconvolution analysis of the expression matrix. In this study, the CIBERSORT algorithm was applied to patient data to estimate the relative proportions of these 22 infiltrating immune cell types. Additionally, Pearson correlation analysis was conducted to assess the relationship between gene expression levels and immune cell content.

Statistical analysis

Reliable MR analysis is contingent upon three key assumptions: (1) the relevance assumption, which posits that instrumental variables must be strongly associated with the exposure of interest but not directly linked to the outcome; (2) the independence assumption, which requires that instrumental variables are not correlated with confounding factors; and (3) the exclusion-restriction assumption, which stipulates that instrumental variables can influence the outcome only through the exposure of interest. Any influence of instrumental variables on the outcome through alternative pathways indicates the presence of horizontal pleiotropy. In this study, the MR analysis was conducted using R software (version 4.2.2). All statistical tests were two-sided, and a p-value of less than 0.05 was considered statistically significant.

Results

Differential expression analysis

The AF dataset GSE79768 was obtained from the GEO database, containing expression profile data from 13 patients, with 6 patients in the control group and 7 in the disease group. DEGs between the control and disease groups were identified using the limma package, applying the thresholds of p-value < 0.05 and |logFC|> 0.585. A total of 503 DEGs were identified, comprising 312 upregulated and 191 downregulated genes (Supplementary Table S1–S3, Fig. 1A and B).

Fig. 1.

Differentially expressed genes identified by AF screening and functional correlation analysis. (A) Volcano plot of GSE79768. Red signifies upregulated genes, blue signifies downregulated genes, and black signifies no genes. (B) Heatmap of the GSE79768 dataset. Blue represents the normal group, and red represents the AF group. (C) Gene Ontology (GO) biological function enrichment evaluation of the DEGs. The X-axis represents the enrichment of DEGs in GO entries. (D) KEGG pathway enrichment analysis of the DEGs. The color of the bar represents the adjusted P value. The redder the color is, the lower the adjusted P value, and the bluer the color is, the higher the adjusted P value. The length of the bar serves as a proxy for the number of enriched genes

Pathway analysis of these DEGs was subsequently conducted. GO enrichment analysis revealed significant enrichment in pathways such as positive regulation of response to external stimuli, positive regulation of inflammatory responses, and positive regulation of defense responses (Fig. 1C). KEGG enrichment analysis demonstrated that the DEGs were predominantly associated with cytokine − cytokine receptor interaction and the renin − angiotensin system pathways (Fig. 1D).

Mendelian randomization analysis

To identify key genes influencing AF, the DEGs identified earlier were analyzed in the context of data from 484,598 patients, comprising of 3,537 controls and 481,061 cases with conditions related to AF. The outcome data were obtained using the summary statistics corresponding to the ID ebi-a-GCST90038689. A total of 286 causal relationships between genes and outcomes were extracted (Supplementary Table S4).

MR analysis and screening was conducted, identifying two significant causal relationships between genes and eQTL-positive outcomes The genes identified were ST8 Alpha-N-Acetyl-Neuraminide Alpha-2,8-Sialyltransferase 4 (ST8SIA4) and Secretory Leukocyte Protease Inhibitor (SLPI). The causal estimates revealed that ST8SIA4 (odds ratio: 0.9991; 95% confidence interval: 0.9984–0.9999; p-value = 0.039) and SLPI (odds ratio: 0.9990; 95% confidence interval: 0.9982–0.9999; p-value = 0.041) were associated with a lower risk of AF.The MR-Egger intercept term showed non-significant values (ST8SIA4: p-value = 0.106; SLPI: p-value = 0.131; both > 0.05 threshold), suggesting absence of detectable horizontal pleiotropy (Supplementary Table S5, Fig. 2A–F).MR Egger and inverse variance weighted analyses showed no evidence of heterogeneity for ST8SIA4 and SLPI (Supplementary Table S6).

Fig. 2.

The relationships between exposure variables (ST8SIA4 and SLPI) and the outcome variable (AF). (A) A scatter plot illustrating the trend of the effect of ST8SIA4 on AF in the presence of SNPs. (B) B displays a scatter plot illustrating the trend of the effect of SLPI on AF in the presence of SNPs (IVW p-value < 0.05)

Sensitivity analyses using the leave-one-out method were conducted to assess the robustness of these causal relationships. The results indicated that excluding any single SNP did not significantly affect the overall error line, confirming the reliability of the identified causal associations (Fig. 3A–B). These findings indicate that ST8SIA4 and SLPI are critical genes for subsequent research.

Fig. 3.

Sensitivity analysis of the causal relationships between two pairs of genes to determine their reliability using the leave-one-out method. (A) ST8SIA4; (B) SLPI

We conducted a coloc co-localization analysis of SLPI and ST8SIA4 at the eQTL-GWAS level, and the PP.H4 of the co-localized SNPs was higher than 0.75 (Fig. 4).

Fig. 4.

Colocalization posterior probabilities (PP.H4 > 0.75) for causal variant sharing between heart tissue eQTLs and AF GWAS associations. (A) ST8SIA4; (B) SLPI

GSEA pathway enrichment analysis

Subsequently, we explored the specific signaling pathways associated with the two key genes to elucidate their potential molecular mechanisms in the progression of AF. GSEA revealed distinct pathways enriched for each gene.

For ST8SIA4, the enriched pathways included nitrogen metabolism, the PPAR signaling pathway, spliceosomes, and additional pathways (Fig. 5A and B). For SLPI, the enriched pathways included the cGMP − PKG signaling pathway, the phagosome pathway, the Rap1 signaling pathway, and other pathways (Fig. 5C and D).

Fig. 5.

GSEA and GSVA of high and low expression of ST8SIA4 and SLPI. (A-B) GSEA of ST8SIA4; (C-D) GSEA of SLPI; (E) GSVA of ST8SIA; (F) GSVA of SLPI

Enrichment analysis of the GSVA pathways

GSVA results revealed that ST8SIA4 was associated with genes involved in the Notch, PI3K/AKT, Kras, and other pathways (Fig. 5E). Similarly, SLPI was linked to pathways such as IL6/JAK/STAT3, Wnt, Myc, and additional signaling cascades (Fig. 5F). These findings indicate that these key genes may influence AF progression through their involvement in these critical signaling pathways.

Single-cell sequencing analysis

Single-cell analysis was conducted using the Seurat package and tSNE algorithm on the GSE224995 dataset, resulting in the identification of 16 distinct subtypes (Fig. 6A). Each subtype was annotated using the SingleR package, which grouped the 16 clusters into five major cell categories: endothelial cells, smooth muscle cells, macrophages, tissue stem cells, and T cells (Fig. 6B). The expression patterns of the key genes across these five cell types were visualized (Fig. 6C and D). Furthermore, area under the curve for single cell (AUCell) functions were used to quantify immune and metabolic pathway activity within the single-cell dataset. Differences in the activity of key genes and their associated immune and metabolic pathways were detected (Fig. 6E).

Fig. 6.

scRNA-seq analysis of GSE224995. (A) The t-SNE plot revealed that all cells were classified into 16 distinct clusters. (B) The t-SNE map was used to identify 5 different cell types. (C-D) The distribution of hub genes in the five types of cells. The deeper the blue color is, the greater the expression. The larger the circle is, the more significant the difference. (E) Bubble map showing the differences in the activity of key genes and immune and metabolic pathways. The redder the color is, the stronger the positive correlation, and the deeper the blue color is, the stronger the negative correlation. The larger the circle is, the more significant the difference

Transcriptional regulation analysis of key genes

The two key genes identified in this study were applied to the gene set for further analysis, revealing that they are regulated by shared mechanisms, including multiple transcription factors. To investigate this, enrichment analysis of the transcription factors was conducted using cumulative recovery curves. Motif-TF annotation and selection analysis of significant genes identified the motif with the highest normalized enrichment score (NES: 7.89) as cisbp M3036. A comprehensive analysis of all enriched motifs and their corresponding transcription factors for the key genes was conducted (Fig. 7A and B).

Fig. 7.

A-B: All enriched motifs and corresponding transcription factors of key genes

Immunoinfiltration analysis

The microenvironment primarily consists of immune cells, the extracellular matrix, various growth factors, inflammatory factors, and distinct physicochemical properties. These components collectively exert a substantial influence on disease diagnosis, survival outcomes, and sensitivity to clinical treatments. To clarify the relationship between key genes and immune infiltration in the AF dataset, and to further examine the potential molecular mechanisms through which these genes affect AF progression, the proportion of immune cells in each patient and the correlations between different immune cell types were examined (Fig. 8A and B). Statistically significant differences were observed in the numbers of resting mast cells and neutrophils between the groups (Fig. 8C).

Fig. 8.

Immune cell infiltration analysis and correlation analysis of the hub genes with immune cells. (A) Percentage of immune cells between the normal group and the AF group. Green: the normal control group; purple: the AF group. (B) Interaction analysis among 22 different immune cells in AF patients (* represents p-value < 0.05, ** represents p-value < 0.01 and *** represents p-value < 0.001). (C) Comparisons of immune cells between the normal control group and the AF group (* * represents p-value < 0.05, ** represents p-value < 0.01 between the two groups). (D) The correlations between the two hub genes (ST8SIA and SLPI) and infiltrating immune cells. * represents p-value < 0.05; ** represents p-value < 0.01. The redder the color is, the stronger the positive correlation is; the deeper the blue color is, the stronger the negative correlation is

Additionally, analyses of the associations between key genes and specific immune cell populations revealed that ST8SIA4 was positively correlated with γδ T cells and negatively correlated with M2 macrophages and CD8 + T cells. In contrast, SLPI demonstrated a significant negative correlation with eosinophil counts (Fig. 8D).

Discussion

AF is a prevalent arrhythmia, and epidemiological data indicate a progressive increase in its incidence worldwide. The risk of AF rises with advancing age and a growing burden of chronic diseases [1–3]. Patients with AF face a higher likelihood of stroke and myocardial infarction, underscoring the importance of careful prognostic management [4–6]. However, due to the complexity and heterogeneity of AF pathogenesis and progression, current pharmacological and procedural interventions are insufficient for addressing disease-specific mechanisms. Genetic factors have been implicated in AF onset [23, 24]. Identification of key genes can enhance understanding of drug responses, disease progression, and potential genetic risk factors. Such insights may facilitate the development of personalized treatment strategies and improve treatment outcomes and prognosis. A deeper understanding of AF-related genetic variants may also provide new therapeutic targets.

In this study, a total of 503 DEGs were identified by comparing gene expression profiles between AF and normal samples. Among these DEGs, 312 were upregulated and 191 were downregulated. Pathway enrichment analysis revealed that these DEGs were primarily associated with immune response, energy metabolism, inflammation, apoptosis, and coagulation. These genetic changes ultimately contribute to both electrical and structural remodeling in the atrial tissue. Inflammation, may contribute to AF development through electrical remodeling, in which changes in ion handling disrupt the cardiac action potential [25–28]. Fibrosis, another inflammatory consequence, can create heterogeneous atrial depolarization that facilitates electrical reentry and subsequent AF [25, 29]. The immune system is also involved in AF pathophysiology, as immune remodeling has been proposed as a contributing factor [30]. Immune remodeling in AF encompasses the recruitment and activation of immune cells, as well as changes in immune molecular components, ultimately shaping a distinct atrial and systemic immune environment [31].

MR analysis identified two pairs of genes corresponding to eQTL-positive outcomes, specifically ST8SIA4 and SLPI. Both genes appear to be involved in the pathogenesis of AF, and their presence may be associated with reduced disease risks. Subsequent GSEA and GSVA revealed that ST8SIA4 was primarily enriched in the Notch, PI3K/AKT, Kras, and additional signaling pathways, while SLPI was enriched predominantly in IL6/JAK/STAT3, Wnt, Myc, and related pathways. Previous studies have demonstrated that these enriched pathways can, either directly or indirectly, increase susceptibility to AF [32–37].

The α−2,8-sialyltransferase facilitates the transfer of sialic acid from CMP-Sia [38]. Changes in the expression of specific α−2,8-sialyltransferases have been documented [39]. In knockout models of ST8SIA4, post-translational modification of NCAM-1 is inhibited, leading to aberrant expansion and disruption of protein transport between connexin membranes and extracellular fluid, ultimately resulting in functional defects [40]. To date, no studies have established an association between ST8SIA4 and AF. This report identifies the downregulation of ST8SIA4 as a contributing factor in the pathophysiological mechanisms underlying AF.

SLPI, an inflammatory serine protease inhibitor, plays a key role in maintaining the protease/antiprotease balance. Produced by host defense effector cells during inflammation, SLPI prevents tissue damage induced by proteolytic enzymes [41]. It is commonly found in mucosal tissues, monocytes, and neutrophils and is expressed in epithelial cells during injury, where it mitigates excessive tissue destruction along with delayed wound healing [42, 43].

Single-cell sequencing analysis was conducted, and immune and metabolic pathways were quantitatively assessed using AUCell. ST8SIA4 is overexpressed and highly enriched in macrophages, T cells and other immunoinflammatory response pathways. As early as 2008, studies reported a significant increase in atrial CD45⁺ cell infiltration [26, 44]. Cardio-resident macrophages, originating from bone marrow lineage cells, represent a major immune cell population within the heart, contributing to immune regulation and the maintenance of cardiac homeostasis [45].

AF induces the polarization of proinflammatory macrophages, which secrete various inflammatory mediators and modulate the expression of proteins associated with AF [46, 47]. Lymphocytes within the heart also participate in the immune response, expressing T-cell markers such as CD4 and CD8 [48, 49]. CD4⁺ T cells differentiate into specific subsets of helper T (Th) cells that promote inflammation and cardiac fibrosis, contributing to AF pathogenesis. The infiltration of CD8⁺ T cells in the left atrium was significantly elevated [50].

SLPI is expressed at low levels in endothelial cells, smooth muscle cells, macrophages, tissue stem cells, and T cells. Its primary involvement is observed in the coagulation and epithelial-mesenchymal transition pathways. SLPI regulates the production of TGF-β and IL-10 while inhibiting the production and activity of monocyte matrix metalloproteinases (MMPs) [51–53]. Additionally, SLPI exhibits significant anti-inflammatory properties by interfering with signal transduction pathways leading to MMP production, thereby playing a regulatory role in the pathogenesis and progression of AF [51, 54]. Studies have also demonstrated that FXIIIa directly targets SLPI. [51].

Analysis of key genes and immune cell infiltration in AF revealed a potential molecular mechanism whereby ST8SIA4 is positively correlated with γδ T cells and T cells, while being negatively correlated with M2 macrophages and CD8 + T cells. SLPI, in contrast, showed a significant negative correlation with eosinophils. M2 macrophages, activated by cytokines such as IL-4, IL-13, and IL-18, induce the upregulation of IL-10 and TGF-β1, promoting the transformation of fibroblasts into myofibroblasts [55, 56].

In early-stage AF, the activation of cytotoxic CD8⁺ T cells is a hallmark feature in newly diagnosed patients. This activation is closely associated with myocardial fibrosis and atrial dysfunction, suggesting that CD8⁺ T cells may represent a key cellular subtype driving AF progression [50]. Moreover, the TF-FXa-FIIa axis has been implicated in thrombotic inflammation via PAR1 signaling in CD8 + T cells, with PAR1-mediated CD8 + T cell activation further contributing to disease progression [57]. The fraction of γδ T cells has also been significantly increased [58, 59]. However, studies investigating the relationship between γδ T cells and AF development remain limited, warranting further experimental exploration.

A case–control study demonstrated that patients with paroxysmal solitary AF exhibited higher eosinophil and neutrophil counts compared to controls, with left atrial diameter and eosinophil counts identified as independent associations [60]. These findings collectively indicate that ST8SIA4 and SLPI are associated with a reduced risk of AF, consistent with results from MR analyses. Consequently, ST8SIA4 and SLPI hold promise as potential diagnostic biomarkers and therapeutic targets for AF.

This study has several limitations. First, the findings are based on secondary analysis of previously published datasets, and the robustness of these results requires further validation through laboratory experiments and clinical trials. Second, the data were derived exclusively from European populations, which may limit the generalizability of the findings to populations of other ethnicities or populations.

Conclusions

This study presents a comprehensive analysis of AF, identifying critical genes and pathways through advanced bioinformatics and statistical approaches. It highlights the significance of immune cells and genetic factors, with a particular focus on ST8SIA4 and SLPI. The findings offer valuable insights into the intricate molecular mechanisms underlying AF and propose potential avenues for novel therapeutic strategies.

limitations

These findings require further validation, particularly given inherent limitations in data selection and analytical methodologies. Our primary analysis employed blood-derived eQTL data from the eQTLGen Consortium. Although tissue-specific eQTL data (e.g., GTEx atrial appendage) holds significant potential for elucidating AF-related gene regulatory mechanisms and strengthening the biological relevance of Mendelian randomization inferences, currently available cardiac tissue resources remain limited (as discussed). We therefore emphasize that future research must prioritize developing larger-scale, higher-resolution cardiac tissue-specific eQTL datasets—particularly for atrial tissue. Advances through expanded GTEx initiatives, dedicated cardiac eQTL projects, or single-cell sequencing applications will enable more robust assessment of causal relationships between atrial tissue-specific gene expression and AF risk, yielding deeper biologically contextualized insights. Ultimately, validation of our results requires more comprehensive cardiac tissue eQTL resources.

Abbreviations

AF

Atrial fibrillation

DEG

Differentially expressed gene

MR

Mendelian randomization

eQTL

Expression quantitative trait loci

AUCell

Area under the curve for single cell

GWAS

Genome-wide association study

GSEA

Gene set enrichment analysis

GSVA

Gene set variation analysis

GEO

Gene expression omnibus

NCBI

National center for biotechnology information

EBI

European bioinformatics institute

dbSNP

Database of single nucleotide polymorphisms

SNP

Single nucleotide polymorphism

GO

Gene ontology

KEGG

Kyoto encyclopedia of genes and genomes

IVs

Instrumental variables

LD

Linkage disequilibrium

IVW

Inverse variance weighted

ST8SIA4

ST8 Alpha-N-acetyl-neuraminide alpha-2,8-sialyltransferase

SLPI

Secretory leukocyte protease inhibitor

tSNE

T-distributed stochastic neighbor embedding

PCA

Principal component analysis

NES

Normalized enrichment score

AUC

Area under the curve

γδ T cells

Gamma delta T cells

Th cells 

Helper T cells

MMP

Matrix metalloproteinase

Author contributions

Conception and design of the research: Mei-juan Zheng, Jiao Wang, Yu-chun Yang, Muhuyati·Wulasihan Acquisition of data: Mei-juan Zheng, Zhen Bao Analysis and interpretation of the data: Mei-juan Zheng, Ruo-nan Wang Statistical analysis: Zhen Bao, Ruo-nan Wang Obtaining financing: Muhuyati·Wulasihan Writing of the manuscript: Mei-juan Zheng, Jiao Wang Critical revision of the manuscript for intellectual content: Yu-chun Yang, Muhuyati·Wulasihan All authors read and approved the final draft.

Funding

Tianshan Talent Program-Science and Technology Innovation Leading Talent Project (no. 2022TSYCLJ0065). National Natural Science Foundation of China, Research on the Mechanism of PDCD4 Regulating Mitophagy via LDHA/Beclin-1/PINK1 to Promote Energy Metabolism Remodeling in Atrial Fibrillation (no. 82560064).The Project for Cultivating Major Scientific Research Achievements of Xinjiang Medical University, Exploring the Mechanism of PDCD4-Modified Mesenchymal Stem Cell-Derived Exosomes in Improving Atrial Remodeling after Acute Myocardial Infarction Based on AKT/PFKFB2-Mediated Macrophage Lactate-Dependent Glycolysis (no.: XYD2024ZC02).

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Code/Software availability

eQTL [14], GWAS Catalog [16], TWOSAMPLEMR [61] (0.5.7), Limma [17] (3.58.1),Clusterprofiler [18] (4.6.2), GSEA [18], GSVA [19] (1.50.0), CIBERSORT [22], RcisTarget [21] (1.19.2), Celldex [20] (1.12.0).

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.