Identification of DEGs
Gene expression levels of merged GEO series that have been adjusted batch effects were standardized and the results of pre- and post- standardized were presented in Supplementary Figure 1 and 2. The 54675 probes corresponding to 21654 genes in GSE115574, GSE31821, GSE79768, GSE41177 and GSE14975 datasets were identified and DEGs of AF were confirmed. Twenty-seven of DEGs with |log2 FC| ≥ 0.58 in LAA samples of AF patients compared with SR was identified, including 19 up-regulated genes and 8 down-regulated genes (Supplementary Table 1). Volcano plot and Heatmap plot of 27 DEGs enrolled in subsequent analyses was showed in Figure 1 and Supplementary Figure 3. Using a screening criteria of |log2 FC| ≥ 1, there were 5 genes identified, with 4 of these genes being up-regulated and 1 down-regulated (Table 2). Boxplots for the 5 selected genes were shown in Figure 2.
Functional enrichment analyses of DEGs
To further investigate the biological functions of the 27 DEGs, functional enrichment analyses were performed and results were shown in Table 3. The result of molecular function in GO revealed that two up-regulated DEGs (IGFBP2 and IGFBP3) were enriched in insulin-like growth factor I binding process (adjusted P value = 0.0168 and Q value = 0.0120) and insulin-like growth factor binding process (adjusted P value = 0.0432 and Q value = 0.0310). Using screening criteria of adjusted P value < 0.05 and Q value < 0.05, no pathway was enriched in KEGG. The ‘mineral absorption’, ‘calcium signaling pathway’ and ‘proximal tubule bicarbonate reclamation’ pathways were enriched (P values = 0.0033, 0.0331 and 0.0343 respectively). Specifically, SLC26A9 and ATP1B4 genes were enriched in mineral absorption, while two up-regulated genes (CXCR4 and HTR2B) were correlated with calcium signaling pathway. Only up-regulated ATP1B4 gene was enriched in proximal tubule bicarbonate reclamation pathway. However, these enrichments did not remain significant after multiplicity adjustment by BH.
Pathway enrichment using the REACTOME database identified that three up-regulated DEGs (IGFBP2, IGFBP3 and CHGB) were enriched in Regulation of Insulin-like Growth Factor (IGF) transport and uptake by Insulin-like Growth Factor Binding Proteins (IGFBPs) (P value = 0.0008 and Q value < 0.0435, Supplementary Figure 4). DO enrichment analysis revealed that DEGs were enriched in 29 biological processes (adjusted P value < 0.05 and Q value < 0.05), but not in AF associated process (Supplementary Figure 5). The result of GSEA analysis showed that no gene was enriched under the specific cutoff of P value.
PPI network construction and potential crucial genes selection
Using the STRING platform, PPI analysis of these DEGs identified 18 nodes and 26 interactions. In addition, one significant module with 5 nodes and 9 edges was screened out via MCODE (Supplementary Figure 6 and 7). CXCR4, IGFBP2, IGFBP3, SNAI2 and ANGPTL2 were hub nodes in module. Only CXCR4, IGFBP2 and IGFBP3 were selected for hub genes, all of which were involved in playing pivotal regulatory roles in PPI network, due to the high degree of connectivity (degree ≥ 5, Supplementary Figure 7). Furthermore, after combining with the results of differential expression, enrichment analyses and PPI, IGFBP2, IGFBP3, CHGB, CXCR4, HTR2B, FHL2, C1orf105, ATP1B4 and SLC26A9 were considered potential crucial genes for further analyses.
Identification of functional and pathway enrichment among predicted microRNAs and potential crucial genes
Bioinformatic prediction tools including mirDIP, miRDB, TargetScan, and DIANA were used to identify microRNAs targeting potential crucial genes involved in AF and these data were displayed in Table 4. There were 16 microRNAs met the inclusion criteria, of which 5 targeting IGFBP3 and 8 FHL2.
In order to understand how predicted microRNAs are related to AF, functional and pathway enrichment analyses were performed using Diana-miRPath. GO analysis revealed hsa-miR-197-3p, hsa-miR-19a-3p, hsa-miR-19b-3p, hsa-miR-340-5p and hsa-miR-9-5p targeting IGFBP3 were mainly enriched in insulin-like growth factor binding protein complex, protein tyrosine phosphatase activator activity, to mention a few. Hsa-miR-25-3p, hsa-miR-32-5p, hsa-miR-363-3p, hsa-miR-367-3p, hsa-miR-4325, hsa-miR-661, hsa-miR-92a-3p and hsa-miR-92b-3p targeting FHL2 were mainly involved in atrial cardiac muscle cell development, ventricular cardiac muscle cell development and heart trabecula formation, among others. KEGG analysis result revealed that microRNAs targeting IGFBP3 were enriched in p53 signaling pathway and transcriptional misregulation in cancer, while microRNAs targeting FHL2 were involved in osteoclast differentiation.
Identification of potential crucial genes associated with AF
CTD database was employed to explore the interaction between potential crucial genes and AF. As shown in Supplementary Figure 8, potential crucial genes targeting AF, left ventricular dysfunction, heart diseases and cardiovascular diseases. Inference scores in CTD reflected the association between chemical, disease and genes. The interaction results showed that CXCR4, IGFBP2, IGFBP3 and FHL2 have a higher score with AF.