Acta Medica Mediterranea, 2019, 35: 2315 IDENTIFICATION OF POTENTIAL KEY GENES ASSOCIATED WITH CARDIAC FIBROSIS BY RNA SEQUENCING DATA ANALYSIS DANDAN LIU, HAIZHU WANG, XIAO HAN, CAIPING HAN, FENGBO REN Department of Cardiology, Zhoukou Central Hospital, Zhoukou 466000, Henan, China ABSTRACT Introduction: Cardiac fibrosis is central to a broad constellation of cardiovascular diseases with similar pathophysiologic companions, and is associated with cardiac dysfunction, arrhythmogenesis, and adverse outcome. However, the option of effective treatment strategies is limited due to the insufficient understanding of the mechanisms for cardiac fibrosis. Materials and methods: The RNA sequencing data (GSE97358) comprising 84 TGF-β1-stimulated samples and 84 paired unstimulated samples of cultured primary human cardiac fibroblast from GEO database was used to explore crucial genes and pathways involved in cardiac fibrosis. The differentially expressed genes (DEGs) were identified using edgeR package in R. Pro- tein-protein interaction (PPI) network and module analyses were performed and visualized using STRING and Cytoscape. GO (gene ontology) and KEGG (Kyoto Encyclopedia of Genes and Genomes) enrichment analyses were performed by clusterprofiler. The hub genes extracted from PPI were identified by the CytoHubba plug-in and the transcription factor(TF)-hub gene network was further constructed by the iRegulon plug-in. Results: Totally, 647 DEGs were initially screened out in TGF-β1-stimulated primary human cardiac fibroblast. Twenty hub genes (9 up-regulated: S1PR5, F2RL3, GPR68, CXCR5, KISS1, GAL, LPAR5, HTR1D, PLCB4; 11 down-regulated: CXCL1, GPR65, CYSLTR2, EDNRA, CXCL6, F2R, GNG2 F2RL2, SSTR1, TAS2R1, HTR2B) were further identified. Wnt signaling and neu- roactive ligand-receptor interaction signaling pathways enriched were ultimately identified as the key pathways involved in cardiac fibrosis. Seven TFs (RELB, FOS, SREBF2, PURA, TBX21, IRF1 and IRF4) were identified for the TF-hub gene networks. Conclusions: Our results may provide novel insights into the molecular mechanisms and treatments of cardiac fibrosis. How- ever, further molecular biological experiments are required to confirm these findings. Keywords: cardiac fibrosis, differentially expressed genes, bioinformatics analysis, interaction network. DOI: 10.19193/0393-6384_2019_5_360 Received November 30, 2018; Accepted February 20, 2019 Introduction ed in response to some pathophysiologic stimuli, such as pressure and/or volume overload, metabol- Cardiac fibrosis is characterized by the net ac- ic disorder, ischemic insults or aging which may cumulation of extracellular matrix which involves result in interstitial and perivascular fibrosis (3,4). unbalanced collagen turnover and excessive dif- Yet, activated myofibroblasts are the main effector fuse collagen deposition in the interstitial spaces cells in response to these pathophysiologic stimu- of the myocardium(1). As documented by previous li in cardiac fibrosis (1). In this regard, some fibrot- studies, cardiac fibrosis is central to a broad con- ic factors, such as cytokines, chemokines, growth stellation of cardiovascular diseases with similar factors, hormones, and reactive oxygen species, pathophysiologic companions, and is associated are responsible for the activation of fibroblasts and with cardiac dysfunction, arrhythmogenesis, and the alteration of extracellular matrix(5). A growing adverse outcome(2). In many cases, cardiac fibrosis of studies have revealed multiple signaling path- is the result of a reparative process that is activat- ways and biological processes involved in cardiac 2316 Dandan Liu, Haizhu Wang et Al fibrosis, such as transforming growth factor-beta fibroblast from patients receiving coronary artery (TGF-β), Wnt/β-catenin, mitogen-activated protein bypass grafting(16). The RNA sequencing data were kinase (MAPK) signaling, epithelial-mesenchymal normalized and analyzed by the edgeR package transition (EMT), endothelial-mesenchymal transi- in R. Differentially expressed genes (DEGs) were tion (EndMT), inflammation, oxidative stress pro- identified with the cut-off criteria |log2FC| ≥1, cesses, etc(1,6-8). P-value < 0.05 and adjust P-value < 0.05. Although some conventional drugs, such as angiotensin-converting enzyme inhibitors (ACEIs), Functional and pathway enrichment analysis aldosterone antagonists, β-blocker, and statins, have To elucidate potential biological processes, been shown to alleviate cardiac fibrosis in clinical molecular functions and signaling pathways cor- trials(9-13), most of these traditional therapies are not related with the DEGs, the Gene Ontology (GO, directed towards alleviating fibrosis but secondary http://www.geneontology.org) and Kyoto Ency- to the correction of the underlying cardiac dysfunc- clopedia of Genes and Genomes (KEGG, http:// tion mechanisms and do not effectively hamper the www.ge¬nome.ad.jp/kegg/) pathway enrichment progression of cardiac fibrosis (2). Despite the ad- analyses were performed(17,18), which were carried vance in exploring the pathogenesis and treatment, out with clusterProfiler for the up-regulated and the exact mechanisms of fibrosis accounting for the down-regulated genes respectively(19). The enriched cardiac dysfunction and adverse outcome are not GO and KEGG terms were considered significant fully understood. Therefore, careful dissections of with the cut-off criteria of adjust P-value < 0.05 the cell biological mechanisms are of primary im- portance in the development of effective therapies. Protein-protein interaction network con- In recent years, the advancement of microar- struction and module analysis ray and high throughput sequencing technologies In order to interpret the molecular mecha- has provided an efficient tool to decipher critical nisms of key cellular activities in cardiac fibrosis, genetic alternations in cardiac fibrosis and to iden- the online Search Tool for the Retrieval of Interact- tify various key genes, molecular pathways, bio- ing Genes (STRING, http://string-db.org/) database logical processes, and cellular behaviors (14-16). In was used to construct a protein-protein interaction the present study, a large-scale RNA sequencing (PPI) network of the DEGs(20), which was selected data (GSE97358) downloaded from Gene Expres- and visualized with confidence score ≥0.4(medium sion Omnibus (GEO) database were employed to confidence score). According to the degree of im- acquire the differentially expressed genes (DEGs). portance, significant modules of PPI network were We further explored the development of cardiac fi- screened out using the plug-in Molecular Complex brosis by a way of DEGs functional enrichment and Detection (MCODE) with the degree cutoff =2, interaction network analysis. The hub gene-tran- node score cutoff=0.2, k-core=5 and max dept =100 scription factor interaction network was also con- in the Cytoscape (version 3.6). Moreover, the func- structed. The present study aimed to identify crucial tion and pathway enrichment analyses were also genes and pathways involved in cardiac fibrosis by performed for DEGs in the significant modules. using bioinformatics analysis, which may result in a better understanding of the pathological mecha- Construction of hub gene-transcription fac- nisms of cardiac fibrosis. tor interaction network The hub genes were extracted from the PPI Materials and methods network and identified by using the Cytoscape plug-in CytoHubba which provides a user-friend- Identification of differentially expressed ly interface to explore important nodes in biolog- genes ical networks, and the Maximal Clique Centrality The public raw RNA sequencing data (MCC) method which has a better performance was (GSE97358) was obtained from GEO data- used(21). The transcription factor (TF) which may base (https://www. ncbi.nlm.nih.gov/gds/?ter - target hub genes were predicted by using the Cyto- m=GSE97358)(16). The dataset was deposit- scape plug-in iRegulon. As reported, iRegulon can ed by Schafer et al. in 2017 and comprised 84 enrich TF motifs based on their direct targets with TGF-β1-stimulated samples and 84 paired unstim- the position weight matrix method(22). The gene sets ulated samples of cultured primary human cardiac and TF-target pairs on iRegulon were derived from Identification of potential key genes associated with cardiac fibrosis by rna sequencing data analysis 2317 ENCODE ChIP-seq data, and the TRANSFAC Gene ontology and pathway enrichment and JASPER databases. In the present study, the analysis potential TFs corresponding to the hub genes were Through GO analysis, the enriched go terms identified by the motif enrichment analysis with the were classified into biological process (BP) and criteria that the between orthologous genes ≥0.05 molecular function (MF). As shown in Figure 1, and false discovery rate (FDR) on motif similarity 13 terms in BP ontology for up-regulated DEGs ≤ 0.001, and normalized enrichment score (NES) were enriched, such as extracellular matrix or- >5. Finally, the hub gene-transcription factor inter- ganization (17 genes), extracellular structure or - action networks were visualized by the Cytoscape ganization (17 genes), muscle system process (19 software. genes), cardiac muscle tissue development (19 genes) and muscle contraction (15 genes), etc. In Statistical analysis the MF ontology, 6 terms for up-regulated DEGs Statistical analyses were performed using R were enriched, including receptor regulator activ- software v3.4.3 (R Foundation for Statistical