Int J Clin Exp Pathol 2020;13(7):1698-1706 www.ijcep.com /ISSN:1936-2625/IJCEP0107763 Original Article Identification of transcriptomic markers for developing idiopathic pulmonary fibrosis: an integrative analysis of gene expression profiles Diandian Li1, Yi Liu2, Bo Wang1 1Department of Respiratory and Critical Care Medicine, West China Hospital of Sichuan University, Chengdu 610041, China; 2West China School of Medicine, Sichuan University, Chengdu 610041, China Received January 12, 2020; Accepted March 6, 2020; Epub July 1, 2020; Published July 15, 2020 Abstract: Idiopathic pulmonary fibrosis (IPF) remains a lethal disease with unknown etiology and unmet medical need. The aim of this study was to perform an integrative analysis of multiple public microarray datasets to inves- tigate gene expression patterns between IPF patients and healthy controls. Moreover, functional interpretation of differentially expressed genes (DEGs) was performed to assess the molecular mechanisms underlying IPF progres- sion. DEGs between IPF and normal lung tissues were picked out by GEO2R tool and Venn diagram software. Data- base for Annotation, Visualization and Integrated Discovery (DAVID) was applied to analyze gene ontology (GO) and Kyoto Encyclopedia of Gene and Genome (KEGG) pathway. Protein-protein interaction (PPI) of these DEGs was visu- alized by Cytoscape with Search Tool for the Retrieval of Interacting Genes (STRING). 5520 DEGs were identified in IPF based on six profile datasets, including 3714 up-regulated genes and 1806 down-regulated genes. Using Venn software, a total of 367 commonly altered DEGs were revealed, including 259 up-regulated genes mostly enriched in collagen catabolic process, heparin binding, and the extracellular region. For pathway analysis, up-regulated DEGs were mainly enriched in ECM-receptor interaction, protein digestion and absorption, and focal adhesion. Fi- nally, 24 DEGs with degrees ≥10 were screened as hub genes from the PPI network, which were enriched in protein digestion and absorption, ECM-receptor interaction, focal adhesion, PI3K-Akt signaling pathway, amoebiasis, and platelet activation. The present integrative study identified DEGs and hub genes that may be diagnostic biomarkers or therapeutic targets, and provide novel insights into the pathogenesis of IPF. Keywords: IPF, differentially expressed gene, co-expression network, transcriptomic markers, bioinformatic analy- sis Introduction efficacy is still unclear. Particularly, patients with severe IPF often develop acute exacerba- Idiopathic interstitial pneumonia (IIP), with tions resulting in the rapid deterioration of unknown etiology, has major implications for lung function, requiring lung transplantation [4]. prognosis and management [1]. Idiopathic pul- Thus, it is essential to elucidate the mecha- monary fibrosis (IPF), a chronic and progres- nisms of IPF pathogenesis and explore poten- sive interstitial fibrotic lung disease, where no tial biomarkers for improving the treatment cause can be identified, is one of the most effect of IPF. common types of IIP and remains a fatal dis- ease with a median survival time from diagno- Currently, studies have demonstrated that sis of 2 to 5 years [2]. Although the prognosis is environmental (such as smoking, metal or poor, it is difficult to predict due to limited clini- wood dust, sand, and spores from soil) and cal biomarkers reliably reflecting disease pro- genetic factors variably contribute to disease gression. With regard to management, limited susceptibility and outcomes [5]. A number of pharmacologic options--pirfenidone and ninte- common gene variants with modest effect size danib are supposed to be promising for IPF increase the risk of disease in patients with based on clinical trials [3], but the long-term sporadic IPF, while rare variants with large Transcriptomic markers in IPF effect size influence disease risk in patients gene expression data. Ethical approval for this with familial interstitial pneumonia [6]. In the study was not required because the data were past decades, rapid growth of high-through- freely available in public datasets. put transcriptomic data largely enables quickly identification of differentially expressed genes Data processing of DEGs (DEGs) in diseases and has been proved to be a reliable technique [7]. There are a large GEO2R (http://www.ncbi.nlm.nih.gov/geo/geo- number of microarray gene expression datas- 2r/), a web tool, can perform sophisticated ets that are publicly available from the Gene R-based analyses of GEO data and presents Expression Omnibus (GEO) database, and many the results as a table of DEGs [8]. In the bioinformatic studies on IPF have shown that present study, GEO2R online tools were applied DEGs and biologic functional pathways partici- to identify DEGs in lung tissues between IPF pated in the development of this disease. and healthy controls with absolute log2 Fold However, single microarray analysis often Change (logFC) >1 and adjusted P-value < brings a high false-positive rate, while analysis 0.05. The DEGs with logFC >1 were consider- of multiple transcriptomic datasets may shed ed as up-regulated genes, while the DEGs light on discovering robust candidates for diag- with logFC <-1 were considered as down-regu- nosis and treatment. lated genes [9]. The raw data downloaded in TXT format were imported in Venn software Therefore, the aim of present study was to per- online to detect the commonly DEGs among the form an integrative analysis of multiple pu- six datasets. blic microarray datasets and investigate gene expression patterns in IPF lung tissue in IPF Gene ontology and pathway enrichment analy- patients and healthy controls. Moreover, DEGs sis identified in this study were further interpreted by enrichment analysis and co-expression net- Functional interpretation/gene ontology (GO) work construction to assess the molecular biological process analysis and Kyoto Ency- mechanisms underlying IPF progression. clopedia of Gene and Genome (KEGG) path- way analysis) of DEGs was further performed Materials and methods using Database for Annotation, Visualization and Integrated Discovery (DAVID 6.8; available Datasets search and selection online: http://david.ncifcrf.gov). This online bioinformatic tool was designed to identify a Microarray data from IPF-related mRNA ex- large number of genes or proteins function pression profiles were retrieved and down- [10]. By gene ontology (GO) analysis of high- loaded from the National Center for Biotech- throughput transcriptome or genome data, nology Information (NCBI) GEO database genes and their RNA or protein product can (http://www.ncbi.nlmNih.gov/geo). We search- be defined to identify unique biologic proper- ed public microarray datasets till Jun 1, 2019 ties [11]. KEGG is a collection of databases using the keyword “idiopathic pulmonary fi- dealing with genomes, diseases, biologic path- brosis” or “IPF” restricted to Homo sapiens. ways, drugs, and chemical materials [12]. We The datasets obtained were further selected used DAVID to visualize the DEGs enrichment by our inclusion criteria: (1) case-control stu- of biologic process (BP), molecular function dy; (2) sample size larger than 5 in each group; (MF), cellular component (CC) and pathways (3) dataset using lung tissues for gene ex- (P<0.05). pression analysis; (4) raw data or processed data were available in these datasets; (5) Protein-protein interaction network construc- subjects with IPF included in this study met tion the diagnostic criteria for IPF based on the American Thoracic Society (ATS) and European Information of protein-protein interaction (PPI) Respiratory Society (ERS) consensus state- of the DEGs was evaluated by STRING (Search ment. Two investigators (J.H. and Y.L.) ex- Tool for the Retrieval of Interacting Genes) tracted the data and reached a consensus online tool [13], with interactions of a combin- on all items. Data retrieved from the dataset ed score >0.4 considered statistically signifi- included GEO accession, publication year, sam- cant [14]. Cytoscape (version 3.6.1) was used ple size, sample source, platform, and raw to visualize molecular interaction networks 1699 Int J Clin Exp Pathol 2020;13(7):1698-1706 Transcriptomic markers in IPF Table 1. Microarray datasets included in this analysis Publication Sample size GEO accession Source Platform year IPF Healthy control GSE110147 2018 22 11 Lung tissue Affymetrix Human Gene 1.0 ST Array GSE53845 2014 40 8 Lung tissue Agilent-014850 Whole Human Genome Microarray 4x44K G4112F GSE24206 2011 11 6 Lung tissue Affymetrix Human Genome U133 Plus 2.0 Array GSE10667 2009 23 15 Lung tissue Agilent-014850 Whole Human Genome Microarray 4x44K G4112F GSE32537 2013 119 50 Lung tissue Affymetrix Human Gene 1.0 ST Array GSE2052 2005 13 11 Lung tissue Amersham Biosciences CodeLink Uniset Human I Bioarray GEO, Gene Expression Omnibus; IPF, idiopathic pulmonary fibrosis. [15]. In addition, the MCODE app in Cytoscape um-dependent phospholipase A2 activity, and was used to check the most significant module arachidonic acid binding. Moreover, changes in in the PPI networks (degree cutoff = 2, node CC of up-regulated DEGs were mainly enriched score cutoff = 0.2, max depth = 100, and k-core in the extracellular region,space and matrix. = 2). Table 2 presented a summary of the GO bio- logic process analysis results. Results of KEGG Results analysis are shown in Table 3, which suggest that up-regulated DEGs were mainly enriched Identification of DEGs in IPF in ECM-receptor