Lu et al. Cancer Cell Int (2019) 19:52 https://doi.org/10.1186/s12935-019-0770-9 Cancer Cell International PRIMARY RESEARCH Open Access Identifcation of DNA methylation-driven genes in esophageal squamous cell carcinoma: a study based on The Cancer Genome Atlas Tong Lu1, Di Chen2, Yuanyong Wang1, Xiao Sun1, Shicheng Li1, Shuncheng Miao1, Yang Wo1, Yanting Dong1, Xiaoliang Leng1, Wenxing Du1 and Wenjie Jiao1* Abstract Background: Aberrant DNA methylations are signifcantly associated with esophageal squamous cell carcinoma (ESCC). In this study, we aimed to investigate the DNA methylation-driven genes in ESCC by integrative bioinformatics analysis. Methods: Data of DNA methylation and transcriptome profling were downloaded from TCGA database. DNA methylation-driven genes were obtained by methylmix R package. David database and ConsensusPathDB were used to perform gene ontology (GO) analysis and pathway analysis, respectively. Survival R package was used to analyze overall survival analysis of methylation-driven genes. Results: Totally 26 DNA methylation-driven genes were identifed by the methylmix, which were enriched in molecu- lar function of DNA binding and transcription factor activity. Then, ABCD1, SLC5A10, SPIN3, ZNF69, and ZNF608 were recognized as signifcant independent prognostic biomarkers from 26 methylation-driven genes. Additionally, a fur- ther integrative survival analysis, which combined methylation and gene expression data, was identifed that ABCD1, CCDC8, FBXO17 were signifcantly associated with patients’ survival. Also, multiple aberrant methylation sites were found to be correlated with gene expression. Conclusion: In summary, we studied the DNA methylation-driven genes in ESCC by bioinformatics analysis, ofering better understand of molecular mechanisms of ESCC and providing potential biomarkers precision treatment and prognosis detection. Keywords: Esophageal squamous cell carcinoma, Methylation, Biomarker, TCGA Background adenocarcinoma (EAC) and esophageal squamous cell Esophageal carcinoma (EC) is one of the most com- carcinoma (ESCC). Among them, ESCC is the predomi- mon malignant tumors in the digestive system. It occurs nant subtype and accounts for 80% of all patients [2]. For mostly in the esophageal epithelium, and there are no the moment, the mechanism of ESCC is still not fully typical clinical symptoms in the early stage of the patient. characterized, and the early symptoms of patients are Terefore, more than 80% of EC patients have pro- atypical, which brings great difculties for clinical diag- gressed to the advanced stage when they are diagnosed, nosis and therapy [3]. Similar to other malignancies, the which afects the prognosis of patients [1]. Esophageal progression of ESCC is also a complex process involv- cancer has two major histological subtypes, esophageal ing multiple factors and multiple gene mutations. Stud- ies have shown that the changes in the molecular level of ESCC tissues are earlier than the clinical features. Tere- *Correspondence: [email protected] fore, early diagnosis and intervention are signifcant for 1 Department of Thoracic Surgery, Afliated Hospital of Qingdao University, No. 16 Jiangsu Road, Shinan District, Qingdao 266003, China reducing the incidence of ESCC [4]. Full list of author information is available at the end of the article © The Author(s) 2019. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creat iveco mmons .org/ publi cdoma in/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Lu et al. Cancer Cell Int (2019) 19:52 Page 2 of 13 Epigenetic changes are identifed as signifcant con- r.gov/). Of them, the DNA methylation data was using tributors to cancer progression [5]. Te abnormal DNF the Illumina Infnium HumanMethylation450 platform, methylation is one of the most important and common and beta values, ranged from 0 to 1, was quantifed to epigenetic modifcations, and plays key roles in regulat- indicate the levels of DNA methylation. Te DNA meth- ing genome function [6]. Selective hypermethylation ylation data included 3 normal samples, 96 ESCC sam- or hypomethylation of genes to regulate the expression ples. And we used transcriptome profling data without of genes and form specifc tissue types during develop- isoform expression and miRNA expression quantifca- ment are considered to be a hallmark in developing many tion, for analyzing the gene expression of ESCC. Ten, carcinomas [7]. In recent years, studies on methylation R software and packages were utilized to analyze and and tumors have gradually drawn more attention. For normalize the downloaded data to obtain diferentially instance, Roy et al. analyzed the lymph node metastasis expressed genes (DEGs) and diferentially methylated in esophageal squamous cell carcinoma and built a com- genes (DMGs). Furthermore, a total of 96 ESCC sufers prehensive methylation signature for predicting the prog- had recorded clinical data and were used in further sur- nosis of patients [8]. Genes including NNK, MSH3 and vival analysis (Additional fle 1: Table S1). Te data from P16, which were stated to be methylated, associated with TCGA is open-ended and publicly available. tumors progression [9–11]. Identifcation of abnormal methylated genes can explore the redundancy and insta- Integrative analysis bility of the esophageal carcinoma genome and provide Te DEGs and DMGs were integrated for performing an the basis for risk prediction and targeted therapy. analysis via the R package MethylMix [15]. MethylMix is Te wide DNA methylation arrays and advent of deep a program used for automatically analyzing the correla- RNA-Seq approach has signifcantly contributed to study tion between methylation events and gene expression the interaction between methylation and gene expres- [13]. Tree datasets are required as input for analysis: sion during tissue carcinogenesis and development. normal DNA methylation data, cancer DNA methylation An integrative analysis of mRNA expression and DNA data and matched gene expression data. Ten, the Meth- methylation studied by Kim et al. stated out the function ylmix identify cancer specifc hyper and hypo methylated of epigenetic changes on malignant mesothelioma cell genes, which named transcriptionally predictive genes, [12]. Furthermore, in order to identify the mechanism and compute the correlation between methylated genes contributed to oncogenesis, Olivier Gevaert et al. devel- and related genes. A Wilcoxon rank sum test was adopted oped a novel computational algorithm called Methylmix in this algorithm. And the fnal output of MethylMix is to study abnormal methylated genes and predict tran- genes that are both transcriptionally predictive and dif- scription [13]. As a well-known cancer genome data- ferentially methylated states. Additionally, the diferential base, Te Cancer Genome Atlas (TCGA) [14] provides methylation (DM) value where a negative DM value sig- a great genomic data with patients information, which nifes hypomethylation and a positive DM value signifes can translate molecular information into potential clini- hypermethylation can be used in subsequent analysis. cal information. In this study, ESCC-related expressed and abnormally methylated genes were recognized based Methylation‑driven genes functional enrichment on TCGA database, and the related diferential genes and pathway analysis and expression of abnormally methylated genes in ESCC Gene ontology (GO) analysis was conducted on identifed patients were clarifed. We analyzed RNA-Seq tran- methylation-driven genes with methylation/expression scriptomes and DNA methylation data of ESCC samples using the DAVID database. DAVID provides integrative from 99 cases in TCGA. Five candidate genes (ABCD1, and systematic annotation tools for unraveling biological SLC5A10, SPIN3, ZNF69, ZNF608) were identifed from meaning of genes. Gene ontology (GO) analysis includes 26 driven genes (p < 0.05), which could be served as inde- the molecular function, biological process and cellular pendent prognostic biomarker. Additionally, ABCD1, component [16]. And we used Goplot to visualize the CCDC8, FBXO17 were identifed to be meaningfully result. correlated with prognosis by further integrative survival Pathway analysis was conducted for the methylation- analysis. Besides, we found the signifcant correlation driven genes with ConsensusPathDB [17], which is a between methylated sites with gene expression. functional molecular interaction database, integrating information on genetic interacting signaling, protein Methods interacting, drug-target interactions, metabolism and Data acquisition and preprocessing gene regulation in humans. Over-representation analysis In this study, all data were obtained from TCGA data was based on neighbourhood entity sets or biochemi- portal accessed on 20181108 (https ://porta l.gdc.cance cal pathways, and the pathway analysis was performed Lu et al. Cancer Cell Int (2019) 19:52 Page 3 of 13 on the basis of imputed gene list. Lists of hypomethyl- Besides, biological process (BP) indicated enrichment ated genes and hypermethylated genes were analyzed predominantly