Identification Of A Diagnosis Panel Of DNA Methylation And The Relative Gene's Function Study In Esophageal Squamous Cell Carcinoma (ESCC)

At present,esophageal cancer is a kind of malignant tumor with its incidence ranked on the eighth,and its death rate on the sixth.In China,the incidence of esophageal cancer is more higher than other countries and regions with over 220,000 new cases added every year,which means we have the half of the new cases worldwide.Esophageal squamous cell carcinoma(ESCC)is the major type of esophageal cancer in our country,genetic and epigenetic factorshave great importance in ESCC development and prognosis.DNA methylation was taken as the most promising biomarker for caner diagnosis in recent years.Compared to cancer specific single nucleotides polymorphisms(SNPs),the number of specific DNA methylation sites is much more,so that it is more effective in diagnosis.Tumor tissue DNA has high heterogenicity and the status of DNA methylation may be different,but with the rapid development of Genetics and Molecular Biology many methylation-based diagnosis panel were found in a few type of cancer,and some of these panel have achieved a high diagnostic accuracy.In ESCC,no such diagnosis panel has been developed,so that it is urgent to search the most suitable methylation biomarkers to develop a diagnosis panel for ESCC.In our study,6 novel genes(ZNF132,ADHFE1,SALL1,TFPI2,TMEM132 C,ZNF790)are selected to construct the prediction group according to prophase screening in TCGA database consist with CD4&.CD8 white cell database in normal population,UCSC Encode histone database and paper searching.The DNA methylation status of these genes are validated in pairwise ESCC tumor tissues and adjacent tissues of chinese han population by bisulfite sequencing.The result shows 6 genes are all significantly hypermethylated in tumor tissues than in adjacent tissues.Seven statistical models(Logistic Regression,Random Forest,Support Vector Machine SVM,Naive Bayes,Neural Network,Linear Discriminant Analysis,Mixture Discriminant Analysis,Flexible Discriminant Analysis)are used to evaluate the performance discriminatory algorithms according to the sequencing result.We can see that in train set,SVM has thebest diagnosis efficiency,its sensitivity,specificity and accuracy are 70%,86%,78%.And in test set logistic regression has better result than others,its sensitivity,specificity and accuracy are 75%,88%,81%.All statistical models have the similar results in both set.This suggests that combining these 6 genes can be a effective panel to assistthe diagnosis of ESCC.It has potential value in clinical diagnosis.Among these 6 genes,ZNF132's expression is significantly downregulated in ESCC tumor tissues in TCGA ESCC RNA_Seq database.As we treated ESCC cell lines with 5-Aza,the expression of ZNF132 has a significant upregulated expression after demethylation in cell lines.Then ZNF132 promoter hypermethylation in cell lines lead to its significantdownregulated expression.In pairwise tissues,comparing to adjacent tissues,the expression of ZNF132 in tumor tissue is also significantly downregulated.These results suggest that ZNF132's expression is regulated by its DNA methylation status.Based on the finds above,we make ZNF132 overexpressed in ESCC cell lines,find that cell's proliferation rate is significantly reduced and its migration ability is also decreased.These results suggest that ZNF132 may play an important role in ESCC tumorigenesis as a tumor suppressor gene.Our results lay a foundation for ZNF132's further research in ESCC,hoping to provide a new way to explain the molecular mechanism for ESCC and optimize the clinical treatment.