Identification And Functional Analysis Of Feed-forward Loops As Well As Feed-back Loops Centered On Enhancers In Multiple Tissues

Feed-forward loops(FFLs)have been shown to be one of the most common and important classes of transcriptional network motifs that involved in various diseases such as cancer.It usually enclosed in a three-node pattern,which is composed of an input element regulating another element and a jointly regulated target gene.Enhancers are cis-regulatory elements that positively regulate protein-coding genes or microRNAs(miRNAs)by recruiting DNA-binding transcription factors(TFs).However,it is still unclear that how enhancers function in the way of network by interacting with other elements.In this study,feed-forward loops in human 86 tissue/cell lines and 5 mouse cell lines are constructed by integrating the regulatory relationships between TFs,enhancers,miRNAs and genes.Then,loops are exemplified by hypergeometric test.As a result,there are 53,280 TF-enhancer-miRNA loops,10,233 enhancer-miRNA-gene loops,6,754,346 TF-enhancer-gene loops and 1,494 TFenhancer feedback loops.Finally,we obtain regulatory networks centered on enhancers in 16 human tumor cell lines.Subsequently,Betweenness Centrality values are calculated and 17 pan cancer miRNAs are identified which function as key elements in at least 14 tumor cell lines.Functional enrichment analysis based on target genes reveals that 16 pan cancer miRNAs are related to tumor pathways.In order to browse and search loops identified,EnhFFL(http://lcbb.swjtu.edu.cn/EnhFFL)is constructed.Users can browse loops by selecting species,type of tissue/cell line(normal or cancer),types of FFLs.EnhFFL offers searching elements by name/ID,genomic location as well as the conservation of miRNA target genes.We also developed tools for users to screen customized FFLs using threshold of q value as well as confidence score of miRNA target genes.Finally,we focus on p53,a widely accepted tumor suppressor,to construct enhancer regulatory networks involving p53 in 8 human tissue/cell lines.Result shows that the regulatory networks are highly tissue-specific,while they tend to be much more similar when it comes to samples obtained from same tissue.In addition,most of the 5 genes and 6 miRNAs,which are widely engaged in the networks identified,are reported to function in cancer pathways.Similarly,among the 14 genes and 14 miRNAs involved in all 3 lung samples,there are 7 genes and 7 miRNAs contributing to lung cancer.At last,by differential expression analysis,we get 7 genes and 2 miRNAs in the HepG2 p53-enhancer network significantly in correlation with the level of p53,among which there are 3 genes and miRNAs function in liver cancer.