Defining novel cancer genes, sage tags and co-regulated regions of the human genome evolving from the search and identification of novel Wnt/TCF/B-catenin targets

Abstract

The Wnt signaling pathway is involved in many differentiation events during embryonic development and can lead to tumor formation after aberrant activation of its components. β-catenin, a cytoplasmic component, plays a major role in the transduction of canonical Wnt signaling. The first part of this study aimed at identifying novel genes that are regulated by active β-catenin/TCF signaling in hepatocellular carcinoma-derived Huh7 cells with high (transfected) and low β-catenin/TCF activities. High TCF activity Huh7 cells led to earlier and larger tumor formation when xenografted into nude mice. SAGE (Serial Analysis of Gene Expression), genome-wide microarray and in silico promoter analysis were performed in parallel, to compare gene expression between low and high β-catenin/TCF activity clones, and also those that had been rescued from the xenograft tumors. SAGE and genome-wide microarray data were compared and contrasted. BRI3 and HSF2 were identified as novel targets of Wnt/β-catenin signaling after combined analysis and confirming experiments including qRT-PCR, ChIP, luciferase assay and lithium treatment. In the course of analyzing SAGE and microarray data of Huh7 cells, we felt the need to overlay this data with different data sources. Starting with this motivation in the second part of the study, we performed the first meta-analysis of SAGE and microarray data together, which we termed common gene expression patterning (COGENT). By using COGENT data, we introduced the “being multi-cancer” concept for genes being differentially expressed in several types of tumors. There are not only multi-cancer genes but also multi-cancer co-regulated regions (RIDGEs) on the genome. “Multi-cancerness” of a gene correlates well with the number publications stating that gene in a cancer context, its rank within a single study and the number of its orthologs. There are many multi-cancer genes which have not been related to tumorigenesis as yet. In addition, there are 81 tumor associated RIDGEs (TA-RIDGEs) on the human genome which change similarly in multiple types of tumors.