A key goal of functional genomics is to identify transcriptional regulatory elements throughout the entire human genome that are of physiological significance. Despite the importance of this goal, unbiased, genome-wide identification of transcriptional regulatory elements has never been described in an experimental fashion. Many transcriptional regulatory proteins play causal roles in human cancer by acting as oncoproteins and tumor suppressor proteins. The combination of chromatin immunoprecipitation (ChIP) and microarray analysis is a powerful tool for identifying in vivo target sites on a genome-wide level, and this approach has often revealed unexpected and crucial insights into the biological functions of transcriptional regulatory proteins. In collaboration with Tom Gingeras at Affymetrix, we have used tiled microarrays representing all of human chromosomes 21 and 22 to perform an unbiased and comprehensive ChIP analysis on the p53 tumor suppressor, the Myc oncoprotein, and Sp1. This analysis has identified numerous physiological targets, many of which are in unexpected genomic locations. We intend to continue this collaboration to define, on a whole-genome scale, physiological targets of many transcriptional regulatory proteins directly involved in human cancer. Two experimental systems will be employed. First, we will use MCF-10A cells, a non-transformed human breast epithelial cell line, in which expression of the Src oncogene (fused to the ligand-binding domain of estrogen receptor) can be rapidly induced, thereby permitting a comparison of """"""""normal"""""""" cells with cells at different states along the transformation pathway. Second, we will analyze h-TERT immortalized fibroblasts that are or are not transformed with T antigen + activated Ras. The initial experiments will be performed on approximately 50 proteins (including multiple members of the p53, Jun-Fos, Myc, Myb, Ets, Rb, E2F families of oncoproteins and tumor suppressors) on the ENCODE array representing 1 % of the genome. From these ChIP results and gene expression profiles, target sites of approximately 35 proteins (average 1.5 conditions/protein) of highest interest will be determined on a whole-genome basis using arrays that should be available when this proposal would begin. The results will provide a massive amount oi unbiased and comprehensive information such as 1) the location of physiological target sites with respect to known and unknown genes, 2) links between oncogenically-regulated genes and specific factors, 3) whether factor occupancy at regulated genes is regulated or unregulated, 4) specific combinations of factors that preferentially associate with oncogenically-regulated genes, 5) the specificity and redundancy of binding by proteins in multiprotein families, 6) identification of regulatory circuits. This detailed molecular comparison of normal versus transformed cells during a defined transition between the two states should be extremely useful both to numerous investigators who work on many different aspects of cancer. More generally, the results should represent a new dimension in functional genomics, and they should spawn a variety of experimental and computational approaches to understand how oncoproteins and tumor suppressor proteins cause cancer. ? ? ?
Showing the most recent 10 out of 22 publications
