With the sequence of the human genome now complete, and the sequence of many other model organism genomes moving rapidly forward, there is a pressing need to identify the sequences involved in regulation of expression of the 25,000 ? 30,000 human genes. That includes promoters, enhancers, insulators, silencers, and other structures that signal the transcriptional apparatus to recognize (or not) a particular gene for expression. For more than 20 years, the gold standard method for identifying the genomic location of such regulatory signals has been the use of DNAse hypersensitivity analysis. Histones that normally coat the DNA strand are stripped off in regions of the genome that are the site of active binding of specific regulatory factors, rendering these regions accessible to the action of dilute concentrations of DNAse. While this method has produced valuable data for a few dozen genes, it is very laborious and has not previously been amenable to genome-wide application. We have devised a method to apply this approach to the entire genome of a particular tissue or cell line. Nuclei are exposed to DNAse in the traditional fashion, but then the digested ends are polished, the DNA is cut with a restriction enzyme, and the specific fragments that have one blunt end (from DNAse) and one sticky end (from the restriction enzyme) are captured and sequenced from the blunt end. This can be accomplished by direct sequencing, by using a modification of the SAGE approach, or even by adapting the bead-based method known as massively parallel signature sequencing (MPSS). Even 20 base pairs of tag sequence is sufficient to map most sequence signatures back to the genome. In a pilot experiment that generated 5000 such tags from primary human CD4+ T-cells, the captured sequences were significantly enriched for segments that lie just upstream or in the first exon or intron of known genes. Further validation of the captured sequences by a real-time PCR approach indicated that about 40% of the sequences correspond to genuine DNAse hypersensitivity sites. We are now exploring the scale up of this effort to generate an entire profile of regulatory sequences in the genome of any cell line or tissue. With truly massive sequencing throughput, one should even be able to quantify the degree of hypersensitivity of each site, by how many times that site appears in a deep collection of tags.
