Investigators say that there are a lot more binding events between regulatory proteins and genes, as measured by ChIP-chip, than expected. They add that most of them appear not to be involved in gene expression.
The blueprints for the transformation from embryonic cells to adults are encoded in the genes of every cell and are read out by vast networks of transcription factors, which determine where and when genes are expressed. It has been generally assumed that transcription factors are targeted to a limited set of genes and that they regulate expression wherever they are bound.
A team of researchers found, however, that there are thousands of regions reproducibly bound by each factor. “This is several orders of magnitude more genes than these factors are thought to regulate, raising the question of what the function of the binding is,” points out author Michael Eisen, Ph.D., a computational and evolutionary biologist at the Lawrence Berkeley National Laboratory (LBL) and the University of California at Berkeley.
Rather than classify regions as bound or unbound, as other researchers have done, the Berkeley team examined the full scope of binding observed in their ChIP-chip data. They focused on differences in the amount of factor bound to each gene.
The team found a clear relationship between the number of factor molecules bound at a given site and the site’s role in gene regulation. DNA sites that bound the most molecules were those already thought to be key in regulation during development. Much of the low-level binding detected at thousands of genes, while clearly representing real molecular interactions, appears to play no role in regulating gene expression.
“Realizing that much of the binding detected in these assays may be nonfunctional significantly impacts how the results of these experiments should be interpreted,” explains Mark Biggin, Ph.D., head of the genome science department at LBL. “The analysis and conclusions of published ChIP-chip studies should be reexamined with this possibility in mind.”
The investigators involved in this research were from Lawrence Berkeley National Laboratory, University of California, Berkeley, Affymetrix, and the California Institute for Quantitative Biosciences. The study is published in PLoS Biology.