Microarray analysis has become a widely used technique for observing the transcriptional profile of genes at a genome-wide level. But its strength reveals one of its limitations. Observing the patterns of transcriptional activity that occur under different conditions reveals genes that are co-expressed, but cannot distinguish between genes that are under common regulatory control and those whose expression patterns just happen to correlate. What we'd really like to know is, if two genes have matching transcriptional profiles, are they also under the direct transcriptional control of a common upstream regulator? A group led by Richard Young has taken us one step further towards answering this question by developing a microarray-based technique that can monitor the chromosomal locations at which protein–DNA interactions occur across the entire yeast genome. Researchers should now be able to identify all the genes whose transcription is directly affected by any DNA-binding protein.

The new method involves combining chromatin immunoprecipitation (also known as ChIP) with microarray analysis. After crosslinking a protein of interest to genomic DNA, the complex is immunoprecipitated. Two pools of DNA fragment — those released from the protein and those from control DNA — are amplified and labelled using different fluorescent dyes. These are then hybridized to a microarray plate that contains all (6,361) intergenic sequences in the Saccharomyces cerevisiae genome. The greater the difference in the fluorescent intensity at any fragment on the array, the stronger the binding of the protein to that fragment.

To prove that the method works, the authors performed a genome-wide location analysis to identify the sites bound by Gal4 and Ste12, two well-characterized yeast transcription factors.

Gal4 activates genes required to metabolize galactose, from which yeast cells derive energy when glucose is limiting. Only ten loci were bound by Gal4 (and were also activated in the presence of galactose), seven of which were previously known downstream targets of Gal4. The three new genes, such as PCL10 , revealed new information about the coordination of different metabolic pathways. For instance, by preventing glycogen formation, the activation of PCL10 could be required to maximize the energy that is obtained from galactose.

In response to mating pheromone, Ste12 induces the activation of over 200 genes, but it is not known how many are affected directly. Ste12 was found to bind to the upstream regions of 29 pheromone-induced genes, 11 of which are known to be involved in various steps of the mating process. Again, the unknown genes could plausibly have mating-related functions — in cell morphology or in nuclear fusion, for example.

The ability to identify all the genes whose upstream regulatory regions are bound by a certain DNA-binding protein should greatly improve the functional studies of proteins, especially where our understanding of their function has been limited by restricted knowledge of their targets.