Connections between factors and genes change drastically in different conditions.


Micrographs came before movies, and for systems biologists the transition to a dynamic picture seems even more daunting: assembling a global picture of transcription connections in one state was plenty of work already. But now, Nicholas Luscombe, Mark Gerstein (Yale University, New Haven, CT), Madan Babu (MRC, Cambridge, UK), and colleagues have pooled data on multiple growth conditions in yeast. They find that transcriptional connections vary wildly between the states, suggesting that the cell faces a major task when switching from one state to another.

Half of the active interactions (transcription factor to regulated gene) are replaced for every change in condition, and only 66 of 2,476 interactions are retained across 4 or more conditions. The logic of organization also changes depending on whether the cell is responding to purely intracellular changes or to a signal from outside the cell. The latter response is characterized by pathways that are simpler (less transcription factors per target gene), more decisive (more targets per factor), and more direct (fewer sequential steps in a pathway and fewer connections between pathways). By contrast, endogenous pathways are more cautious; they feature more of the feed-forward motifs that buffer conditions before proceeding.

One prominent feature of previous static pictures was hubs of activity. Hubs were thought of as constant, but most (78%) are now found to be transient. Plenty of transcription factors do stick around for other tasks—the vast majority participate in more than one process and unique regulation relies on combinations.

“This is really a first view,” says Luscombe. He hopes to extend the approach to protein interactions, post-translational modifications, and other organisms with complex developmental programs. More detailed time courses would also illuminate whether there are unique pathways by which cells move from one state to another, hubs that are critical in propagating changes, and particular bottlenecks or vulnerabilities.


Luscombe, N.M., et al.