irst came the gene. Then the genome, the transcriptome, the proteome, and the metabolome. The geyser of biological information continues to gush and at times more information seems to preclude greater knowledge, so complex are living systems. Cells respond to their environment, make use of different food sources, repair internal defects, and co-ordinate all these activities in an orchestrated cycle of growth and development. How? A team led by Richard Young of the Whitehead Institute for Biomedical Research in Cambridge, Massachusetts, has a paper at #8 that points the way to a new understanding of how the organism as a whole works.

The thousands of genes that contribute to smooth functioning are controlled by transcriptional regulatory proteins, which bind to genes and increase or decrease the rate at which the gene is transcribed. Young’s group used the yeast genome data to identify every transcriptional regulator’s target genes. The method is called genome-wide location analysis. First they identified all 141 of the transcription regulator genes and constructed strains in which one of these genes was tagged with a marker. Unfortunately 17 of the genes resisted tagging, and a further 18 were not expressed at useful levels in the cell, but that still gave a set of 106 strains with tagged, expressed, regulator sequences.

A microarray of yeast DNA was used to identify each of the targets that the tagged sequences recognized, in total just under 40% of yeast’s 6,270 genes. The group then turned its attention to the detailed links between regulators and targets, and uncovered six different patterns or motifs. The simplest is an autoregulation motif, in which a regulator binds to itself. This kind of control enables a system to respond very quickly to a signal. For example, a gene called Ste12 is activated by a mating signal from another yeast strain and is autoregulated. That means that the amount of Ste12 product increases very rapidly in response to a mating signal: Ste12 also controls several other genes associated with mating.

More complex motifs included the multi-component loop and the feedforward loop. The group uncovered three multi-component loops, in which A regulates B and B regulates A. Such a loop (which can contain more than two genes) allows feedback control and can switch the metabolism into one of two stable states. In a feedforward loop A controls B and both A and B control C. This motif was very common, with 10% of the transcriptional targets part of such a motif. Feedforward loops can help to time events, because expression of the target may require the accumulation of sufficient amounts of both the primary and secondary regulators. They can also ensure that a pathway responds to a sustained signal input rather than a temporary blip.

There were also single-input and multi-input motifs. A single regulator might bind to several targets, all of which are part of a single metabolic pathway. The regulator Leu3, for example, binds to three genes involved in leucine synthesis. Or an entire pathway may be controlled by many regulator genes. A multi-input motif would be able to coordinate a pathway under several different conditions, so that this part of the cellular apparatus would function, for example, in the presence of several different sources of food.

Finally, the group uncovered regulator chains, in which one transcription regulator controls a second, which controls a third, and so on to the final gene target. These are also common, with 188 chains ranging in length from 3 to 10 regulators. Such chains represent the simplest logic for ensuring that genes are activated in a specific temporal sequence.

The basic motifs represent logical units, which can be assembled into larger networks. To demonstrate the power of this approach Young’s team set out to map the network that controls yeast’s reproductive cycle. Cell-cycle control is rather well understood, but the team made no assumptions about it. Instead, they developed an algorithm to identify genes that were both bound together and expressed together. The resulting sets of regulators and genes are multi-input motifs that were then used to find genes whose expression fluctuated during the cell cycle. This yielded 11 regulator genes that could be slotted correctly into the known cell-cycle on the basis of the time of peak expression of the genes each controlled. Moreover, two regulators that had been previously implicated in cell-cycle control, but without well-defined functions, were given their correct place in the network.

As the authors note, "this approach should represent a general method for contructing other regulatory networks," a belief borne out by the mass of citations and the development of tools, for example a web-based suite of programs designed to sniff out transcriptional regulatory motifs. Although the individual DNA details that distinguish two species differ, the motifs used to control genes, and the way those motifs are assembled into networks, are likely to be very similar. The possibility of moving from the particular of one species’ genome sequence to the general of a network built of simpler motifs is likely to assist the understanding of all biological systems.