Here is a plumbing problem: Pump water through a series of flexible microscopic tubes in a way that helps identify which pieces of DNA a protein will stick to. It may sound esoteric, but researchers have devised just such a system to help solve a fundamental problem in biology—how a cell takes shape from a fluctuating network of genes and proteins.

The device is a unique example of microfluidics technology, sometimes called a lab-on-a-chip, that pushes water around in microscopic tubes and reservoirs made from the same cellophanelike plastic as soft contact lenses. Powered by hydraulics and flexible valves, the system presses down with small "buttons" of plastic to trap DNA and proteins in the middle of bonding together. Besides being able to perform hundreds of simultaneous measurements, it can detect relatively weak, fleeting interactions between molecules that other experiments can miss, says the system's co-designer, biochemist Sebastian Maerkl of the California Institute of Technology.

"It is really important to know what these interaction strengths are," Maerkl says. "It's basically what makes the cell function." Genes become more or less active at the touch of proteins called transcription factors, each of which can influence hundreds or thousands of other genes. As a factor becomes more concentrated in a cell, it will begin to activate genes that it has less intrinsic affinity for, potentially changing the cell's behavior. Researchers would like to have a map of these interactions they can fiddle with on computers, but getting a bird's-eye view of a transcription factor's effects can be laborious.

To facilitate this research program, Maerkl and his PhD adviser Stephen Quake gave their device a series of dumbbell-shaped water-filled reservoirs [see photo above]. The two sides of each reservoir were separated by a valve. One side of each dumbbell was filled with a different short sequence of DNA labeled with a fluorescent molecule. In the other side, the researchers synthesized the transcription factor they wanted to study and fixed it in place. When the protein was ready, they opened the valve and the DNA spread to the other side.

The researchers measured the strength with which the protein and DNA bonded by pressing down on the protein with a plastic "button." Any DNA bound to the protein was caught in place, and they measured its quantity by the amount of fluorescence in the reservoir. They tested their system on a pair of yeast transcription factors and used the data to predict which yeast genes the proteins would target, they report in this week's Science. "We managed to really take the first comprehensive measurement of a transcription factor binding-energy landscape," Maerkl says. "The advantage over most other methods is it's easy to scale."

"I think it's a wonderful paper," says systems biologist Carl Hansen of the University of British Columbia. "An integrated platform like this ... would be a quantum leap" for studying interactions between proteins, too, he says. It promises a global view of transcription factor–DNA binding, agrees Curtis Callan of Princeton University, a theoretical physicist who studies biological systems. "In principle," he says, "they can do really comprehensive experiments that'll just dominate the field."