Biologists know a lot about the development of body parts—how cells become specialized for certain functions and why, say, the heart develops in the chest and not the head. But the body isn't assembled on a production line that adds one component after another. As an organ forms, one tissue's development must be coordinated in time and space with that of all the others.

To understand the genetic programs that orchestrate this complicated task, Norbert Perrimon studies organ development in Drosophila, the fruit fly. "I wanted to use molecular tools to identify gene function in the context of development," he remembers thinking when he became a graduate student in Paris.

When the right tools didn't exist, Perrimon had to invent them. One of his major achievements is a method, developed with Andrea Brand, for controlling when and where genes are turned on or off. Called the GAL4/UAS system, it is one of the basic tools in molecular genetics. It allows scientists to determine when a gene springs into action, when it shuts down, and its role in development. Before GAL4/UAS was invented, it was impossible to study genes that proved lethal if left on too long because there was no way of turning them off. But Perrimon's system keeps genes silent except when researchers want them to be expressed. Now thousands of different fly lines contain GAL4/UAS. "This is a pretty important contribution to the field," Perrimon says, "because it gives us a lot of flexibility and precision in the way we control gene expression."

Using GAL4/UAS and other methods he developed, Perrimon has studied cell signaling pathways. For example, he demystified the Wnt pathway, a network of proteins that controls developmental processes such as embryonic induction (one cell influences the fate of other cells), cell polarity (which end of a cell is which), and cell fate (what a cell will become). It was previously known that Wnt proteins secreted from specific sites during development form a gradient in neighboring tissue. The proteins affect genes in target cells, but to different degrees, depending on the distance from the secretion site.

Before Perrimon's studies, the signaling pathway that links Wnt secretion to target genes was a black box. Among the genes his group identified was porcupine, which is essential for Wnt secretion, and two key members of the pathway: dishevelled and GSK3. When dishevelled protein receives the go-ahead from Wnt, it inhibits GSK3 protein. When not inhibited, GSK3 helps degrade a protein that switches on certain genes. Therefore, the net effect of Wnt signaling is to activate those genes. With these landmarks in place, the research community was able to identify the Wnt pathway's other components. Because some members of the pathway have been implicated in several cancers, this detailed knowledge is suggesting new ways to treat the disease.

Perrimon's interests expanded beyond signaling pathways after Drosophila's genome was sequenced in 2000. "This started to make us think about new methods that would allow us to screen the genome more systematically for gene function," he says. "Drosophila has about 14,000 genes, and we wanted to interrogate every one."

To develop a mass screening technique, Perrimon appropriated a technology called RNA interference (RNAi) to prevent each gene in turn from working. Then he determined how suppressing that gene affected Drosophila cells. "Before, when we were doing a genetic screen in the fly, we would focus on only a few genes," he says. "But because of the power of RNAi screening, when we study a problem we now come up with hundreds of genes of interest. This forces us to think more globally about the contributions of all those genes."

For the past eight years, Perrimon's group has been making the technology more and more robust, and they established the Drosophila RNAi Screening Center at Harvard (www.flyRNAi.org). "We transferred all the technologies that our lab has developed into this center, which is open to the scientific community," Perrimon says. "People can go there and do genomewide RNAi screens very effectively."

Perrimon is using the center to screen specific types of cells. For example, some members of his group are identifying genes that assemble the interlaced filaments of proteins that allow muscle cells to expand and contract. "It's like having all the different parts of a house in front of you," Perrimon says. "We want to understand the assembly of those components."

By expressing human genes in the fly, the group is also developing models for diseases such as muscular dystrophy. And they are investigating the interplay between nature and nurture by studying the effects of insulin or diet on muscle gain and loss. "So now we're getting more relevant to humans," Perrimon says. "There are a number of ways in which muscle can atrophy, and we're going to characterize the different pathways that are involved. Muscle is turning out to be a very good system to study with our tools."