Researchers have designed a new type of DNA computer that works in human cells, perhaps paving the way for a distant technology capable of picking out diseased cells from otherwise healthy tissue. The system runs on a process called RNA interference (RNAi) in which small molecules of RNA prevent a gene from producing protein.

The goal is to inject human cells with DNA that can determine whether a cell is cancerous or otherwise diseased, based solely on the mix of molecules inside the cell. Sensing disease, the DNA might trigger a pinpoint dose of treatment in response. That technology, however, is a long way off. For now, researchers are testing different ways of turning DNA into versatile computers that can detect certain combinations of molecules and respond by producing other molecules.

"The central challenge is how do you create a 'molecular computer' capable of making decisions," says bioengineer Yaakov Benenson of Harvard University. Researchers have designed powerful test tube DNA computers that could play tic-tac-toe or perform the basic tasks of logic, but getting them to work in human cells was likely to be tricky, Benenson says.

RNAi is something that cells do naturally. Cells produce what are known as short interfering RNA (siRNA) molecules, which recognize corresponding DNA sequences in genes and cause them to shut down.

Benenson and colleagues engineered a target gene to be sensitive to several different siRNAs of their own design. In the simplest case, they introduced a single siRNA molecule to switch off a target gene that encoded a fluorescent protein. In more complex cases, a pair of siRNAs or either of two siRNAs switched off another target gene, which in turn switched off a gene for a fluorescent protein. To make sure the system worked as intended, the researchers based their siRNAs on those of other species, they report in a paper published online today by Nature Biotechnology.

In principle, the RNAi technique can reach great heights of complexity, Benenson says, by making genes sensitive to more and more siRNAs in various combinations. "The scalability is very important, because eventually you want to make complex decisions," he says.

He says the next step is figuring out how to make the molecules inside a cell—such as those that are overproduced in cancer—trigger the production of siRNAs.