Such devices could one day allow humans to manipulate biological systems directly, said the California Institute of Technology's Christina Smolke, who co-authored the study, which will be published tomorrow in the journal Science.
Bio-computers might eventually serve as brains for producing biofuels from cells, for example, or to control "smart drugs" that medicate only under certain conditions.
For example, a smart drug could sample a cellular environment and trigger a self-destruct sequence if disease is detected, Smolke said.
The new bio-computer consists of snippets of engineered RNA assembled inside a yeast cell.
RNA is a biological molecule similar to DNA, which encodes genetic information, such as how to make various proteins.
In engineering terms, the bio-computer's "inputs" are molecules floating around inside the cell. The "output" manifests as changes in protein production.
For example, an RNA computer may be able to bind with two different molecules. If both target molecules attach to it, they trigger the device to change shape.
The altered bio-computer is now the right shape to bind to DNA, where it can directly affect gene expression and ramp up or slow down the making of desired proteins.
Those proteins can affect the cell in various ways, such as killing it if it is cancerous.
The team designed the different parts of their RNA computer to be modular, so the pieces can be mixed and matched.
"Depending on the combination that we put together, we'll get different," Smolke said.
Nature tends to evolve specialized molecules that perform single functions extremely well.
Creating a few interchangeable components to carry out multiple functions is a different, but highly effective, design approach, the researchers say.
Not for Wii
Most scientists agree that biological computers are unlikely to surpass or even match the power and efficiency of electronic ones.
"[They're] not to run Microsoft Windows any faster or run your latest game of Wii," commented Ron Weiss, an electrical engineer and molecular biologist at Princeton University in New Jersey.
Rather, the power lies in bio-computers' potential ability to patch into and directly interact with, cellular processes.
"It's basically speaking the language of the cells," said Weiss, who was not involved with the study.
Weiss said the new finding "pushes the boundary" of what can be done with biological computers.
"Previously [RNA computers] weren't quite as complex," Weiss said, adding that Smolke "is expanding the library of things that are available."
Ehud Shapiro is a computer scientist and biologist at the Weizmann Institute of Science in Israel. He was not involved in the study.
Shapiro's team had previously created a bio-computer using DNA that worked inside a test tube and could perform simple calculations, such as determining whether a list of zeros and ones contained an even number of ones.
But unlike the new RNA computer, Shapiro's test tube bio-computer was "oblivious" to its surroundings and could not interact with or be affected by its environs in any meaningful way, he said.
"The work of Smolke shows a computer that can respond to molecules inside a cell," said Shapiro, who wrote a review of the new study for Science.
Shapiro looks forward to a day when RNA computers are replaced by more sophisticated devices made from proteins.
"Proteins are the most efficient natural devices we know of," he said. "We know how to evolve RNA to do simple tasks, but do not know yet how to engineer proteins." NG