Injectable Medibots: Programmable DNA could diagnose and treat cancer

Scientists have created a miniature medical computer out of DNA that can detect cancer genes in a test tube and respond by releasing a drug. Proving what had been only a concept, the feat offers a vision of how medicine might look in the future.

A few years ago, Ehud Shapiro and his colleagues at the Weizmann Institute of Science in Rehovot, Israel, developed a molecular computer out of DNA. It was capable of performing simple computations (Math Trek, Foogue Online: Computers by the Trillions). In this biological nanocomputer, strands of DNA serve as software that control the activity of enzymes. The tiny device is listed in the 2004 Guinness Book of World Records as the smallest biological computing device. Trillions of these DNA-based computers could fit into a single drop of water.

Even so, it takes sophisticated lab equipment to extract results from the nanocomputers, so they’re unlikely to outdo silicon-based electronic computers, says Shapiro. That’s why the Israeli team of computer scientists and biochemists pursued a different application: a DNA computer that could by itself diagnose and treat disease.

The researchers programmed their computer for two types of cancer, prostate cancer and a form of lung cancer. For each cancer, the team targeted four genes that become either overactive or underactive in people with the disease.

To detect changes in gene activity, the researchers designed their computer to have three components. The first consists of short strands of DNA, called transition molecules, that bind to a segment of the messenger RNA that each cancer gene produces. For their experiments, the scientists synthesized those segments and put various amounts of them into test tubes to simulate the presence or absence of cancer.

The second component is a computation module made up of a long DNA strand. It contains a series of nodes, each of which participates in a logic operation that determines a diagnosis from the RNA in the test tube. Each operation relies on a series of reactions in which the transition molecules direct an enzyme to cut the module in one place or another. This long DNA strand also harbors the computer’s third component, a therapeutic fragment of DNA that binds to and suppresses the activity of a disease-causing gene.

In a positive diagnosis of malignancy, the computer’s transition molecules detect changes in the activity of all four of a cancer’s genes. When the molecules determine that all four genes have abnormal activities, the enzyme cuts the computation module so that it releases the drug.

However, even if the activity of only one of the four genes is normal, the diagnosis is “not cancerous.” In these cases, the enzyme cuts off a different strand of the computer’s DNA, which neutralizes the drug. If the computer releases the drug by accident, a separate component keeps the system in check by simultaneously releasing the drug suppressor.

The researchers describe their computer in an upcoming Nature.

Ron Weiss, an electrical engineer at Princeton University, rates the work as “an important feat.” He says the research “provides a very nice way of engineering logic behavior [using DNA].”

So far, the researchers have run the computer only in a test tube, but they consider that an amazing step forward. An injectable version would have to work inside cells, and that accomplishment could take decades. Says Shapiro: “I’m not sure it will be within my lifetime.”

Noting people’s fears of freewheeling nanoscale robots circulating in the body, Weiss says, “You have to build safeguards into these systems. Once researchers are able to design reliable DNA computers that make a mistake only once every 10 billion times, say, then I think people will become comfortable with the idea.”

More Stories from Foogue on Tech