The Wall Street Journal February 28, 2003

DNA's Double Helix Isn't So Golden Now


By Sharon Begley

If only more geneticists were Sherlock Holmes fans. They might find in the triumphs of the caped detective a much-needed epiphany.

As Holmes realized in "The Adventure of Silver Blaze," it was the failure of a watchdog to bark on the night of the murder that provided a solution to the mystery. Geneticists face a similar puzzle. Sure, they can study people who both carry a disease gene and have the disease (breast cancer , melanoma, depression ...). But it might be a lot more illuminating to look at people who carry the gene but never get the disease -- a gene that doesn't bark. For these folks might show the way to prevention and cure.

Last week, I wrote about a nascent revolution in which "systems biology" is overthrowing the reductionist, molecular-biology paradigm that has reigned for half a century, ever since James Watson and Francis Crick discovered that the DNA molecule is a double helix, 50 years ago Friday. In particular, the new approach promises to explain why even the most publicized disease genes fall short of their billing.

Of women who carry mutations in BRCA1 and 2, widely known as breast-cancer and ovarian-cancer genes, for example, 56% to 87% get breast cancer and 28% to 44% get ovarian cancer . Mutations in a gene called p16 lead to melanoma in 76% of those who carry it. The likelihood that a gene will lead to a trait or disease is called its penetrance. Anytime penetrance is less than 100%, something interesting is afoot.

I could go on, but the point is that although geneticists and their media enablers give people the impression that DNA is destiny, it isn't so. Systems biology, which abjures the one-gene-at-a-time paradigm to incorporate all the genes and proteins in a cell, may explain why.

Take penetrance. Biologists ascribe functions to genes based on studies of "inbred organisms with identical genotypes," says Lee Hartwell, who shared the 2001 Nobel Prize in medicine for discoveries in yeast genetics and is now president of the Fred Hutchinson Cancer Research Center in Seattle. "The problem is, things are very different in natural populations."

Sure, you can say Gene X causes diabetes in an extended family, but what you are really saying is that Gene X causes diabetes when it interacts with precisely the genes those people share. Put the gene into 100 people with different genetic backgrounds, and maybe only a few dozen will get the disease.

"This notion the public has been given that we'll genotype a person at birth and tell them their attributes is hogwash," says Dr. Hartwell. "A mutation against one genetic background will produce no effects, while a mutation against a different genetic background will produce a disease."

To figure out how genes interact, biotech start-up Gene Network Sciences of Ithaca, N.Y., has produced what it calls the most detailed model ever of a human cancer cell. Incorporating more than 500 genes and proteins, "it connects processes that have traditionally been studied in isolation," says Colin Hill, GNS's chairman. "In reality these genes and processes are connected, and you can't understand the behavior of the cell without knowing this biological circuitry."

The model shows that "drugs hit more than one target," says Mr. Hill. "That's why you have side effects. We can also show that even when you knock out one gene or protein another can take its place, with the result that the cell doesn't die." That may explain why controversial cancer drugs like Erbitux and Iressa help few patients.

That kind of knowledge could provide a desperately needed boost to drug discovery, which for too long has focused on single targets. As economists say, it's impossible to change only one thing; zap the target that you think causes disease and some other gene or protein might assume its function. Your patient is still sick.

In contrast, a systems-level approach can identify feedbacks that neutralize drugs or cause side effects, says biologist Hiroaki Kitano of the government-funded Kitano Symbiotic Systems Program in Tokyo. As a result, a plethora of systems-biology companies are also drug-discovery companies.

Systems biology isn't only about things going wrong. One of its singular accomplishments shows how things go right. Researchers led by Eric Davidson of the California Institute of Technology in Pasadena have produced a "gene regulatory network" that explains the embryological development of the sea urchin. Incorporating 55 genes so far, it shows that if enough molecules flow through the circuit, they bind to DNA and turn on genes.

With this diagram, you can predict how any tweak will change the sea urchin. Prof. Davidson's group figured out, for instance, how to make a sea urchin develop two guts -- not exactly a holy grail of biology, but a proof of principle: It is a network of genes, not a single "gut" gene, that matters. A comparable wiring diagram for human development might show how to tweak stem cells so they differentiate into any of the hundreds of types of cells in the body.

Systems biology is still in its infancy. But its early successes suggest traditional molecular biology, launched by the discovery of the structure of DNA, has run its course.

Happy 50th Birthday anyway, double helix.

File Date: 03.01.03