Wired.com Dissection column, March 21, 2008Link
We like to tell ourselves that it's easy to distinguish between the natural and the artificial, but they have a knack for fooling us. When European colonists traveled through the patchwork of forests and meadows of New England, they thought they were exploring primeval nature. In fact, Native Americans had been tending it carefully with fires for centuries. When the Viking probe snapped a fuzzy picture of a mountain on Mars in 1976, some people were sure it showed a giant face carved by Martians. When another probe took a sharper picture in 2001, all trace of the face had vanished.
Today the mystery of the natural versus the artificial is moving from mountains and forests down to the microscopic realm. Scientists can now synthesize DNA from scratch. They regularly add new genes to bacteria, plants and animals. They are learning how to manufacture whole genomes. Can we tell the difference between our growing menageries of engineered organisms and natural ones? A fascinating new study from scientists at Lawrence Livermore National Lab in California shows that we can -- at least for now.
Despite the philosophical nature of their study, the Lawrence Livermore researchers had a very practical goal in mind. They wanted to advance the science of tracing bacteria to their source -- what's sometimes called "microbial forensics." When someone commits bioterrorism -- like the anthrax attacks of 2001 -- it is no simple matter to trace the bacteria to their source. The rise of genetic engineering raises the possibility, remote for now, that someone will unleash even more dangerous plagues. Another potential risk of genetic engineering is that a modified microbe may slip out of a lab and wreak ecological havoc. Should the day ever come when such a disaster does happen, it would be vital to quickly figure out if the cause is man-made. Yet no one has ever demonstrated a systematic way to tell genetically modified bacteria from natural ones.
You might well imagine this was an easy thing to do. Consider the genetically engineered E. coli that produces much of the insulin that diabetics use these days. It makes insulin because scientists have inserted a ring of DNA, called a plasmid, into the microbe. On that plasmid is the human gene for insulin. If scientists were handed a beaker of these weird chimeras, it wouldn't take too long for them to identify the genes and figure out that the bacteria were engineered.
But now imagine a different kind of genetic engineering. Imagine that some scientists decide to make the bacteria that cause bubonic plague easier to spread. Imagine that they manage to do exactly that by adding plasmids carrying a gene from a different pathogen. It would be a lot harder to determine whether this new strain was the work of humans, because different species of bacteria will sometimes naturally swap plasmids.
Some researchers have speculated that it might be possible to tell the difference between natural and artificial life, if scientists added "watermarks" to their engineered DNA. In January, for instance, genome guru Craig Venter and his colleagues made news when they rebuilt a microbe's entire genome. It wasn't a carbon copy of the original, however, because the scientists also inserted small segments of DNA to spell out their names in the genetic code.
There are three problems with watermarking, though. One is that it probably doesn't last very long. Once an engineered strain of bacteria starts breeding, mutations will probably degrade their signatures into gibberish.
Watermarking also suffers from false positives. DARWIN, for example, already exists in lots of genomes of bacteria, fungi, plants and animals. But I'd bet the house that Darwin did not put his name there.
The third and biggest problem is: This process depends on people being nice enough to watermark their handiwork in the first place. Someone who wants to cause harm and not get caught probably can't be counted on for that sort of courtesy.
The Lawrence Livermore scientists decided to use a different strategy. They took advantage of the fact that not just any plasmid will do for genetic engineering. In order to work reliably, plasmids have to be easily sliced open to receive new genes, for example, and they have to be able to move obediently into new hosts. Scientists also like to add genes to vectors that make bacteria resistant to a certain antibiotic. By dousing their colonies with the drug, they can kill off the microbes that didn't take in the vector.
The Lawrence Livermore scientists searched public databases and gathered DNA sequences from 3,799 plasmids currently used for genetic engineering, along with every natural plasmid and every sequenced bacterial genome. The scientists then broke up each set of DNA into short segments and used computers to see whether those segments were distinctive to the vectors. They ultimately succeeded. There are sets of DNA segments measuring just a few dozen base pairs long that are found in almost every known vector and in no natural genomes. The scientists tested these sets on vectors that they hadn't used in their analysis and could identify vectors 98 percent of the time.
Now the scientists are hoping they can use these sets of DNA to build sensors for genetically modified bacteria. They envision a microarray studded with tens of thousands of genetic probes, each able to snag one of the segments they have identified. Scientists might be able to use a Star Trek tricorder-like device to determine whether an outbreak is caused by a natural microbe or an engineered one.
But this week's accomplishments are only temporary. There's a world of natural plasmids out there to be discovered, and some of those plasmids are probably going to be naturally well suited to be vectors. In order for a tricorder to continue to tell the difference between the artificial and the natural, it will need to be constantly updated. Eventually, it may become so easy to synthesize genomes to order that plasmids will become obsolete. No one knows if synthetic genomes will have the same distinctive signature as plasmid vectors. It would be a good idea to find out as soon as possible. The dividing line between natural and artificial is real and important, but science's broom will have to sweep it constantly clean.
Copyright 2008 Carl Zimmer