This story starts in 1987, with the skin of a frog.

Michael Zasloff, a scientist then at NIH, was impressed by how well a frog in his lab recovered from an incision he had made in its skin during an experiment. He kept his frogs in a tank that must have been rife with bacteria that should have turned the incision into a deadly maw of infection. Zasloff wondered if something in the skin of the frog was blocking the bacteria. After months of searching, he found it. The frogs produced an antibiotic radically unlike the sort that doctors prescribed their patients.

Most antibiotics kill bacteria by jamming up their enzymes. The bacteria can no longer copy its DNA or expand its membrane as it grows or do some other task essential to their survival, and they die. Zasloff and his colleagues figured out that the antibiotics in frog skin worked entirely differently. These small molecules were attracted to the positive charge on the surface of many species of bacteria. Once they stuck to the membrane, the frog molecules changed shape, so that they punched a hole through the membrane. The bacteria's innards spilled out of the hole, leading to their death.

The antibiotics from frog skin proved to be just a tiny sampling of a huge natural pharmacy. Antimicrobial peptides, as these molecules are known, can be found in all manner of animals. We humans make a lot of them, both on our skin and in the lining of our guts and lungs. One reason that cystic fibrosis is so devastating seems to be that it monkeys with our ability to make antimicrobial peptides in our lungs. The microbes that swarm into the unprotected tissue cause the lungs to become inflamed, loading them with fluids. Many of the antimicrobial peptides found in one species are not produced by any other animal, and yet they are all remarkably lethal to bacteria.

Zasloff recognized a promising opportunity for inventing a new drug. The science of antibiotics hasn't moved forward much since the 1940s, when penicillin and other drugs were first introduced. These antibiotics, most of which were produced by fungi or bacteria, were miracle drugs at first. They can still clear up all sorts of infections in a manner of days--provided you're infected with a vulnerable strain of bacteria. Within a few years, every antibiotic that has been put on the market has triggered the evolution of resistance. Some bacteria acquire the ability to pump the drugs out, others to change the shape of their enzymes to make them harder to grab, and others do all sorts of other remarkable evolutionary tricks.

Before the antibiotic era, the mutations that help make bacteria resistant to drugs didn't bring a big benefit. In fact, they may have had nasty side-effects, slowing down the growth rate of microbes. As a result, they remained rare. But once bacteria began regularly to face these drugs, the evolutionary balance tipped. People often don't take enough antibiotics to wipe out their infections, allowing bacteria with a little resistance to survive and acquire new mutations. People sick with viruses regularly get antibiotics, even though the treatment is useless. Bacteria also encounter antibiotics in livestock, which get loaded with antibiotics to grow faster. Resistance genes can spread as microbes reproduce, and can get traded between different species. The situation has gotten so bad that scientists are now warning surgeons may soon be operating in conditions not seen since the Civil War, unable to stop bacteria that get into open wounds.

The secret of frog skin promised a solution to this disaster. A drug based on antimicrobial peptides might be able to wipe out bacteria that had evolved resistance to other drugs. And even more exciting was the possibility that these new antibiotics might be resistance-proof. Bacteria might theoretically able to evolve resistance to antimicrobial peptides by changing the charge on their surface so that the molecules wouldn't be attracted. But that wouldn't be just a tweak to an enzyme or some other series of small changes: it would be a fundamental alteration of the beast. Experiments seemed to back up this hunch. Some scientists tried to produce resistant bacteria by randomly mutating their genes and then seeing whether any mutants could survive a dose of antimicrobial peptides. No luck.

But a Canadian evolutionary biologist named Graham Bell suspected that bacteria--and their evolutionary potential--might be more powerful than others thought. Michael Zasloff for one didn't think so. But as a good scientist, he was willing to put his hypothesis to the test. Remarkably, it failed.

The researchers began by exposing bacteria to low levels of antimicrobial peptides. They would then use a few of the survivors to start a new colony and then expose the bacteria to slightly higher levels of the poison. As they report in the Proceedings of the Royal Society of London, 30 out of 32 colonies evolved to be resistant to a full does of antimicrobial peptides. It took only about 600 generations for them to do the impossible.

The new paper doesn't offer any evidence for what the evolved bacteria are doing to escape antimicrobial peptides. It is hard to pinpoint mutations that produce new traits, and even harder to figure out exactly how they change the workings of a microbe. So we may have to wait to learn the trick that bacteria have discovered. But the results are enough to raise serious concerns about the future of antimicrobial peptides. People who take full doses of the drugs might wipe out all the bacteria infecting them, but microbes that are exposed to low levels--in people who don't take full prescriptions, in animals, or even in the environment--could evolve resistance. As the bacteria became stronger, they would be able withstand higher doses. They might gradually invade a new ecological niche: the world of full-strength antimicrobial peptides.

The new research shows yet again that it's pointless to rely on personal incredulity to understand the workings of evolution, despite what some creationists may claim. But it also reveals a paradox: if resistance can emerge so easily, why are bacteria susceptible to antimicrobial peptides in nature? Clues to the answer lie in the evolutionary history of the peptides themselves. Scientists have compared peptides to figure out how they evolved from common ancestors. The peptides have been evolving at high speed for millions of years. Ancient genes were accidentally duplicated, and mutated so that they produced molecules with different structures. But different parts of the genes evolved at different rates. Each gene for an antimicrobial peptide contains a signal sequence that acts like a mailing label: once the DNA code of the gene is translated into a protein, the signal sequence tells a cell where the protein should go. The signal sequence in antimicrobial peptides barely changes over hundreds of millions of years--presumably because all of these molecules need to go to the same place, out of the cell. But the portion of the gene that codes the bacteria-fighting end of the protein has changed drastically over time. In fact, some research even suggests that this part of the gene is more prone to mutating than typical DNA.

This pattern suggests that antimicrobial peptides are effective only if they are continually reinvented. When a population of frogs begins to colonize a new habitat, for example, they may encounter new microbes. Their old antimicrobial peptides may not be very effective against these pathogens, but they can rapidly evolve new ones. But even if animals stay put, they may evolve new peptides. That's because pathogens can evolve the ability to knock out these weapons. Certain bacteria, for example, can produce enzymes that neutralize antimicrobial peptides. Hosts that can evolve new peptides that can't be knocked out so easily may be more likely to survive. And so the arms race continues.

There's another lesson for the drug industry in the evolution of these molecules. Not only have animals repeatedly experimented with new versions, but they never rely on just one. Each species may produce ten different kinds of antimicrobial peptides, and the molecules are often most effective in combination (for reasons scientists don't yet understand). By using a range of different peptides at once, animals may thwart the evolution of resistance, because bacteria never get intensely exposed to a single drug.

It would be absurd to model man-made antimicrobial peptides too closely on natural ones. After all, natural selection produces remarkable antibiotics only through the different levels of success of different genes. Some animals die, in other words, and some don't. But it does offer some guidance. Just because microbes can evolve resistance to antimicrobial peptides doesn't necessarily mean they will if these drugs enter the marketplace. If doctors use them sparingly, combine several kinds of antimicrobial peptides, and continue to invest in new versions (like this extremely powerful one from a mushroom reported by Zasloff in October), they may be able to stay one step ahead of the bacteria. We just need to face evolution with our eyes open.

Update: link to Bell and Zasloff paper fixed, I hope.