A highly controversial research paper on bird flu was released today by the journal Nature. It shows that a particularly troublesome strain of avian influenza, designated H5N1, which has been worrying public health officials for more than a decade, has the potential to become a human pandemic. In other words, H5N1 bird flu, which so far has been highly lethal to humans but has not acquired the ability to spread easily among us, could do so at any time.

The researchers, under the direction of Yoshihiro Kawaoka at the University of Wisconsin at Madison, crossed an H5N1 virus with the H1N1 pandemic virus of 2009, which spread like wildfire from one end of the world to another. The 2009 pandemic, you'll recall, caught public health officials by surprise but luckily turned out to be mild. Kawaoka's lab-made hybrid virus spreads among ferrets by airborne droplets expelled during the course of respiration--just as human influenza viruses such as the 2009 pandemic strain spread from person to person. Kawaoka's concoction does not kill ferrets, and probably wouldn't kill humans, but the feat is troubling because it demonstrates that an H5N1 virus that can spread among humans is most likely possible. (We don't know for sure because it was tested only on ferrets, not humans, of course.)

Whether an H5N1 virus could acquire the two deadly traits--transmissibility and lethality--at the same time is a burning question. (Kawaoka's paper does not address it.) If one could, it would be bad news indeed. The 1918 flu virus, an H1N1 type, killed about 2 or 2.5 percent of the people it infected but spread so readily the death toll reached 20 million to 50 million. By contrast, about 60 percent of the 600 or so people who have caught H5N1 in the past few years have died. A highly transmissible H5N1 virus with a 60 percent fatality rate could kill hundreds of millions of people, perhaps billions.

That's not likely to happen, fortunately. The known H5N1 viruses that currently exist in nature do not attach to the upper respiratory tracts of humans, so most victims caught the virus from close contact with birds. And there is reason to think that a highly lethal but poorly transmissible virus like H5N1 would change in the process of becoming transmissible—that in acquiring the ability to spread, the virus would have to make genetic tradeoffs that compromise its ability to kill. For instance, one reason that the Ebola virus doesn’t spread widely among humans is that it is too efficient—mortality is as high as 90 percent—which means victims die before they can infect many others. We do not know, however, to what extent a human-transmissible H5N1 virus would have to make this tradeoff. Scientists differ widely in their opinions on this point.

The strain that Kawaoka concocted in his lab seems to be mild—it made ferrets sick, but didn’t kill them. Even if his virus is not itself dangerous, however, it demonstrates that an H5N1 strain could one day arise that turns into a worrisome human pathogen. As Michael Osterholm, director of the Center for Infectious Disease Research and Policy at the University of Minnesota, has pointed out, a human-transmissible virus with a lethality rate 20 times lower than current wild strains of H5N1 would still kill more efficiently than the 1918 flu. The degree to which H5N1 is potentially dangerous to humans is the subject of much speculation.

In part because of all the uncertainty, the biosecurity community has been in an uproar ever since the Kawaoka paper came to their attention last year. (Things really got stirred up when Ron Fouchier of Erasmus Medical Center in Rotterdam made remarks at a conference in Malta about similar work on H5N1 transmissibility, which he submitted to Science but has been held up by the Dutch government.) Many security experts believe no work on H5N1 transmissibility among mammals should have been conducted in the first place, and that the world would be a safer place if information about the results of the Kawaoka and Fouchier experiments and their methods were never disseminated. (Although the National Science Advisory Board for Biosecurity voted to green-light the publication of both papers, that vote came only after the results and the methods used to make the viruses had been disseminated to hundreds of people in the course of the standard pre-publication review process; holding up publication at that point, in this age of Wikileaks, probably would have been futile.) Recently the U.S. government has called for risk-benefit assessments of pathogens on its select agent list before research is conducted (H5N1 is on that list), but compliance is left to the funding agencies. Some biosecurity experts are calling for further restrictions in H5N1 research on transmissibility among mammals, such as limiting the work to labs with the highest biosafety standards.

Many scientists believe this would be a mistake. They think that research on what could possibly make H5N1 deadly to humans is too urgent to be put under onerous security constraints. I interviewed Kawaoka in 2010, during the course of research for my book, The Fate Of the Species (which Bloomsbury releases on May 22). At the time, Kawaokoa was deeply worried about the danger of H5N1 bird flu. There had been outbreaks among poultry in 1997 and again in 2004 in Hong Kong, and the virus had shown other signs of restlessness—it had killed wild birds, which usually carry H5N1 without symptoms, and it had shown a propensity to spread outward from east Asia, despite the culling of millions of chickens and other poultry. How long would it be before nature's roulette wheel produced a human version of H5N1?

Kawaoka got a taste of how nasty bird flu can be when he arrived in the United States in 1983 as a young researcher in the lab of Robert Webster at St. Jude Children’s Hospital in Memphis, Tennessee. Webster was investigating the theory that birds provide a reservoir for human influenza viruses (which has since been confirmed and is now accepted as conventional wisdom). Webster got wind of an outbreak of avian influenza on poultry farms in Pennsylvania and raced to the scene, sending tissue samples back to Kawaoka, who analyzed them in the lab.

The 1983 demonstrated to the young Kawaoka how changeable and dynamic a flu virus can be. The outbreak started out in April 1983 as a mild virus—it make chickens a little wheezy, but didn’t kill many of them. By November, the virus (an H5N2 type) had given rise to a sub-strain that was highly lethal. Chickens started keeling over, and farmers worried for their livelihoods. The U.S. Department of Agriculture came in to handle the outbreak, imposing a military-style culling of 17 million birds from Pennsylvania to Virginia. (No people died in the outbreak, just poultry.)

While all this was going on, Kawaoka, back in Memphis, isolated samples of the virus and infected chickens in the lab to see how the virus worked. The mortality rate among his chickens was 100 percent. When he performed autopsies, he found that the virus had ravaged not just the birds’ gastrointestinal tracts, which you’d expect in a bird flu, but almost every organ—the kind of total attack documented in human victims of Ebola.

Later, Kawaoka decided to compare the April strain and the November one to find out what had changed to make the virus suddenly lethal. He found that the two were identical except for a single protein. “What this tells you,” Kawaoka told me in 2010, “is that a highly pathogenic virus was generated from a single mutation. And it tells you there are many sources of highly pathogenic influenza viruses. It’s all out there in birds.”

Researchers have already found natural strains of H5N1 that have acquired some of the genetic changes that Kawaoka’s recent paper identifies as necessary for transmissibility among ferrets. The research that Kawaoka and other influenza scientists are conducting certainly poses a risk, as some biosecurity experts have pointed out. The experiment that nature is conducting every day, as H5N1 viruses mutate and borrow genetic material from other viruses, is also potentially dangerous. Which should we fear most? On this question there are many opinions, but no definitive answers.