The researchers started with a strain called influenza virus A/Indonesia/5/2005, which has infected and killed a number of humans. They first changed three proteins to make it more compatible with mammals' tissues and body temperatures. And then they introduced it into the nose of a ferret.
Ferrets are the best model we have for studying influenza—they are similarly susceptible to it and can spread it by sneezing—and have been used to study the infection since the 1930s. But that does not mean that they are a perfect flu foil for humans. "An H5N1 virus strictly adapted for ferret transmissibility may not be entirely relevant to humans," Fauci and Collins wrote in their essay.
The experimental ferrets Fouchier used did not pass the disease directly to one another, as would occur among humans in a real-world outbreak. Instead, during the first six of the 10 transmissions, researchers dropped the strain into the ferret's nose and then later swabbed the nasal area to collect the virus—with any new mutations—to be given to the next ferret. In the final few transmissions, instead of swabbing the nose, researchers made each animal sneeze, then collected the contents and placed them directly in the nose of the next ferret. Follow-up experiments showed that when an uninfected ferret was placed near one of these infected subjects—sharing air but not able to make physical contact—the uninfected ferret often caught the illness (looking tired and not eating much) but did not die from it.
As Fouchier and his co-authors noted, none of the ferrets that received the newly transmissible virus died from the infection. (And only one in eight of the animals given an earlier form of the virus, via the nose inoculation, died from the illness.) This shift in virulence mirrors the pattern that has often played out in human pandemics: as a virus becomes more easily transmissible, it becomes less deadly. In the experiments, the viruses were also treatable with common antiviral drugs.
What were the mutations that made the virus able to infect other ferrets just by breathing the same air? Four of the five mutations changed the virus's surface protein (hemagglutinin), which plays a role in allowing H5N1 to gain entry into host cells. And the fifth mutation boosted its ability to reproduce its genetic information.
Two of the mutations identified by Fouchier are already quite common in wild strains of H5N1 in birds and occasionally show up in the same strain, according to Smith's study of some 4,000 strains. As Smith and his colleagues pointed out in their paper, that means that some strains might need only three more mutations to become easily contagious among mammals. These three other mutations are rare in H5 strains (although two of them were present in the H2 and H3 pandemics of 1957 and 1968, respectively), so they are likely harmful to the virus while it resides in its bird host. Smith and his team created mathematical models to try to figure out how likely it was that these three mutations could occur in a human host. But without further research, they could only speculate that it was possible, they noted.
Trying to assess the likelihood of an outbreak of this virus in humans is about as easy as "predict[ing] an earthquake of tsunami," Smith said at the press briefing. "We now know that we're living on a fault line," thanks to Fouchier's research. "And it's an active fault line."
That prediction problem, however, suggests that much more research needs to be done on just how these viruses mutate within hosts. They also suggest increasing surveillance to do deeper genetic sequencing of circulating strains to look for less common mutations that might increase the odds the virus will jump to a new species host.