The H5N1 ‘bird flu’ virus can evolve to spread through the air between ferrets after picking up at least five mutations, according to a long-awaited study from Ron Fouchier from the Erasmus Medical Center in Rotterdam, the Netherlands1. The paper is published today in Science after months of debate about whether the benefits of the research outweighed the risks.

H5N1 can cause lethal infections in humans but cannot spread effectively from person to person. Fouchier’s paper is the second of two publications describing how the virus could evolve this ability. The first, from Yoshihiro Kawaoka at the University of Wisconsin–Madison, involved a hybrid virus with genes from both H5N1 and the H1N1 strain behind the 2009 pandemic2 (see Mutant flu paper published). By contrast, Fouchier’s mutant is a pure H5N1 virus.

“Coming after the Kawaoka manuscript, I thought it might be anticlimactic, but it does not fail to deliver,” says Vincent Racianello, a virologist at Columbia University in New York.

There is only one common mutation between the combinations that Fouchier and Kawaoka identified. “It just means that there are multiple ways to skin a cat,” says Robert Webster, a virologist at St Jude Children’s Research Hospital in Memphis, Tennessee.

But despite their differences, both sets of mutations confer similar properties upon H5N1. Some mutations allow the haemagglutinin (HA) protein on the virus’s surface to stick to receptors in human upper airways. Others stabilise HA, an unexpected property that now seems important for transmission between mammals. “Two very different strategies led to common functional changes,” says Malik Peiris, a virologist at the University of Hong Kong, who notes that these changes are “not uncommon in field strains already”.

Fouchier now wants to know if these traits could make any flu virus go airborne, and if previous pandemics started this way. “The virus could evolve in a thousand different ways,” he says. “But if we can show that any receptor binding mutations, or any that change the stability of HA, can yield airborne viruses, the story becomes completely different. We can then do surveillance for mutations that cause the same phenotype.”

Fouchier’s team began by adding three mutations to an H5N1 strain isolated from Indonesia in 2005. Two of these – Q222L and G224S – changed the HA gene so that its protein preferentially sticks to human-type receptors over those found in bird cells. The third—E627K—changed the PB2 gene, which is involved in copying the virus’ genetic material, so that the virus can better replicate in the cooler environment of mammalian cells.

The team allowed this mutant virus to evolve naturally by passing it from one ferret to the next, inoculating their noses and collecting samples four days later. After 10 rounds, the virus could spread between ferrets housed in separate cages. These airborne strains spread less effectively than the 2009 H1N1 virus, and are sensitive to current antivirals and potential vaccines. They are lethal if delivered directly to the ferrets’ airways in high doses, but not if the animals catch the infections naturally through the air.

Fouchier’s airborne viruses carried a diverse array of mutations, but all of them shared just five mutations, including the original three that Fouchier inserted and two more in the HA gene—T156A, which affects receptor binding, and H103Y, which stabilises the protein. These five mutations might allow H5N1 to spread between ferrets, but it may take nine or even more. “Until the changes are introduced in combinations into H5N1 virus, we won’t know the answer,” says Racianello.

In a related paper, also published in Science, Derek Smith from the University of Cambridge has shown that wild H5N1 viruses already have many of the mutations that Fouchier and Kawaoka identified3. Some of them are two to four nucleotide substitutions away from the complete sets. However, it is unclear if the same mutations would allow different H5N1 strains to spread between ferrets, let alone humans.

Fouchier now wants to work out the airborne virus’s basic reproduction number, or R0 – the number of individuals who catch the virus for every one who is infected. “R0 needs to be higher than 1 to cause a pandemic,” he says. “We need more quantitative transmission models to start addressing it.”

But Racianello is not convinced that such studies should be prioritised. “I suggest we spend research money developing more flu antivirals… and a universal flu vaccine.”