They say the need to understand how flu viruses cause lethal pandemics outweighs any safety risks. But the risks may not be negligible.

By painstakingly piecing together viral fragments from hospital specimens and a victim buried in Alaskan permafrost, Jeff Taubenberger and colleagues at the US Armed Forces Institute of Pathology in Rockville, Maryland, have now sequenced all eight coding regions of the 1918 flu virus’s genome. They published the last three - coding for the polymerase complex that allows the virus to replicate - on Wednesday (Nature DOI: 10.1038/nature04230).

Meanwhile, Terrence Tumpey at the US Centers for Disease Control in Atlanta and colleagues used the sequences to rebuild the virus itself, and infect mice with it. They report this week that unlike other flu viruses, 1918 does not need a protein-splitting enzyme from its surroundings to replicate, instead using some hitherto-unknown mechanism. And as in 1918, it rapidly destroys lungs (Science, vol 310, p 77).

Deadly genes

By replacing the genes for either the 1918 virus’s surface protein, haemagglutinin (HA), or for its polymerase complex, with genes from a tamer, related flu, the team showed that both contribute to its deadliness, with the HA especially damaging for lungs.

“This work will help us make vaccines and antivirals against pandemic strains,” says Tumpey. It is unclear how, as the next pandemic is likely to be a different kind of flu.

But, says Taubenberger, the 1918 sequences are already helping in another way: they prove that a bird flu can go pandemic without combining with a human flu, and suggest which mutations it needs. The most likely pandemic candidate, H5N1 bird flu, already has some of the mutations. We should watch out for more, he warns.

The CDC handled the 1918 virus at only the second-highest level of biological containment, in which lab workers wear breathing hoods but not completely enclosed “spacesuits”.

Existing immunity

If such a virulent virus escaped it could cause serious illness and death. But CDC director Julie Gerberding says that in the event of an accidental release, drugs, vaccines and people’s existing immunity would limit the risk.

Most people have been infected with milder flu viruses related to the 1918 strain. Gerberding says that they, and mice vaccinated against these viruses, have antibodies that give “cross-protection” to 1918 flu. Similar prior exposure to previous flu viruses protected some people in 1918. But, she admits, “we don’t know to what extent this would protect people from harm”.

Gerberding told New Scientist that the highest level of containment is for pathogens for which there are no drugs and vaccines, which is not true of 1918 flu.

But while a specific 1918 vaccine is possible, none is now available, and Tumpey says no vaccine has been tested against the resurrected virus. Likewise, he says, while the 1918 virus has no mutations known to confer resistance to antiviral drugs, these drugs have not been tested on infected mice.

H5N1 also has no mutations giving resistance to the antiviral oseltamivir. But in studies published in July 2005, even high, prolonged doses of oseltamivir saved only 80 per cent of infected mice. If the drug has similar efficacy against 1918 flu, it might not stop the virus from spreading further if it escaped.