PHILADELPHIA -- New research from the University of Pennsylvania is beginning to crack the code of which strain of flu will be prevalent in a given year, with major implications for global public health preparedness. 

Joshua Plotkin, the Martin Meyerson Assistant Professor of Interdisciplinary Studies at Penn, conducted the research with Penn post-doctoral fellow Sergey Kryazhimskiy and researchers at McMaster University and the Institute for Information Transmission Problems of the Russian Academy of Sciences.

Plotkin believes that his group’s computational study of 40 years worth of flu genomes offers a new way of looking at mutations: by cataloging pairs of genetic changes that have occurred in rapid succession, observing that a mutation in one half of the pair can act as an early warning sign of a mutation about to occur in the other.    

The findings will be published in the Feb. 17 edition of the open-access journal PLoS Genetics.

The influenza virus is a perfect model for studying rapid evolutionary change.  The virus has a very small genome, and because it infects so many people every year — up to a fifth of the people on Earth — it’s subjected to a tremendous amount of natural selection.  Only the virus strains that have mutated to withstand human antibodies can replicate themselves and return in the next flu season.

But tracking single mutations in a vacuum is not always enough to understand how the flu virus evolves. 

“Sometimes a mutation is functional or adaptive only if it’s in the context of a certain genetic background – that is, if the protein already has some other mutation,” Plotkin said.

The influence such combinations have on an organization’s adaptive fitness is known as epistasis.

Because of the flu’s major relevance to world health, there is a comprehensive historical record of mutations for Plotkin and his colleagues to draw upon.  A statistical analysis of the places on the flu’s genome where these mutations occur revealed strong correlations between pairs of sites.

“If you see a mutation occur in Site A and then very soon after you see a mutation in Site B, and this pattern happens repeatedly, then you have some evidence that A and B influence fitness epistatically,” Plotkin said.  “The first mutation might be useless on its own, but it might be a prerequisite for the second mutation to be useful.  The first mutation is like giving you a nail, and the second one is like giving you a hammer.”

Researchers have long studied epistasis, though they have historically focused on pairs that change more-or-less simultaneously.  Plotkin’s study is groundbreaking in that it examines changes that are separated by one or more generations of the virus. 

Plotkin feels that such knowledge is far more useful in that it can provide foresight.

“If you know that Site A interacts epistatically, and you observe a mutation in Site A this year, then you can predict with some accuracy that Site B will mutate within the next several years.  That’s genuine predictive power at the molecular level of the virus,” he said.  

Because these mutations generally affect the surface proteins that determine whether the virus can enter and infect human cells, being able to predict what mutations are likely to happen in the near future has lifesaving applications.  Tens of thousands of Americans, and hundreds of thousands worldwide, die of seasonal flu complications every year.

Flu vaccine production is labor intensive and time consuming; to have enough supplies ready for the flu season, public health groups like the Centers for Disease Control and the World Health Organization must make an educated guess as to which strain is likely to be the most active several months in advance. Observing the leading site of an epistatic pair could give them a head start.

“If you could understand the mechanisms that are responsible for the types of evolution that occur, you could maybe choose that strain more accurately, meaning you could have a better match every season between the vaccine strain the prevalent strain of flu,” Plotkin said.  “And it allows you to choose vaccine strains not only amongst the ones that have existed up until now, it possibly allows you to predict a future strain that may not even exist in the population today.”

The same method could show also when a pandemic strain, such as the H1N1 swine flu, is about to develop resistance to the one non-vaccine based treatment that has been effective in preventing infection: oseltamivir, or TamiFlu. Epistatically linked mutations in seasonal flu conferred widespread resistance to TamiFlu in 2008, so observing a mutation in the equivalent leading site in the pandemic flu would give public health officials an early warning of this multibillion dollar drug’s impending uselessness. 

But beyond public health applications, this new way of studying epistasis opens the door for a future-oriented approach to genetics. 

“Evolution is viewed as a historical field; we dig up dinosaurs and build phylogenic trees back into the past,” Plotkin said.  “I think that’s interesting, but, in order to understand evolution more fully, we can hope to have predictive power as well.”

The research was done by Plotkin and Kryazhimskiy along with Jonathan Dushoff of the Department of Biology, McMaster University, and Georgii A. Bazykin of the Institute for Information Transmission Problems of the Russian Academy of Sciences.  It was supported in part by funding from the Burroughs Wellcome Fund, David and Lucile Packard Foundation, James S. McDonnell Foundation, Alfred P. Sloan Foundation, Defense Advanced Research Projects Agency and U.S. National Institute of Allergy and Infectious Diseases.

Plotkin holds appointments at Penn in the Department of Biology in the School of Arts and Sciences and the Department of Computer and Information Science in the School of Engineering and Applied Science.