A large stockpile of smallpox vaccine recently was rediscovered by a large pharmaceutical company. It had been left over, deep frozen a quarter-century ago when the last smallpox case was eradicated from the earth. Now, in the shadow of bioterrorism, those resurrected vaccines are a welcome trove for potential homeland protection.

Most viral vaccines, like those for smallpox, have a long, undetermined shelf life. Influenza is a strikingly sinister exception. It leads every winter in mid-February to a meeting of flu vaccinologists at World Health Organization headquarters in Geneva.

Their counsel of war must second-guess the annual quick-change artistry by which the virus keeps a jump ahead of the human immune system. “The typical changes,” observed mathematical biologist Joshua Plotkin at Princeton University, “are the replacement of one amino acid with another by point mutation. The question is,” he added, “Where are they likely to happen the next year? Over the past 30 years we’ve see the point mutations occur in most of one viral gene. And it’s not clear which ones are going to be the active epitopes [vaccine targets] the following season.

“There’s a huge variety among all the different viral pathogens,” Plotkin continued, “the speed at which their strain mutations evolve. Roughly every two to five years, influenza virus changes strains significantly. Many classical diseases don’t evolve at all. The reason they survive is that they infect fewer people, so don’t drive themselves to extinction. But flu is so infective that it needs to evolve in order to escape the immune antibodies it provokes.

“Human immune deficiency virus is on the other end of the spectrum,” Plotkin pointed out. “HIV is even faster than influenza at switching strains. In the course of one infection in one individual person, HIV evolves at a dramatic rate. So on the spectrum of viruses, ranging from those that are barely mutating at all to those changing on a time scale like flu, far upstream we have HIV. The main reason there is still no HIV vaccine is that there’s no such thing as a fixed HIV sequence. It changes so quickly.”

Clustering Gene Sequences Into Swarms

Plotkin, a graduate student at Princeton, is lead author of a paper in today’s Proceedings of the National Academy of Sciences (PNAS), dated April 30, 2002, but released online April 23. Its title: “Hemagglutinin sequence clusters and the antigenic evolution of influenza A virus.”

“Our principal finding,” Plotkin told BioWorld Today, “is that using the sequence databases of hemagglutinin RNA nucleotides, we found that we can cluster these sequences on the basis of how similar they are to one another genetically. Once clustered, these groups of sequences can be interpreted as swarms of related genotypes, which are essentially strains of influenza. We could see in this database one strain replacing another every two to five years, and that on the basis of all of the sequences collected up into a given year, once we clustered those sequences we have a fairly good shot at predicting the kind of strains that will be present the following year, when vaccines must be chosen.

“That’s the essence of the PNAS paper,” Plotkin observed, “to look at these sequences of RNA, which had been historically examined either from a phylogenetic [evolutionary tree] viewpoint, or using some molecular assay. What we did that’s new was to look at them in clustered sequences, sort them into groups in a mathematical way, then use those groups to understand and to some extent predict the flu virus’s evolution.

“The hemagglutinin gene we looked at is only one of several this virus possesses. It’s the most important gene for viral interaction with the immune system. So it’s the gene that is most important for a vaccine to be calibrated to. Within that gene,” Plotkin explained, “there are five main epitopes, and those are the parts of the gene that the immune system sees in particular. They are the regions of the sequence that change the most, which we identify as the glue that holds the cluster together.

“Physically, the five epitopes are five regions located on the surface of the hemagglutinin protein, which the human immune system has the ability to bind to and then neutralize. From year to year, gradual mutations continue to change just slightly the physical structure of these five regions. And those alterations are the reason why this vaccine has to be updated yearly, because once it changes one or two epitopes, the virus has moved sufficiently far that last year’s old vaccines will be ineffective.”

Flu Virus Evolution Pattern Recognition

“What we actually did,” he recounted, “was take the database, which has 560 sequences in it, downloaded them, and assigned a distance between every pair of sequences. These were simply the point mutations, which would be required to move from one sequence to another. And we found the exciting fact that we’d clustered these sequences purely on the basis of their RNA nucleotides, but the clusters are localized in time. You see one cluster rise up to prominence and slowly be replaced by the next cluster. And that basic pattern occurs over and over; it’s really the pattern of influenza evolution.

“Once a cluster of viruses has come to prominence and infected a lot of the human population,” Plotkin went on, “the viruses themselves stimulate immunity in the whole population they are infecting. And after a period of time, those viral strains actually drive themselves to extinction by stimulating immunity to themselves. They’re making room for new variants to enter the pool of circulating flu strains. And this happens on a time scale of every two to five years.”

Then Plotkin and his team applied mathematical analysis metrics as notions of distance between two sequences and algorithms for clustering genetic sequences into distinct groups.

“A big component of our future work, and something we’re actively pursuing, is more empirical how can we actually help inform vaccine choices on a year-to-year basis. What we’ve found,” he concluded, “is certainly not the best answer to what is the best vaccine, but what is a useful algorithm to help in choosing the vaccine every year.”