Ingmar Bergman’s famous 1957 movie The Seventh Seal takes place during the 14th century, when Europe is in the midst of a major epidemic of the bubonic plague—the Black Death—which ultimately killed about half the population. A Swedish knight, Antonius Block, returns from the Crusades and finds Death waiting for him. He challenges Death, later seen disguised as a priest, to a chess match, hoping to stave off his own death by devising what he hopes is a winning next move.

For the past three decades, researchers and health workers have engaged in a similar battle against one of the most cunning viruses to afflict humanity and much of the animal world: the dread influenza virus. This pathogen is even smarter than Death; it continuously changes the appearance of its chess pawns—the proteins on its coat—so that im­-mune systems do not recognize the new disguise.

Every year the World Health Organization and other institutions try to predict the next change in the virus’s coat. Once the WHO decides on the likeliest alterations, drug manufacturers then have only a few months to develop vaccines. “The whole infrastructure required for the preparation of seasonal vaccines has enormous disadvantages,” remarks Walter Fiers, a molecular biologist at Ghent University in Belgium. “It is slow—sometimes we miss the strain that becomes predominant—and if a pandemic should arrive, we will not be prepared.” Fiers’s goal: a universal vaccine that, like some childhood immunizations, would confer lifelong immunity.

Scientists have dreamed for decades of a one-shot approach to stop the flu—particularly influenza A, the most serious type. But the task is daunting. The appearance-changing coat of the influenza virus is studded with mainly two proteins: hemagglutinin, which allows the virus to attach to and enter a cell; and neuraminidase, which boosts the virus’s ability to pass to other cells. (These proteins serve as the basis for influenza nomenclature; for instance, the H5N1 virus refers to specific classes of hemagglutinin and neura­minidase, which in this example correspond to an avian flu subtype.) The genes responsible for these proteins undergo frequent point mutations, resulting in genetic “drift”; moreover, the genes from different animal and human strains may also interchange, resulting in genetic “shift.” Both drift and shift make these proteins unrecognizable to the antibodies present in people that were previously inoculated against the flu virus, which now circulates as more than 90 strains.

Unlike the hapless knight Block, the 77-year-old Fiers believes that he has found his adversary’s Achilles’ heel: although the virus is good at disguising its pawns, there is one on its coat that it cannot change. That pawn, the external part of a protein called M2, should be the target for vaccination, he says.

Fiers has come to this conclusion after five decades of work in molecular biology—in particular, decoding genomes. In 1972 he and his team were the first to publish the nucleotide sequence of a complete gene. This gene codes for the coat protein of a bacteria-infecting virus, or bacteriophage. Four years later they published the bacteriophage’s complete genome—all four genes of it. “This was the first complete genome that was sequenced,” Fiers recalls. Because of its medical importance, he decided around that time to focus on the influenza virus.

In 1980 Fiers first sequenced the gene for hemagglutinin derived from the human influenza strain H3N2 that circulated in 1965. He then compared the hema­g­glut­inin gene of this strain with a similar hem­ag­glut­inin gene derived from the strain that had started the 1968 Hong Kong pandemic. His analysis proved that point mutations account for genetic drift.

Equally important, his studies led him to see how the virus can jump species through genetic shifts. At that time scientists knew that antibodies from people infected by the 1968 pandemic virus also reacted with an influenza strain isolated in 1963 from flu-ridden ducks. Fiers investigated the nucleotide sequence of the hemagglutinin gene of this duck virus and found that it was indeed very closely related to the strain that started the 1968 Hong Kong outbreak. Today infectious disease experts recognize that an avian flu virus could genetically change enough to trigger a human pandemic.

If such a jump occurred, the virus “would not be hindered by any preexisting immunity in the human population,” Fiers explains. “It would spread fast over the world because when it arrived here it would look like an entirely different virus. Therefore, you need a vaccine that is not invalidated by drift or shift.”

More specifically, he needed a vaccine based on a part of the influenza virus that does not change. Fiers found it in the viral coat protein M2, which creates a pore in the coat. Specifically, he noticed that a section of that protein, called M2e, remains stable even as the other viral surface proteins mutate. The M2 protein, however, occurs only in small numbers on the virus, which is too low to set off a good immune response.

The obvious solution was to amplify the number of M2e segments. But how? Fiers turned to the liver-attacking hepatitis B virus. This pathogen has an inner protein core called HBc, and something intriguing happens when the gene for HBc is inserted into the bacterium Escherichia coli. The bacterium starts producing HBc proteins and assembles them to produce viruslike particles. Fiers found that by linking M2e genes to HBc genes, the bacteria would produce viruslike particles studded with M2e.

In tests with mice and, later, ferrets, the M2e-HBc particles caused the formation of antibodies directed against M2e, thereby protecting the animals from a lethal dose of influenza. The vaccine works differently from conventional vaccines in that it does not prevent infection directly. “The target is not the virus, but the target is the virus-infected cell,” Fiers says. “If at an early stage, you can kill off these cells, then you will counteract the infection.”

In 1997 Fiers received a patent for the technology, and in 1999 he published a paper in Nature Medicine expounding his approach. The British-American company Acambis, based in Cambridge, Mass., and Cambridge, England, has obtained the license to start production of the vaccine. In a phase I trial completed last year, Acam­bis found that among the 79 volunteers who received the vaccine, 90 percent developed antibodies to the M2e segment.

Whether the antibodies protect against influenza now needs to be determined—no guarantee if the past is any guide. A decade ago a drug based on an internal protein of the flu virus, called NP (for nucleoprotein), set the immune system’s killer T cells into action, but it only partially protected mice from the flu.

Because intentionally infecting a volunteer to see if a vaccine works is unethical, the compound will have to face large field trials. “We have to find an area where there is a higher probability that an influenza epidemic will take place,” Fiers says. The likeliest places are where dense human populations live near farm animals. Thousands of people will have to be vaccinated to obtain statistically acceptable results. (The current version of the M2 vaccine would protect only against influenza A, the type that has launched pandemics.)

The pharmaceutical industry is showing strong interest, and Arnold Monto, an epidemiologist at the University of Michigan at Ann Arbor, thinks that this universal vaccine is promising. Still, “whether it will be sufficient without having other proteins in there, I’m not sure,” Monto says, referring to a possible need to use seasonal flu vaccines as supplements. The chess battle against influenza will have to continue for a while longer, but for now hopes are high that playing the M2 vaccine might be the right strategy to checkmate the virus.