Manmade H5N1 Paper Finally Published, And Worrying

Posted on by Laurie Garrett

One of the two notorious manmade flu virus papers that spawned massive concerns regarding biosecurity was finally published today in Nature magazine. Extensively edited and rewritten by U.S. government reviewers, a vast number of scientific peers and the editors of Nature, the paper by University of Wisconsin's Dr. Yoshihiro Kawaoka offers genuine grounds for concern.

At the top of list of worries is Kawaoka's description of airborne spread between caged ferrets of his genetically altered flu virus noting, "this transmission pattern is comparable to that of the 1918 pandemic virus when tested under the same experimental conditions..."

As worrying as the Kawaoka findings are, the more serious biosecurity concerns that sparked a chain of controversial U.S. and Dutch government responses are believed to be contained in the other, as yet unpublished paper by Erasmus University scientist Ron Fouchier. We await its publication in Science, possibly quite soon given the Dutch government has just lifted all its restrictions on release of the information.

Even more worrying: what comes next? What are the likely next step experiments? Why can’t U.S. and European authorities find ways to put the proverbial genie back in the bottle?


Yoshi Kawaoka has been in the influenza virology game for decades, operating enormous laboratories in Madison and three other locations in Japan. As one rival virologist put it to me, “Kawaoka runs a flu machine,” that can address virtually any question one might wish to pose about influenza. Not surprisingly, then, his lengthy paper is a carefully edited series of questions and answers. Though it reads like a committee report, reflecting some nine months of editing and controversy, the work is very clear.

Kawaoka started by asking what mutations might be necessary to turn an avian flu into one that can spread among human beings via coughing, sneezing, handshakes and the other casual methods that typically make flu epidemics rapid global phenomena. He began with the assumption, based on a large body of prior work, that the key lay with the haemagglutinin (HA) proteins of the virus, which are bulbs that sprout out all over the surface of the round RNA virus and play a crucial role in gaining the microbe entry into target cells. A second assumption concerned sialic acid, a compound on cells that the HA flu proteins attach to in the infection process. Typically, forms of flu that only infect birds have a section of HA that is dubbed Sia-alpha2,3Gal. But the equivalent in mammalian cell-targeting flu is called Sia-alpha2,6Gal. The names can get confusing, but the point is rather simple – there is a minor difference in the avian versus mammalian-infectious forms of HA, and Kawaoka made loads of deliberately mutated viruses with changes in their codes aimed at this specific difference. He snipped off the HA “bulbs” of a strain of avian H5N1 flu found in Vietnam, removed genes that coded for cell killing or virulence, and manufactured loads of manmade viruses that combined the nownonvirulent H5N1 bulbs with the mildly virulent swine flu H1N1 remaining viral segments. He used the 2009 swine flu as his base both because it was a fairly benign virus to which most of humanity is now immune, and it was infectious to humans.

Crucial point: Kawaoka started his question/answer process by deliberately emasculating the virus, with the goal of producing a nonlethal but highly infectious germ. Because he removed virulence genes, and linked his H5N1 segments to the mild swine flu that most of us caught back in 2009, Kawaoka felt his first questions could be experimentally addressed in low security facilities – BSL-2 (BioSafety Level-2).

After making 370 synthetic viruses, Kawaoka’s team identified nine that could lock onto the correct form of sialic receptors on human/mammalian cells. Out of a large battery of mutants, the nine showed best adaptation to the 2,6 sialic form. But was the binding the key? Next experiment: binding the nine mutants to actual human cells. One of the mutant viruses bound tightly to human cells – tighter than wild flus. Kawaoka’s group compared it to an astonishing two million analogous segments of wild flu viruses, finding this mutant (cumbersomely named Q226L/G228S – a moniker unlikely to capture memorable attention) a contender for the sort of attachment to human cells seen in powerful epidemic strains.

OK, Kawaoka next asked, will my mutant attach to the upper airways of human beings – locations from which viruses get coughed out onto other people and spread in contagion? To answer that, the team grew human tracheal cells in the lab, exposed them to the mutant virus, and voila! Yes, the mutant attacks the upper airways.

What makes this mutant, with its Sia-alpha2,6Gal binding capacity so potentially contagious? The team submitted it to complicated protein analysis, including examination of the HA protein’s three-dimensional structure, and compared it to garden variety avian flu proteins. They found that the shape of the HA bulb’s sialic binding site was a perfect fit – like a hand grabbing onto a doorknob – while wild H5N1 viruses could not make that handshake.

OK, Kawaoka thought, I’ve made the mutant. But could such a thing arise in nature, or is this only possible in the laboratory? To solve this puzzle he moved his work into a higher security, BSL-3+ lab. (A great deal of space is devoted in the paper to describing the security conditions of the Wisconsin BSL-3+ facility and training of its staff, inspections by Federal authorities and Wisconsin State examinations.) The team encouraged reassortments of wild H1N1 swine viruses with the mutant partial H5N1. In nature flu viruses are very sloppy with their reproductive process. The genetic material is loosely stored in three chromosomes inside the viral capsule. When the virus infects a cell and commandeers cellular machinery to produce billions of “baby” viruses, the master “sire” viruses essentially fall apart, exposing their chromosomes to whatever other genetic material may be in the neighborhood. The process is so sloppy that it inevitably results in mutations, and in reassortments – occasioned by the co-presence in the same cell of other flu viruses. They essentially swap their genes around.

Kawaoka demonstrated that yes, this gene swapping reassortment can, indeed, occur, allowing for the possibility that his contagious mutant H5N1 could arise naturally.

OK, but would such a mutant spread among mammals? To answer that, Kawaoka’s team built special cages, located just far enough from one another that the trapped ferrets could not touch one another, but close enough to allow a sneeze to project from cage-to-cage. One set of animals were deliberately infected by stuffing the mutant viruses up their snouts, and then the waiting game began.

Yes, the virus spread readily among the animals. Because Kawaoka deliberately used a nonvirulent form of the microbe, none of the infected ferrets died and illnesses were mild. But autopsies revealed extensive binding of the mutant viruses (compared to control wild swine H1N1 viruses) to the animal’s respiratory tract cells. In the end four minor mutations were the key to transmission.

From one so-infected ferret a super-transmitter form of the manmade mutant virus emerged. When the entire ferret experiment was repeated using this virus, the results were more dramatic --- that is where in the paper Kawaoka states that its transmission was comparable to the 1918 pandemic influenza (which killed somewhere between 75-100 million human beings in 18 months).

Study of this super-mutant showed that the crucial transmissibility, coupled with its unusually strong binding ability across a range of temperatures and pH conditions, was the result of just three nucleotide changes. In other words, for H5N1 viruses in nature to attain this level of contagion potential, only three miniscule genetic switches must be altered. Given the sloppy replication process influenza goes through routinely, three nucleotides isn’t much to ask for.

Worried? Well, Kawaoka isn’t, because he tested this super mutant on cells from people that had participated in H5N1 vaccine trials, and their immune systems blocked the mutant, “indicating that current H5N1 vaccines would be efficacious against the H5 transmissible reassortant mutant virus.”


In the end Kawaoka showed that a contagious form of bird flu could be generated in the lab, or through a reassortment process akin to that which occurs in nature. Offering comfort, he also showed that currently available experiment H5N1 vaccines could block that mutant virus.

But the Kawaoka experiment fails to answer the sorts of questions a politician or policymaker needs to know in order to decide what steps should be taken to prevent a global H5N1 human pandemic.

Firstly, Kawaoka acknowledges in his paper that the particular set of mutations he produced is probably not the only way nature could transform a bird flu into a human one. He lists other potential mutation sites, and other routes to viral evolution. Therefore, it would be dangerous for any policy flowing from this paper to assume it is now possible to narrow humanity’s surveillance or threat assessment to the mutant Kawaoka developed.

Kawaoka points out one especially worrying set of mutations now seen in strains of H5N1 found in the Middle East, especially in now-chaotic Egypt. These H5N1 strains show four mutations that affect sialic binding to mammalian cells and another part of the virus entirely, called PB2. Combined, the mutational drift seen in wild Middle Eastern strains prompt Kawaoka to say, “we cannot predict whether the four mutations in the H5 HA identified here would render a wholly avian H5N1 virus transmissible,” but the changes seen in the Middle Eastern viruses, “may be several steps closer to those capable of efficient transmission in humans and are of concern.”

Similarly, all but the HA bulb section of Kawaoka’s mutants came from the relatively benign HiN1 swine flu that spread worldwide in 2009. Most of humanity was exposed to that virus – including the individuals who got the H5N1 vaccine and donated cells that proved impervious to super mutant infection. Were the cells impervious because the mutated HA bulbs from bird flu triggered an immune response, or because the entire rest of the reassortant viruses were H1N1?


Committee after committee has tried in the U.S. and Europe to come to grips with the implications of this work, and that in the the soon-to-be-published Fouchier experiments. All have failed to come up with consistent, meaningful policies. In the case of the National Scientific Advisory Board on Biosecurity (NSABB) in the U.S. the entire process is now besmirched by allegations of deliberate obfuscation on information provided by the staff managing the NSABB meetings.

When Kawaoka presented his work last month in fair detail at the Royal Society in London he was repeatedly asked by myself and meeting chair Simon Wain-Hobson, “What’s next?” And he repeatedly dodged the question.

But therein is the challenge.

The well-prepared and extremely detailed Kawaoka paper is, indeed, a roadmap that can and will be followed. A new standard of flu virology research has been set, the bar has been raised, and scientists will now pursue ambitious questions, including:

  • How lethal would such a highly transmissible mutant virus be? If the mutated HA bulbs were attached to an H5N1 body, instead of the benign H1N1 flu segments, would the result be a lab full of dead ferrets?
  • Would transmissibility and virulence be sustained over time? In 1918 the first wave of infection to circle the world was fairly benign – but the second wave was a mutant that killed and spread with ferocity. In contrast, many evolutionary biologists contend that germs adapt to their hosts, becoming less lethal with the passage of infections and time. So wouldn’t it make sense to answer that by infecting hundreds of ferrets sequentially with a variety of manmade H5N1 viruses?
  • Can the experimental H5N1 vaccine work against a mutant that is fully H5, not H1?
  • If the mutational process commences with the troublesome forms of H5N1 now circulating in the Middle East, would the outcome in ferrets be better? Worse? Terrifying?
  • Skeptics insist the ferret is a lousy model, though the U.S. Centers for Disease Control and most flu experts feel ferret respiratory cells are a reasonably close match to those of humans. But are there experiments that can now be performed on other species, satisfying the ferret skeptics?

We are only at the very beginning of this process. As Kawaoka put it at the Royal Society meeting when asked if his experimental roadmap posed intrinsic dangers either with further legitimate experimentation, or in the wrong hands, “It’s not creating, it’s evaluating risk in nature. The danger is out there, so we have to deal with it. Now we experimentally identified the virus that is transmissible. Now next take wild viruses to study – it if transmits, we have found a dangerous wild virus.”

The most important speech delivered at the Royal Society came from Dr. Paul Berg, the 89-year-old Stanford biologist who led the first great review of biology experimental safety back in 1975. That was at the dawn of the genetic engineering era, when scientists were just beginning to manipulate life forms and use viruses to carry genetic changes into cells.

“Hubris runs high among scientists,” Berg sternly warned the London meeting via live videoconference from Palo Alto. “Scientists have an incredible ability to ignore the risks of our own work.”

Hubris also pushes scientists to believe in self-regulation, excluding outsiders and demeaning the worries of the general public and political leaders. That is dangerous, warned Berg. “Not enough has been done to keep the public informed and aware,” in the H5N1 case. “A social contract between science and the public is great. But now it is under strain.”

And it is, as Congress is now investigating, the White House is compelled to respond, and the NSABB process appears to have lost its credibility.

Kawaoka Scicence Paper on H5N1.PNG
H5N1 spread.PNG