Playing to Win:
An Interview with Ian Wilson on Project Checkmate
In 2006, The Scripps Research Institute and IBM announced the creation of Project Checkmate to develop methods to anticipate, manage, and contain infectious diseases that could become global pandemics—most notably avian flu virus or "bird flu." Mika Ono of News&Views catches up with Scripps Research Professor Ian Wilson, who is heading the initiative.
How did Project Checkmate come into being?
Richard Lerner [president of Scripps Research] thought up the entire idea, and talked to IBM and key people in Florida to see how we could work together initially on H5N1 influenza virus. The goal of Project Checkmate is to set up strategies to deal with the possibility of a new pandemic influenza virus coming into the human population. Using the power of modern computing and current advances in biology, chemistry, and virology, we want to figure out what we could do to protect against a potential pandemic. If we could predict how a virus, such as H5N1, might evolve, we could have antibodies or vaccines ready, and be prepared for whatever move the virus might make. Although the project is currently focused on the imminent threat of the bird flu virus strain H5, the same strategy could be used for any other influenza virus—or even for other viruses.
How great a threat is H5?
It's difficult to say. So far, we have only had sporadic episodes of individuals who have been infected from bird flu. With H5, many people who have been identified as infected have, in fact, died, but we really don't know how many people have been infected and not shown symptoms. The H5 virus has been around for some 10 years but hasn't yet made the transition to become a transmissible human disease. That's a very good sign, but that's not to say that one day it couldn't develop that capability.
We know that there is a huge reservoir of potential flu viruses out there in the bird population, particularly in aquatic birds, so we have to be prepared for any of these. It could be H5. It could be H9. We're overdue for an influenza pandemic—way overdue if you look at history. It's like waiting for an earthquake.
What would it take for a bird flu virus to become pandemic among humans?
To become a pandemic, a bird flu not only has to get into the human population, but also develop the ability to be transmitted among humans. This requires that the bird flu changes receptor specificity. What we have seen in past pandemics is that the virus usually goes through some sort of reassortment, so there's a mixing and matching of an avian virus with a human virus. The influenza virus contains a segmented genome with eight RNA segments so you can actually take out two segments from the bird flu and keep six from the human to get a largely human-adapted virus that has altered characteristics, such as new surface antigens, that our immune system doesn't recognize. Such reassortment happened in the flu pandemics of 1957 and 1968. We're not certain what happened in 1918, the year of the deadly Spanish Flu, because in 1918 we didn't even know that influenza was a virus. Viruses weren't isolated until the 1930s.
How do you go about making predictions?
We're taking a combined computational and experimental approach. We know, for example, how the 1968 pandemic has evolved over the last 30 years, so we can therefore look at the 1968 pandemic at an early stage, and see if we can find ways to predict how the virus has already evolved. In addition, we can try to evolve the virus in the test tube. Once we make predictions, we can set up experiments as a reality check to see: 1) do we get a viable virus, 2) does it infect, and 3) does it transmit. All these are really important considerations. When new information comes in about how an existing virus is changing, we can again see if our predictions have been correct. If not, we change the algorithm to better emulate what is actually happening in nature.
Has anyone tried to do this kind of thing before?
Not on this scale and magnitude. This is a completely new approach to try to predict in advance what will happen to an emerging virus. Usually, you are working after the event, not before.
How far have you gotten in the last year, since you came on board with the project?
We have made a lot of progress assembling a fantastic team. We have a big effort here at Scripps in La Jolla, California. Richard Lerner and Kim Janda are working on experimental phage display-type approaches. Dennis Burton is examining antibodies that could cross-neutralize against different strains and sub-types. My group is looking at crystal structures of viral components. We have already determined the structure of the 1918 influenza virus hemagglutinin, as well as the H5 HA from the highly pathogenic Vietnam 2004 strain. We are also generating libraries of influenza strains that will be very useful for what we need to do. Peter Palese, a renowned influenza virologist at Mount Sinai who was recently involved in the reverse genetics of the 1918 virus, is also a key member of the team. Computationally, there has already been preliminary work at IBM to see how much computation power will be needed for the initial research to look at antigenic drift in the virus to see how variation occurs from year to year. Scripps Florida is well positioned to work with IBM using the institute's high-throughput screening capabilities as well as the expertise in bioinformatics there.
What are currently the biggest roadblocks for Project Checkmate?
We have been trying to get adequate funding to get the project going full steam. These sophisticated and state-of-the-art computers are not cheap, and the project requires computation of an unprecedented magnitude. We need support to put that computation in place, as well as to further the experimental work. We've been talking to the State of Florida, to Palm Beach County, as well as to national funding agencies. So far, the response has been encouraging.
Ultimately, do you see Project Checkmate working hand-in-hand with a public health approach, say as part of an early warning system?
Absolutely. If we saw changes that would alter the receptor specificity of an influenza virus, we would be able to give that information to public health officials at the Centers for Disease Control and other agencies. If those sequences then appeared in natural isolates, then one could consider imminent action, such as using that strain as a starting point for a new vaccine. Similarly, we would expect to have access to the most current influenza sequences to see what is happening in the field and how circulating and emerging viral strains correlate with the models we are generating.
Do you find Project Checkmate dovetailing with your other big projects, such as the Joint Center for Structural Genomics [JCSG]?
The projects all work together very well. The experience that we have gained working as an integrated team has proved invaluable for bringing other diverse groups together. You get used to dealing with doing high-powered science in multi-institutional and multi-group consortia. Information management and flow are key. Within our projects, we have meetings—weekly and monthly as a group, and daily within the subteams. Teleconferencing and videoconferencing make things easier. We also know how to use a public web site to disseminate information and new findings to the community as quickly as possible.
What are the advantages of big projects? Pooling together technology and brain power?
Exactly. That is really the strength of Project Checkmate. We've got expertise in influenza virus and antibodies, as well as experimental approaches in structure, and phage display, that can be combined with awesome computational power. Integrating these component projects is key. As in the JCSG, we will probably create databases not only to look at our successes, but also to learn from our failures. That's really important. It's difficult to store negative data, but often it's often the negative results that give the clues as to what to do next. You've got to have both successes and failures to chart out a sensible path.
What do you find new and exciting in the field of influenza research?
There's much more information nowadays and we have many more tools than we used to. Hundreds of flu sequences are available now, and we're getting them at an increasingly fast rate, enabling us to correlate sequence changes with disease.
One valuable new tool is the glycan array, which was developed by [Scripps Research Professor] James Paulson's Center for Functional Glycomics. It was originally designed for looking at glycan binding proteins, but it is also extremely useful for looking at the binding of influenza to its sialylated receptor. You can put influenza receptor analogues onto the surface of this array and detect binding, either to the hemagglutinin antigen of the virus or even to whole virus.
Computational power has also increased orders of magnitude. The type of calculations we're talking about would have been impossible even a couple of years ago. Hemagglutinin is a huge molecule. It's got a molecular weight of 200,000 daltons, including 25 percent carbohydrate, so the question is how to simulate that structure computationally. These are enormous calculations that require huge computer resources. That's why we teamed up with IBM, to take advantage of their Blue Gene computer developments and to harness the power of these computers for experimental and predictive purposes to track influenza virus evolution.
Is there anything you would like to add?
Project Checkmate addresses a major global health threat, that harnesses the combined resources of both Scripps campuses, Mt. Sinai, and IBM. It's a tremendously exciting opportunity to apply our combined skills and resources to such an important medical problem.
Send comments to: firstname.lastname@example.org