The Bill & Melinda Gates Foundation last week announced that it had awarded $287 million to 16 multi-institutional teams with the goal of developing a vaccine for HIV in five years.
While the majority of the awards support wetlab-based research, one of the collaborative teams, led by Leo Stamatatos of the Seattle Biomedical Research Institute, was awarded $19.4 million to apply computational methods to the problem of HIV vaccine design.
This project will use the Rosetta protein structure-prediction platform developed by David Baker at the University of Washington to design synthetic molecules that will trigger antibodies against HIV.
Rosetta is best known in the computational biology community for its winning record in the CASP (Critical Assessment of Techniques for Protein Structure Prediction) experiments that take place every two years, but Bill Schief, a Rosetta developer, said that Gates Foundation grant will help the software evolve from a computational exercise to a platform for solving "a real medical problem."
CASP and other evaluations "are critical for developing the infrastructure in Rosetta and developing the energy function and the whole underlying technology, so all that has to come first," Schief said. "But now Rosetta — and particularly the energy function — is at a point where we can describe the interaction between atoms accurately enough that we can design proteins of different types to do different jobs."
"There's no guarantee, but we don't feel like this is just playing around with computers."
Schief said that he and Baker have already been developing computational vaccine design methods with researchers from the National Institutes of Health's Vaccine Research Center, "and we have some preliminary data with those guys that suggest that the approach is quite viable," he said.
"From what we can tell, it's not a crazy approach. It's quite sensible and quite plausible, so we're really excited that it actually has a chance of working," Schief said. "There's no guarantee, but we don't feel like this is just playing around with computers."
Stamatatos agreed, saying that while his lab hasn't tested Rosetta yet, "it looks extremely promising."
Stamatatos' research focuses on using the HIV envelope protein as an immunogen to provoke an immune response in the host. The high degree of variation in amino acid sequence and glycosylation patterns that this protein exhibits across different HIV isolates makes it difficult to design an immunogen that is universally effective, however.
"It's hard to find an immunogen that will elicit a response against what is conserved among all these variants," he said. "Most of the antibody responses target everything that's variable — not what is conserved."
The Rosetta-based approach, Stamatatos said, will isolate the structure of conserved HIV epitopes in a "scaffold" of a non-HIV protein. "This non-HIV protein now presents a single conserved HIV epitope as an immunogen," he said. "Will antibodies then be generated against that structure?"
In order to find out, the collaborators plan to use Rosetta to computationally design scaffolds that will present these conserved epitopes. The proteins will then be synthesized and tested as immunogens.
Stamatatos said that the consortium plans to screen around 200 new scaffolds that express conserved HIV epitopes. To do this they plan to use an iterative approach that will cycle between computational design and experimental validation.
Schief said that immunogen design is a new application for Rosetta, so his team has a bit of work ahead of it. "The software development involves using some existing pieces of existing protocols and stringing them together, but also writing a lot of new pieces that just aren't in there now," he said.
However, he added, much of the initial groundwork has already been carried out in collaboration with the NIH, "so we're just refining the software and making it more aggressive. We started off using fairly conservative protocols, and the data suggests that those have worked, and now we're trying to go for more aggressive protocols."
As the project ramps up, Schief said that it will likely turn to [email protected], a distributed version the Rosetta package that is currently running as a screensaver on more than 70,000 PCs worldwide. According to the [email protected] website, this is the equivalent of a 38.6-teraflop supercomputer, which would put it at the No. 6 spot on the current version of the Top500 supercomputer ranking [BioInform 06-30-06].
"And as we make our vaccine design protocols more and more aggressive, we will rely more and more on [email protected]," Schief said. "So I think over the next couple of months we're going to start using it, and then over time there will be some types of vaccine design where we just can't do it except on [email protected] because we'll be testing so many different conformations for a particular piece of the protein and we'll be really trying to search the conformational space at high density."
Stamatatos noted that the computational method is a "totally new approach in attacking this problem" that may not have been possible to support with federal funding.
"Federal funding in general will not allow us to use this high-throughput approach," he said. "We have to test hundreds of different immunogens. It's both risky and expensive, it needs collaboration among a lot of different people, and that's why forming this consortium was central."
Stamatatos said that the Gates Foundation is encouraging more of an "industrial approach" toward research. "Here, basically what they want is a vaccine at the end of this five year funding, and that is the main goal. We're not just doing research for the sake of doing research — we're gearing up to develop a vaccine," he said.
One of the requirements of the Gates Foundation grants is that researchers share their data via centralized facilities for testing vaccine candidates and statistical analysis — an aspect that Stamatatos described as "crucial" to the success of the project.
"Everything is an open book here. All the results are available to the participants, the statistical analysis will be open to all of us, and that will facilitate the selection process," he said.
— Bernadette Toner ([email protected])