Stopping HIV in Its Tracks
PSC’s Bridges Helps Scientists Understand Monkey Protein that Confers Immunity to HIV
Nov. 12, 2018
Scientists have scored a number of victories against HIV, the virus that causes AIDS. But these victories are incomplete. We can hold the virus in check, but not cure it. We can reduce the chances someone will be infected, but do not have a surefire way to prevent infection. Scientists from the University of Delaware and the University of Pittsburgh are using the XSEDE resource Bridges at the Pittsburgh Supercomputing Center to investigate how one protein prevents HIV from infecting monkeys. Understanding how it does that, and why the human version of that protein doesn’t, promises a completely new avenue for stopping HIV in its tracks.
Why It’s Important:
Globally, nearly a million people died of AIDS in 2017, according to the World Health Organization. That number is down from its peak in 2004-2005. Still, it’s fair to say that we have HIV, the AIDS virus, on the ropes but have not yet knocked it out. We can control the virus in infected people and extend their lives. But we can’t cure them. We can give drugs that reduce the risk of infection to people at high risk. But we can’t prevent infection. The drugs we have already are lifesavers. Still, new avenues for attacking the virus might be needed before cures, and true preventives, are possible.
“There’s this protein called TRIM5α. It’s an HIV restriction factor, which means that it inhibits infection. The protein is naturally occurring in old-world monkeys and rhesus monkeys. The big question is that we [humans] also have this protein; it’s just that it doesn’t make us immune to HIV.” —Juan Perilla, University of Delaware
Possible new targets for therapy are the reason why Juan Perilla of the University of Delaware and colleagues there and at the University of Pittsburgh are interested in a protein called TRIM5α (“trim five alpha”). In old-world and rhesus monkeys, TRIM5α provides a hard stop to the HIV’s ability to infect cells. Its ability to destabilize HIV’s capsid, the protective shell around the virus’s genetic material, is what makes HIV unable to infect monkeys. On the other hand, humans do have a version of TRIM5α—but for reasons we don’t yet understand it can’t stop the virus like the monkey version does. Perilla and his collaborators would like to find out how monkey TRIM5α works, and possibly how the human version can be helped to do the same trick. This knowledge could provide a new way of attacking the virus that might stop HIV much more fully than the current generation of drugs.
How PSC and XSEDE Helped:
The scientists studied the problem in two ways. Angela Gronenborn of Pitt and Tatyana Polenova of Delaware used a lab technique called nuclear magnetic resonance (NMR) to study which parts of the virus’s capsid protein, called CA, are affected when TRIM5α is present. NMR can tell what parts of the protein are affected, but not how they’re affected. So, alongside the lab work, Perilla and his graduate students studied the system using simulations on PSC’s Bridges supercomputer.
“Bridges allows two things. It performs large-scale simulations, but because Bridges is equipped with large-memory nodes we can use it to study high-dimensional trajectories—there are so many particles in the simulation, the dimensionality of the problem increases significantly. On another supercomputer, you’d have to start [rewriting your code]. Bridges’ memory really reduces the time to solution. It also allows us to give parts of the project to graduate students who are not computer engineers. And … it quickly allows us to do [additional,] more creative analyses of the dataset.” —Juan Perilla, University of Delaware
The simulation was truly massive. Studying the interaction between a single copy each of the CA and TRIM5α proteins would be a significant computational problem. But Perilla’s group took it much farther than that. They created a virtual version of the entire viral capsid, containing more than 1,000 copies of CA—a total of roughly four million atoms. In turn, the scientists embedded their simulated proteins in a box of simulated water molecules. The result was a system containing more than 64 million atoms.
The researchers simulated the molecules’ interactions, watching how TRIM5α affects the capsid. Analyzing the many interactions in the simulation required repeating the simulation under different conditions, which in turn required massive computer memory (RAM). The XSEDE-allocated Bridges’ “large memory” nodes were perfect for the job. They contain 3 terabytes of RAM, which is 96 times the RAM in a high-end laptop.
“We discovered that the pentamers and hexamers had very different dynamical properties. In the presence of TRIM5α in the assembled capsid, pentamers were a lot more rigid than hexamers … The capsid becomes less stable and therefore unable to complete its function once inside the infected cell.” —Juan Perilla, University of Delaware
The simulated results obtained from Bridges agreed perfectly with the NMR lab results. That gave the scientists confidence that the simulations were accurately capturing the system. The simulation also painted an intriguing picture. Just as a soccer ball needs six-sided and five-sided panels to be round, the HIV capsid needs CA proteins grouped in five-sided “pentamers” and six-sided “hexamers” to achieve its normal, oblong shape. The effect of TRIM5α on the CA proteins was different depending on which of the two shapes they were forming, though. Normally the CA proteins, even when they’re linked together to form the capsid shell, wriggle and writhe. But in the presence of TRIM5α, the CA proteins in the pentamers were unusually stiff, moving far less than they normally do. This disruption of the pentamers makes the capsid unstable. In turn, that makes the virus unable to deliver the its genetic material properly when it enters a host cell. The infection is stopped in its tracks. The scientists reported their results in the Proceedings of the National Academy of Sciences, USA, in October 2018.
The researchers are now studying destabilization of the capsid in more detail. Their hope is that, in the long term, better knowledge of the system and how the small differences between human and monkey TRIM5α lead to such large differences in effect can offer clues to helping the human version work better, and possibly prevent infection in humans.