Anton 2 Simulations Explain mechanism for Loading DNA into Virus


Two-phase Motor Packages DNA into Viral Capsids, May Offer Drug Target

To cause disease, viruses must package their DNA into new virus particles to infect more cells. An important step—and possible drug target—in some viruses is the insertion of the virus’s DNA into its exterior shell, or capsid, by a biomolecular “motor.” Using simulations on the Anton 2 supercomputer designed by D. E. Shaw Research (DESRES) and hosted at PSC, in coordination with supercomputers in the National Science Foundation’s (NSF’s) XSEDE network, a collaboration of scientists across the U.S. showed how five copies of the ATPase protein making up the motor in the asccϕ28 virus may store chemical energy for a coordinated shove to push viral DNA into the capsid.

As the five ATPase proteins (numbered 1 to 5) absorb energy from binding ATP, their hinges open one by one, extending their far ends into a helix (F through J, at bottom). Once the energy in this relatively long dwell phase is stored, ATP is split into ADP and phosphate, which causes the helix to collapse in a short burst phase that pushes the DNA strands through the portal and into the viral capsid (A through E, top). From Pajak J et al. (2021) Atomistic basis of force generation, translocation, and coordination in a viral genome packaging motor. Nucleic Acids Res. 49(11):6474-6488.

Why It’s Important

Viruses make us sick when they reproduce faster than our immune systems can clear them from our bodies. And viruses are fast at procreating.

But every system has a weak point. For some viruses, a critical moment comes when its DNA must be inserted into its capsid. Threading DNA into a tiny container is a complex process that takes energy, which is supplied by ATP, the “alkaline batteries” of the cell. If scientists could learn how this system works and how to derail it, they could design treatments that trap a virus in an infected host cell, unable to infect more cells.

“What we were interested in was how do certain viruses—bacteriophages [viruses that infect bacteria] and other viruses such as adenovirus or herpes virus—package their DNA into near-crystalline densities through the action of powerful molecular motors, some of the strongest found in nature.”
—Gaurav Arya, Duke University

Previous science on the closely related ϕ29 virus offered some clues. The virus’s ATPase, so named because it splits the high-energy ATP molecule into lower-energy ADP plus phosphate, transfers energy from ATP into its own molecular structure, storing that energy to use it to push the DNA into the capsid. Results from the Carlos Bustamante lab at the University of California, Berkeley, suggested that the ATP energy is stored step by step in a slow “dwell phase” so it can be released at once, in a fast “burst phase.” The three-dimensional structure of the ATPase motor, discovered by Marc Morais and colleagues at the University of Texas Medical Branch in Galveston, provided a first look at how the ATPase’s atoms are arranged. But the motor works through its movement. How its parts cooperate dynamically to move the DNA was a mystery.

Graduate student Joshua Pajak, working in the lab of Gaurav Arya at Duke University, decided to see if molecular dynamics computer simulations of the ATPase of ϕ28 could offer clues as to how the dwell and burst phases happen. In collaboration with the Morais group and colleagues at the University of Massachusetts Medical School and the University of Minnesota, he simulated the system using the Anton 2 supercomputer hosted at PSC, as well as supercomputers in the NSF’s XSEDE network of advanced research computing centers, in which PSC is a leading member.


How PSC Helped

The simulations would require a lot of computing power, at two different scales of complexity. Using cryo-electron microscopy lab experiments, the group showed that the five copies of the ATPase in ϕ28 arrange themselves in a ring on the outside of the portal that opens into the capsid’s interior. The ring surrounds the double-helix DNA strand that needs to be pushed through the portal.

Somehow these five proteins had to coordinate, like a tug-of-war team. Pajak hoped the simulations would reveal how. But recreating the individual ATPases in the computer would require lots of processing power. This power came from the San Diego Supercomputer Center’s Comet system, which is a part of XSEDE. The time delay of the dwell phase, though, suggested that the five-ATPase complex would store the ATP-derived energy in a coordinated way that wouldn’t be visible when simulating only one copy. The huge scale of this second simulation would involve hundreds of thousands of atoms on a timescale of multiple microseconds. That task lies beyond the abilities of general-purpose supercomputers.

Anton 2, designed and made available without cost by DESRES and hosted at PSC thanks to operational funding from the National Institutes of Health, offered a unique, specialized ability to carry out such a massive molecular dynamics simulation. The system allowed the scientists to simulate five ATPase copies and their association with the capsid portal and the DNA strands for several microseconds.

“For this system, we needed to consider the five ATPase subunits, DNA, and all the water that surrounds them. The end result was a massive system, almost impossible to simulate using standard computational resources. Anton 2 allowed us to reach the microsecond timescale, where most of the interesting bio-molecular motions happen. We’d never have been able to simulate this system over several microseconds without Anton 2.”—Joshua Pajak, Duke University

At the scale of individual ATPase proteins, the Comet simulations showed how releasing the spent ADP molecules transfers energy to the ATPase’s structure, by opening a hinge in the protein. The Anton 2 simulations showed how opening these hinges makes the five ATPases expand away from the portal, one by one. While their ends attached to the portal remain in place, their opposite ends extend into a spiral—a helix—that coordinates the energy storage and grabs hold of the DNA. The extended formation also sets the protein up for the burst phase, collapsing the helix in a single, synchronized shove that pushes the DNA through the portal. The collaborators reported their findings in a series of 2021 papers in the journals Nucleic Acids Research; the Proceedings of the National Academy of Sciences, USA; and Science Advances.

The simulations offer a compelling explanation for how the five ATPase proteins coordinate. The step-by-step movement of the hinges explains Bustamante’s dwell phase. But it is still a computer simulation, and the collaborators are busy confirming the computer results with further lab experiments. They’d also like to use artificial intelligence (AI) to zoom in for more detail on the proteins’ movements. This task may call for PSC’s NSF-funded, XSEDE-member Bridges-2/Bridges-AI advanced research computers, which were designed to carry out large-data AI work. Ultimately, the lessons learned may help provide targets for antiviral drugs as well as engineering possible phage-based drugs to attack bacterial infections.