Snapping into Place

Anton 2 Simulations Give Clue to Poorly Understood Role of Cellular “Bricks”

Tubulin proteins are the cell’s “Lego bricks,”connecting with themselves into tubular structures that help give living cells shape and stiffness. But the protein may also play a role in energy generation by the cell’s mitochondria, with individual tubulin “bricks” sticking to the surface of mitochondria by a poorly understood mechanism. Scientists at the National Institute of Standards and Technology and elsewhere used the Anton 2 system hosted at PSC to unravel how tubulin snaps into place—offering clues to phenomena as different as chemotherapy side effects, cancer, and brain development.

Why It’s Important

The protein tubulin is the cell’s Lego brick. When stacked it forms tubular structures called “microtubules.” These structures connect with other cell structures to form the cytoskeleton that gives living cells shape and stiffness. Microtubules also play a role in moving things around the cell, for example, pulling the chromosomes apart when a cell divides so that each daughter cell has a full set of DNA.

This puts tubulin at the center of some important medical conditions, including cancer. The chemotherapy agent Taxol works by preventing tubulin from disassembling into individual bricks, preventing cancer cells from dividing.

“Why do we need Anton 2 for this particular problem? We were not entirely sure what the binding surface of the protein was … You need to worry about the time it takes the tubulin to randomly explore a complete rotation … You have enormous numbers of atoms you’re trying to simulate over microseconds, and you can’t do it in a reasonable amount of time on other systems.”—David Hoogerheide, National Institute of Standards and Technology

But there’s more: tubulin also does a poorly understood dance with the cell’s powerplants, called the mitochondria. When not stacked into microtubules, individual tubulin blocks can stick to the mitochondria’s outer membrane. There, they can do things like interfere with the entry and exit of ATP, the cell’s basic fuel. Scientists aren’t sure whether interfering with this process might be a cause of chemotherapy side-effects. They don’t even know exactly how tubulin sticks to the mitochondria. David Hoogerheide of the Center for Neutron Research at the National Institute of Standards and Technology and colleagues elsewhere decided to use simulations on the D.E. Shaw Research Anton 2 supercomputer hosted at PSC, in concert with laboratory measurements, to figure out this basic question.

How PSC Helped

Hoogerheide and his fellow scientists had a powerful tool for looking at tubulin membrane binding in the real world: neutron scattering. This method can produce an image of the protein stuck to a membrane in conditions that imitate the surface of a mitochondrion. But neutron scattering produces pictures that are too blurry to see individual atoms. It provides important clues to how tubulin is interacting with the mitochondrial outer membrane, but not the complete answer.

To get that answer, the investigators paired their neutron scattering results with computations on Anton 2, a system specially designed to simulate the motions of large molecules. Anton 2 can simulate the motions of the individual atoms in tubulin and the membrane, giving a view of these molecules for longer timescales than possible with a traditional supercomputer. This was important for the group’s tubulin simulation, because they weren’t sure exactly what part of tubulin was sticking how, and to which of the lipids that make up the membrane. They had to start with the tubulin completely unattached and maybe in the wrong position, and give the protein time to rotate and move into place. To do this they needed simulations of a microsecond or longer, which would have taken at least ten times longer on another supercomputer.

“I’d like to emphasize the happy marriage between neutron scattering and the [Anton 2] simulations. With neutron scattering, we get low-resolution information on a system that we think is mimicking a cellular environment rather well. Then we use the simulations to match up with that and get high-resolution information … These two together can be a really powerful combination of techniques.”—David Hoogerheide, National Institute of Standards and Technology

The results with Anton 2 mirrored the lab measurements exactly, showing that the “alpha” half of the tubulin protein was the part that stuck to the membrane, and that a specific lipid called DOPE was necessary for it to stick well. Anton 2, then, gave the scientists the exact geometry of the interaction. Neutron scattering gave compatible results in the real world that helped increase their confidence that the simulation was producing realistic results. The team published their findings in the Proceedings of the National Academy of Sciences USA in April 2017. Future work may involve unentangling the therapeutic effects and side effects of chemotherapy agents that interact with tubulin. Another avenue for future work will be investigating how other proteins with similar interactions with the mitochondria, such as alpha synuclein, may play a role in conditions like Parkinson’s disease.