With PSC'S HP MARVEL, an exceptional machine for quantum computations, researchers attacked a key problem in protein biology

Energy in the body comes from millions and millions of tiny power generators, each equipped with a crankshaft that spins round and round 24/7, producing the fuel that makes us go.

Right. And the moon is made of Gouda cheese.

Photo of Klaus Schulten and Markus Dittrich.

© DaimlerChrysler AG

“You can take a spoonful of that protein and it generates as much torque as a Mercedes engine.”

Suspend your disbelief. The protein adenosine triphosphate synthase, better known as ATPase, is nature’s smallest rotary motor. “You can take a spoonful of that protein,” says biophysicist Klaus Schulten of the University of Illinois Urbana-Champaign, “and it generates as much torque as a Mercedes engine.”

A remarkable molecular motor that in the laboratory produces torque from chemical fuel, ATPase works the other direction in humans — converting torque into ATP, the basic fuel of life, the chemical energy that fuels muscle contraction, transmission of nerve messages and many other functions. Probably the most abundant protein in all living organisms, ATPase is the power plant of metabolism. In an active day, an adult human can produce and consume its body weight or more of ATP, nearly all of it produced by ATPase.

The 1997 Nobel Prize in Chemistry recognized Paul Boyer and John Walker for their work in assembling, for the first time, a detailed picture of ATPase and how this tiny molecular machine does its job. Subsequent research has added to the picture, but many challenging questions remain.

Imagine we’re from Mars, says Schulten, director of the Theoretical and Computational Biophysics Group at the University of Illinois Beckman Institute, and we want to understand how a car engine works, but the engine is infinitesimally small. “It’s nearly impossible to see the details in motion. The only way is to use the computer to simulate it, and then we can recognize the combustion process driving the car engine. It appears that ATPase is kind of a combustion engine too.”

Photo of Klaus Schulten and Markus Dittrich.

Klaus Schulten (left) and Markus Dittrich, University of Illinois, Urbana-Champaign.

To see what they could see, Schulten and graduate student Markus Dittrich turned to Jonas, PSC’s 128-processor HP Marvel system, which is dedicated to biomedical research. With the exceptional capability of this system, they were able to simulate “combustion” in ATPase with quantum theory — to get a picture of how electrons move from atom-to-atom during chemical reactions.

“If you want to do careful calculations,” says Schulten, “you have to invest a lot of computing power. It would make no sense to approach this important problem with cheap methodology. We decided to do the most advanced calculation that people do today when simulating a biological reaction.” Jonas’s very fast EV7 processors and large shared memory made the quantum calculations feasible, and the resulting study turned up invaluable new information about nature’s tiniest motor.

The Wheel Spins Round and Round

ATPase schematic.

This cartoon schematic shows the F0 part of ATPase (blue) sitting in a membrane, where proton (H+) concentration outside (below) the membrane greater than on the inside causes F0 to rotate. The axle (red) of F1 also rotates and triggers rotary "combustion" in three active sites, contained within alternating subunits (green) of F1's hexagonal active-site complex.

Like most motors, ATPase has moving and non-moving parts. There’s a wheel that spins, similar to a millwheel, to turn an axle that revolves inside a hexagonal cluster, in which there are three combustion chambers (active sites), each of which, in sequence, charges up with chemical raw materials — adenosine diphosphate (ADP) and phosphate — and “fires” to produce ATP.

The wheel part of the protein is called F0. In humans and other mammals, F0 resides in the membrane of mitochondria, microscopic structures inside the cell. In bacteria, where ATPase works reversibly both as an ATP generator and an ATP-fueled motor, F0 sits in the cellular wall. In both cases, it forms a channel for protons to flow through the wall, a flow which — much like water turning a millwheel — causes F0 to rotate.

The other main part of ATPase’s structure, called F1, extends into the cellular or mitochondrial interior. F1 includes a central stalk — the axle — that is coupled to and turns with F0. The other end of this axle revolves inside F1’s hexagonal cluster, which contains the three active sites where “combustion” occurs.

“F0 spins, and this spins the F1 central stalk,” says Dittrich, “and this leads to ATP synthesis.” The axle spins in one direction (clockwise when viewed from the membrane side) during ATP production. What’s not only fascinating — but also a large benefit for understanding the protein — is that the reverse reaction also works in the laboratory. Put F1, a very large protein by itself, in solution with ATP, and ATP will “hydrolyze” inside F1’s active sites into ADP and phosphate and the axle will spin counterclockwise.


Because ATP hydrolysis is a chemical mirror of synthesis and more amenable to laboratory study, it offers an invaluable back-door approach to gaining knowledge about ATPase, and this reaction is what Schulten and Dittrich set out to simulate.

Zooming-In on the Spark Plug

One of the intriguing questions about F1-ATPase is how its three combustion chambers cooperate with each other. During each 360-degree rotation of the F1 axle, each active site, one after another, changes structure and thereby alters its function. For the hydrolysis reaction, each site is open to bind with an ATP molecule, closes to hold it during breakdown into constituent products, and opens again during product release. With each rotation, three ATP molecules hydrolyze, one at a time. Schulten and Dittrich hoped to shed light on the atom-by-atom details of this rotary process.

Another challenging question concerns how the hydrolysis reaction causes the F1 axle to spin. “We don’t understand how the chemical reaction is coupled to rotation of the axle,” says Schulten. “That’s still a big mystery.”

To get at the crucial details of bonds breaking and reforming during a chemical reaction requires quantum theory. Schulten and Dittrich therefore used a method called QM/MM (quantum mechanics/molecular mechanics), which made it possible to simulate the entire subunit of F1 that houses the active site, but used quantum theory selectively like a zoom lens to zoom-in on the active site itself, where “combustion” occurs.

Wishbone Gyroid Structure

In this representation of the molecular structure of F1-ATPase, the rotating stalk (red) protrudes from a hexagonal cluster composed of alternating alpha (yellow) and beta (green) subunits, three of each, within which the stalk rotates. The three active sites are in the beta subunits.

The active-site subsystem for the QM part of the simulations (blue) includes water molecules. The blowup represents this active-site subsystem with ATP.

In total, they simulated two different configurations of the F1-ATPase active-site subunits, more than 8,000 atoms each, while the QM calculations zoomed in on the combustion chamber. Employing up to 32 of Jonas’s processors at a time, with an extensive series of simulations, they used over 12,000 hours of computing time. The outcome is several key findings on a crucial biological system.

ATP in solution without ATPase is extremely slow to hydrolyze, taking as long as a week. With ATPase, this reaction goes about 100 billion times faster, a huge speedup that researchers have been hard pressed to explain. The Jonas simulations reinforce Schulten and Dittrich’s finding (in a prior simulation study) that ATPase hydrolysis — in which water attacks ATP to break it down — is carried out by two water molecules in concert, rather than only one, as had been thought. This finding — unobservable in laboratory experiment — is a big step toward explaining ATPase’s remarkable catalytic efficiency.

“In order to have this kind of concerted action of two water molecules,” explains Dittrich, “they have to be arranged in a particular way. This is accomplished by ATPase and won’t happen in solution because there it’s unlikely the water molecules will assume this special conformation. If we didn’t have this mechanism, the reaction would take place on a much slower time-scale and wouldn’t lead to the observed physiological rates.”

The simulations also show, unexpectedly, that there’s no energy change in the active site as ATP breaks down into its reaction products — a finding that goes further than experiments in establishing that the reaction itself doesn’t provide any force toward making the F1 axle rotate. “It’s not the chemical event that drives the rotation,” says Dittrich, “which means we have to look at other possibilities.” The two remaining possibilities are when ATP binds to the active site or when the reaction products are released.

Perhaps most importantly, the simulations reveal that one of the amino acids in the active site of F1-ATPase — arginine — appears to play a key role in coordinating the timing among the three active sites. This amino acid — referred to as the arginine finger — seems to operate like a spark plug. It shifts its position depending on whether ATP or the reaction products are in the combustion chamber. “We think that the findings,” says Schulten, “particularly with respect to the arginine finger, could well prove to be a crucial part of this puzzle.”

© Pittsburgh Supercomputing Center, Carnegie Mellon University, University of Pittsburgh
300 S. Craig Street, Pittsburgh, PA 15213 Phone: 412.268.4960 Fax: 412.268.5832

This page last updated: May 18, 2012