Coming to a screen later this century: Molecular portrait of motor movement
Capturing the essence of an object sounds like the lofty goal of an artist who's accustomed to using a paintbrush and palette. Michael Levitt, PhD, however, prefers a different medium—the 1's and 0's of computer science.
For almost 40 years, Levitt, professor of structural biology, has been merging the worlds of biology and computers to create dynamic models of proteins, nucleic acids and other macromolecules.
In the last six months, Levitt has undertaken one of his most ambitious projects to date: representing the varied and complex behavior of the motor protein, myosin. This protein drives a number of critical cell processes including muscle contraction, cell division and the movement of materials within cells. It does this by releasing the chemical energy stored in adenosine triphosphate (ATP), the molecular fuel for most cellular reactions, and converting it into mechanical energy.
Of the 13 known types of myosin, Levitt is focusing on myosin II or conventional myosin, which possesses an arm-like ability to reach out and pull itself forward along filaments of actin, a major protein component of the cell cytoskeleton.
Myosin II molecules cause the contraction of skeletal muscle by working together in chains, which simultaneously pull themselves along actin tracks from opposite ends of each muscle fiber. By representing the behavior of a single molecule, Levitt's model will offer an up-close view of the force-generating agent underlying all motor movement.
Once they've mastered myosin II, Levitt and his team hope to extend their work to the molecule's numerous other forms, said Dahlia Weiss, a graduate student in Levitt's lab who is working on the model as her thesis project.
"That's the nice thing about computational work," Weiss said. "If you solve a problem well enough, then you can apply what you've done to similar situations."
Scientists have known the three-dimensional architecture of myosin's 10,000-plus atoms since the 1990s. Using a technique called crystallography, which scans crystallized samples of a molecule with X-rays, scientists have also described the precise structures of myosin II at the beginning and end of its pulling action. The task now before Levitt and his team is filling in the motion between these pictures.
That's where computers come in.
"If you have certain atoms sitting in certain locations and if you know something about the forces between the atoms, then you can create computer-based simulations that predict how everything will interact and move," said Levitt.
But programming these simulations is anything but straightforward. "You need a lot of biological background and intuition to know if what you are doing makes any sense at all or if it is computer-generated junk," said Weiss. "Right now we are not sure what is the best approach for modeling myosin. We are figuring this out as we go along."
While unclear, this process of "figuring out" has been far from uninformed. Indeed, bringing together experts in biophysics, mechanical engineering and other fields, the myosin project reflects the interdisciplinary spirit of the Clark Center where Levitt and his team work. Home to the university's Bio-X initiative, the center's open workspaces encourage collaborations between scientists from far-ranging disciplines.
Jim Spudich, PhD, professor of biochemistry and developmental biology, who has studied myosin for much of his career, has helped explain the biochemistry behind myosin's ATP-driven movement, among other properties. And robotic specialists have worked with Levitt to create a schematic of myosin's motion, which depicts the plausible intermediates between its starting and ending points.
Still in the early stages of the project, researchers are grappling with one of the key challenges of model building: defining an appropriate level of complexity.
"The goal is to build a model that is as simple as possible but still reproduces all of the mechanical properties of myosin," said Levitt.
So far the researchers have reduced the molecule's 10,000 atoms to 60 rigid groups of atoms, which they would ultimately like to distill to five or six groups. To validate each step of this process, researchers use normal mode calculations, a mathematical technique for determining the vibrational properties intrinsic to any shape.
The trick is making sure any simplified model retains the vibrational properties of myosin's atomically detailed shape. "If I can say the model behaves well with these five pieces, then I know we have somehow captured the essence of myosin," said Levitt.
Beyond reflecting a refined understanding of the molecule's form and function, this approach also addresses more practical concerns—the constraints of today's computing power.
At the speed of today's processors, running a simulation that accounts for all of myosin's 10,000 atoms would take millions of years, said Levitt.
Once researchers have honed the contours of their model, they'll begin incorporating myosin's biophysical and mechanical properties and defining system variables for water and other solvents. As the movie of the molecule's action takes form, they will fine-tune the simulation by testing how well it can mimic known experimental findings.
The true mettle of a computer model, however, lies in its ability to go beyond the data upon which it is based to fill in missing information or predict new behavior, said Levitt, who would ultimately like to see the model used as a tool to inform research at the bench.
To accommodate the wide-ranging needs of experimentalists, Levitt plans on creating multiple models for each type of myosin. Each model will offer its own level of complexity to help experimentalists study different questions about the molecule and its behavior.
For example, once Levitt's team has finished modeling the arm-like action of myosin II using five moving parts, they'll expand the simulation into a family of related models, containing anywhere from three to 10 pieces.
Today this goal is still a long way off and with all of the challenges that lay ahead, Weiss remains focused on the task at hand: defining a virtual object that captures myosin II's key traits.
"[This project] certainly won't be finished by the time I am done with my PhD," she said. "For me, it would be great to build a model that has the same vibrational and biophysical properties as myosin. Beyond that, I'll leave the rest to someone else."
Research on the myosin project is supported by the National Institutes of Health through the Simbios National Center for Biomedical Computing at Stanford.