Projects in Scientific Computing: New Understanding of Life and Its Processes
The Rude Mechanicals

Opening A Rich
New Field Called
Protein Mechanics

As with people, some proteins are more well known than others. Among the established stars are the enzymes, which make things happen by catalyzing reactions. Other proteins act as security agents, protecting the body from toxic compounds. Others are transport vehicles, picking up ions like iron and calcium and moving them where they're needed.

Less well known are the mechanical proteins. These are the heavy-lifters, the brawny, hard-hat, working-class-hero proteins. Their job, in some respects, is simple: exert force. When stretched, they pull back. It's a job no less essential than other protein tasks, but it's only in the last few years that effective research methods have evolved to study them.

"Enzymes have been studied for many decades," says physicist Klaus Schulten, "but for mechanical proteins there wasn't a good method. Today we have techniques — atomic force microscopy (AFM) and optical tweezers — that allow you to take individual proteins, stretch them and measure their responses."

Schulten directs the Theoretical Biophysics Group at the University of Illinois Beckman Institute for Advanced Science and Technology. In 1999, he and his colleagues carried out a series of molecular dynamics studies of a mechanical protein called titin. Working hand-in-hand with AFM studies, the computer simulations provide a detailed picture of how titin stretches. This is new information about this important protein, which is ubiquitous in muscle tissue and plays a role in cardiac muscular disease.

Image illustrating titin in muscle fibers.

Download larger version of this image.

"AFM tells you what force you need to extend the protein to a certain length in a certain time," says Schulten. "This is crucial knowledge, but it doesn't tell you how nature controls these properties. Using the atomic-level structure of titin's components in simulations complements the AFM observations. Together these research techniques open the door to an aspect of the cell, protein mechanics, which couldn't be investigated before. This is an exciting and rich field."

Tugging on Titin

Like the similarly named giants of Greek mythology, titin is big. A long filament of roughly 30,000 amino-acids, it's the largest known protein. In muscle, it plays important roles in extension and elasticity, helping to form the sarcomere, the integral unit of muscle fiber. When muscle is stretched, titin extends — holding the sarcomere together and providing passive force, like a stretched rubber band, that pulls it back to its unstretched state.

While many large proteins are an aggregate of separate component proteins, titin is a single strand of about 300 tightly-folded domains, each like a molecular coiled spring. There are variant types of these spring-like domains spaced along titin's length, and the Schulten group's recent simulations focus on one called I27, related in structure to the antibody immunoglobulin.

AFM experiments with titin show that, compared to proteins that don't have a mechanical role, it requires considerable stretching force to unfold the molecule. As titin stretches, moreover, the pulling force rises and falls in a series of peaks, like a sawtooth pattern — as if you're pulling on a knotted rope and the knots let go one at a time. Similar results occur with cloned sections of titin made of connected I27 domains. The spacing of the force peaks matches the length of an unfolded I27 domain, indicating that each force-peak corresponds to a single I27 unfolding and that they unfold one-by-one.

AFM experiments with I27 also show a phenomenon called pre-stretch. With a weak force (about 50 picoNewton), the I27 extends slightly (5 Angstrom). With more force (above 100 pN), I27 stretches further (to about 15 Angstrom). Only with further added force (about 200 pN) will I27 let go and unfold all the way (about 300 Angstrom).

This shows, says Schulten, that I27 comes with some built-in protection against unfolding. While the AFM experiments provide key information, however, they don't give structural detail — don't show how the protein's molecular architecture allows it to resist stretching, but only to a certain point. For that, Schulten and his colleagues turned to molecular dynamics.

Opening a Black Box with Steered Molecular Dynamics

Image from simulations of I27in water

Simulations of I27 in Water
The I27 domain is shown as simulated in its unstretched state (upper left), pre-stretch state (upper right) and, at smaller scale, after unfolding and fully extended (bottom). In simulations, the domain is fixed at one end (left) and a stretching force applied at the other end. I27 is a structural motif known as a beta-sandwich: two sheets of ribbon-like strands (orange & green), one on top of the other.

Download larger version of this image.

How has nature designed these proteins to control what's happening as they're stretched? "Without the simulations," says Schulten, "you have a black box. You know what's happening but not why or how." To fill-in the missing atom-by-atom details, Schulten and colleagues used a technique developed by his group called steered molecular dynamics.

If you think of the structure of a protein as a photograph that records the spatial relationships among all the atoms in the molecule, a molecular dynamics simulation is the movie version — it records how the atoms move from one tiny moment to the next. Steered molecular dynamics is a novel approach that applies a relatively large force to the molecule. SMD can induce molecular changes to occur rapidly that would otherwise take too long to simulate, and it's especially useful for mimicking AFM experiments.

To look more closely at AFM results with the I27 domain, Schulten's team carried out an extensive series of SMD simulations. To realistically depict the living environment, the I27 structure was represented as immersed in water, and simulations were done as numerical experiments applying a range of forces to one end of the I27 domain, with the other end fixed.

Analysis of these simulations showed that I27's resistance to unfolding arises from a patch of six hydrogen bonds that bridge between two of the protein's folded strands. When enough force is applied, all six of these bonds rupture simultaneously — the critical event that allows I27 to fully unravel.

These six hydrogen bonds, explains Schulten, can be thought of as an energy barrier between folded and unfolded I27. The AFM observations reveal that the barrier arises at a stretch of about 15 Angstrom (with a height of about 20 kilocalories/mol). With access to Pittsburgh Supercomputing Center's CRAY T3E, the researchers did simulations — up to 18 for each numerical experiment — to investigate this energy barrier and compare the computational results quantitatively with AFM data.

"This is one of the first times," says Schulten, "that researchers did sampling with multiple simulations. This allows us to be much more firm in our comparison with experiment." The spread in results among individual simulations corresponded within 10 percent to the experimentally observed positions and height of the energy barrier, giving the researchers a high degree of confidence.

The Pre-Stretch Intermediate

Image of I27 domain before and after hydrogen bonds break

Hydrogen Bonds
The I27 domain in its native state (left), pre-stretch at 10 Angstrom, and unfolded to 25 Angstrom (right). The simulations reveal that unfolding occurs when six hydrogen bonds (dotted lines) between two of I27's strands (A' and G) rupture simultaneously. Two hydrogen bonds (between A & B strands) break at pre-stretch.

Download larger version of this image.

The simulations also showed that two other hydrogen bonds, between two different folded strands, break before the six-bond rupture that precipitates unfolding, contributing to a small energy barrier at an extension of about 5 Angstrom. Does this account for the pre-stretch from AFM experiments?

To clinch whether this intermediate I27 state corresponds to the pre-stretch, Schulten's experimental collaborator J. Fernandez at the Mayo Clinic in Rochester, MN cloned a mutant I27, with a changed amino-acid that eliminated the two hydrogen bonds. AFM experiments showed that this mutant I27 unfolds without an intermediate state.

This pre-stretch, explains Schulten, gives the I27 domain finer control over three stages of forced stretch: straightening of the domain with a stretch of about 5 Angstrom, prestretch to about 15 Angstrom, and complete stretch to about 300 Angstrom. In normal functioning, pre-stretch allows the protein to avoid complete unfolding, saving the relatively long time needed to refold and shorten. Nevertheless, says Schulten, I27's ability to provide a slack up to 300 Angstrom is critical for its biological function.

Compared to enzymes it's a late start, but titin and another mechanical protein, fibronectin, which Schulten and his collaborators have also studied, are rapidly gaining celebrity status. Research shows these proteins play several crucial roles: organizing chromosomes in the nucleus, cell-to-cell communication, and the movement of cells relative to each other. By revealing the mechanical properties of I27, SMD simulations have proven themselves as a full partner with AFM experiments in learning about these proteins. With their work, the Schulten group has opened a new chapter, protein mechanics, in the ever expanding book of biomolecular science.

Download PDF PDF: The Rude Mechanicals
Researchers Klaus Schulten, University of Illinois Beckman Institute for Advanced Science and Technology
Hardware CRAY T3E
Software NAMD
Related Material
on the Web
Unfolding Titin Immunoglobulin Domains.
Theoretical Biophysics Group, University of Illinois.
Steered / Interactive Molecular Dynamics.
References Piotr E. Marszalek, Hui Lu, Hongbin Li, Mariano Carrion-Vazquez, Andres F. Oberhauser, Klaus Schulten & Julio M. Fernandez, "Mechanical unfolding intermediates in titin modules, " Nature 402, 100-103 (1999).

Hui Lu & Klaus Schulten, "Steered molecular dynamics simulation of conformational changes of immunoglobulin domain I27 interpret atomic force microscopy observations," Chemical Physics 247, 141-153 (1999).

Andre Krammer, Hui Lu, Barry Isralewitz, Klaus Schulten, Viola Vogel, "Forced Unfolding of the Fibronectin Type III Module Reveals a Tensile Molecular Recognition Switch", Proceedings of the National Academy of Science, USA, 96, 1351-1356 (1999).

Hui Lu, Barry Isralewitz, Andre Krammer, Viola Vogel, and Klaus Schulten,"Unfolding of Titin Immunoglobulin Domains by Steered Molecular Dynamics", Biophysical Journal, 75, 662-671 (1998)
Writing: Michael Schneider
HTML Layout/Coding: R. Sean Fulton
© Pittsburgh Supercomputing Center (PSC), Revised: June 27, 2000
URL: /science/2000/schulten/rude_mechanicals.html