Proteins are the molecules of life. They form the framework of our muscular systems and other tissues, serve as antibodies, function as hormones, and perform many other vital tasks in the human body. Proteins are also the enzymes that carry out most of the chemical reactions in the human body.
Each protein comes from a different gene sequence. The sequence determines the structure of each individual protein. There are an astronomical number of three-dimensional configurations that proteins can form–scientists estimate that there may be more possible structures than the number of grains of sand on a beach. Because proteins derive their functionality from their structure, which is based on their gene sequence, each gene sequence produces a different kind of protein destined to complete a unique task.
However, surprisingly little is known about the physical structures of proteins and how they are created from those gene sequences. Scientists know that proteins undergo processes called folding and unfolding through which they are built and torn down, but they don't know how or why naturally occurring proteins consistently display a particular shape, or native state, after the folding process.
Finding out why proteins fold into a consistent native state is important because occasionally the natural folding process breaks down and the proteins form the wrong structures. When that happens, the proteins' functions suffer– DNA may not replicate properly or drugs sporting proteins as main components may not work. Protein misfolding diseases, such as bovine spongiform encephalopathy (Mad Cow disease), Parkinson's, or Alzheimer's, can occur.
To prevent proteins from misfolding and causing serious problems, scientists need to know more about why they fold and unfold the way they do. If researchers could discover what stimuli cause certain structures to appear, new solutions to protein misfolding diseases, such as structure-based drugs and a better understanding of gene mutations, could be possible.
Carlos Simmerling, a professor of biochemistry at Stony Brook University in New York, and a team of students and colleagues are working to understand why proteins fold and unfold and what happens during the process. Using NCSA's Tungsten and Platinum supercomputing clusters, they are simulating the folding and unfolding processes of small proteins and comparing their findings to experimental results from other research groups.
Simulating The Folding Process
How proteins fold and unfold has long been an issue of debate. Simmerling says, “Nobody knows what the unfolded protein looks like.” This is due mostly to the speed with which a protein transitions between folded states and the fact that there aren't many unfolded proteins to observe in normal physiological conditions. In fact, the idea that many naturally occurring proteins fold quickly and reliably to their native state despite an enormous number of possible structural configurations is a paradox that has confounded researchers for decades. If each protein worked its way through all the possible variations in its structure before returning to its native state, folding could take billions of years. Yet many proteins can unfold and fold back into their original structure in seconds or less.
In their quest to understand the folding and unfolding process and predict protein structure from amino acid sequences, Simmerling and his research group began by looking at the smallest protein known with normal folding properties. The 20 amino acid sequence Tryptophan cage, or Trp cage, was isolated and developed by Niels Anderson, professor of chemistry and biomolecular structure and design, and his colleagues at the University of Washington at Seattle. Trp cage is a simple version of a protein derived from Gila monster saliva that could have potential as in biomedical applications for misfolding diseases. Other reseachers in the field have used simpler models of Trp cage to predict the final structure of the protein, but none of them have previously simulated the full folding process.
The researchers carried out a series of molecular dynamics simulations, with an allocation of 960,000 hours on NCSA's Tungsten and Platinum clusters to look at Trp cage and other small proteins. They predicted the structure of Trp cage from its amino acid sequence and compared their computer simulations to Anderson's results from physical experiments in which Trp cage was folded in a lab and then measured to plot where individual atoms are naturally positioned. The level of accuracy in the groups' predictions created a stir in the field of biochemistry when they published their findings in the Journal of the American Chemical Society in 2002 because no researchers had ever before simulated protein folding in such a fine level of detail. The final structure of the simulated Trp cage was virtually indistinguishable from the experimental data.
Using the supercomputers, the group can now run multiple simulations at the same time. Simmerling says that the advantage of running multiple simulations in parallel is that the simulations can “talk” to each other, communicating information that can be shared and cutting down on computational time. This benefit enables the researchers to study slightly larger molecules on which there are existing physical experiments, rather than the smaller mini-proteins in very simplified water environments they were previously limited to in simulation.
The new simulations, demonstrating how Trp cage folds when submerged in thousands of individual water molecules, allow the researchers learn what structure emerges from the protein in a particular environment and how the protein forms that structure in specific conditions. The team found out that because the protein has a hydrophobic core it folds rapidly in water to protect its inner atoms from the surrounding water molecules.
Simulating what causes protein folding may enable researchers to put that knowledge to use. For example, Trp cage's transformation to a particularly stable structure is consistently sparked by submerging it in water at a specific temperature threshold. That information could be important to researchers who design drugs, helping them. produce structure-based drugs that are more stable in water-based solutions at specific temperatures, such as normal human body temperature.
Cracking The Code
Simmerling says the work of his group and others also will help make resources such as genome databases more readily useable. A genome represents all the DNA sequences for an organism–genetic codes responsible for the sequence of amino acids that constitute proteins. While genome databases currently provide information about the sequences of amino acids, they don't provide information about the structures of the proteins. Since the functionality of the proteins is determined by the configurations they form upon folding, the lack of structural information makes genomic sequencing difficult to apply to real- world medicine.
A sequenced genome is “like having a book written in another language,” Simmerling says. “You can see patterns, but may have no idea of their meaning.” The simulations produced by Simmerling and his colleagues on NCSA supercomputers may someday allow researchers to translate all those books of genomes and begin to understand why certain amino acid patterns produce diseases and how to design the structure-based drugs that will counteract them.