INDUSTRY

by Kathleen M. Wong -- Molecular dynamics simulations on an Alliance cluster help biochemists understand how damaged DNA is recognized by a cell's genetic repair system. Deep in the heart of every living cell, the DNA police are out in force. Patrolling precincts and checking IDs, they keep tabs on any suspicious movements, any deviant behaviors. One squad, known as the nucleotide excision repair system (NER), is so indispensable that no cell -- whether from a bacterium or a blue whale -- can go on for long without them. The forte of NER is an uncanny ability to recognize friend from foe. From a dizzying range of healthy DNA sequences, it can recognize an impressive lineup of bad-apple mistakes, snip out the errors, and repair the chain good as new. At the same time, it doesn't step on the toes of other DNA repair systems, which fix other types of damage such as base mismatches. Keeping the message error-free is critical so that when the time comes for a cell to divide, it will pass on DNA free of potentially dangerous mutations. In fact, bacteria and higher organisms have evolved two entirely different NER systems that can recognize and repair similar types of DNA damage. This suggests how important it is to keep an organism's genetic inheritance intact. But just how NER systems can find a few bad bases in a lengthy chain of sound DNA remains unknown. Because these systems are so complex and poorly understood, two biochemistry researchers at the University of Kentucky are approaching the problem from the opposite direction. "What we're concerned with is, how does it recognize damage? And under what circumstances does the NER miss a problem?" says professor of biochemistry H. Peter Spielmann. Together with doctoral biochemistry student R. Jake Isaacs, Spielmann is working with Alliance supercomputers housed at the University of Kentucky to determine what it is about damaged DNA that makes it recognizable to NER systems. The weakest link Because DNA damage comes in so many forms, NER probably doesn't home in on specific errors. Instead, scientists theorize, these systems may be detecting changes in the stability of the DNA itself. In the watery environment of a cell, DNA is never still. Its component base pairs twist and roll, its molecular moieties spin and whirl, and its phosphodiester backbone flexes and stretches with all the action of a disco-dancing John Travolta. Scientists suspect that NER operates by gauging the range of motion in a given section of DNA. Working its way down a strand of DNA, NER may bind to, then bypass, sequences with the more sedate movements of stable, healthy DNA and stop to fix weak, wobbly sections bearing errors. By simulating the intricate, moment-by-moment gyrations of both damaged and undamaged DNA sequences on a supercomputer, Isaacs and Spielmann hope to determine why some are recognized and repaired by NER and others are not. "We're trying to approach this from an experimental side and are using molecular dynamics to flesh out what's happening, to fill in the physical picture for us," Isaacs says. "Ultimately, we would like to develop a predictive model so that we could look at the molecular dynamics of some damage types and say, 'This is going to be recognized by NER, whereas this might not be.'" Isaacs and Spielmann have already conducted experiments to determine how a few sample DNA sequences behave in real life. Some of their sequences represent DNA damaged by known mutagenic chemicals; others represent common coding errors such as the substitution of one base for another; and still others represent types of damage not recognized or fixed by NER. Using nuclear magnetic resonance (NMR) spectroscopy, the researchers gained a snapshot of the equilibrium conformation, or shape, each molecule assumes most often. Spielmann likens it to a photograph of his daughter. "It's an accurate representation, except that she usually is moving around." The researchers also subjected the sample sequences to a more advanced type of NMR imaging. This technique yields a fuzzy picture of how tightly coupled the movement of certain carbon-hydrogen pairs in the molecule are to the molecule as a whole. "It tells us that this atom moves more independently, or is more disordered than, its neighbors," Spielmann says. It also shows how much area each carbon-hydrogen group sweeps out in space, but not the exact coordinates of its movements. "It could be going around in a circle, a cone, up or down; you don't really know," Isaacs says. Most importantly, NMR cannot demonstrate that any given atom is moving in concert with several others. Known as a vibrational mode, this information is necessary to reconstruct the dynamic responses of a DNA sequence to an NER inspection. Molecular moviemaking Armed with their experimental data, the researchers have turned to Alliance supercomputers to fill in the gaps. "We're making a movie of how this DNA molecule is wiggling around in solution, how it's changing its conformation over time," Isaacs says. They're tracking the enormously complicated movements of between 10,000 and 20,000 atoms over the course of five to ten nanoseconds. Most are the water and sodium solvent molecules so important to the behavior of DNA within a cell. Like cartoon animators obsessed with detail, the scientists are saving a million snapshots of the molecule's coordinates for every nanosecond of the simulation to get a reasonably realistic record of the action. "That's what's great about computers -- we can see every atom. From that, we can calculate believable vibrational modes," Spielmann says. To run their simulations, the scientists first used the University of Kentucky's Hewlett-Packard N-4000 complex. Now they use the university's HP Superdome, which replaced the N-4000 complex last December. Initial runs have taken up about a third of the 100,000 hours of supercomputing time they have been allotted. Spielmann and Isaacs are using the Amber 6.0 software package developed by researchers at the University of California, San Francisco, to simulate so-called molecular dynamics trajectories. The team has already completed runs for each of their sample molecules with the program's default settings. However, the results only loosely agreed with their experimental data. Isaacs is now tweaking the program to get outputs that match the team's NMR data. Once he's finished fine-tuning the model, the team hopes to use it to derive some general principles about the effects of nucleotide sequences and DNA damage on DNA motion. Someday researchers will be able to plug in sequence information about a given stretch of DNA sequence and expect a model program to simulate its behavior quite realistically. By then, perhaps, the secret workings of the NER cell police will be out in the open. ----- Our thanks to NCSA's Access: http://access.ncsa.uiuc.edu/Stories/dynamicdna/