To the mind of Dr. Christopher Gomez, computer models seemed to be just so much mumbo-jumbo. A neurologist at the University of Minnesota, he was uneasy with the notion that a computer, fed a series of equations and raw data, could realistically mimic something as complex as the interaction of a nerve and a muscle.
 |
 |
|
Dr. Joel Stiles, using a computer simulation, was able to pinpoint why a young patient was not showing the muscle degeneration typical of his disease. (John Beale/Post-Gazette) |
So it was less out of conviction than exasperation a couple of years ago that Gomez sought the help of Dr. Joel Stiles, a computational neuroscientist then at Cornell University, in figuring out what was causing a young patient to grow progressively weaker.
The boy was a puzzle. Since birth, he had suffered a rare neuromuscular disorder called slow channel syndrome. First his head was floppy, then his eyelids drooped and, years later, the steady loss of muscle strength put him in a wheelchair. But the boy's muscle biopsies, performed annually, didn't look typical of patients with the syndrome. The boy was now entering his teens, yet Gomez, an international authority on the disorder, still wasn't sure why the boy's nervous system sent weak signals to his muscles.
Despite Gomez's initial reservations, the numerical model that Stiles developed of the boy's neuromuscular junctions -- the points where nerves and muscles meet -- proved insightful. Computer simulations using the model pinpointed a possible culprit: balky proteins, called receptors, lining the muscle side of the junction.
Laboratory tests later confirmed that Stiles and his computer model were correct and that the boy had a defect never before seen in a patient with slow channel syndrome.
The finding, due to be reported soon in a neurology journal, hasn't yet benefited the boy, now 15 and living in Los Angeles. But it demonstrates how computer models of the nervous system might be used to identify the mechanisms that underlie diseases or to predict how a disease might respond to therapy.
Powerful new computers are making it possible for scientists to construct highly detailed, three-dimensional models that can keep track of a dizzying number of interactions between tissues and body chemicals.
"At this stage of the game, we're just scratching the surface," said Stiles, who joined the staff of the Pittsburgh Supercomputing Center a year ago. The neuromuscular junction, or synapse, that he modeled in the boy's case is relatively simple compared to the synapses that link nerve cells in brain, the site computational neuroscientists would most like to simulate.
"You'd like to see this applied to more complex sites in the central nervous system," agreed Mark Ellisman, a professor of neuroscience at the University of California, San Diego. There, the models might be used to show what goes wrong in the nervous system to cause such devastating diseases as Parkinson's, dementia and schizophrenia.
Promise, and challenge
Use of computer models for "rational detection" of disease mechanisms is as promising as "rational drug design," a much-heralded approach to developing drugs by using computers to custom-design molecules that treat disease without the dangerous side effects of drugs derived from nature.
But despite years of hoopla over rational drug design, most drugs continued to be derived from plants, not from computer programs. Realizing the promise of rational detection will be similarly difficult, said Ellisman, neuroscience coordinator for the National Partnership for Advanced Computational Infrastructure, a group affiliated with the San Diego Supercomputer Center.
"Here you've got a case where rational detection ... has led to a better understanding of a disease," Ellisman said. "There's not many examples of that yet."
Computers have made major inroads in biology, notably in gene sequencing and in predicting protein structures. But simulating the function of physiological systems "is still pretty rare," Stiles said. Even the neuromuscular junction, though far less complex than the connections between brain cells, can be a challenge for computer models.
At this junction, tiny electrical impulses traveling through thin nerve cells must be amplified and translated into chemical signals that cause the more massive muscle cell to contract. And because the muscle must react rapidly, the signal must be one that can be turned on and off quickly.
It's a multi-step process involving a number of players. When an electrical signal travels down the long nerve, it stimulates the release at the nerve tip of a chemical called acetylcholine. This chemical, a neurotransmitter, enters the tiny space between the nerve and the muscle, called the synaptic cleft. The muscle side of the cleft, formed into a series of deep, narrow canyons, is lined with proteins called receptors.
When an acetylcholine molecule makes its way across the cleft and binds with an acetylcholine receptor, the receptor is activated and becomes a channel that allows electrically charged sodium atoms -- also present in the synaptic cleft -- to pour into the muscle. The sodium infusion causes the muscle to contract. Another denizen of the synaptic cleft, an enzyme called acetylcholinesterase, subsequently breaks acetylcholine apart, deactivating the receptor and relaxing the muscle.
When any part of this process is disrupted, bad things can happen. For instance, nerve gas, such as sarin, works by inhibiting the acetylcholine enzyme, so the muscles never relax; the resulting paralysis leads to asphyxiation.
Another example is myasthenia gravis, a disease that causes droopy eyelids and other muscle weakness. This is an autoimmune disease in which the body's immune system attacks the acetylcholine receptors. As the number of receptors declines, the signals sent to the muscles become weaker.
Slow channel syndrome -- or, slow channel congenital myasthenic syndrome -- also leads to muscle weakness. In this collection of disorders, children are born with acetylcholine receptors that are slow to deactivate. Thus, the channel that allows sodium ions to enter the muscle sticks open long enough to allow other ions -- calcium ions -- to also enter the muscle. The calcium causes tissue deterioration, making it difficult for the muscle to receive signals.
What puzzled Gomez about his young patient from Los Angeles is that the boy displayed all of the symptoms of slow channel syndrome, but his initial muscle biopsies revealed little of the tissue degeneration normally seen. Why did he resist this deterioration and what would account for the muscle weakness?
"We had some clues before I met Joel," Gomez recalled. He isolated a genetic mutation responsible for the boy's problem, but didn't know precisely what effect the mutation had. It was an exceedingly complex puzzle and he and his colleagues could never put it together.
Gomez was aware of Stiles' work with computer models, however, and suspected that perhaps he might be able to help.
"I had no idea what a perfect person he was to call," Gomez said. "Stiles is just a brilliant guy."
The detective work begins
While working with the late Cornell neurobiologist Miriam Salpeter, Stiles and a colleague, Thomas Bartol, now at the Salk Institute in La Jolla, Calif., had developed a software program called MCell. The program allows researchers to numerically represent the shapes of structures, populate them with specific molecules and simulate chemical reactions in those spaces.
Gomez had noted some abnormalities in the structure in the boy's clefts, so Stiles first performed simulations to see if those altered shapes would affect the strength of the signals. The MCell simulations showed little, if any, effect.
So Stiles played a hunch. Normally in patients with slow channel syndrome, the receptor channels are quick to open and slow to close. But what if the channels were balky --slow to open as well as close? That might limit the amount of calcium that entered the muscle and account for the reduced amount of early damage.
Stiles' simulations showed that it could indeed account for what Gomez was seeing.
"He understood how the synapse worked far better than us," Gomez said. "We could never appreciate [such a subtle variation] while thinking about it holistically." Measuring the speed at which receptors open is not an easy or obvious experiment; it required cloning parts of the receptors and injecting them into frog eggs. But the measurement confirmed Stiles' prediction.
Whether this insight will help the boy is not clear yet. "We're still kinda thinking about the meaning of all this," Gomez said. Progression of the disorder has slowed and the boy could live for many more years. Quinidine, a drug used to treat heart rhythm disturbances, is used in some patients with slow channel syndrome because it boosts nerve signals to the muscles. But doctors are still trying to figure out if the drug would help or hurt, given the unusual performance of the boy's receptors.
"From a scientific perspective, this was a unique opportunity," Stiles said of the study. "But these things are never cool from the patient's perspective."
