New Understanding of Life and Its Processes: Cellular Biology

spacer Where Nerve and Muscle Meet spacer

Photo: Joel Stiles Joel Stiles, Pittbsurgh Supercomputing Center.


Modeling the nerve-muscle junction identifies a previously unknown disease process.

From birth, John's muscles didn't work the way they should. At a month old, he couldn't hold up his "floppy" head. At six months, his eyelids were droopy. At 10 months, he still couldn't sit without support. By age nine, he needed a wheelchair to get around.

Now 15, John — not his real name — still lives with the difficulties of his condition. His symptoms, muscle weakness and fatigue that progress over time, are characteristic of a neurological disorder called slow-channel congenital myasthenic syndrome. In many cases of SCCMS, symptoms aren't as pronounced at an early age as they were for John, and some doctors fail to recognize it as a disorder. They may even think it's behavioral — a lazy kid. Others diagnose SCCMS as a related disorder, myasthenia gravis, a mistake that can worsen SCCMS by treating it with the wrong drug.

With John, doctors suspected SCCMS, but a series of lab tests on biopsies from his muscle tissue revealed something puzzling. The electrical current that triggers muscle cells to fire continued longer than it should before shutting off — the "slow-channel" response that gives SCCMS its name. But even at 16 months, the current was also very weak, like a much more advanced case. And the structure of his muscle cells at the nerve-muscle junction, the synapse, was inconsistent with the weak current.

"What was unusual," says PSC senior scientist Joel Stiles, "is that he presented early in life with currents that were long and small, but the synaptic structure wasn't terribly abnormal. He had muscular weakness but mild structural change compared to classical slow-channel syndrome."

SCCMS is a catch phrase for a number of different but related genetic defects, all of which lead to mutations in a protein, AChR, that plays a key role in muscle movement. A receptor protein in muscle cells, AChR works like a combination watchman and gate. It receives chemical signals from the nerve cells that tell it to open and flips the switch to become a channel for the flow of ions — mainly sodium — into the muscle cell, a bioelectrical current that triggers movement. Lab studies showed a mutation in John's AChR different from any other documented case of SCCMS. Could this novel mutation lead to different AChR behavior, something not seen before that might explain John's unusual condition?

A medical doctor with a Ph.D. in physiology, Stiles is a specialist in computational neuroscience. With Thomas Bartol of the Salk Institute, he authored software called MCell that's used in two dozen labs around the world to simulate the microphysiology of nerve cells interacting with other cells. John's situation led Stiles to thinking. "If the synaptic structure isn't terribly abnormal, but the currents are still smaller than normal as well as longer than normal, there has to be something else going on in the receptor, other than just getting stuck in the open state."

With MCell, Stiles had a tool to test his hunch. From a series of simulations, he deduced that the AChR receptors were not only slow to close, but also slow to open. Following from this lead, lab studies have confirmed that, among other subtle changes, this novel AChR mutation causes the channel to open slowly — invaluable new knowledge that can help in arriving at appropriate drug therapy as well as in research to develop new and better treatments.

Model Neuromuscular Junction

Model Neuromuscular Junction
The normal architecture (top) contrasts with the focally deformed architecture (bottom). The slightly widened clefts and dimpled depressions of the deformed muscle membrane represent the microphysiology of the novel SCCMS. In both cases, the model includes a nerve membrane (semi-transparent sheet) and 30 synaptic vesicles, five arrayed above each cleft in the muscle membrane. Four of these (red) are release sites for neurotransmitter molecules. Two other release sites (blue and black) are adjacent to focal deformations.

Download a larger version of this image.

Sweeps in Parameter Space

MCell is a powerful biological research technology that fills a gap between smaller-scale simulations that model molecules atom-by-atom and larger-scale approaches that model whole cells or groups of cells. MCell models the space and boundaries within and between cells, and then tracks the release and dispersion of diffusing molecules and their reactions with cellular membrane proteins, such as AChR.

MCell starts with a highly realistic 3-D model of the cellular space — reconstructed from electron microscope scans. It simulates molecular diffusion by means of algorithms that model the random motion of the molecules, and at each succeeding slice of time — often a microsecond or less — it uses statistical methods to test for all the possible reactions.

With an Information Technology Research grant from the National Science Foundation, Stiles and Bartol are collaborating with computer scientists at the Salk Institute, the University of California, San Diego and the University of Tennessee — under the direction of Francine Berman of UC San Diego — to expand MCell's usefulness. The idea is to create a "virtual instrument," software with an easy-to-use interface that can run on a "grid" of many different computers at the same time and keep track of thousands of simultaneous computations.

"What's embedded in this kind of work," say Stiles, "is the need to examine the parameter space that the model sits in. Many unknowns go into this detailed a model — for example, the reaction between the neurotransmitter and the receptor and other proteins that contribute to the cellular system. In many cases, these inputs are poorly understood, and the modeler needs to examine how the output signal varies with changes in input. If you have a model with 20, 50 or 100 input parameters, they may all need to be varied in some systematic way, and with all those simulations you get into what we call gigantic parameter sweep scenarios."

Stiles and Bartol and PSC systems engineer Stuart Pomerantz are working on MCell project-design issues and on a graphical user interface that will allow researchers to easily control these parameter-sweep simulations. At the same time, the team of computer scientists is developing sophisticated ability to "steer" them. With software that can monitor simulations in progress, the researcher will be able to shift mid-stream, with no loss of data, to refine the parameters and test possibilities suggested by interim results.


An MCell Closeup
<-- The globally deformed junction during a simulation with neurotransmitter molecules releasing from a vesicle (red). Another vesicle (black) isn't releasing at this moment. The simulation tests for the presence of neurotransmitters at two sampling regions (shown as boxes). The colored markers on the muscle membrane indicate AChR receptors, in five different states of interaction with the neurotransmitters, ranging from unbound (blue), to intermediately bound (red & purple), fully bound (green) and activated (yellow). Other markers (white and black) represent an enzyme, AChE. "In a normal muscle membrane," says Stiles, "yellow would predominate because a large fraction of the receptors would be open and making current."

Download a larger version of this image.

Model Neuromuscular Junction

Thin sections through normal (left) and focally deformed (right) model neuromuscular junctions. Red, green and blue coding of the muscle membrane indicates crest, intermediate and deep regions of the muscle's folded structure, which correspond to decreasing density of receptor sites.

Download a larger version of this image.

Model Neuromuscular Junction

A representative thin section through the globally deformed model neuromuscular junction. Increasing the distance between the nerve and muscle membrane (150 nanometers) modeled deformation similar to severe, late SCCMS.

Download a larger version of this image.

Zeroing in with MCell

With the novel case of SCCMS, the key parameters were relatively well defined. "The question," says Stiles, "was what combination of factors in the face of relatively normal structure gives us both small and slow currents."

The muscle-cell membrane at the synapse is an irregularly curved surface — deep clefts lined with thousands of AChRs and other molecules. In these spaces, neurotransmitter molecules disperse and react with the receptors and with enzymes. Stiles ran a series of MCell simulations that compared AChR response in three different structural states — normal, focally deformed and globally deformed. Focal deformations, a relatively mild widening of some of the clefts, mimicked the biopsy at 16 months. The globally deformed structure represented advanced SCCMS.

As expected, MCell showed that slowing the closing rate of the AChR channel, so that more AChRs are open longer, extends duration of the electrical signal. MCell also showed, however, for both focally and globally deformed structure, that slowing the channel-closing rate doesn't reduce current amplitude. To the contrary, current peaks on average get slightly higher. Something else had to account for John's unusual condition.

Stiles knew from prior modeling that channel opening could interact with channel closing to affect current amplitude. When he modeled a combination of slow closing and opening rates for the focally deformed structure, he found an AChR response similar to the biopsy data.

"The modeling led us to say 'Now wait a minute,'" says Stiles. "There has to be something else, something beyond the classical picture of this disease, which is just stuck in the open state. It has to have trouble getting into the open state as well. Both of these defects have to be present to explain the progression of measurements in this patient." Lab work by Stiles's collaborators subsequently confirmed this insight and demonstrated that this novel AChR mutation leads to a previously unknown disease mechanism.

Download PDF Where Nerve and Muscle Meet
from Projects in Scientific Computing 2001
Researchers Joel Stiles, Pittsburgh Supercomputing Center
Software MCell
Related Material
on the Web
An impulse deconstructed: Scientist's software simulates intricate functions of the central nervous system — Pittsburgh Post Gazette
Biomedical Initiative, Pittsburgh Supercomputing Center
Myasthenic Syndromes
References Gomez, Maselli, Vohra, Navedo, Stiles, Charnet, Schott, Rojas, Keesey, Verity, Wollmann & Lasalde, "A Novel Delta Subunit Mutation in Slow-Channel Syndrome Causes Severe Weakness by Delayed AChR Opening and Widening Synaptic Cleft," Annals of Neurology 51 (1), 102-12 (2002).

Henri Casanova, Thomas M. Bartol, Jr., Joel Stiles & Francine Berman, "Distributing MCell Simulations on the Grid," International Journal of High-Performance Computing Applications 15 (3), 243-57 (2001).
Writing: Michael Schneider
HTML Layout/Coding: R. Sean Fulton
Graphic: PSC Logo