Left: Microscope image of the healthy neuromuscular junction (NMJ). Bright spots are where a fluorescent-label has stuck to muscle-cell-surface proteins that respond to neurotransmitters. Upper Right: MCell model of healthy mouse NMJ active zones (AZs). Six AZs are positioned in a simulation box, and each AZ consists of voltage-gated calcium channels (VGCCs) (small red dots) other cell-surface proteins (blue dots) and two neurotransmitter-containing synaptic vesicles (gold spheres). Lower Right: LEMS AZs, showing how the geometry has been disrupted and some calcium channels are now “L-type” channels with less activity (yellow).

Bridges-2 Sims of Lambert-Eaton Myasthenic Syndrome Identify Multiple Factors that May Need to Be Addressed

In Lambert-Eaton myasthenic syndrome (LEMS), patients’ immune systems attack a critical nerve cell protein, causing muscle weakness and degeneration. But not always — some patients lack antibodies to this protein, the calcium channel. Using the MCell simulation software developed in collaboration with PSC, running on PSC’s Bridges-2 system, a team from the University of Pittsburgh and PSC showed that just blocking the calcium channel doesn’t fully reproduce the syndrome. They identified several factors that may contribute to LEMS, via the mix of antibodies in each individual patient. The results may help explain the historically limited success in treating LEMS and point to possible better strategies.


Lambert-Eaton myasthenic syndrome, or LEMS, is a devastating autoimmune disease that causes muscle weakness and degeneration. LEMS is caused by the immune system’s own antibodies attacking the neuromuscular junction (NMJ) between nerve cells and the muscle cells they are meant to signal. Patients with the syndrome often get progressively weaker, losing the ability to walk or even get out of bed. To make matters worse, FIRDAPSE, the only FDA-approved drug for LEMS, helps patients, but only partly. In most patients with LEMS, their antibodies attack the calcium channel protein. This protein allows calcium to cross the nerve-cell membrane, in turn making the cell release neurotransmitters that fire the muscle cell on the other side of the NMJ. One prevailing theory was that blocking calcium channel activity and the signal to the muscle cell caused LEMS. But wait: 10 to 15% of patients with LEMS don’t have measurable levels of anti-calcium-channel antibodies. Also, in the only animal model for LEMS, the mouse, you can mimic the disease by introducing the mix of antibodies from individual patients into the mouse NMJ. But you can’t fully replicate it with antibodies to only the calcium channels. Some doctors and scientists wondered what that meant. Were there other problems in LEMS patients besides calcium channels that might need to be addressed for a more effective treatment?

“The reality is that every patient has a different mix of antibodies … Some don’t have the antibodies to the calcium channel, but they have antibodies to other nerve-cell proteins. And so this has kind of been percolating among the scientists for years … Everybody just assumed, ‘Yeah, OK, they make a lot of different antibodies. But the calcium channel antibodies are the important ones.’ Well, turns out we never really were sure that was the case, and we wanted to investigate that more carefully.”
— Steve Meriney, University of Pittsburgh

The University of Pittsburgh’s Steve Meriney and PSC’s Rozita Laghaei wondered whether their successful computer simulations of the NMJ could be used to see what factors contribute to the full clinical picture in LEMS. These simulations used software called MCell as well as PSC’s NSF-funded Bridges-2 supercomputer. MCell was developed by the NIH-funded Multiscale Modeling of Biological Systems Center, established by the University of Pittsburgh, PSC, the Salk Institute, and Carnegie Mellon University.


MCell, which stands for Monte Carlo Cell, uses a specialized Monte Carlo algorithm to simulate the movements and reactions happening inside and around the cells to produce a realistic version of living cells and their component parts in the computer. In this type of simulation, the parts of each cell act as they do by simple rules that copy how they act in life. The trick is that, like a gambler at Monte Carlo making many bets to try to “win,” MCell simulates the cells’ behavior many times over. By averaging the cells’ behavior over many trials, MCell can both produce a realistic average behavior as well as give an idea of how consistent that behavior is. The team ran their NMJ simulations on MCell using Bridges-2’s regular memory nodes. With 256 megabytes of memory apiece, the RM nodes would qualify as “large memory” in many supercomputers. The scientists needed this memory, because the simulations would have to track the behavior of dozens of cell components over thousands of trials. The team got access to Bridges-2 through ACCESS, the NSF’s ecosystem of supercomputing resources, as well as the previous NSF-sponsored XSEDE, in both of which PSC is a leading member.

We thank XSEDE, and now ACCESS, [for giving] us this really amazing supercomputer. We ran all the simulations and analyses, which need a lot of memory, at least 6,000 times [each] … It is impossible to do all of this in a computer like a desktop.

— Rozita Laghaei, PSC

Meriney and Laghaei simulated the behavior of the mouse NMJ assuming that different things were causing the disorder. Would blocking calcium channels mimic the disease? Would stopping the activities of other crucial nerve-membrane proteins replicate the disorder? What about disrupting the physical organization of proteins and transmitter-releasing vesicles? Electron microscope studies have shown such disruptions in LEMS patients.

The results of the simulations were eye-opening. Stopping the calcium channels alone replicated the disease only partially. That matched the lab experiments in mice. Blocking other proteins in the nerve cell membrane, particularly the calcium sensor protein synaptotagmin, also partially reproduced the syndrome. Disrupting the position of the calcium channels in the cell membrane made a surprisingly large difference. Movements of as little as 5 nanometers — about one five-millionth of an inch, and comparable to the size of the calcium channel protein — caused LEMS-like dysfunction in the NMJ. The simulations needed to mix and match the effects of nearly half a dozen possible factors to fully replicate LEMS.

The results supported researchers’ suspicions that individual patients’ LEMS symptoms arise from different combinations of factors, stemming from their particular mix of autoimmune antibodies. This isn’t exactly good news. It reveals that LEMS is more complex than doctors would have liked. But it puts doctors and scientists on firmer ground, with a better understanding of how the disorder actually develops. The findings, which the Pitt/PSC team reported in the Journal of Neurophysiology in May 2023, promise a more complete approach to treating LEMS that more fully reverses the effects of the disease.