Drug Discovery
Computer Screening Uncovers Compounds Against ALS

Neuroscience
Optic Nerve Regrown in Mice

Public Health
Stats Tool Puts Health Disparities on the Map

Statistical Method Detects Disease Outbreaks Without Background Population Data

Molecular Modulator of Synaptic Plasticity Revealed

Hot Spots for Genetic Recombination Different in Chimps and Humans

Proceedings of the HMS Faculty Council

New Appointments to Full Professor

Nominations Invited for Biostatistics Alum Award

Grants Offered in Women’s Health

Honors and Advances

Madras Receives Public Service Award

Students Bag Pharma in Second Year Show

Front Page

Computer Screening Uncovers Compounds Against ALS

Findings May Recharge in silico Screening in General and Targeted Research on Amyotrophic Lateral Sclerosis

Working alone in his apartment on a passel of past-their-prime computers, an HMS researcher has made a discovery that he and his colleagues believe could lead to promising therapies for inherited forms of the muscle-crippling disease amyotrophic lateral sclerosis (ALS). In contrast to sporadic forms, which have no obvious distinguishing hallmarks, familial ALS tends to be characterized by the appearance of dense jots of mutant protein in motor neurons of the spinal cord. Using a computer-based method for matching molecules to specific targets—dubbed in silico screening—Soumya Ray identified a set of virtual compounds that when synthesized and tested in actual lab dishes, prevented the mutant protein, SOD1, from clumping.

Soumya Ray appears next to the “Beowulf” computer cluster he and colleagues used to find molecules that are potential leads in their drug-discovery program for amyotrophic lateral sclerosis. (Photo courtesy of Soumya Ray)

The findings by Ray and his colleagues Peter Lansbury, Robert Brown, and Richard Nowak could bring some fresh air to two stifled research areas, ALS and in silico screening. The lack of progress in understanding and treating ALS, which currently affects 35,000 Americans, is well known. In silico screening has created a buzz in the rarefied drug discovery world, and yet few in silico compounds have successfully made the transition from the virtual world to the lab. “I think, to be honest, there are many in silico screens that have been tried, but they ask too much of the method. In our case, we were limited in resources. We could only screen a couple of million compounds,” said Ray, HMS instructor in neurology at Brig-ham and Women’s Hospital. “It is kind of crude, most theoretical people will tell you,” said Lansbury, HMS professor of neurology at BWH. Their study appears in the Feb. 14 online Proceedings of the National Academy of Sciences.

What their bare-bones approach also had going for it was a stripped-down working hypothesis. One of the big puzzles in familial ALS research is why mutant forms of SOD1 are so prone to clumping. The researchers embarked on their in silico screen with a simple idea in mind: mutant SOD1, a two-part, or dimeric, structure is unstable, and its tendency to fall apart is what leads to aggregation. Ray identified what he thought might be SOD1’s weak spot, a cavity between the two halves of the dimer, and set his computers to the task of selecting compounds from a public database that would fill it, essentially soldering the molecule. He and his colleagues bought real versions of the 100 most promising compounds and mixed them with mutant SOD1 in the hope that they would inhibit aggregation, which 15 of them did.

From Clumps to Cures
If the compounds do the same thing in living animals and if they also slow disease, it could help solve a second ALS puzzle: are the sinister-looking clumps actually causing disease and, if so, how? Lansbury, who is cofounder of the Laboratory for Drug Discovery in Neurodegeneration in the Harvard Center for Neurodegeneration and Repair (HCNR), is eager to make the move into animals, though his reasons are more pragmatic than theoretical. “We may never know what it is about aggregation that causes disease—it could be five different things. We want to prevent all of it,” he said. “The drive in the basic science community is so much to understand the details before one goes forward. If that attitude was followed throughout history, we would not have any medicines.”

Keeping it together. Mutant SOD1, the protein associated with familial forms of amyotrophic lateral sclerosis, is a highly unstable dimer (top) that tends to fall apart (bottom left). Soumya Ray, along with colleagues including Peter Lansbury and Robert Brown (inset, right and left) were able to stabilize the mutant and prevent it from forming aggregates, which potentially cause disease, by filling its cavity with compounds discovered by an in silico screen. (Image adapted by Rachel Meyer; inset photos by Steve Gilbert)

Though Lansbury and his HCNR colleagues are prepared to conduct animal trials themselves, developing the compounds into actual drugs would be another story. Most academic labs simply do not have the resources to bring a drug to market. And there would not be much financial payoff even if they could, since the compounds they found are in the public domain. He is hoping to persuade a pharmaceutical company—or better yet, a bunch of them—to undertake in silico screenings of their own compounds. “The molecules in the basements of Hoffman–La Roche or Pfizer are unbelievable,” he said. “They have completely proprietary collections with really interesting biological properties. I guarantee you that not one of them has been screened for an ALS target.”

Suspect SOD1
Mutant SOD1, the protein at the center of the inquiry, is a quirky character. Normally, it performs the crucial role of scavenging potentially damaging superoxide ions. What is odd is that it is so vulnerable to mutation. Mutant SOD1, which Brown, HMS professor of neurology at Massachusetts General Hospital, discovered in 1993, comes in 114 forms. “Other things about this protein are fishy, like why it is not more stable than it is,” Lansbury said.

He and his colleagues suspected that it was the mutant’s instability, rather than a loss of the scavenging function, that might, via its propensity to aggregate, lead to disease. He began looking for ways to stabilize it. In an early attempt to correct the defect, Ray designed a mutant SOD1 in which the two halves were essentially stapled together by a disulfide bond. The strapped together molecule did not aggregate in vitro. Upon closer inspection of his own creation and of a crystal structure of the most common SOD1 variant, A4V, solved by a group of British researchers, Ray noticed that there was a cavity where the two subunits meet.

“We thought if we could fill this cavity with a druglike molecule, maybe that would stabilize the whole protein,” Lansbury said. An experimental chemist, Lansbury was planning to do a robotic screen when Ray approached him with the idea of using a computer. Lansbury was both intimidated and incredulous. “I never thought it would work. I just did not believe him. I am from the computer card era,” he said.

“We may never know what it is about aggregation that causes disease—it could be five different things. We want to prevent all of it.”

But Ray summoned his resources. “Many of my friends donated their old computers that were sitting around,” he said. “I happen to know how to assemble them to make a parallel processer.” In his apartment, working till late at night, Ray began his computer simulation by hitting the cavity of a computer model of the A4V mutant with 2 million virtual molecules, drawn from a public database. His computer program automatically ranked the molecules according to fit.

When tested in the lab, 80 of the top 100 hits had some effect, and 15 had a significant effect in stabilizing SOD1. Not only did these 15 slow or inhibit aggregation, they also made it more difficult for the A4V mutant to unfold in the presence of a denaturing agent. Interestingly, they did not have that effect on Ray’s early cavity-occluded SOD1 version. “So eliminating the cavity, these molecules had no effect, which confirmed that they were binding at the site we chose.” What is more, the molecules appear to bind not just A4V but other SOD1 mutants.

Such versatility could be yet another selling point in Lansbury’s bid to get pharmaceutical companies to raid their own libraries for mutant SOD1 binding molecules. This could be done quickly and cheaply. “If all the drug companies spent one day of person time, they could potentially turn up some real leads,” he said. But there are sticking points as well. Though there is much public and even industry sympathy for ALS, which typically kills patients in five years, it is a relatively rare disease and the number of familial cases is just five percent of the total, so an effective medication would have a limited commercial market. Still, Lansbury is pursuing his campaign with born-again zeal. “He is now a believer, I can tell you that,” said Ray.

—Misia Landau

top