ALS: A Mystery Almost Solved?

by Jim Schnabel

August 19, 2015


Ann Whitman
(212) 223-4040

Precisely how amyotrophic lateral sclerosis (ALS) kills people has long been mysterious. Even for cases driven by known gene mutations, scientists still don’t have a complete picture of how those mutations lead to illness. 

That’s not for want of trying. Long before the Ice Bucket Challenge, ALS research was a busy field, funded amply relative to the number of patients with the disease. In fiscal year 2013, it received about $3,000 in NIH funding per patient, compared to about $100 per patient for Alzheimer’s.

That ALS has been such a strong draw for researchers and research funds may be due, in part, to the keen sense of injustice it creates among patients and their loved ones. No other disease on its scale strikes seemingly healthy people in the prime of life and kills them so quickly, so implacably.

The good news is that decades of ALS research, helped by broad advances in research technology, are showing signs of paying off. Scientists seem to be zeroing in on the once-elusive mechanisms of the disease, and are starting to design and test therapies that target those mechanisms.

“We now understand much more about the pathogenic process than we did ten years ago,” says Don W. Cleveland, a long-time ALS researcher at the University of California—San Diego and member of the Dana Alliance for Brain Initiatives (DABI).


ALS has been in medical literature since 1869, but almost all the advances in understanding it have come since the development of modern molecular biology techniques in the 1980s.

In 1992, for example, researchers reported finding that ALS patients’ brain and spinal cord tissue have defects in the management of glutamate, the principal neurotransmitter that motor neurons use to signal each other. In particular, the scientists found evidence of reduced “glutamate transport,” which normally helps keep glutamate concentrations at safe levels in the synapses of motor neurons. Too much glutamate can kill neurons via a mechanism called excitotoxicity.

A clinical trial soon showed that riluzole, a compound that (among other things) inhibits glutamate signaling, extended the lives of ALS patients by a few months on average. Riluzole was approved by the US FDA in 1996 and remains the only officially sanctioned therapy for ALS. However, its benefit appears to be very modest at best, and other trials of drugs with anti-glutamate effects, including ceftriaxone, have failed in clinical trials with ALS patients.


ALS researchers made another apparent breakthrough in the early 1990s by finding that a small proportion of inherited, “familial” ALS cases are caused by mutations in SOD1—a gene that codes for an antioxidant enzyme called superoxide dismutase.

It takes only a single mutant copy of the gene to cause ALS. That fact, and studies in SOD1-mutant mice, soon established that SOD1 mutations cause disease by altering the SOD1 enzyme in a way that somehow makes it toxic to cells.

The mechanism of that toxicity hasn’t yet been found. But scientists have shown that in both “SOD1 mice” and in humans with SOD1-ALS, clumps or “inclusions” of the mutant enzyme form within motor neurons in disease-affected parts of the brain and spinal cord.

Researchers studying SOD1-ALS eventually found evidence that these clumps of mutant-SOD1 contain particularly tough aggregates known as amyloids. This suggested that ALS, or at least SOD1-ALS, might be part of the wider family of amyloid-aggregate diseases that includes Alzheimer’s disease, Parkinson’s disease, Huntington’s disease, and Creutzfeldt-Jakob disease.

Scientists have begun to appreciate that the amyloid aggregates in these diseases can be self-propagating: They can make new copies of their aggregate structures, or enlarge existing copies, by co-opting individual proteins.

That means that these aggregates can spread their corrupted state if they happen to be secreted from one cell (or released from a cell as it dies) and taken up by another—and that phenomenon might explain the progress of cell death and symptoms that is seen in ALS. Indeed, in 2011, researchers in the group of Anne Bertolotti at the Laboratory of Molecular Biology in Cambridge, UK reported that mutant SOD1 aggregates can spread from cell to cell in the lab dish.

“At first there was strong opposition, but since then some of the most skeptical people have seen this phenomenon in their own labs,” says Bertolotti.

Helper cells stop helping, start hurting

Studies of SOD1 mice helped provide another important clue to ALS. As Cleveland’s laboratory found, for example, in studies beginning in the late 1990s, mutant SOD1 hits helper cells called astrocytes particularly hard—and thereby harms the motor neurons that depend on these cells’ support. We were able to double the survival time of SOD1 mice after disease onset by lowering synthesis of mutant SOD1 just in astrocytes,” says Cleveland.

About eight years ago, researchers began to find evidence that astrocytes also drive ALS more directly, by somehow becoming toxic to nearby motor neurons—possibly by their production of SOD1 or other aggregates. This evidence initially came from studies of different types of SOD1-ALS, but in a pivotal finding in 2011, researchers showed that astrocytes from patients with common, sporadic ALS also have this toxic effect: healthy motor neurons cultured with them start to die off within days.

Microglia, another “helper cell” type, are the brain’s resident immune cells—cousins of the pathogen-gobbling macrophage cells of the ordinary immune system. In the early 1990s, researchers began noting that microglia in ALS-affected parts of the brain and spine are activated and surrounded by other signs of inflammation. Since then, studies have found that these inflamed microglia are not just there to clean up the damage caused by the disease, but, like astrocytes, directly worsen the disease process. Exactly how they do that isn’t yet clear, but there is already some evidence from ALS mouse models that one can block their harmful influence and slow the disease process significantly.

In 2012 came a report that oligodendrocytes, helper cells that build and maintain the myelin-protein insulation around neuronal axons, have a similar adverse role in ALS. Scientists found that oligodendrocytes in a commonly used type of SOD1 mouse die early in the disease course, and new replacement cells fail to mature normally—leading to a progressive loss of myelin insulation from the axons of motor neurons. The researchers found similar changes, including the loss of myelin from axons, in autopsied tissue from familial and sporadic ALS patients.

Affected motor neurons are big and vulnerable

One of the central questions about ALS has always been: why does it kill motor neurons (and nearby helper cells, we now know) but have much less—albeit someeffect on other types of neuron?

A likely factor is that the motor neurons most affected by the disease are gigantic in comparison with most other neurons—indeed with most other cell types. “They are 5,000 times the volume of a typical human cell,” says Cleveland.

Being much larger means having much greater needs for energy as well as general repair and maintenance systems. It also likely means having less leeway when those systems fail.

Moreover, in recent years, as ALS research has focused increasingly on protein aggregation, scientists have found that this is a factor against which motor neurons lack strong defenses. “Although they’re enormous, these motor neurons seem to have relatively poor mechanisms for dealing with protein aggregates and associated stresses,” says Cleveland.

And of course like almost all other neurons in mammals, motor neurons are virtually irreplaceable once they have matured and formed connections—connections that in some cases extend for more than a meter, from spine to feet.


Ninety-eight percent of ALS cases do not feature SOD1 gene mutations or SOD1-containing inclusions in motor neurons and support cells. Yet virtually all these non-SOD1 cases do feature inclusions of some type of aggregate. In 2006, University of Pennsylvania researchers John Q. Trojanowski and Virginia Lee, both DABI members, reported in Science that they and their colleagues had identified the major constituent of these non-SOD1 aggregates: a protein called TDP-43. 

In follow-up studies, the Trojanowski/Lee laboratory and others found these TDP-43 aggregates in the vast majority of ALS cases examined. Moreover, in 2008 researchers made a key finding that mutations in the TDP-43 gene account for some cases of familial ALS—as well as some cases that had been deemed sporadic.

“The fact that TDP-43 pathology is almost always found in areas of degeneration, and the fact that TDP-43 mutations can cause ALS, strongly suggest that TDP-43 is central to ALS,” Trojanowksi says.

How could TDP-43 drive ALS? The protein’s putative cell-harming role, and its normal function in cells, have both been hard to study, in part because of early difficulties in creating mice that have no TDP-43 (the mice die) or that successfully model ALS-like TDP-43 pathology.

Scientists do know that normal TDP-43 has its primary role in the cell nucleus, where it binds to both DNA and RNA, and—broadly speaking—helps in the handling of RNA-based gene transcripts, which will often go on to be translated into proteins. In ALS, however, TDP-43 proteins in affected cells tend to be found not in the nucleus but instead in the main compartment (cytoplasm) of the cell, where they form abnormal, granular or threadlike aggregates.

“The loss of TDP-43’s nuclear functions when it is trapped in the cytoplasm leads to changes in the levels and the splicing patterns of about 1,500 RNAs, which can’t be good—in fact we know that chronic loss of TDP-43 from the nucleus is lethal to mice,” says Cleveland.

“It is possible too,” says Trojanowski, “that TDP-43 aggregates become toxic by, for example, blocking molecular traffic within neurons and displacing organelles.”

There is some evidence that TDP-43, like mutant SOD1 and other neurodegenerative disease proteins, can form amyloid-type aggregates, which may then spread from cell to cell, thus helping to explain the typical progression of neuronal loss and symptoms in ALS. Work by Trojanowski and colleagues suggests that TDP-43 only rarely forms large, fibrous, insoluble amyloids like those seen in Alzheimer’s and Parkinson’s. But other researchers have reported that TDP-43 can form smaller, soluble aggregates (oligomers), which may be more relevant to the disease process.

Cleveland notes that many labs now are trying to establish that this aggregate-mediated spread really does happen in ALS. “It would astonish me if it weren’t occurring,” he says.

The broader question for TDP-43 research is whether the protein causes neuronal death mainly by moving from its usual functional location in the nucleus, or alternatively by acquiring some toxic property, involving aggregates or not, when it gathers in the cytoplasm.

Scientists now seem tantalizingly close to being able to resolve this question. Lee, Trojanowski, and their colleagues recently developed a mouse model in which human TDP-43 is produced in brain and spinal cord neurons, in a form that prevents it from localizing to the nucleus. Both the human TDP-43 and the mice’s own TDP-43 drift out into the cytoplasm, with the result that the mice develop a fast-moving degenerative condition remarkably like ALS, including TDP-43 inclusions and muscle weakness. However, their disease essentially stops progressing if the human TDP-43 production is shut off and mouse TDP-43 returns to its normal location in the cell nucleus.


In 2009, two teams of researchers reported that mutations in a gene called FUS are yet another cause of ALS. As in the case of TDP-43, the protein encoded by the FUS gene had been studied hardly at all, but was known as an RNA-binding protein that normally works in the cell nucleus. And like the mutant TDP-43 that causes ALS, the mutant FUS protein turned out to be depleted in the nucleus and aggregated in the cytoplasm. As one of the research teams noted in describing its discovery, “Neuronal cytoplasmic protein aggregation and defective RNA metabolism thus appear to be common pathogenic mechanisms involved in ALS and possibly in other neurodegenerative disorders.”


Big developments in ALS research were now coming more frequently. In October 2011, another two teams of researchers reported that they had linked many cases of familial ALS to a particular mutant gene on chromosome 9. Again, almost nothing was known about the gene in question, called C9ORF72. But while a normal copy of the gene contains a few dozen repeats of a six-nucleotide sequence, GGGGCC, the two teams found that people with ALS often have hundreds of GGGGCC repeats. Genetic repeat expansions are known to underlie many disorders, including the neurodegenerative Huntington’s disease.

Follow-on research has found the C9ORF72 repeat expansion in about 40% of familial ALS cases from a worldwide sample, and 7% of more common, non-familial cases—making it by far the largest single known cause of ALS. Remarkably, genetic analysis suggests that the C9ORF72 repeat-expansion arose only about 1,500 years ago, in Finland, probably in a single individual, and that it was disseminated to other parts of Europe—perhaps via Viking raids and conquests—and later made its way even to parts of Asia.

Analyses of how C9ORF72 repeat expansions harm cells seem broadly consistent with the idea that pathological aggregates and defective RNA processing drive ALS. For example, in a study published in Nature last year, Jiou Wang and colleagues at Johns Hopkins School of Medicine found that the mutant C9ORF72 gene forms abnormal loops of DNA, that it is transcribed into RNA that also forms abnormal loops, and that these DNA and RNA loops stick together and grab other gene transcripts and proteins in their vicinity, thus probably preventing them from functioning normally.

“These abnormal structures appear to affect cells at many levels, from the transcription of DNA into RNA, to the translation of RNA into proteins,” says Wang.

One of the trapped proteins they identified as nucleolin, named for its important role in the nucleolus, where the essential RNA-to-protein-translation machines called ribosomes are constructed. A subsequent study from the University of Texas Southwestern concluded that nucleolin is also—or alternatively—bound and trapped by the abnormal protein encoded by mutant C9ORF72.

Researchers now are working to determine more precisely and conclusively how the C9ORF72 mutation kills motor neurons and helper cells. They also want to know why TDP-43 aggregates show up in C9ORF72-ALS cases.

“A number of genetic mutations, involving genes other than TDP-43, culminate in the production of TDP-43 aggregates or inclusions; and that’s still totally mysterious,” says Trojanowksi.

An emerging hypothesis

Nevertheless some researchers have a hypothesis for why these aggregates form. The idea is that there is initially some stress—presumably exacerbated by aging in most cases—that leads to an “unfolded protein response” in affected cells.

The unfolded protein response in a cell includes a drastic reduction in protein synthesis and in the RNA processing that supports that synthesis. This is thought to help the cell by de-cluttering it and taking the load off systems that handle abnormal aggregates and other “unfolded” or “misfolded” proteins. But the response, while it needs to be robust, is apparently meant to be only temporary—among other reasons because the long-term shutdown of protein production would leave the cell unable to maintain itself in good repair, and thus in effect would kill it.

There is some evidence that TDP-43, at least, plays a role in this stress response by rapidly aggregating and moving out of the nucleus, in order to facilitate the shutdown of RNA processing and protein production. It then shows up in the cytoplasm—in conglomerations of nucleus-originating proteins and RNA that are known as “stress granules” because they tend to appear during episodes of cell stress.

Some ALS researchers now suspect that this process, meant to be temporary, lasts longer than it should in certain situations, such as when the TDP-43 gene is mutated, or when there are other long term stresses on a cell—from aggregating mutant SOD1, for example, or from the various derangements caused by mutant C9ORF72, or perhaps even from chronic inflammation following head injury.

“I would say that as a consequence of some provoking event, TDP-43 forms these cytoplasmic stress granules, which then convert from a reversible form by some mechanism into an irreversible form, so that now you have a cascading disease,” says Cleveland.

A continuum of neurodegenerative disease

Solving the ALS mystery will likely provide the keys to other neurodegenerative mysteries. In fact, the more that scientists study ALS and other neurodegenerative diseases, the less distinct they  appear.

ALS doesn’t merely share its broad pathological features, such as protein aggregation, with other neurodegenerative diseases. It also shares at least three key proteins. Lee and Trojanowski and colleagues reported in 2006 that TDP-43 aggregates are found in many cases of fronto-temporal lobar degeneration (FTLD), which as the name suggests, kills neurons in the frontal and temporal lobes.

“That discovery prompted us to start referring to these diseases as TDP-43-proteinopathies, even though they were perceived as distinct disorders,” says Trojanowksi.

TDP-43 mutations also were linked to FTLD, as—later—were FUS mutations and aggregates, and C9ORF72 repeat expansions.

Even in cases that initially seem like ALS, TDP-43 aggregates and associated cell death are often seen in the fronto-temporal lobes and other parts of the brain, with accompanying clinical signs of cognitive and behavioral problems. About one in five ALS patients gets a diagnosis of “ALS-FTLD.”

Other researchers have since found TDP-43 aggregates in association with the head-injury-related degenerative syndrome Chronic Traumatic Encephalopathy (CTE), which also features tau aggregates. As Trojanowski suggests, the most important defining factors in these diseases may not be the clinical signs and symptoms but the abnormal structure-forming molecules, such as TDP-43, that drive the disease process.


Trojanowski and colleagues have been looking into the possible use of antibodies that grab aggregated (but not normal) TDP-43, which in principle—even if they only blocked the cell-to-cell spread of TDP-43 aggregates—would slow the course of ALS as well as TDP-43-linked FTLD and CTE.

Other laboratories, including Cleveland’s, are using gene-silencing nucleotide or small-protein strategies to try to reduce the expression of ALS-causing genes and the production of their harmful protein products, and thereby slow or stop or prevent the disease. “Already in trial are designer DNA drugs to silence ALS-causing mutations in the SOD1 gene,” Cleveland says. “We expect the clinical trial of our C9ORF72-targeting therapy to start by the third or fourth quarter of 2016.”

Still others are targeting the unfolded protein response. Working from the hypothesis that ALS is caused by an unfolded protein response that stays “on” too long, shutting down the synthesis of essential proteins, a group from the University of Pennsylvania, including Trojanowski, reported last year that inhibiting the response with a small-molecular compound seemed to block or slow the disease process in fly and cell models of ALS.

On the other hand, Bertolotti’s laboratory reported in Science this April that a compound designed to enhance the protein synthesis shutdown in the unfolded protein response—and thus clear mutant SOD1 more effectively—successfully protected SOD1 mice from the deaths of motor neurons as well as physical deficits, and without noticeable side-effects. The same compound also worked in mice that model a different misfolded-protein disease, Charcot Marie Tooth 1B. Whether this approach will work in non-SOD1 ALS remains to be seen, but Bertolotti hopes to answer that and other questions before long. “While there is still much more to do, I am pleased to say that our work receives a lot of interest from pharma companies and investors who may help in developing this further,” she says.

A different approach, now being pursued by several groups, is to inject neural stem or progenitor cells into the spines of ALS patients, in hopes that they will mature into neuron-nourishing helper cells that help prevent the loss of motor neurons. DABI member Eva Feldman of the University of Michigan heads one such group. “To date we have injected up to 16 million such cells directly to the spinal cords of ALS patients, and a subset of participants are experiencing a stabilization in the progression of their disease,” she says. “Our application for a broader trial is under way and we are cautiously optimistic about future results.”

“It is a time of optimism,” says Cleveland. “Not optimism for solutions tomorrow necessarily, but for success in the next decade, at least against ALS caused by mutant genes such as SOD1 and C9ORF72. For other types of ALS I’m optimistic that there’ll be a successful or partly successful therapy in my lifetime.”

SOD1 [ital.] = the human gene

SOD1 = the protein encoded by the gene

SOD1 mice = conventional term for transgenic mice that express human mutant SOD1 and develop ALS-like disease.

SOD1-ALS = conventional term for ALS caused by mutant SOD1.