In the spring of 1981, a geneticist from Indianapolis and a neurologist from Chicago got in a car and for three days they crisscrossed the yellowing cornfields of Iowa. Every now and then, they stopped at a house to draw blood from the inhabitants — more than 30 members of one extended family spread across the state. The scientists sent the blood 950 miles east to Massachusetts General Hospital in Boston, to a tiny lab (recently converted from a storeroom) where a 28-year-old postdoc named James Gusella and his 23-year-old research technician, Rudolph Tanzi, got to work.
Using old hamster cages, chair cushions purchased from Kmart, and a Polaroid camera given to Tanzi for his 11th birthday, they began taking pictures of pieces of DNA carrying the genetic code of each member of that family. With enough pictures, and enough pieces, the scientists hoped to do something many of their colleagues considered crazy: find the location of a lethal gene that had sentenced more than five generations of this family, and thousands of other people around the world, to the brain-ravaging disease known as Huntington’s.
Find it they did. Clues from the Iowans, later confirmed by similar analyses of blood from members of an extended family living in the remote northwestern part of Venezuela, mapped the gene to somewhere on chromosome 4. Over the next decade, those families would prove critical to unraveling the sequence of the gene itself. The landmark discovery, in 1993, promised to unlock the molecular underpinnings of the disease, thus providing a road map for developing a potential cure.
Now approaching three decades later, effective treatments — let alone curative ones — have yet to materialize. In the last two years, three once-promising drugs have failed in clinical trials, and at least as many prospective therapies have been abandoned before reaching human testing. With companies either stumbling or shutting down, researchers increasingly believe the drug industry, in an echo of its disappointing record with Alzheimer’s, has been pursuing a solution based on an incomplete understanding of the disease. Instead, an observation made 25 years ago — and then ignored by most of the Huntington’s field — may hold the answer.
Like with many other complex neurodegenerative conditions in which the brain’s circuits begin to fall apart in early to middle adulthood, Huntington’s researchers long focused their efforts on the mutant proteins produced by the problematic gene. Named HTT, it produces a protein called huntingtin. People with the disease carry a genetic stutter in HTT — repeats of a sequence of three DNA “letters,” CAG — which makes a larger, more brittle version of huntingtin that is prone to breaking apart and piling up in neurons. In later stages of the disease, those neurons, especially ones in the movement-controlling striatum, die off in huge numbers.
The devastation it leaves behind is striking. Individuals who inherit this gene — about 3 people per 100,000 worldwide — usually begin having trouble walking by their late 40s, involuntarily jerk their limbs, lose coordination, and die prematurely. Where the brains of deceased Alzheimer’s patients show overall shrinkage, Huntington’s patients have a hole where the striatum should be. “It’s hard to even look at,” said neurobiologist Lisa Ellerby of the Buck Institute.
When researchers recreated that DNA stutter in mice, their brains not only became littered with protein fragments, they also developed symptoms of Huntington’s disease patients. Studies in these mice in the 2000s showed that suppressing the production of mutant proteins improved their symptoms. These observations would guide the majority of drug development efforts for nearly two decades. Lower levels of mutant huntingtin, the thinking went, and you could stop the striatal neuron slaughter.
But so far, at least, it hasn’t turned out that way. In March 2021, Roche cut off a Phase 3 study of its promising huntingtin-lowering therapy tominersen after it proved ineffective. A week later, Wave Life Sciences shelved its two lead candidates after they failed to sufficiently slash mutant huntingtin levels in early-stage trials. In August this year, Novartis halted a Phase 2 study of its oral huntingtin-lowering drug, branaplam, over indications of side effects including peripheral nerve damage. In October, the U.S. arm of a Phase 2 study of a similar drug being developed by PTC Therapeutics was paused after the Food and Drug Administration asked the company to add monitoring for such side effects.
Some scientists who spoke to STAT believe the problems might just involve dosing — either too much drug too fast, causing a cascade of inflammation and other damage, or not enough. Others wonder whether some of the approaches are too blunt, taking out both the mutant protein as well as healthy versions of it essential to a variety of cellular processes (most people with Huntington’s inherit just one copy of the lethal gene so produce both versions of the protein). But a growing number of researchers are questioning the underlying premise of the therapeutic strategy.
“Going after preventing toxicity with huntingtin-lowering makes the assumption that it’s the full-length protein that matters,” Gusella told STAT. “And I don’t think we’re certain what actually causes the toxicity.”
What he is certain of is that there are ways of delaying symptom onset, by years in some cases, not by lowering levels of huntingtin protein, but by stabilizing the gene itself. He’s certain, because by analyzing the DNA of thousands of Huntington’s patients over the past 20 years, he and his colleagues have found people who defy the canonical calculus of Huntington’s onset.
A hallmark of the disease is that the number of CAG repeats a person has determines when in life symptoms start. Three dozen repeats of the sequence are normal. Forty or more generally result in disease onset in the person’s 40s. More than 60 repeats can bring symptoms as early as adolescence.
But even in his earliest studies of Huntington’s disease pedigrees, Gusella noticed something that didn’t quite fit. In each family, he found individuals who started experiencing symptoms much earlier or much later than their relatives. Because the number of CAG repeats varies generation to generation, some differences are to be expected. These examples, though, were so extreme that something else had to be going on. Somewhere in these outliers’ genomes, other rare bits of DNA were doing something to give or take away years of disease-free living. He and his colleagues just needed genetic material from enough people to find them.
The mid-1990s were heady days for Huntington’s disease researchers. An international consortium, wrangled by Columbia University psychologist Nancy Wexler (who had led medical teams to collect the Venezuelan family’s blood samples) and funded by her Hereditary Disease Foundation, had just identified the HTT gene. Now a great rush was on to reproduce the mutation in mice so its effects could be teased apart in the lab.
The first to do so was a team from United Medical and Dental Schools in London, led by biologist Gillian Bates. They had spent 18 unsuccessful months trying to put a version of the human HTT gene with very long stretches of CAGs — around 150 repeats — onto an artificial chromosome to make a mouse model of the disease. But it proved too big to be stable. The chromosome kept falling apart and recombining.
To try to understand why that was, Bates’ team switched to just the first piece of the gene, a protein-coding area known as exon 1 with about 130 CAG repeats. To their surprise, this fragment was enough to produce mice that suffered many of the progressive neurological symptoms of Huntington’s.
Because the animals got so sick, Bates’ animals would go on to become the most heavily studied mouse model, used in labs all over the world to this day. Those rodents held another surprise. Bates and her team observed in 1997 that their mice showed signs of a phenomenon called somatic instability. A healthy person inherits 30 or fewer CAG repeats in their HTT gene, and however many they’re born with, that’s how many they will have when they die. But in the mice with Huntington’s, the HTT gene seemed to grow over time, at least in some types of cells. Bates’ group showed that CAGs were expanding most rapidly in mouse brains, specifically the striatum.
This somatic expansion was seen as not much more than a quaint oddity by all but a few labs. The majority of the research community latched on to something else happening in the animals’ brains. They were being sown with sticky clumps of HTT protein containing extra-long tracts of the amino acid glutamine, coded by the DNA letters CAG. Influenced by work going on with other neurological disorders including Alzheimer’s disease, with its plaques of misfolded amyloid protein, that’s where the field stalled for the next two decades.
“Somatic instability is a story that’s been around for a long time,” Bates said. “The pattern we saw in mice then is pretty much what we see in humans now. But most of the work has focused on protein rather than the DNA.”
It’s not that researchers didn’t recognize the potential importance of somatic expansions, Bates said. It just kind of got lost among the many other aspects of the disease that people were uncovering. Partly that’s because scientists assumed the extra CAGs were getting added on during DNA replication that takes place during cell division. And neurons, the cells that are impacted in Huntington’s disease, stop dividing very early on in development. It must be in other brain cells where the gene is expanding, and then the mutant protein they produce winds up being toxic to neurons, the thinking went.
But puzzles remained. For one thing, autopsy studies of deceased patients showed that the mutant huntingtin was present in similar levels in many different types of tissue. But only one kind of brain cell, the medium spiny striatal neurons, died off in droves. Why just them?
To answer that question, much of the field honed in even deeper on the protein — trying to figure out how different cells processed the mutant version, and which configurations might make it toxic.
Fewer than half a dozen labs thought the answer lay in the dynamics of the gene itself. And for decades, genomic instability slowly cranked along at the margins, “sort of as a cottage industry,” said Albert LaSpada, a longtime neurodegenerative disease researcher who directs the Institute for Neurotherapeutics at the University of California, Irvine.
“Somatic instability is a story that’s been around for a long time. The pattern we saw in mice then is pretty much what we see in humans now.”
Gillian Bates, Kings College London
In a 2000 study, a lab helmed by Peggy Shelbourne of the University of Glasgow looked at the brains of Huntington’s mice and found that the older they got, the more repeats of CAG showed up in the striatum. A few years later, they saw the same thing in the brains of deceased patients.
They compared two young individuals, who’d inherited 41 and 51 repeats and who had shown no signs of symptoms or neuronal loss at the time of their deaths, with an older man who’d inherited 75 repeats and died of late-stage disease. The younger brains showed massive expansions in the striatum; some cells had ballooned massively, carrying mutations more than 1,000 CAGs long. No such expansions were seen in cortical neurons, which also die off in Huntington’s, but only later in its progression. In contrast, the older man’s cortex showed significant expansion; the HTT gene had doubled in size in at least 25% of cells. The researchers couldn’t actually find many CAG repeats in the striatum, but that was because most of those neurons had already disappeared.
Their discoveries revealed that somatic expansions happened early in the disease, and suggested they mapped better to the affected areas of the brain than protein aggregation.
“It was a very elegant study,” said Ricardo Mouro Pinto, a neurologist and Huntington’s researcher at Harvard Medical School and Massachusetts General Hospital. “But it was dismissed by most of the field. There was little traction behind those ideas back then.”
One person who didn’t dismiss it was Mouro Pinto’s former Ph.D. adviser, Vanessa Wheeler. She had worked as a postdoc down the hall from Gusella, trying to understand why Huntington’s mice passed down longer CAG repeats to their litters. When she set up her own lab at Mass. General a few years later, she devoted it to unwinding the mechanisms for somatic expansion.
An early clue had come in 1999, when researchers at SUNY Albany had shown that a key component of the cell’s DNA repair systems — protein complexes that roam the genome looking for errors and trying to fix them — played an important role. Specifically, it was an enzyme called MSH2, a protein involved in a particular kind of proofreading known as DNA mismatch repair that kicks in when the two strands get out of sync. When they crossed Huntington’s mice with a strain missing MSH2, the CAG expansion disappeared. Wheeler’s lab, and then Mouro Pinto’s, took that even further, mapping out the activity of a handful of mismatch-repair enzymes all involved in various steps of the process.
For much of the 2000s and 2010s, they presented these ideas at conferences, but were usually the only ones talking about somatic instability. That all changed in 2015, when a group helmed by Gusella — the Genetic Modifiers of Huntington’s Disease, or GeM-HD Consortium — published a genetic analysis of nearly 5,000 Huntington’s patients. Somatic expansion wasn’t just a quirk, the data showed, it was a driver of disease.
While other researchers had spent the decades post-gene discovery trying to understand the protein it made and the problems wrought by mutations in it, Gusella had set his sights on a slightly different goal: amassing a giant collection of DNA from Huntington’s patients. Because the disease was found in diverse populations, nature was, in a way, already running a clinical trial at a global scale. People didn’t inherit a mutated HTT gene in isolation. From their parents, they also received constellations of genetic modifiers — a unique DNA letter here, an interesting combination of variants there — all of which might be changing the odds of when someone develops the disease.
“What’s a genetic modifier? It’s an intervention that changes the onset of clinical disease,” said Gusella. “And you already know it’s valid because it works in people.”
Working with clinical trial registries, brain banks, and natural history studies being conducted in the U.S. and Europe, and scraping together funds from the National Institutes of Health and CHDI, the Cure Huntington’s Disease Initiative, Gusella and his colleagues finally got enough genomes to run their genetic analysis. It took about 10 years, in part because no one thought it would be worth the money. All it would turn up, many believed, were chaperone molecules that would help refold all the aggregated mutant HTT proteins. What they got instead sent shockwaves through the field.
“It immediately became clear that DNA instability mattered, at least in part,” said Gusella.
People who had developed the disease later than expected, sometimes up to eight years later, had inherited certain genetic variants of DNA repair molecules, including mismatch-repair enzymes. Mimicking their protective effects with a drug might provide an avenue for addressing Huntington’s at an earlier stage of the disease, before symptoms set in.
“There’s no doubt, that was truly a watershed moment,” LaSpada said of the publication.
Over the last seven years, better understanding somatic expansion and finding ways to combat it has become a focus of dozens of labs around the world, major funders, and several startups. Many Huntington’s researchers who spoke to STAT now believe it represents perhaps the best chance of finally producing a therapy that could delay if not prevent the worst ravages of the disease. The question on their minds now is how long it’s going to take Big Pharma to catch up.
”Data from thousands of patients is difficult to deny,” Mouro Pinto said. “Now, can you rally the field to develop tools to mimic that therapeutically? That’s a different story.”
Although large pharmaceutical firms like Roche and Novartis are still looking for ways to treat Huntington’s disease by lowering levels of mutant proteins, the 2015 paper did spur the formation of at least two small companies with somatic expansion in their sights.
“It was a bit like joining the dots,” said Nessan Bermingham, an operating partner at Khosla Ventures and the founder and CEO of one of those companies, Boston-based Triplet Therapeutics. “There were a lot of dots that were out there and frankly they’d been out there for a while, and for industry, many of us had not really put the pieces together and saw this as a viable target until Jim and Vanessa’s data came out. That was the clincher for us.”
Triplet launched in 2019 with $59 million to develop treatments for repeat expansion disorders, of which Huntington’s is just one of dozens. (Gusella and Wheeler served on the company’s scientific advisory board.) Its approach was to design an antisense oligonucleotide — a small, single strand of synthetic RNA — that could bind to the piece of RNA that manufactures one of those well-intentioned but ultimately problematic DNA-repair enzymes, in this case one called MSH3.
Somatic expansion, as Bermingham explains it, is caused by a series of probabilistic events that starts when a CAG-heavy stretch happens to loop back in on itself and ends when a DNA repair complex sees one of those loops, grabs on, and adds another repeat. It’s difficult to influence the likelihood of the first of those events. But formation of the DNA repair complex is a competition between proteins, and that can be pushed one way or another. Reducing the amount of MSH3 by about half should be enough to prevent the complex from forming and the expansion from taking place, Bermingham said. Stop expansion, treat the disease.
That’s the hypothesis, anyway. And Triplet was set to be the first company to test it in patients, with plans to begin a trial at the end of this year. Which was why so many were disappointed when it quietly shut down last month, after finding early signs that its lead Huntington’s drug candidate was toxic to neurons in studies with mice and non-human primates. Although the company’s data didn’t explain what exactly was behind the problems, Bermingham believes that it’s something about the chemistry of the antisense oligonucleotides that’s leading them to accumulate in the central nervous system and cause damage.
Both Roche and Wave Life Sciences are also pursuing antisense oligonucleotide drugs, and Bermingham said that the combination of all the lackluster data and safety concerns soured investors on the approach. In that environment, Triplet wasn’t able to raise enough money to go back and reformulate a new candidate. The company’s few remaining employees are sticking around just long enough to shepherd a natural history study it had embarked on — intended to follow 60 patients over two years to better understand changes in somatic instability and DNA damage repair over time — under a new steward, CHDI.
“That’s certainly one thing that still needs to be better understood — what is the right point of intervention,” Bermingham said. “Because, clearly neurons can survive the expansion getting bigger and bigger, but there’s a tipping point where now those cells have been pushed over the edge and the expansion is causing cell death. So what is that tipping point?”
Now that Triplet is out of the race, answering that question falls to other companies, like U.K.-based LoQus23 Therapeutics, which exited stealth last year with $13.5 million in financing from the Novartis Venture Fund and the Dementia Discovery Fund. It is also going after mismatch repair pathways, but unlike Triplet, it’s focusing its efforts on developing small-molecule drugs that can gum them up.
Academic researchers, meanwhile, continue to be occupied by understanding how, exactly, DNA repair fits into the pathology of the disease. And they don’t all agree.
In the last few years, a handful of labs have shown that the normal, healthy version of the HTT protein is itself involved in DNA repair, acting as a scaffold for other DNA-patching enzymes to assemble into a complex that responds to breakages caused by oxidative stress. Neurons, because they burn a tremendous amount of energy — more than 50% of metabolism in our bodies takes place above the neck — generate massive amounts of reactive oxygen species, which careen around the cell and blast DNA apart if they’re not dealt with. And the older people get, the worse those free radical-smothering repair systems get.
The mutant HTT protein, these labs argue, can’t draw together the repair complex, resulting in runaway DNA damage that eventually leads to cell death. If that’s true, what neurons might actually need is both less of the mutant protein and more of the healthy one.
“We know that our protein is literally involved in DNA damage repair as a first responder and something’s going wrong there that we need to fix,” said Ray Truant, a Huntington’s researcher at McMaster University in Ontario whose lab pivoted after the 2015 study from studying HTT protein aggregation to how it’s involved in DNA repair.
Other groups that have been at this longer, like those led by Wheeler and Mouro Pinto, have come up with a different model for what’s going on. It’s based around DNA mismatch-repair proteins, which are constantly patrolling cells’ genomes, checking for small spelling errors, as opposed to large breaks. Most of the time they do their job just fine. But in highly repetitive areas they struggle.
When there’s a long stretch of CAGs all in a row, the DNA starts to form these looping secondary structures, where the strand bubbles out and binds to itself. Wheeler and Mouro Pinto believe that these mismatch repair enzymes read these loops as errors to be snipped open and fixed, but once inside these CAG tangles, their navigation systems crash. Disoriented, they reach for whatever is around to fill in the gap, and end up introducing more CAGs.
“These vigilant repair enzymes are just trying to do their job, but instead wind up making it worse,” Mouro Pinto said.
That idea was bolstered in 2019 by a larger follow-up study from the GeM-HD consortium. Looking at data from nearly 9,000 patients, Gusella and his colleagues found about 100 people who had a rare genetic hiccup. Most of the time, the HTT gene repeats stop when a G flips to an A, terminating in a sequence that spells CAACAG. But rare individuals don’t get the flip, creating a longer unbroken stretch of CAGs. And those people develop symptoms earlier than individuals who possess the CAA interruption, even though CAA also codes for glutamine. In other words, they produce the same protein, but they get sick sooner. That suggests that there’s something about the longer uninterrupted spans of CAGs that make it unstable and lead to further expansion as people age.
“These vigilant repair enzymes are just trying to do their job, but instead wind up making it worse.”
Ricardo Mouro Pinto, Harvard and Massachusetts General Hospital
This sort of discovery could be made only by studying very large groups of people. That’s been possible in Huntington’s because so many patients and their families have been willing to participate in research. Jenna Shae, who does not have the disease, and her mother, who does, are two of 21,000 people enrolled in a long-term natural history study of Huntington’s. “At times it can seem so insignificant,” Shae said of their role. “But when many people come together and do their part to just do that one visit each year, it does have a significant impact.”
The work of the GeM-HD Consortium has also brought standout scientists from adjacent fields into Huntington’s research for the first time.
This summer, Steve McCarroll, director of genomic neurobiology at the Broad Institute of MIT and Harvard, whose lab invented a method for measuring the gene expression of thousands of cells at once, presented unpublished data from a study applying that method to postmortem Huntington’s patients’ brains. McCarroll’s team found that in people who’d inherited between 40 and 50 CAG repeats, the cells that make up large swaths of their brains carried the same number. In astrocytes, they saw no expansion. In oligodendrocytes, no expansion. Interneurons, no expansion. But in medium spiny neurons — the ones that die off in mass extinctions in the striatum — CAGs swelled to 300, 400, sometimes 500 repeats. It showed, in the finest resolution yet, the link between somatic instability and the cells impacted by the disease.
That’s not to say that mutant proteins with extra-long polyglutamine tracts might not turn out, ultimately, to be the cause of toxicity to those cells, McCarroll told STAT in an interview. It could be that the expansions hit a point where they trigger a short toxic phase that hastens the end of each neuron’s life. Or it could not be glutamine at all. The point is, researchers should be open to other possibilities.
“That Huntington’s arises from the slow, lifelong accumulation of polyglutamine in cells is a model that I think a huge fraction of the field has bought into,” McCarroll said. “And it’s a model that we need to look at much more critically than we do.”
Bates, now at King’s College London, has an idea for a model that better fits much of the new data that’s now emerging.
In 2013, her lab discovered that somatic expansion doesn’t just make bigger HTT proteins, it also makes a much smaller, deadlier one. As more CAG repeats get added, transcription of the gene sometimes gets cut off early, resulting in a tiny protein fragment. And not just any fragment — the same one had proved highly pathogenic and prone to aggregation in Bates’ mouse models.
A few years later, when they looked in the brains of deceased Huntington’s patients, her team found that the longer the CAG repeats, the more of this tiny, problematic protein got produced. So her working hypothesis is that it’s the CAG expansion that prematurely pushes the transcription closed, creating a very nasty little protein that kicks off aggregation and causes cell toxicity.
It’s still just an idea, and Bates’s lab is still slowly conducting experiments trying to link all those steps together. But she said that working on it has given her the same rush she felt 20 years ago, when she and others first mapped the HTT gene. “It’s just as exciting as that time, because we’re at a point where we’ve got a really good handle on very good new therapeutic targets,” Bates said. “It won’t be easy because there are a lot of hurdles with delivery into the brain, but with luck, all of that will come along in parallel.”
Ultimately, there may be a need to develop strategies that go after targets both new and old. Drugs that stall or even stop somatic expansion could help young people buy more time without symptoms, while therapies that remove mutant HTT protein might slow down the disease’s progression in patients already experiencing behavioral and physical changes.
Maybe that will be enough to finally provide the Huntington’s community the long-awaited breakthrough it’s been fighting for. Or maybe neuronal toxicity is caused by some other process yet to come to light. At this point, researchers just don’t know. Which is why Gusella and the GeM-HD consortium are continuing to analyze even more DNA being collected by large clinical and natural history studies, in the hopes of better understanding some of the genes that don’t cleanly fit into the somatic instability hypothesis and uncovering new ones.
“I don’t think we’re certain what causes the toxicity, but we know enough now to know that the earliest target is DNA instability and I’m hopeful about that,” said Gusella. And though all the trials to date have been mired in failures, Gusella said they have succeeded in one very important way: generating momentum.
“When we cloned the gene in ’93, the idea that lots of biotech companies would actually be interested in developments didn’t exist,” he said. “At that point, nobody thought anyone in industry would be interested. And now there are lots of companies that see this as a disease worth treating and they’re developing technologies that can potentially do it. That will help in getting across the finish line.”
Create a display name to comment
This name will appear with your comment