AlphaFold’s predictions pave the way for new treatments that could impact more than 10 million people worldwide.
It was a source of hard-earned satisfaction after what had often seemed like an uphill battle. David Komander and his colleagues had finally published the long-sought structure of PINK1. Mutations in the gene that codes for this protein cause early onset of Parkinson’s disease, a neurodegenerative disease with a wide range of progressive symptoms, including body tremors and difficulty moving. But when other scientific teams published their own structures for the same protein, it became clear that something was wrong.
“The other two structures that emerged were very different from the structure produced by our group,” explains Zhong Yan Gan, doctoral student in Komander’s laboratory, co-supervised by Associate Professor Grant Dewson, at WEHI (Walter and Eliza Hall Institute of Medical Research) in Melbourne, Australia. Theirs was a strange model, with unique features that didn’t seem to exist in others. The stakes were high: understanding PINK1 could help unlock new treatments targeting the fundamental cause of Parkinson’s disease, which affects more than 10 million people worldwide.
Although Komander’s team was confident in its own findings, the contrasting results raised big questions. And in a competitive research field, they knew they wouldn’t be alone in searching for answers. “Not only were these problems very difficult to solve, but, once they were solved, you suddenly open up a whole world where everyone is doing very similar things,” Komander says.
The team eventually solved the mystery, but it took several more years of research, a chance discovery, and help from DeepMind’s protein structure prediction system, AlphaFold.
THE symptoms of Parkinson’s disease develop when a person’s brain can no longer produce enough of the chemical dopamine. Most people with Parkinson’s disease don’t know the specific cause, but about 10% of patients can point to the cause. a particular genetic mutation. In these cases, Parkinson’s disease tends to develop early, affecting people before reaching the age of 50.
One of these genetic mutations is found in the gene that codes for PINK1 protein. PINK1 plays a key role in the breakdown and removal of mitochondria, often called the powerhouses of our cells. “As we age, mitochondria can become old and damaged,” says Gan. “PINK1 is part of the body’s mechanism for recycling old mitochondria to make room for new ones.”
When this mechanism fails, damaged mitochondria accumulate, leading to the loss of dopamine-producing nerve cells and eventually Parkinson’s disease. So one way to find better ways to treat this disease is to better understand PINK1 and its role.
When researchers discovered that PINK1 could cause Parkinson’s disease in 2004, finding its structure became a key goal – but this was not possible, in part because human PINK1 was too unstable to be produced in the laboratory. Driven to cast a wider net, scientists discovered that insect versions of PINK1 – like those in human body lice – were stable enough to be produced and studied in the laboratory.
Which brings us back to the beginning of our story. The Komander team has published its ROSE1 structure in 2017. But when other researchers published different structures for the same protein from a different insect (flour beetles), they knew they only had part of the story. This wasn’t entirely surprising. After all, proteins are dynamic molecules. “They are like machines and can take different forms,” says Gan. What if the published structure was just one of these forms – a snapshot of PINK1 during a single step of a longer process?
Gan took on the ambitious task of determining what PINK1 looks like at each stage of its activation process as part of his doctoral project. It was during this work that he spotted something strange: a molecule that seemed far too large to be his target. “Normally, you would ignore it as something that just clumped together, like a scrambled egg white kind of thing,” Komander says.
But Gan had a hunch that this cluster was worth studying in more detail and decided, with the help of Dr. Alisa Glukhova, to probe the molecule on the atomic scale using cryo-electron microscopy (cryo-EM), in which a frozen sample is examined using an electron beam. “I remember saying to Zhong, ‘Yeah, you can try, but it’ll never work,’” Komander admits.
Gan’s perseverance paid off. What he discovered was the very molecule the researchers were looking for: PINK1. But why so big? It turns out that PINK1 likes company. Instead of a single protein, it was grouped into pairs of molecules called dimers, which arranged themselves into even larger formations. “Six PINK1 dimers assembled into large bagel-shaped structures,” says Gan.
This serendipitous discovery meant he could use cryo-EM, which would not work for a molecule as small as a single PINK1, to solve the physical structure of the protein. The team had their answer.
The previously published structures of PINK1 were not a mistake: they were different shapes that the protein takes at different stages of its activation process. But there was a catch. All of this experimental work was carried out using insect-derived PINK1. To understand the implications of their findings for humans with Parkinson’s disease, they would need to test whether their findings extended to the human version of the protein.
Komander and his team turned to AlphaFold. “We had these new structures, and at the time we were the only ones on the planet who knew what PINK1 looks like when it activates,” Komander says. So they used AlphaFold to call its prediction about the structure of PINK1 of human origin, and moments later, it was there on the screen. It was “completely shocking” to see how accurate AlphaFold’s predictions were, he said.
Later, when Gan inserted two protein sequences into AlphaFold to predict the structure of a PINK1 dimer in humans, the result was almost indistinguishable from his experimental work on the insect protein. “This dimer showed exactly how these two proteins interacted so that they could act and work together to form some of these complexes that we had seen,” Komander says.
This close alignment between several experimental results and the structures predicted by AlphaFold gave the team confidence that the AI system could provide meaningful insights beyond their empirical work. They then used AlphaFold to model the effect that certain mutations would have on dimer formation – to explore how these mutations might lead to Parkinson’s disease, and their suspicions were confirmed.
“We were able to immediately generate real information about people with these particular mutations,” says Komander. This knowledge could ultimately lead to new treatments. “We can start to think about, ‘what kind of drugs do we need to develop to repair the protein, rather than just deal with the fact that it’s broken,'” Komander says.
They submitted their findings on PINK1 activation mechanism to the journal Nature in August 2021 and the paper was accepted in early December 2021. It turned out that researchers at the Trempe Lab in Montreal, Canada, had reached similar conclusions, and when that team’s paper was published in December 2021, the WEHI authors had to expedite final revisions. “We were asked to finish the diary three days before Christmas, so that it could be published in 2021,” says Komander. “It was a brutal timeline.”
Ultimately, these high profile papers were released within weeks of each other, both providing vital insights into the molecular basis of Parkinson’s disease.
Of course, many questions remain for researchers in the field, and AlphaFold is available for free to help them find some of the answers. For example, Sylvie Callegari, a senior postdoctoral researcher in Komander’s lab, used AlphaFold to discover the structure of a large protein called VPS13C – known to cause Parkinson’s disease – by piecing together smaller protein fragments.
“Now we can start asking different questions,” she says. “Instead of ‘What does it look like?’ we can begin to ask ourselves, “How does it work?” », “How do mutations in this protein cause diseases? »
One of AlphaFold’s many goals is to accelerate medical research. It is also being applied at WEHI to genetic sequences from people with early-stage Alzheimer’s disease, to allow researchers to study the causes of individual cases. “AlphaFold allows us to do this based on fantastic, correct human models,” says Komander. “It’s very powerful.”