Targeting Early-Stage Parkinson’s Disease with AI

AlphaFold’s predictions pave the way for novel treatments that could impact more than 10 million people worldwide

It was a source of hard-earned satisfaction after what often seemed like an uphill battle. David Komander and his colleagues finally published the long-sought structure of PINK1. Mutations in the gene that codes for the protein cause early-onset Parkinson’s disease, a neurodegenerative disorder with a wide range of progressive symptoms—notably tremors and difficulty moving. But when other research teams published their own structures of the same protein, it became clear that something was wrong.

“The other two structures that were created looked completely different from the structure our group created,” says Zhong Yan Gan, a PhD student Commander’s Laboratoryco-supervised by Associate Professor Grant Dewson, at WEHI (Walter and Eliza Hall Medical Research Institute) in Melbourne, Australia. Theirs was special, with unique features not found in others. The stakes were high: understanding PINK1 could lend a hand unlock novel treatments for the underlying cause of Parkinson’s disease, which affects over 10 million people worldwide.

While Komander’s team was confident in their own findings, the contrasting results raised some crucial questions. And in a competitive field of research, they knew they wouldn’t be alone in their quest for answers. “Not only were these really tough nuts to crack, but once you cracked them, suddenly this whole realm opened up where everyone was doing very similar things,” Komander says.

The team eventually solved the mystery, but it took several more years of research, one accidental discovery, and lend a hand from DeepMind’s AlphaFold protein structure prediction system.

This symptoms of parkinson’s disease develops when a person’s brain can no longer produce enough dopamine. Most people who get Parkinson’s disease don’t know the specific cause, but about 10% of patients can point to specific genetic mutationIn such cases, Parkinson’s disease usually develops early, affecting people before they reach the age of 50.

One such genetic mutation is found in the gene encoding PINK1 protein. PINK1 plays a key role in breaking down and removing mitochondria, often referred to as the power plants inside our cells. “As we age, mitochondria can become senescent and damaged,” Gan says. “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, ultimately, Parkinson’s disease. So one way to find better treatments for the disease is to better understand PINK1 and its role.

When scientists discovered that PINK1 may cause Parkinson’s disease In 2004, discovering its structure became a key goal—but it never materialized, in part because human PINK1 was too unstable to produce in the lab. Forced to cast their net wider, scientists discovered that insect versions of PINK1—such as those from human body lice—were stable enough to produce and study in the lab.

Which brings us back to the beginning of our story. The Komander team has released their PINK1 structure in 2017. But when other researchers published different structures of the same protein from a different insect (flour beetles), they knew they only knew part of the story. That wasn’t entirely surprising. Proteins are vigorous molecules, after all. “They’re like machines, and they can take on different shapes,” Gan says. But what if the published structure was just one of those shapes—a snapshot of PINK1 during a single step in a longer process?

Gan took on the ambitious task of figuring out what PINK1 looked like at each stage of the activation process as his doctoral project. It was during this work that he noticed something odd: a molecule that seemed far too huge to be his target. “Normally, you would dismiss it as something that just clung together, like scrambled eggs,” Komander says.

However, Gan had a hunch that this lump was worth examining in more detail and decided, with the lend a hand of Dr. Alisa Glukhova, to examine the molecule at the atomic level using Electron cryomicroscopy (cryo-EM), in which a frozen sample is examined with an electron beam. “I remember telling Zhong, ‘Yeah, you can try, but it’ll never work,’” Komander admits.

Gan’s persistence paid off gigantic time. He found exactly the molecule the researchers were looking for: PINK1. But why so gigantic? It turned out that PINK1 liked company. Instead of being a single protein, it was grouped into pairs of molecules known as dimers, which arranged themselves into even bigger formations. “Six PINK1 dimers formed big bagel-shaped structures,” Gan says.

This accidental discovery meant he could employ cryo-electron microscopy, which wouldn’t work on a molecule as diminutive as a single PINK1, to solve the protein’s physical structure. The team had their answer.

The previously published structures of PINK1 were no mistake—they were the different forms the protein takes at different stages of the activation process. But there was a catch. All of this experimental work was done using insect-derived PINK1. To understand the implications of their findings for people with Parkinson’s, they would need to examine 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 people on the planet who knew what PINK1 looked like when it was activated,” Komander says. So they used AlphaFold to trigger his predictions of the structure PINK1 of human originand a moment later it was on the screen. It was “absolutely shocking” how true AlphaFold’s predictions were, he says.

Later, when Gan put the two protein sequences into AlphaFold to predict the structure of a PINK1 dimer in humans, the result was nearly indistinguishable from his experimental work with the insect protein. “This dimer basically showed exactly how these two proteins interact so that they can work and cooperate to form some of these complexes that we’ve seen,” Komander says.

This close correspondence between several experimental results and AlphaFold’s predicted structures gave the team confidence that the AI ​​system could provide significant insights beyond their empirical work. They then used AlphaFold to model the effect that certain mutations would have on dimer formation—to investigate how those mutations might lead to Parkinson’s disease, and their suspicions were confirmed.

“We were able to immediately generate some real insights for people who have these particular mutations,” Komander says. Those insights could ultimately lead to novel treatments. “We can start thinking about, ‘What drugs do we need to develop to fix this protein, rather than just deal with the fact that it’s broken,’” Komander says.

They presented their findings on the matter PINK1 activation mechanism to the journal Nature in August 2021, and the paper was accepted in early December 2021. It turned out that scientists 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 rush the final revisions. “We were told to finish the paper three days before Christmas so it could be published in 2021,” Komander says. “It was a brutal timeline.”

At the end these documents of great importance were released within a few weeks of each other, with both studies providing crucial insights into the molecular basis of Parkinson’s disease.

Of course, researchers in this field still have many questions, and AlphaFold is freely available to lend a hand them find some answers. For example, Sylvie Callegari, a senior postdoctoral researcher in Komander’s lab, used AlphaFold to find the structure of a huge protein called VPS13C—known to cause Parkinson’s disease—by joining smaller pieces of the protein.

“Now we can start asking different questions,” he says. “Instead of, ‘What does this look like?’ we can start asking, ‘How does this work?’ ‘How do mutations in this protein cause disease?’”

One of AlphaFold’s many goals is to accelerate medical research, and it’s also being used at WEHI to sequence the genes of people with early-onset Alzheimer’s disease to allow researchers to investigate the causes of individual cases. “AlphaFold allows us to do that based on fantastic and valid human models,” Komander says. “That’s very powerful.”