AlphaFold solves one of the biggest mysteries in biology

AI system helps scientists piece together one of the largest molecular structures in human cells

When Pietro Fontana joined the Wu Lab at Harvard Medical School and Boston Children’s Hospital in May 2019, he faced what has been called one of the world’s most hard, giant puzzles: the task of assembling a model of the nuclear pore convoluted, one of the largest molecular machines in human cells.

“It was very difficult from the very beginning,” he explains. The convoluted was named monster and for good reason: it is made up of over 30 different protein subunits, called nucleoporins, and in total contains over 1,000 of them, intricately linked to each other.

So when he sat down to operate AlphaFold for the first time in his work two years later—with Alexander Tong of the University of California at Berkeley, who was more familiar with the AI ​​system—he wasn’t sure it would assist. But what came in the summer of 2021 was a somewhat unexpected breakthrough. AlphaFold predicted structures of nucleoporins that hadn’t been determined before, revealing more of the nuclear pore convoluted in the process. With AI, they were able to generate a nearly complete model of the cytoplasmic ring convoluted.

“A lot of the components were already well-known, but with AlphaFold we also built some that were structurally unknown,” he says. “I started to realize that this was a really big and useful tool for us. I think AlphaFold completely changed the idea of ​​structural biology.”

Molecular scientists like Fontana have dedicated themselves to deciphering the nuclear pore convoluted for decades. It is vital because it is the gatekeeper of everything that enters and exits the nucleus and is believed to hold the answers to the growing number of serious human diseasesincluding amyotrophic lateral sclerosis (ALS) and other neurodegenerative diseasesKnowing how this convoluted is built could open the door to other groundbreaking, even life-saving, discoveries.

The size of the convoluted is challenging enough, but its many, varied parts add to the complexity. “This is one of the main difficulties in achieving a solution [clear enough] that we can interpret the sequence and structure of the complex,” says Hao Wu, the lab’s principal investigator. Even with a wealth of data, the team had previously managed only moderate-resolution structural images.

Missing pieces of the puzzle have also hampered progress. Without the full set, it’s difficult to tell how the puzzle fits together, Wu says. “To figure out how the different protein subunits fit together, you really need to have help with their individual structures,” Wu explains.

It was here that AlphaFold changed the game for Wu’s lab, which also included Ying Dong and Xiong Pi. By testing proteins found in the eggs of the African clawed frog (Xenopus laevis)—used as a model system—the team managed to map all the different subunit structures that had been previously unknown. “When we started trying, we didn’t really know if the predictions would fit the map well,” Wu recalls. “But they did. It was really extraordinary.”

Of course, science is a team effort. When it comes to solving a puzzle as convoluted as the nuclear pore convoluted, it is not just a team effort, but the culmination of diligence and perseverance many teams around the world. Across the Atlantic, scientists from the Max Planck Institute for Biophysics (MPIBP) and the European Molecular Biology Laboratory (EMBL) in Germany have used AlphaFold in combination with electron cryotomography to model the human NPC. What they have achieved so far is new model twice as complete as the previous one. Now covering two-thirds of the NPC, a huge part of the puzzle has been solved, and a large step has been taken towards understanding how it controls what enters and exits the cell nucleus.

There’s still a long way to go—the final third remains. And while AlphaFold will make solving the remaining puzzle easier, the researchers are also aware of its limitations. According to Wu, the AI ​​system worked well for the nuclear pore convoluted because its subunits contained repeating helical structures that are easier to predict. But it may not be as simple for other proteins.

It’s vital not to treat AlphaFold — or any other AI tool — as the one and only solution, Tong said. “In reality, AlphaFold can give you very strange results,” Wu says. “But if you understand how it predicts, you can take that into account.” [in the analysis].”

Still, it’s clear that AlphaFold not only pushed the boundaries of science, but did so in a time previously thought impossible. “I’m glad AlphaFold came along at the right time, because it sped things up a lot,” Fontana says.

Fontana P., Dong Y., Pi X., Tong A. B., Hecksel C. W., Wang L., Fu T. M., Bustamante C., Wu H. Structure of the cytoplasmic ring of the nuclear pore convoluted by integrated cryo-electron microscopy and AlphaFold. Science 376, 6598, (2022). DOI:10.1126/science.abm9326.