AlphaFold, the AI system that can accurately predict 3D models of protein structures from their amino acid sequences developed by DeepMind, has revolutionized proteomics research. Since its creation, it has accelerated research in many fields: drug discovery, vaccine development, food safety, bioinformatics, ecology…

Deepmind proposed a 1st version of AlphaFold to the CASP (Critical Assessment of protein Structure Prediction) in 2018, it obtained the 1st place just as AlphaFold 2, the second version to which we also dedicated an article in our magazine ActuIA N°3, did it in 2020.

In 2021, the company published the scientific paper and source code explaining how it had created this AI system and partnered with the EMBL European Bioinformatics Institute (EMBL-EBI) to create the AlphaFold DB platform to make these predictions freely available to the scientific community. The latest version of the database contains over 200 million entries, providing broad coverage of UniProt, the standard repository of protein sequences and annotations.

Some of the progress made with AlphaFold shared by DeepMind

DeepMind shared advances from the work of several scientists.

Matthew Higgins and his team: developing an effective drug to eradicate malaria

Malaria, or malaria, killed an estimated 627,000 people in 2020, mostly children under the age of five, with Africa being hit hardest by far.

Plasmodium, the parasite responsible for malaria, is transmitted to humans by an infected female mosquito. It then infects liver cells and circulates in the bloodstream by colonizing red blood cells. It produces proteins that bind to the surface of the host cell, only one of which appears on the surface of the red blood cell at a time, preventing the immune system from reacting properly.

Biochemist Matthew Higgins, a professor of molecular parasitology at Oxford University, created a research group in 2006 with the goal of discovering a truly effective vaccine, but he faced the challenge of the metamorphic nature of malaria parasites.

Currently, RTS,S, better known as Mosquirix, is the only approved inoculation. It is a recombinant protein-based vaccine active against Plasmodium falciparum, the main parasite responsible for malaria. Designed for children and approved by the WHO in October 2021, it targets only the first stage of infection, when the malaria parasite is carried to the liver, and is only about 30% effective.

The Jenner Institute, another team at Oxford University, reported promising results from another similar vaccine recently. Its approach, which consists of three doses followed by a booster a year later, has an efficacy rate of 77%. However, like Mosquirix, this vaccine addresses the early pre-hepatic stage of the malaria parasite’s life cycle

The goal of Matthew Higgins and his Oxford-based collaborators Simon Draper and Sumi Biswas was to develop vaccine immunogens for a multi-stage vaccine that could work simultaneously at each phase of the infection cycle, targeting the blood cell invasion that follows infection, but also the final reproductive stage of the parasite life cycle. This stage is very important because infected humans can transmit the parasite in turn to previously uninfected mosquitoes if they are bitten again, thus continuing the cycle.

Until AlphaFold came along, their models were flawed and incomplete, but according to Matthew Higgins “AlphaFold has allowed us to take our project to the next level, from basic science to pre-clinical and clinical development.”

He further stated:

“I’m sure AlphaFold predictions will get better and better. But for now, combining experimental knowledge with the AlphaFold models is the optimal approach, because it allows us to piece everything together. This is the approach we take for many of our projects.”

Zhong Yan Gan discovers crucial information about the molecular basis of Parkinson’s disease

More than 10 million people worldwide live with Parkinson’s disease, 4% of them are diagnosed before the age of 50, while the symptoms of early Parkinson’s disease would affect 10 to 20%.

The doctoral research of Zhong Yan Gan, a doctoral student in the laboratory of Professor David Komander, co-supervised by Associate Professor Grant Dewson, at the Walter and Eliza Hall Institute of Medical Research (WEHI) in Melbourne, Australia, focuses on understanding the PINK1 protein and how it functions in our cells to trigger the recycling of damaged mitochondria, a process known as mitophagy. Mitophagy is essential for maintaining the health of our cells, and when PINK1 is defective, it leads to the death of neurons in our brain and the development of early Parkinson’s disease.

Understanding PINK1 and its role

A 2004 study showed that PINK1 could cause Parkinson’s disease, although finding its structure became a critical issue, human PINK1 was too unstable to be produced in the lab, so scientists opted for more stable insect versions (like human lice) to study it.

The Komander lab team published their PINK1 structure in 2017, other researchers have published different structures for the same protein from a different insect (flour beetle).

Zhon Yan Gan wondered if the published structure was actually a snapshot of PINK1 during a single step of a longer process and decided as part of his PhD project to determine PINK1 at each step of its activation process. His work showed that the published structures of PINK1 were not a mistake and that they were different forms that the protein takes at different stages of its activation process.

To understand the implications of their findings for humans with Parkinson’s disease, David Komander and his team needed to determine whether their results extended to the human version of the protein and turned to AlphaFold.

Zhon Yan Gan put two protein sequences into AlphaFold to predict the structure of a PINK1 dimer in humans, the result was almost indistinguishable from his experimental work with the insect protein.

David Komander states:

“We were able to immediately generate real information for people who have these particular mutations. We can start thinking, ‘What kind of drugs do we need to develop to fix the protein, rather than just dealing with the fact that it’s broken.”

Melissa Formosa: predicting and combating the onset of osteoporosis

According to the French National Institute of Health and Medical Research (INSERM), osteoporosis is the cause of nearly 400,000 fractures each year in France. Nearly 40% of women over the age of 65 are affected by this disease. Although osteoporosis has a significant genetic component, little scientific research has been done on its causes.

Bone is a living tissue that is constantly rebuilt to maintain its strength. Old, damaged bone is replaced by new, healthy bone. Mutations in the WNT1 gene (an osteoblast, or bone-making cell), disrupt the bone-building process so that carriers have brittle bones and suffer from early osteoporosis. This tends to prove that osteoporosis cannot be considered a disease that affects only the elderly.

A fracture is still often the first indication that osteoporosis is present. According to Melissa Formosa, it is necessary to find biomarkers: a blood test, a gene or a protein to look for a predisposition or a high risk of developing osteoporosis and thus start fighting the disease before it even starts.

To that end, his team has used AlphaFold to try to better understand the genetic causes: ” When we enter the amino acid sequence into the AlphaFold software, it creates a 3D image of what the protein structure looks like and allows us to compare the protein structures encoded by normal and defective genes. With AlphaFold, we can visualize the impact of specific genetic mutations, some of which may cause only subtle structural changes. Others induce significant deformations of the protein, reducing its ability to function properly, contributing to disease.”

The goal is to develop simple blood tests for young adults to better predict the disease and to find new genes and proteins associated with the disease to develop better drugs to treat it. Early detection and the introduction of personalized medicine could mean that osteoporosis can be managed much more effectively, so millions of lives could be significantly improved.

These three AlphaFold use cases are far from the only ones presented by DeepMind, which cites, for example, Drugs for Neglected Diseases (DNDi), which is advancing drug discovery for neglected diseases, such as Chagas disease and leishmaniasis, affecting millions of people in poor and vulnerable communities, or the Centre for Enzyme Innovation (CEI), where researchers are discovering and designing enzymes to break down single-use plastics…

Translated from DeepMind donne la parole aux scientifiques utilisant AlphaFold