The 2024 Nobel Prize in Chemistry is for proteins, the ingenious chemical tools of life. David Baker has succeeded in the almost impossible feat of building entirely new types of proteins. Demis Hassabis and John Jumper have developed an AI model to solve a 50-year-old problem: predicting complex protein structures. These discoveries have enormous potential.
The diversity of life testifies to the incredible capabilities of proteins as chemical tools. They control and drive all the chemical reactions that together form the basis of life. Proteins also function as hormones, signaling substances, antibodies and the building blocks of different tissues. Proteins generally consist of 20 different amino acids, which can be described as the building blocks of life.
The 20 amino acids that makeup proteins
In 2003, David Baker was able to use these building blocks to design a new protein that was unlike any other protein. Since then, his research group has produced one imaginative protein creation after another, including proteins that could be used as pharmaceuticals, vaccines, nanomaterials, and tiny sensors.
David Baker of the University of Washington is a pioneer in computational protein design. He leads the Rosetta project, which uses algorithms to predict and design new proteins from their amino acid sequences. Baker’s work has enabled the creation of proteins that do not exist in nature, with applications ranging from medical treatments to new industrial technologies.
Proteins developed using Baker’s Rosetta program. 2016: New nanomaterials that bind up to 120 proteins spontaneously. 2017: Proteins that bind to an opioid called fentanyl (purple). These could be used to detect fentanyl in the environment. 2021: Nanoparticles (yellow) with proteins that mimic the influenza virus on their surface (green) that could be used as a flu vaccine. Successful in animal models. 2022: Proteins that function as a type of molecular rotor. 2024: Geometrically shaped proteins that can change shape due to external influences. These could be used to produce tiny sensors.
The second breakthrough concerns the prediction of protein structures. In proteins, amino acids are linked together in long chains that fold to form a three-dimensional structure, which is crucial to the protein’s function. Since the 1970s, researchers have been trying to predict protein structures from amino acid sequences, but this has been notoriously difficult.
However, four years ago, there was a stunning breakthrough. Demis Hassabis and John Jumper are behind the revolutionary AlphaFold software, developed by DeepMind. AlphaFold was the first system to accurately predict the three-dimensional structures of proteins from their genetic sequences, solving one of the most challenging challenges in structural biology.
AlphaFold’s predictions are revolutionizing our understanding of protein function and accelerating research in areas such as drug development and synthetic biology.
How does AlphaFold2 work? As part of AlphaFold2’s development, the AI model was trained on all known amino acid sequences and determined protein structures. 1. DATA INPUT AND DATABASE SEARCHES An amino acid sequence with an unknown structure is fed into AlphaFold2, which searches databases for similar amino acid sequences and protein structures. 2. SEQUENCE ANALYSIS. The AI model lines up all similar amino acid sequences—often from different species—and investigates which parts have been preserved during evolution. In the next step, AlphaFold2 explores which amino acids might interact with each other in the three-dimensional structure of the protein. Amino acids that interact coevolve. If one is charged, the other has the opposite charge, so they are attracted to each other. If one is replaced with a water-repellent (hydrophobic) amino acid, the other also becomes hydrophobic. 3. AI ANALYSIS. Using this analysis, AlphaFold2 produces a distance map that estimates how close amino acids are to each other in the structure. The AI model uses neural networks called transformers, which have a great ability to identify important elements to focus on. Data on other protein structures – if any were found in step 1 – is also used. 4. HYPOTHETICAL STRUCTURE. AlphaFold2 puts together a jigsaw puzzle of all the amino acids and tests pathways to produce a hypothetical protein structure. This is run again in step 3. After three cycles, AlphaFold2 arrives at a specific structure. The AI model calculates the probability that different parts of this structure match reality.
Protein structures were determined using AlphaFold2. 2022: Part of a massive molecular structure in the human body. More than a thousand proteins form a pore through the membrane surrounding the cell nucleus. 2022: Natural enzymes that can break down plastic. The goal is to design proteins that can be used to recycle plastic. 2023: A bacterial enzyme that causes antibiotic resistance. The structure is important for discovering ways to prevent antibiotic resistance.
Together, the work of Baker, Hassabis, and Jumper has transformed the field of molecular biology, enabling scientists to design proteins with specific functions and predict their structures quickly and accurately, opening new frontiers in science and health.
Comments