ENGINEERS OF LIFE: Nobel Prize winners in chemistry masters life’s building blocks

Life’s biggest secret lies within the structure of its smallest units. 

The long trek to predict the composition of proteins while creating entirely new blocks has now been unlocked with this year’s Nobel Prize in Chemistry.

The Royal Swedish Academy of Sciences has awarded David Baker, a biochemist, as well as Demis Hassabis and John Jumper, two Google DeepMind scientists, for protein structure prediction and computational protein design.

The 2024 Nobel prize in Chemistry “opens up vast possibilities” as said by the Nobel committee chair, Heiner Linke. 

This is as proteins are involved in almost every biological process–from muscle contraction, food digestion, and neuron signaling–making it critical for the proper functioning of living organisms.

For decades, scientists have attempted to determine protein structure based on their amino acid sequences. 

This includes the Nobel Prize in Chemistry last 1962 with John Kendrew and Max Perutz who successfully used X-ray crystallography to showcase the protein’s first three-dimensional model.

This was followed by the project in 1994 called the Critical Assessment of Protein Structure Prediction (CASP) which challenged researchers to predict the protein structures from its amino acid formations. 

CASP had attracted many researchers but the actual structures hardly changed since the conception of the project.

Baker made his debut in CASP by utilizing crystallographic validation to develop a computer program entitled the Rosetta. 

This was fundamental to assemble short structural fragments and optimization of the organic matter’s sequence and structure.

Rosetta’s success led to Baker and his team thinking to use the software in reverse; entering desired protein structures and obtaining suggestions for the amino acid chain to develop entirely new proteins.

By constructing proteins from scratch, the team would be able to increase the potential of what was once limited to the capabilities of these natural biomolecules.

In 2020, Jumper and Hassabis announced AlphaFold 2 which is an AI model that aims to tackle the “irreducible complexity of biology” by providing predictions of the appearances of these molecules. 

The model itself has been able to predict the structure of 200 million identified proteins with a near-experimental accuracy of 90%.

AF2 works by using a database of similar amino acid sequences then exploring which of these formations could interact with each other in the three-dimensional protein structure. 

The sequence analysis is then further refined for three cycles until it can put together a hypothetical structure – like its matching puzzle pieces with each other.

Besides this, AF2 has also been used to understand antibiotic resistance and to create enzymes able to decompose plastic. 

Bioengineering can also be applied with this discovery as it can advance the design of protein inhibitors (substances that stop or slow the growth of cells) like SARS-CoV-2 which can improve already present drugs and create new medicine.

Such progress from protein structure prediction enables the creation of large databases of predicted protein structures while opening up new areas of research in biomedical applications and synthetic biology.

The deep learning model is proof of artificial intelligence (AI)’s potential for accelerating scientific discovery. 

“I’ve always felt it would be one of the most transformative technologies in human history,” Hassabis mentioned while acknowledging the risks of AI use.

Baker lists products such as universal flu vaccine and greenhouse gas capture as some of the few possible results from the research.

With the discovery, the researchers are optimistic on the impacts of predicting and transcribing protein structure.