The vast majority of protein structures found in nature are just a small fraction of the potential protein structures that could exist. Scientists are now turning to de novo protein design, starting from scratch, to efficiently create new therapeutics customized for specific biological functions. A recent study by the European Molecular Biology Laboratory (EMBL) introduces an AI-driven framework called AlphaDesign, which aims to accelerate the protein design process, paving the way for personalized medicine and custom therapeutics.
The study, titled "AlphaDesign: a de novo protein design framework based on AlphaFold," was published in Molecular Systems Biology. This new approach is a significant step forward in transforming drug discovery and precision medicine by redesigning biology itself, according to Kashif Sadiq, PhD, the founder and CEO of DenovAI.
Creating Functional Proteins
The German research team initiated their study by generating protein sequences and structures to predict how these components change during binding. They tested their designed structures at various levels, from monomers to dimers, and observed conformational changes during binding using different established programs. The authors noted that their method is versatile enough to design proteins with diverse properties, such as those predicted to change conformation upon binding or bind multiple homologues without the need for model re-training or fine-tuning.
Functionality in Living Systems
To assess whether their de novo protein designs could function in living organisms, the team developed active inhibitors targeting bacterial phage proteins. By targeting RcaT-Sen2, a variant with an unknown protein structure, they aimed to inhibit bacterial growth. Testing on E. coli colonies showed that 19.3% of their inhibitor designs significantly inhibited RcaT-Sen2 in vivo. The successful inhibitor designs were confirmed using CD and NMR.
The researchers believe that AlphaDesign marks a significant advancement in computational biology, enabling the creation of new proteins with specific functions. This work showcases the potential of AI not only in predicting biological structures but also in actively generating new molecular capabilities.
Future Implications
While this study represents a groundbreaking achievement in de novo protein design, the researchers acknowledge that further research and testing are necessary before clinical applications can be realized. This work serves as a proof of concept for the potential of functional protein creation, with the ultimate goal of providing better outcomes for patients by designing tailored therapeutic molecules from the outset.
