New Open-Source Platform Improves Protein Design

Physical interactions between proteins are fundamental to processes like cell signaling, growth, and immune responses. Being able to control these interactions is therefore important for therapeutic development. Traditionally, discovering protein binders—proteins engineered to stick to specific targets like antibodies do to pathogens—has involved screening tens of thousands of candidates. This process demands significant experimental capability and computational expertise that not every lab possesses.

Now, researchers at EPFL, in collaboration with MIT, have developed BindCraft, a tool designed to simplify and streamline protein binder creation. Instead of starting with amino acid sequences and hoping to predict a good binder through neural networks, BindCraft reverse engineers the process. It uses the AlphaFold2 protein structure prediction system to generate amino acid sequences based on a desired protein structure that matches specific functional properties, such as binding a particular target.

According to Christian Schellhaas, co-author of the paper published in Nature, BindCraft “essentially reverse-engineer[s] the current pipeline by using the protein structure prediction network right from the start to generate novel binders that have the properties we’re looking for.” This method reduces the number of designed binders that need to be tested to just a handful rather than tens of thousands, making the approach more efficient and accessible.

The research team validated their binders against twelve different biological and therapeutic molecules, including adeno-associated viruses (AAVs) used for gene delivery, the CRISPR-Cas9 gene-editing enzyme, and common allergens. The binders demonstrated an average success rate of 46% in attaching to intended targets. According to first author Martin Pacesa, “For AAVs, the idea is to use these new binders to enable gene delivery only to specific cells and tissues while minimizing the risk of potential side effects. In the case of CRISPR-Cas9, our binders can stop its gene editing activity and keep it from acting when and where it shouldn’t.” 

Since being released as a preprint, BindCraft has been quickly adopted by researchers and industry users, whose feedback has inspired further developments, including plans to adapt BindCraft for smaller therapeutic molecules like peptides.