ľ¹Ï¸£ÀûÓ°ÊÓ

G.E. Folkers, Alexandre Bonvin

Scientific background

The recent progress in machine learning and artificial intelligence has led to a large improvement in predicting structures and protein complexes. This combined the AI-powered protein design tool BindCraft enables quick design of small potential protein binders against any protein. A recent publication on this software tool reveals a success rate of binder design for proteins with high affinity binding for 10 to 100% of the designed protein depending on the target. The group of Alexandre Bonvin has developed the HADDOCK software that uses an information-driven flexible docking approach for the modelling of biomolecular complexes. HADDOCK distinguishes itself from ab-initio docking methods in the fact that it encodes information from identified or predicted protein interfaces in ambiguous interaction restraints (AIRs) to drive the docking process. We would in this research project like to combine protein design through BindCraft, with the HADDOCK approach for scoring protein complexes. Subsequently we will use the software , through selective mutagenesis within the predicted interface of the complex, to improve the binding affinity of the designed complexes.

Project description

Given the progress made recently as described above, we would like to implement the BindCraft methodology to obtain binder proteins and verify the usefulness of HADDOCK in scoring and improving designed proteins. We therefore will use the recently implemented high throughput cloning and expression approach (Wicky et al, Hiller et al). A few target proteins will be selected and proteins will be designed for these proteins. Subsequently using biophysical methods such as fluorescence anisotropy flow-induced dispersion analysis and or ITC and NMR, we will determine the binding affinity. Subsequently using NMR titrations the interaction surfaces for the two p[proteins will be determined. This serves as a basis to verify the quality of the proposed alfa-fold model for the designed binder. On the basis of the adjusted model we will subsequently perform an in silico mutagenesis screen on the binder interface in order to improve the specificity as well as the affinity. This cycle will be repeated few times which we expect will enable us eventually to better predict which designed binder protein will have a good chance to bind specifically with high affinity to a target. If successful integration of the Haddock complex verification could be helpful as an additional scoring criterium for binder proteins designed with BindCraft. And of course we hope and expect to design new binder proteins for relevant target proteins, which could for example be used for possibly diagnostic purposes

Research proposal

Since this project is a joined effort between the computational group of Alexandre Bonvin and the experimental group of Gert Folkers, a more computational or a more experimental project can be designed for you depending on your skills and wishes

Methods used

• Generation of designer proteins using BindCraft
• Recombinant DNA technology including, cloning, mutagenesis and sequencing,
• Protein expression, purification and labelling of proteins in HTP
• Biophysical characterisation using FA, FIDA, ITC
• Solution state NMR spectroscopy including 15N HSQC and 3D experiments for backbone assignment
• NMR titration experiments
• HADDOCK to improve binding interface through in silico mutagenesis