Computational Design & Analysis of Specific Binding Pockets for Natural and Modified Amino Acids

Reagent an`bodies are widely used in research and diagnos`cs, yet more than half fail to recognize their targets or exhibit poor specificity, leading to unreliable results.Their produc`on relies on costly and `me-consuming immuniza`on processes that are difficult to reproduce.These challenges highlight the urgent need for alterna`ve, well-characterized binding systems that are both reliable and customizable.To address this, the mul`disciplinary PRe-ART project aims to replace conven`onal reagent an`bodies with engineered armadillo repeat proteins, which are designed to bind specific amino acid sequences through modular assembly.This thesis contributes to this objec`ve by focusing on the computa`onal design and analysis of binding modules within the designed armadillo repeat protein scaffold, with a par`cular emphasis on phosphorylated amino acids.Given that phosphoryla`on of serine, threonine, and tyrosine residues is cri`cal in signaling pathways and metabolism, the lack of high-performing monoclonal an`bodies for their detec`on presents a major challenge.To overcome this, a computa`onal pipeline was established to design binders for phosphotyrosine, phosphoserine, and phosphothreonine.The pipeline priori`zed key interac`ons while minimizing unfavorable muta`ons, leading to focused binder library sugges`ons to generate ra`onally designed libraries.Although experimental valida`on is ongoing, preliminary results indicate promising binding results.To improve the accuracy of computa`onal predic`ons, this thesis also evaluates different computa`onal methods that are used for binder library designs, providing insights into their predic`on accuracy upon point muta`ons.By evalua`ng strengths and weaknesses of these methods, systema`c tendencies of them were discovered and a complementary approach that enhances binder design reliability is proposed.Finally, to efficiently validate promising computa`onal predic`ons, binding affinity measurements for protein-pep`de interac`ons using dArmRPs within our research group were established.Implemen`ng this setup in our lab enabled faster and more prac`cal tes`ng of designed binders, providing a reliable framework for assessing the accuracy of computa`onal predic`ons.