Computational Design of Proteins and Peptides Harboring Binding Interfaces

Abstract

Computational protein design has advanced rapidly in recent years. Starting from designing for novel protein and peptide structures, the focus of the field has shifted from designing for folds to designing for functions. In this dissertation, I will present the design of de novo miniproteins as antagonists for CD131. CD131 mediates signaling of three different cytokines, IL-3, IL-5 and GM-CSF. The signaling is triggered when CD131 forms a hexamer and/or a dodecamer complex with the cytokine and cytokine specific alpha chains. All three CD131 associated cytokines are implicated in mediating inflammation and three different antibodies against IL-5 have been approved to treat asthma. Using computational design and directed evolution, I have developed 3 different miniproteins that inhibit CD131 mediated STAT5 phosphorylation in a human cell line. These proteins all bind CD131 with sub-100nM to pM Kd and exhibit good thermostability in vitro. A low resolution cocrystal structure confirms the binding site and binding orientation of one of the design proteins. I will also describe using disulfide closed macrocycles to test our ability to design binding interactions with structured loops. Although binding with regular secondary structures has been explored before, the design rules of structured loop interactions are less well understood. Through a case study of designed disulfide macrocycles interacting with DNA Polymerase III β-Clamp, I will discuss findings that should help improve our ability in designing loop interactions.