Development of molecular recognition by rational and combinatorial engineering

Sammanfattning: Combinatorial protein engineering, taking advantage of large libraries of protein variants and powerful selection technology, is a useful strategy for developing affinity proteins for applications in biotechnology and medicine. In this thesis, two small affinity proteins have been subjected to combinatorial protein engineering to improve or redirect the binding. In two of the projects, a three-helix protein domain based on staphylococcal protein A has been used as scaffold to generate so called Affibody molecules capable of binding to key proteins related to two diseases common among elderly people.In the first project, Affibody molecules were selected using phage display technology for binding to Ab-peptides, believed to play a crucial role in Alzheimer’s disease, in that they can oligomerize and contribute to the formation of neural plaques in the brain. The selected Affibody molecules were found to efficiently capture Ab from spiked human plasma when coupled to an affinity resin. The structure of the complex was determined by nuclear magnetic resonance (NMR) and demonstrated that the original helix 1 in the two Affibody molecules was unfolded upon binding, forming intermolecular b-sheets that stabilized the Ab peptide as buried in a tunnel-like cavity. Interestingly, the complex structure also revealed that the Affibody molecules were found to homo-dimerize via a disulfide bridge and bind monomeric Ab-peptide with a 2:1 stoichiometry. Furthermore, Affibody molecule-mediated inhibition of Ab fibrillation in vitro, suggested a potential of selected binders for future therapeutic applications.In the second project, two different selection systems were used to isolate Affibody molecules binding to tumor necrosis factor alpha (TNF), which is involved in inflammatory diseases such as rheumatoid arthritis. Both selection systems, phage display and Gram-positive bacterial display, could successfully generate TNF-binding molecules, with equilibrium dissociation constants (KD) in the picomolar to nanomolar range. Initial characterization of the binding to TNF was evaluated by competitive binding studies between the Affibody molecules and clinically approved TNF antagonists (adaliumumab, infliximab and etanercept) and demonstrated overlapping binding sites with both adaliumumab and etanercept. Furthermore, linkers of different lengths were introduced between Affibody moieties, in dimeric and trimeric constructs that were evaluated for their ability to block the binding between TNF and a recombinant form of its receptor. In the dimeric constructs, a linker length of 20-40 amino acids seemed to have an advantage compared to shorter and longer linkers, and the tested trimeric construct could block the TNF binding at even lower concentration. The results provided valuable information for the design of future Affibody-based molecules that could be investigated in therapeutic or medical imaging applications.In the third project aiming to generate a protein domain with capacity to influence the pharmacokinetics of protein therapeutics, a natural serum albumin-binding domain (ABD) was subjected to an engineering effort aiming at improving the affinity to human serum albumin (HSA), a protein with an exceptional long half-life in serum (19 days). First-generation affinity improved ABD variants were selected using phage display technology from a constructed ABD library. After additional rational engineering of such first generation variants, one variant with a 10,000-fold improved affinity to HSA (KD ≈ 120 fM) was obtained. Furthermore, characterization of this molecule also demonstrated improved affinity to several other serum albumins. When used as a gene fusion partner, this affinity-maturated variant denoted ABD035, should have the potential to extend the half-life of biopharmaceuticals in humans, and several other animal species.