Engineering of a fibronectin type III domain as an improved scaffold for novel binding proteins

• Main Author: Batori, Vincent
• Format: Dissertation
• Language: English
• Published: ProQuest Dissertations & Theses 01-01-2002
• Subjects: Biomedical engineering; Biomedical research; Biophysics
• ISBN: 0493837795; 9780493837796

The understanding of protein-ligand recognition is a central aim in biomedical research. Our laboratory utilizes the tenth type III repeat of human fibronectin (FNfn10) as a scaffold in selection experiments to find novel binding proteins, termed ‘monobodies’. The protein has a β-sandwich fold where mutations are introduced in the loop regions to produce a new binding surface. This design is similar to antibodies and provides a large variety of conformations for a library. In comparison to antibodies, monobodies are smaller proteins that do not exhibit some of the difficulties in purification and handling. Though exceptionally stable at neutral pH, FNfn10 was found to be even more stable at acidic pH. A destabilizing cluster of negative charges on the surface of wild type protein was identified. When the unfavorable electrostatic interactions on the surface of the protein were removed, the mutant protein was stabilized at neutral pH. Therefore, the mutations based on these results provide a more stable scaffold for the monobody technology. To further optimize library design in FNfn10 we addressed systematically how much each loop region contributes to the overall stability. The elongation of the loop connecting E and F strand of FNfn10 was severely destabilizing, illustrating the importance of this loop to the fold. Insertion of residues in the remaining five loop regions in FNfn10 resulted in mutant proteins that remained highly stable, three loop with only the minimal destabilization comparable to the value expected due to increased entropy upon elongation. Greater diversity of a monobody library could be obtained if different, randomized loop regions could be joined combinatorially. This could be achieved by protein fragment reconstitution. Cleavage sites were introduced in two loops. While the EF loop-cleaved fragments exhibited no sign of reconstitution even at high micromolar concentrations, the CD loop-cleavage resulted in fragments that reconstituted a FNfn10 fold with single digit nanomolar affinity. NMR spectroscopy confirmed that no significant change in structure of the complex from the CD loop fragments compared to the uncut protein occurred. The unusually high affinity of the FNfn10 reconstitution indicates a potential for applying a reconstitution-based selection.