Directed evolution of single chain antibody library targeting a disintegrin and metalloprotease‐17 (ADAM‐17)

Abstract

Matrix metalloproteinases (MMPs) and a disintegrin and metalloproteases (ADAMs) are zinc‐dependent metalloproteinases within the metalloproteinase (MP) super family, which play critical roles in remodeling and degrading the extracellular matrix (ECM). Particularly, ADAM‐17’s enzymatic function in the activation or inhibition of cytokines, growth factors and their receptors, and adhesion molecules via cleavage of ectodomains has made this protease a critical target in various inflammatory diseases and cancers. Small molecule therapeutics targeting MPs have caused severe side effects in patients, due to nonspecific binding towards MPs. More specific and selective protein‐based therapeutics, such as antibodies, offer a promising alternative to small molecule therapeutics. Developing a more selective antibody via direct evolution can be labor‐ and time‐intensive. Thus, an algorithm‐based approach, like machine learning, can be utilized to increase mutational analysis, and create a deeper understanding of the correlation between protein sequence and function.Here, a synthetic single chain antibody (scFv) library that was previously engineered to reduce non‐specific binding, was further engineered via directed evolution, yeast surface display (YSD), and fluorescent‐activated cell sorting (FACS), for improved binding affinity and selectivity to the ADAM‐17 catalytic domain. DNA sequence analysis from each round of FACS indicated that frequent antibody mutations within light and heavy chain complementarity‐determining region 3 (CDR‐L3/CDR‐H3), were responsible for improved binding to ADAM‐17. Improved scFv clones also demonstrated decreased binding affinity towards other MPs. Machine learning was implemented in mutational analysis to further analyze the relationship between sequence and function. The knowledge and tools developed in this research can be used to engineer and design selective enzyme inhibitors. This work will ultimately shed light onto protein sequence‐structure‐function relations using a combination of rational and combinatorial protein design.