(634c) A Novel Approach for the Computational De Novo Design of Antibody Structures and Alternative Scaffolds | AIChE

(634c) A Novel Approach for the Computational De Novo Design of Antibody Structures and Alternative Scaffolds

Authors 

Chauhan, V. - Presenter, Auburn University
Pantazes, R., Auburn University
Antibodies play a vital role in aiding the immune system to identify and subdue harmful foreign antigens. They do so by binding to those antigens with high specificity and affinity, and that binding is enabled by six loops in their structures known as Complementarity Determining Regions (CDRs). Since 1975, monoclonal antibodies (mABs) have been synthesized in the lab to bind to specific portions of antigens (i.e. epitopes). mABs have been used for several therapeutic applications, primarily as diagnostic reagents for imaging and analysis of diseases and as direct treatments for cancers, autoimmune diseases and cardiovascular diseases. These applications have led to the development of a $100 billion global market of more than 47 FDA approved therapeutic mABs. The scientific work on mABs has been taken forward towards the development of smaller sized alternative binding structures that retain the high affinity and specificity but offer further advantages. The smaller size of these molecules renders them easier and less expensive to synthesize in lab. Furthermore, they can penetrate tumors faster and are better served to function as simple carriers since they lack the redundant constant domain of mABs. Examples of such alternatives include fibronectin domains, anticalins, and ankyrin repeats. The two conventional approaches of manufacturing such biomolecules in lab, in-vitro evolution and hybridoma technology, while fairly successful, are laborious and inefficient in targeting a specific epitope. To overcome these shortcomings, significant work has been done on computationally designing antibodies. In the past two decades, computational algorithms have been developed to perform specific antibody related tasks such as predicting antibody structure from sequence, designing de novo antibody variable domains, enhancing protein-antibody docking, reducing immunogenicity, improving solubility among others. Programs such as OptCDR and OptMAVEn have been proven, with experimental verification, to design thermally and structurally stable antibodies that bind to target epitopes with nanomolar affinity. While significant work has been done on designing antibodies computationally, no similar work has been done for alternative binding scaffolds mentioned earlier.

In this work, we have developed a novel computational method for the de novo design of both antibody and non-antibody proteins. In addition to being capable of designing more kinds of proteins than existing methods, the algorithm also identifies promising solutions in significantly less time than any current technique. We will present on the details of the algorithm’s development and preliminary results for designed antibodies and fibronectin proteins. We believe that this work can lead to the development of new medicines and experimental reagents and advance the state of art in computational protein engineering.