Our project focuses on engineering antibody domains to increase their stability by introducing novel crosslinking technology adapted from bacteria. To do this we employed a variety of computational methods aimed to either inform rational design of these crosslinks or predict mutants capable of forming crosslinks.
Antibodies are used widely in medicine and research due to their ability to bind specifically to biological molecules. In our Human Practices research, we identified waste of antibodies as a clear issue to their use and sought to target our project on increasing the shelf-life of antibody products. We think increasing the shelf-life will not only reduce the cost of antibody therapies but also widen access, as products with long shelf-lives allow treatment in areas with poor infrastructure such as natural disaster zones. You can find our research on this on our Human Practices page.
In order to increase the shelf-life of an antibody, we need to first increase its stability. The pilin and adhesin proteins found on the surface of gram-positive bacteria possess an intramolecular crosslink known as an isopeptide bond, which conveys a high degree of thermal, mechanical, and proteolytic stability to the parent protein. Our idea was to adapt isopeptide bonds as a protein engineering tool to create highly stable antibody domains. As this has never been done before, we hoped project would serve as a proof-of-concept to the antibody engineering community and lead to new long-shelf-life antibody therapies.
We use a couple of computational techniques to inform how we introduce these crosslinks into an antibody domain. The first one is use structural comparison between antibodies and gram-positive pilins and adhesins to inform rational design of isopeptide bonds. The second is creating a framework based in Machine learning that will allow in silico screening of mutant antibodies for isopeptide bond formation. However, in order to broaden the scope of our project beyond antibodies, we have also undertaken an extensive bioinformatics search to identify previously unknown proteins containing an isopeptide bond, with the intent of broadening the first rational design approach to beyond antibodies. If you are interested in this, you can find the detail on our Modelling page.