Engineering Caspases with Altered Specificities

Maureen Hill, Derek MacPherson, Peng Wu, Olivier Julien, Sergey Savinov, James A. Wells and Jeanne A. Hardy

Caspases are proteases that play important regulatory roles in apoptotic cell death, neurodegeneration and inflammation. They are amongst the most specific of all proteases, and are therefore excellent targets for engineering new specificity. Unfortunately, because they are heterotetramers, they are not amenable to traditional engineering approaches like phage display and are difficult to tackle by rational design due to their flexible active sites.  Such characteristics prompted our lab to develop a GFP-based reporter of caspase activity that has enabled us to perform a directed evolution screen and evolve caspases with new functions.

In our first application of this technology, we engineered caspase-7, which recognizes and cleaves the amino acid sequence DEVD, to recognize and cleave the new sequence VEID, which is recognized by caspase-6 but not caspase-7.  Simple substitution of the homologous residues in caspase-6 dramatically diminished activity, but did not alter the specificity, indicating that a more holistic evolutionary procedure would be needed.  The caspase-7 variants with the most strongly altered specificity displayed kinetic values similar to wild-type caspase-6 and mimicked caspase-6 cleavage patterns with protein substrates. Using N-terminomics we assessed the substrate specificity profile for the most successful engineered caspase.  We found that had completely altered the specificity using our directed evolution approach.  The resulting WebLogos for the engineered caspase were completely superimposable with caspase-6 and unlike the parent scaffold, caspase-7.  Our seven crystal structures of evolved caspases bound to either DEVD or VEID substrates allowed us to observe how the new residues in the active site contributed to the change in specificity.  In particular, F282 plays a critical role by arching inward to create a small pocket around the newly introduced isoleucine at the P2 peptide position.  This unexpected finding underscores why rational design is insufficient for altering substrate recognition.