De novo computational design of a chorismate mutase

EMMANUEL LORENZO DE LOS SANTOS; TIMOTHY WANNIER; BERNARDO SOSA PADILLA ARAUJO; STEPHEN MAYO

San Diego, California

Conferencia; The 26th Annual Symposium of The Protein Society; 2012

The Protein Society

One of the big promises of Computational ProteinDesign (CPD) is its potential for engineering ?designerenzymes.? A general framework to build enzymes isone of the biggest contributions that CPD can provideto synthetic biology. However, rate accelerationsfrom computationally designed enzymes are severalorders of magnitude below those of natural enzymes.Furthermore, attempts at computational design ofcertain, well-characterized reactions have so farproduced only limited results. Chorismate mutase (CM)catalyzes the pericyclic rearrangement of chorismateto prephenate. Studies have shown that enzymaticcatalysis is achieved by stabilizing the chair-liketransition state through electrostatic interactions withsidechains. We have previously demonstrated thattargeted ligand placement (TLP) can be used to placean ab initio transition-state structure into the activesite of E. coli CM with the proper wild-type contacts.Though previous efforts have recapitulated the activesite in a wild-type scaffold attempts to engineer CMactivity into a non-native scaffold have not beenreported. An ideal CM active site would preferentiallybind the diaxial form of chorismate. This requiresseveral catalytic contacts, making TLP in a new scaffolddifficult. Increased computational resources allowingfor large-scale TLP and the use of molecular dynamics(MD) as a prescreen have encouraged us to revisit thisproblem. We developed a partially automated workflowto find suitable scaffolds for the CM reaction. TLP wasperformed on a set of high-resolution crystal structuresthat had ligands of similar size to chorismate andthat had a wild-type arginine in the binding pocket.Sequence design was performed on 485 decoys fromsuccessful active-site searches. MD was performed on121 decoys from this set. Ten of these sequences, on 9different scaffolds were suggested to be active. No CMactivity was detected in both in vivo and in vitro assays.We obtained a crystal structure for one of our designs,based on the PDB scaffold 3QT9. The crystal structureshowed divergence from the design primarily due tobackbone movement at two adjacent arginines. Basedon these results, we are currently creating a secondset of designs within this same scaffold. This work issupported by the DARPA Protein Design Processesgrant and a National Security Science and EngineeringFaculty Fellowship award to S.L.M.