The aspartate-histidine-serine triad is a common hydrolytic motif, convergently evolved by several enzyme superfamilies. Structural alignments show that some related enzymes use different residues to make up their triad implying that it is possible to evolve from one to another.

We studied nucleophile switching in Tobacco Etch Virus (TEV) protease because its structure is similar to the trypsin family of serine proteases, despite having a cysteine nucleophile. Its well characterised structure, specificity and activity make it good model for experimental evolution studies.

When the active site nucleophile is mutated from Cys to Ser in TEV protease, the resulting mutant (TEV^Ser) has 10³ fold lower activity. Directed evolution was performed to recover activity by picking the best variant from an error prone PCR library of 400 per round.

Even with such small libraries, over 8 rounds, 13 amino acid mutations and a truncation give dramatic kinetic improvements over the TEV^Ser starting point. Rather than re-specialising the enzyme, mutations show a weak positive effect on activity with the original Cys nucleophile. Initially there is a tradeoff between activity and soluble expression but in later rounds both properties increase, due in part to a C-terminal truncation. It is therefore possible to evolve an enzyme from one catalytic residue to another via functional intermediates by an unusual, weak-positive tradeoff. Structural data will give information as to how the new nucleophile is accommodated in the active site architecture

Preliminary data shows that the trajectory followed may be one of many equivalent solutions and that mutations facilitating the use of a new nucleophile are a common class. We are currently investigating how the structure of the fitness landscape flanking the experimental pathway changes along the evolutionary trajectory and how that may influence the evolvability of the nucleophile switch.