Examining the role of hydrogen bonding interactions in the substrate specificity for the loading step of polyketide synthase thioesterase domains.

The final step in polyketide synthase-mediated biosynthesis of macrocyclic polyketides is thioesterase (TE)-catalyzed cyclization of a linear polyketide acyl chain. TEs are highly specific in the chemistry they catalyze. Understanding the molecular basis for substrate specificity of TEs is crucial for engineering these enzymes to macrocyclize non-native linear substrates. We investigated the role of hydrogen bonding interactions in the substrate specificity of formation of an acyl-enzyme intermediate for the TE from the 6-deoxyerythronolide B biosynthetic pathway. Thirteen single site-directed mutants were constructed, via removal of side chain hydrogen bonding groups from the binding cavity. Specificity constants for four different substrates with and without hydrogen bond donors and acceptors were determined for the five active mutants. The relative magnitude of specificity constants for substrates did not change for the mutant TEs. Circular dichroism spectroscopy was used to show that the majority of the catalytically inactive mutants did not fold. Two mutations were identified that enabled mutant TEs to form a folded but catalytically inactive tertiary structure. Our data do not support a role for hydrogen bonding in mediating substrate specificity of bacterial polyketide synthase TEs. The highly conserved polar residues in the binding cavity appear to stabilize the unusual substrate channel, which passes through the enzyme. We propose that hydrophobic interactions between the binding cavity and substrate drive substrate specificity, as is seen in many protein-carbohydrate recognition events. This hypothesis is in agreement with high-resolution structural data for nonhydrolyzable acyl-enzyme intermediates from the picromycin TE.