Dissecting complex polyketide biosynthesis

Abstract

Numerous bioactive natural products are synthesised by modular polyketide synthases. These compounds can be made in high yield by native multienzyme assembly lines. However, formation of analogues by genetically engineered systems is often considerably less efficient. Biochemical studies on intact polyketide synthase proteins have amassed a body of knowledge that is substantial but still incomplete. Recently, the constituent enzymes have been structurally characterised as discrete domains or didomains. These recombinant proteins have been used to reconstitute single extension cycles in vitro. This has given further insights into how the final stereochemistry of chiral centres in polyketides is determined. In addition, this approach has revealed how domains co-operate to ensure efficient transfer of growing intermediates along the assembly line. This work is leading towards more effective re-programming of these enzymes for use in synthesis of new medicinal compounds.

Figure 1

6-Deoxyerythronolide B synthase (DEBS). DEBS contains a module for each of the 6 cycles of chain extension. Each module catalyses incorporation of a (2S)-methylmalonyl CoA derived propionate unit into an acyl chain, to form a 2-methylbranched 3-ketoacyl intermediate. The level of processing of the β-ketone group is determined by the reduction domains. Each module also determines the final stereochemistry of methyl- and hydroxyl-bearing centres.

Figure 2

Synthesis of triketide lactones using discrete ACPs and KS-AT didomains. (2S, 3R)-2-methyl-3-hydroxypentanoyl-NAC is used to acylate the KS domain. The AT uses (2S)-methylmalonyl CoA to load the ACP. Condensation proceeds with inversion to give a (2R)-2-methyl-3-ketoacyl-ACP. Triketide lactone products are identified by GC-MS. Reduction of the ketone with borohydride prior to chain release gives a racemic mixture of (3R)- and (3S)- alcohols but fixes the (2R) methyl stereochemistry. Stereospecific ketoreduction can be achieved by adding a KR and NADPH.

Figure 3

In vitro synthesis of diketides. The combinations of KS-AT and ACP used were DEBS KS1-AT1 + ACP1 (epimerising), DEBS KS3-AT3 + ACP3 (epimerising), DEBS KS6-AT6 + ACP6 (non-epimerising), and PICS KS1-AT1 + PICS ACP1 (epimerising). All combinations gave a (2R)-2-methyl-3-ketoacyl-ACP initially. Enzymatic ketoreduction led to formation of a 2-methyl-3-hydroxyacyl chain with the alcohol and methyl stereochemistry characteristic of the module from which the KR was derived.

Figure 4

ACP4 docking interactions for elongation and translocation. A region of loop 1 functions in an elongation docking interaction with KS4-AT4. The N-terminal region of helix I engages in a translocation docking interaction with KS5-AT5.

Figure 5

Schematic representation of KS-ACP interactions in DEBS. ACP_n cannot transfer an extended polyketide back to KS_n because their translocation epitopes are incompatible. This prevents iterative operation of a module. The elongation interactions are also electrostatic and are drawn to depict the preference of ACP3 for KS3 and KS5, the weak preference for KS2 and KS4 and the lack of co-operation with KS1 and KS6. The dashed arcs represent interdimer docking interactions between DEBS1 and DEBS2 and between DEBS2 and DEBS3.