Modular polyketide synthases are intensively studied as exquisite synthetic machines generating bioactive natural products. The enoylreductase, a common component of these machines, has been structurally and functionally characterized, revealing a new complex architecture.
Polyketides constitute a prominent class of natural products, including numerous medicinally important antibacterial, antifungal, immunosuppressant, anthelmintic and anticancer agents. Biosynthesis of the complex polyketides is catalyzed by modular polyketide synthases (PKSs) in a fashion similar to catalysis by mammalian fatty acid synthases (FASs). Modular PKSs are composed of repetitively arranged sets of ketosynthase, acyltransferase and acyl carrier protein (ACP) domains that are responsible for decarboxylative condensation, the selection of the extender units and the retention of growing polyketide chain, respectively. Modular PKSs additionally contain three reductive elements: the β-ketoreductase (KR), dehydratase (DH) and enoylreductase (ER) domains. These processing enzymes are of particular interest as their substrate specificity and domain number determine polyketide structure, and thus their genetic manipulation enables us to create new polyketides with diverse chiral centers. Much is known about these domains, with the exception of the ER domain, for which there is little functional and structural information1,2. In this issue, Zheng et al.3 rectify this situation in their characterization of the structural and biochemical properties of an ER domain from the spinosyn biosynthetic cluster.
The KR, DH and ER domains act on β-carbonyl groups after each chain extension step: KRs stereospecifically reduce the carbonyl to a corresponding hydroxyl group, DHs create double bonds via elimination of the hydroxyl group, and ERs reduce the double bond of the resultant 2-enoyl group. In contrast to FASs, each PKS module contains some, all or none of these reductive domains, leading to a wide variety of polyketides with different reductive states. The KR domains have been well studied structurally and mechanistically4,5,6. A crystal structure of the KR domain from the first module of an erythromycin PKS has revealed that KR shows monomeric contacts across the two-fold axis of the module. The key residues for catalysis and stereochemical control have also been identified by site-specific alteration of the isolated KR domains. Similarly, a representative structure is known for DH domains; the sequence from the fourth module of the erythromycin PKS (EryDH4) forms a dimeric structure and interfaces with the KR and ACP domains7. It is noteworthy that the EryDH4 dimer forms a more flattened architecture than the porcine FAS DH domain, possibly preventing the ER dimerization in the PKS modules. However, the true structure of the ER domains and its similarity to the ER domains of FAS has remained mysterious.
To answer this question, Zheng et al.3 focused on the second module of the spinosyn PKS, which harbors a complete set of domains for the synthesis of the insecticide spinosad8. The SpnER2 domain itself provided no crystals at all. However, the authors noted that, in the mammalian FAS, the ER domain shows a close interaction with the KR domain, suggesting that a Spn(KR+ER)2 didomain construct might yield more success. Indeed, the authors obtained crystals of Spn(KR+ER)2 with 3.0-Å resolution. SpnER2 belongs to the MDR (medium-chain dehydrogenase/reductase) superfamily of enzymes and is composed of two characteristic subdomains, a substrate-binding subdomain (residues 190–307 and 448–501) and a nucleotide-binding subdomain (residues 308–447). The structure shows that the NADPH binding site is located between two subdomains, in which the GGVGMA motif (residues 335–340), Lys360 and Arg375 from the nucleotide-binding subdomain and Phe231 and His494 from the substrate-binding subdomain are responsible for specific interaction with nicotinamide coenzyme.
The crystal structure further reveals that the SpnER2 domain shares a ∼600-Å interface with the SpnKR2 domain. The N-terminal linker connecting SpnKR2 and SpnER2 spans from the C terminus of the SpnKR2 structural subdomain to the β1 strand at the N-terminal SpnER2, whereas the C-terminal linker spans from the β9 strand at the C-terminal SpnER2 to the N terminus of the SpnKR2 catalytic subdomain. It is noteworthy that the length of the C-terminal linker of the Spn(KR+ER)2 didomain (8 residues) was much shorter than that of its mammalian FAS counterparts (29 residues)9, suggesting that the KR+ER PKS didomain cannot adopt the interface observed in the corresponding mammalian FAS domains. A superposition of the SpnER2 domain and the ER domain from the porcine FAS similarly showed different architecture at the βF strand, which is an important element for MDR dimerization. To probe this question directly, the authors performed small-angle X-ray scattering (SAXS) on the didomain construct; these data also did not agree with a superposition of SpnKR and SpnER on the KR and ER domains of porcine FAS (Fig. 1).
The authors then turned to functional characterization of their didomain construct. Up to this point, functional analysis of the ER domain was limited to an in vivo assay in the context of a complete PKS module1,2, as the isolated ER domain could not be obtained as an active enzyme. The authors tested the in vitro catalytic activity of their ER using a trans-α,β-unsaturated substrate analog, crotonyl-pantetheine, which was converted to the expected product, butyryl-pantetheine. Furthermore, mutagenesis of the Spn(KR+ER)2 didomain revealed the importance of Lys422 for catalysis, with a possible function of lowering the pKa of the Tyr241 hydroxyl that sets an S configuration at the α-carbon of an α-substituted substrate.
In combination, the description of the KR-ER interface in Spn(KR+ER)2, the short C-terminal linker region between SpnER2 and SpnKR2, the altered MDR dimerization site in SpnER2 and the flattened DH architecture in modular PKS indicate that the three-dimensional structure of the processing domains in modular PKS is quite different from that in mammalian FAS (Fig. 1). The structural comparison of modular PKSs with mammalian FAS provides a powerful tool for understanding and engineering polyketide biosynthetic machinery. In particular, these data will inform future efforts to design new 'unnatural' natural polyketides by altering the stereoselectivity and configuration during β-carbon modification steps.