There is a practical benefit to understanding the molecular details of natural-product biosynthesis. Combinatorial biosynthesis, a technique in which the enzymes of a biosynthetic pathway are swapped and redesigned to produce new compounds, has proven to be a promising approach to creating molecules having altered biological activities1. The assembly-line logic of type I polyketide biosynthetic pathways seems to be particularly amenable to these metabolic engineering efforts2. Using a combination of chemical synthesis, enzymology and structural biology, Fecik, Smith and co-workers3,4 have provided a strategy for understanding the specificity of a key enzymatic step in polyketide biosynthesis.
Well-known type I polyketides include geldanamycin (an anticancer agent), rapamycin (an immunosuppressant) and erythromycin (an antibiotic)1. Type I polyketide biosynthesis begins with the buildup of a linear polyketide chain by a series of Claisen condensations of malonyl and methylmalonyl substrates attached to carrier proteins via thioester linkages2. Diversity is further introduced during this process by reduction and dehydration of the growing polyketide chain. After synthesis of the linear chain, a dedicated thioesterase (TE) domain catalyzes the formation of a lactone between a hydroxyl group on the chain and the thioester (Fig. 1a).
Figure 1: TE cyclization and inhibition.
(a) A polyketide chain linked via a thioester to an acyl carrier protein (ACP) is cyclized to a macrolactone by a TE domain. Substitution of the thioester linkage with a diphenylphosphonate yields an irreversible enzyme inhibitor. The enone of the polyketide substrate is in green. (b) Inhibitor in the active site of the TE domain, pictured with the active site catalytic triad (red), hydrophilic barrier (blue) and potential anchoring residue (green).
A TE domain functions much like a serine hydrolase, with the typical Ser-His-Asp catalytic triad and an oxyanion hole (Fig. 1b). The thioester-linked chain is delivered from the last carrier protein to the active site serine of the TE to form an ester linkage between the polyketide and the serine residue. Instead of being released from the enzyme by water (hydrolysis), the polyketide chain is liberated by attack of a hydroxyl on the chain to form a macrocycle (lactonization).
In two papers in this issue, Fecik, Smith and co-workers3,4 take advantage of irreversible serine hydrolase inhibitors that also modify the active site serine of the TE. The authors added a diphenylphosphonate inhibitor (Fig. 1) to the termini of several polyketide substrate analogs and then showed that these modified analogs inactivate the TE domain from the pikromycin biosynthetic pathway. The resulting inactivated complexes (Fig. 1b), in which the inhibitor is covalently tethered to the active site serine, mimic the tetrahedral covalent intermediate and are sufficiently stable for crystallographic analysis.
In the manuscript by Giraldes et al.3, structural analysis of relatively short polyketide substrates depicts the arrangement of the TE active site and reveals an unusual oxyanion hole. The authors observed the expected hydrogen bond from a glycine NH backbone amide, as predicted for a serine hydrolase, but no other protein residue makes contact with the phosphonate. Instead, an ordered water molecule functions as the second hydrogen bond donor. The structure also indicates that few protein contacts are made with the substrate. Moreover, the structure of the substrate channel does not change in the presence of the inhibitor, which suggests that an induced fit of the relatively large protein channel around the substrate does not occur.
In the manuscript by Akey et al.4, the diphenylphosphonate moiety is added to longer polyketide substrate analogs (Fig. 1a). Crystallographic analysis shows that this longer substrate analog approaches the exit of the enzyme substrate channel (Fig. 1b). However, a 'soft' hydrophilic barrier consisting of a glutamine residue and several ordered water molecules directs the polyketide chain back toward the active site so that the hydroxyl group is optimally positioned for lactonization. Therefore, the co-crystal structure clearly demonstrates the mechanism by which the enzyme controls the conformation of the substrate to enable catalysis of lactonization. Again, the authors observed few specific contacts between substrate and protein, though a threonine residue may help to anchor the substrate (Thr77, Fig. 1b). Whether other TE domains have more specific substrate interactions remains to be determined.
The enone of the polyketide substrate (Fig. 1a) is critical for lactonization; reduction of this keto group to a hydroxyl group results in a switch from a macrolactone to the hydrolysis product. This can be explained by the newly available crystal structure and accurate model of the polyketide chain's orientation. Modeling these substrates into the crystal structure suggests that the enone conformationally restrains the polyketide chain to optimally position the hydroxyl for attack. In the reduced substrate, the hydroxyl group is not optimally positioned, and water is more likely to hydrolyze the ester. This suggests that substrate assistance is required for efficient enzymatic lactonization.
These studies provide a general methodology that can be applied to TE domains from other polyketide pathways. It will be interesting to learn whether nature uses more than one strategy to facilitate the cyclization of different polyketide chains. For example, the TE domain of the erythromycin pathway seems to have a more hydrophilic substrate channel, which suggests that specific hydrogen bonds between substrate and enzyme may have a more important role in this system5. Co-crystallization of an irreversible inhibitor has also been achieved for a TE domain from a nonribosomal peptide natural-product pathway, but the extended peptide structure was not observed in the electron density map6. Fortunately, the polyketide inhibitor described in this work successfully co-crystallizes with the pikromycin TE domain to provide structural insights into the enzyme–substrate complex.
This study will undoubtedly facilitate structure-based protein engineering experiments. Rational redesign of the substrate specificity of polyketide TE domains will expand the available repertoire of enzyme-catalyzed macrolactonization reactions, thereby improving the prospects for combinatorial biosynthesis of new polyketides.
