Single Amino Acid Substitutions Alter the Efficiency of Docking in Modular Polyketide Biosynthesis

Modular or type I polyketide synthases (PKSs) catalyse the “assembly-line” biosynthesis of a wide range of structurally complex and clinically valuable polyketide natural products, including antibiotics, immunosuppressants and anti-cancer agents. Catalysis is achieved by repeated sets of active domains within the multienzymes, which are grouped into functional units called modules. Each chain-extension module minimally incorporates a ketosynthase (KS), which catalyses carbon–carbon bond formation, an acyl transferase (AT), which recruits the chain-extension unit, and an acyl carrier protein (ACP), which cooperates in chain elongation with the KS, and then chaperones the growing intermediate to domains involved in reductive processing of the keto group. Such reductive activities are optionally present and may comprise a ketoreductase (KR), a dehydratase (DH) and an enoyl reductase (ER). Biosynthesis is initiated by a loading module and is usually terminated by a thioesterase/cyclase (TE) domain. In every PKS cluster sequenced to date, the modules are distributed across multiple polypeptides. For example, the erythromycin PKS (6-deoxyACHTUNGTRENNUNGeryth ACHTUNGTRENNUNGronolide B synthase, DEBS) contains six modules housed in three protein subunits. The modular genetic architecture of PKSs has inspired an approach to the design of polyketide analogues, called “combinatorial biosynthesis”. By genetic engineering, portions of PKS multienzymes can be modified or exchanged among different assembly lines, and the resulting mutant synthases direct the biosynthesis of altered products. One successful strategy has been to construct hybrid PKSs by combining subunits derived from multiple, though highly homologous clusters. For example, subunits from the PKS responsible for the 14membered macrolide pikromycin, have been replaced by multienzymes from both the erythromycin and oleandomycin synthases, and the engineered complexes produced the expected hybrid metabolites. In such cases, the efficiency of chain transfer between the heterologous multienzymes depends both on compatible protein–protein interactions, as well as the inherent substrate tolerance of domains within the acceptor subunit. The relaxed substrate specificity of many PKS domains, including the KS, KR and TE 16] activities, has ACHTUNGTRENNUNGrepeatedly been demonstrated. Therefore, it would seem, in principle, possible to extend this engineering strategy to encompass more distantly related sets of PKS genes by directed engineering of the intersubunit interfaces. The interface between successive multienzymes is formed not only between the ACP and KS domains, which participate in the acyl transfer reaction, but also between the docking domains at the extreme Cand N-termini of the polypeptides (Figure 1A). Alignment of multiple sequences from a large number of docking domains of modular PKS systems supports the presence of three helical regions in C-terminal docking domains (typically composed of 80–100 residues on the C-terminal side of the ACP domain, Figure 2) and a further conserved helical region in the N-terminal partner domains (generally 30–40 residues, on the N-terminal side of the KS domain). Recently, we reported the structure of a docking domain complex from the DEBS PKS that models the docking interaction between DEBS 2 and DEBS 3 (called Dock 2-3). To stabilise their interaction, the docking domains were genetically fused through their termini (Figure 3). The NMR solution structure revealed that the docking complex contains two topologically distinct and noninteracting four-a-helix bundles. The first, an unusual intertwined bundle, is formed by the first two helices of the C-terminal docking domain and is presumed to play a role in the dimerisation of the subunits. The second is formed by the third helix of the C-terminal docking domain, which wraps around a coiled-coil formed by the N-terminal docking domain (Figure 3). As this bundle contains helical elements from both the Cand N-terminal docking domains, it is likely to play a key role in the docking interaction. In support of this view, replacement of specific helical segments within the putative docking bundle broadly reproduces the effects on the biosynthesis of exchanging entire docking domains. Taken together, these results suggest that it might be possible to manipulate the specificity of docking by targeted modification of a few critical residues within the docking four a-helix bundle. In particular, a charge–charge interaction within the Dock 2–3 interface appears to serve as a specificity “code”, in the erythromycin PKS at least. To evaluate this strategy, we have carried out site-directed mutagenesis of specific amino acids within a PKS docking complex (Figure 1A). The model PKS we chose, TKS-JC1, comprises the loading AT–ACP didomain of the DEBS system followed by an engineered version of module 1 in which the ACP domain is replaced by the ACP domain of module 2 (with its associated C-terminal docking domain). The synthase also contains DEBS module 3, the natural partner of DEBS module 2, which has been engineered to terminate in a thioesterase domain. Therefore, this synthase incorporates the native docking interaction between DEBS 1 and DEBS 2 (Figure 1A). In vivo, TKSJC1 assembles two triketide ketolactone products in good yield (Figure 1A): compound 1, derived from acetate as a starter unit, and compound 2 from propionate. As part of a detailed mutagenesis study on the docking complex, we hoped to identify residues that might serve as experimental controls for both single and double mutants within the docking interface. Consideration of the structure of Dock 2–3 led us to chose amino acids Q72 and R95 (residues numbered as in ref. [18]). According to the structure, both amino acids are solvent exposed and so not expected to contribute directly to the docking interaction (Figure 3). The correspond[a] Dr. K. J. Weissman Department of Biochemistry, University of Cambridge 80 Tennis Court Road, Cambridge CB2 1GA (UK) Fax: (+44)1223-766-091 E-mail : kjw21@cam.ac.uk