INTRODUCTION
Modular polyketide synthases (PKS’s) are multidomain, enzymatic assembly lines that play the major role in the production of important pharmaceuticals, such as the antibiotic erythromycin and the anticancer agent epothilone.1-3 Each module is minimally comprised of an acyl carrier protein (ACP) domain that shuttles polyketide intermediates between enzymatic domains using a phosphopantetheinyl arm and a ketosynthase (KS) domain that selects which intermediates to extend and pass along the assembly line (Figure 1). A module may also contain an acyltransferase (AT) domain that selects α-carboxyacyl extender units, such as a malonyl or methylmalonyl group, and processing domains, such as a ketoreductase (KR), dehydratase (DH), or enoylreductase (ER), that perform chemistry on extended intermediates. While the high-resolution structures of each of these domains have been known for many years, the interfaces between them are only now being characterized.4
Each KS of an assembly line is thought to possess distinct docking sites for the upstream and downstream ACP’s with which it collaborates.5, 6 It was hypothesized that the upstream ACP docks at a transacylation site to present intermediates for transfer to the reactive KS cysteine and the downstream ACP docks at an extension site to collect the intermediate through a decarboxylative Claisen-like condensation with a bound α-carboxyacyl extender unit. The extension site has been revealed by cryogenic-electron microscopy (cryo-EM) and x-ray crystallography.7-9 In cryo-EM studies of both PikAIII (the third polypeptide of the Pikromycin synthase)10 and Mycobacterium smegmatisPks13 (PDB: 8CUY, 8CV1),11 other ACP/KS interactions were observed; however, in each of these interactions the phosphopantetheinylated serine of ACP is more than 28 Å from the KS reactive cysteine. As the thioester carbon attacked by the KS reactive cysteine can only stretch 19.2 Å from the phosphopantetheinylated serine oxygen, none of these interfaces represent the transacylation site.
Improvements in folding/docking algorithms are enabling the reliable prediction of physiologically relevant domain-domain interfaces.12-15 Since ACP’s and downstream KS’s collaborate within the PKS module (updated boundary used16, 17 ) to gatekeep for processed polyketide intermediates,18 evolutionarily co-migrate,16, 17, 19, 20 and are best maintained together within engineered assembly lines,21-24 we hypothesized that modern folding/docking algorithms would detect the interface between these domains.
The AlphaFold-Multimer14 predictions for 50 natural ACP/KS pairs from diverse, well-characterized PKS’s are highly similar.2 Half of the top solutions contain a conformational change of the KS TxLGDP motif at the putative docking interface that enables a closer approach (17.2 Å vs. 20.9 Å, on average) of the phosphopantetheinylated ACP serine and the reactive KS cysteine. These solutions provide a molecular description of how acyl-ACP’s dock with KS’s to enable the transacylation reaction. A model triketide lactone synthase composed of the 1st, 6th, and 7th modules of the Pikromycin synthase, P1 -P6 -P7 , was employed to test the predicted transacylation site.24, 25 The relative decreases in the activities of 20 variants containing mutations to surface residues on the KS from the 6thmodule, PikKS6, are consistent with the proposed interface. The data suggest that two residues conserved on the surface of KS, an asparagine in position 275 and a leucine in position 315, play major roles in binding acyl-ACP’s for transacylation.