Strategic Acyl Carrier Protein Engineering Enables Functional Type II Polyketide Synthase Reconstitution In Vitro

Abstract

Microbial polyketides represent a structurally diverse class of secondary metabolites with medicinally relevant properties. Aromatic polyketides are produced by type II polyketide synthase (PKS) systems, each minimally composed of a ketosynthase-chain length factor (KS-CLF) and a phosphopantetheinylated acyl carrier protein (holo-ACP). Although type II PKSs are found throughout the bacterial kingdom, and despite their importance to strategic bioengineering, type II PKSs have not been well-studied in vitro. In cases where the KS-CLF can be accessed via E. coli heterologous expression, often the cognate ACPs are not activatable by the broad specificity Bacillus subtilis surfactin-producing phosphopantetheinyl transferase (PPTase) Sfp and, conversely, in systems where the ACP can be activated by Sfp, the corresponding KS-CLF is typically not readily obtained. Here, we report the high-yield heterologous expression of both cyanobacterial Gloeocapsa sp. PCC 7428 minimal type II PKS (gloPKS) components in E. coli, which allowed us to study this minimal type II PKS in vitro. Initially, neither the cognate PPTase nor Sfp converted gloACP to its active holo state. However, by examining sequence differences between Sfp-compatible and -incompatible ACPs, we identified two conserved residues in gloACP that, when mutated, enabled high-yield phosphopantetheinylation of gloACP by Sfp. Using analogous mutations, other previously Sfp-incompatible type II PKS ACPs from different bacterial phyla were also rendered activatable by Sfp. This demonstrates the generalizability of our approach and breaks down a longstanding barrier to type II PKS studies and the exploration of complex biosynthetic pathways.

Figure 1

Gloeocapsa ACP has a canonical fold and can interact with the B. subtilis PPTase Sfp. (A) The ten lowest energy structural models of WT gloACP obtained by CS-Rosetta. Residues Q31 and T35 important for Sfp-based ACP activation, as well as S34, the attachment point of the phosphopantetheine arm are marked. (B) Comparison of topology models for gloACP (based on AlphaFold) and available structures of Sfp-activatable, prototypical ACP domains: a type II PKS ACP (ActACP, PDB: 2K0Y) from Streptomyces coelicolor, a type I PKS ACP (DEBS ACP2, PDB: 2JU2) from Saccharopolyspora erythraea, and a type II FAS ACP (AcpP, PDB: 1T8K) from Escherichia coli (cylinders represent alpha-helices with red marking indicating the serine point of attachment of the Ppant arm. (C) NMR peak intensities of ¹⁵N-labeled gloACP upon Sfp titration. Severe line broadening at a 1:1 molar ratio (both proteins at 100 μM) indicates tight complex formation (see also Figures S11–S13). Note that helix III found in most carrier proteins is absent in gloACP.

Figure 2

Sfp compatibility can be inferred from ACP sequence, enabling targeted modifications to render ACPs activatable through loosened Sfp interactions. (A) Multiple sequence alignment of carrier proteins known to be modified by Sfp (top four with cyan bar) and non-actinobacterial Sfp-incompatible ACPs (bottom four with magenta bar). Numbering is based on the TycC3_PCP and gloACP sequences, respectively. GloACP, dacACP, and panACP were used in this study (see main text for details), while AntF is a previously characterized non-actinobacterial ACP from Photorhabdus luminescens. The serine carrying the phosphopantetheine arm is highlighted in black in each sequence. Black boxes represent residues with greater than 70% physicochemical similarity. Based on our and previously published functional data (see main text for details) as well as the sequence alignments, we hypothesized that the highlighted residues (cyan and magenta, respectively) direct Sfp-(in)compatibility. (B) CS-Rosetta model (see Figure 1A) of gloACP^Q31G/T35L indicating the position of the two mutations Q31G and T35L. (C) Chemical shift perturbations comparing gloACP WT and the Q31G/T35L double mutant. The proline residue is marked with P and unassigned residues are marked with an asterisk. (D) Relatively higher peak intensity of ¹⁵N-labeled gloACP^Q31G/T35L (100 μM) upon titration with unlabeled Sfp indicates looser interaction compared to WT gloACP (see Figures 1C and S13). Note that the ACP concentration was kept constant and accordingly, increases in linewidth stem from Sfp binding, not sample dilution.

Figure 3

Reconstitution of holo-gloACP^Q31G/T35Land gloKS-CLF yields a functional minimal type II PKS in vitro. (A) Proposed in vitro reaction scheme of gloPKS polyketide production by the minimal gloPKS components and color switch of reaction mixture after addition of ScFabD and malonyl-CoA to holo-gloACP and gloKS-CLF. (B) Chemical formulas and experimental m/z values (¹²C and ¹³C) of putative precursor polyketide metabolites produced by the minimal gloPKS with malonyl-CoA. (C) Extracted ion chromatograms of the putative metabolite C₂₆H₂₀O₁₀. ¹³C₃-malonyl-CoA supplementation leads to a mass shift of +26 amu in both cases (for additional extracted ion chromatograms of putative precursor polyketides see Figure S31).

Figure 4

Salicyl priming expands the product profile of the reconstituted minimal gloPKS. (A) Proposed in vitro gloPKS polyketide production scheme. Red stars represent carbon atoms derived from malonyl-CoA and anticipated to be labeled during ¹³C₃-malonyl-CoA isotope experiments. (B) Chemical formulas and experimental m/z values (¹²C and ¹³C) of putative precursor polyketide metabolites obtained from LC-MS experiments. (C) Extracted ion chromatograms of the putative metabolite C₂₉H₂₀O₁₀. Upon ¹³C₃-malonyl-CoA supplementation, a mass shift of +22 amu is observed (for additional extracted ion chromatograms of putative precursor polyketide products see Figure S33).