Biosensor Guided Polyketide Synthases Engineering for Optimization of Domain Exchange Boundaries

Abstract

Type I modular polyketide synthases (PKSs) are multi-domain enzymes functioning like assembly lines. Many engineering attempts have been made for the last three decades to replace, delete and insert new functional domains into PKSs to produce novel molecules. However, inserting heterologous domains often destabilize PKSs, causing loss of activity and protein misfolding. To address this challenge, here we develop a fluorescence-based solubility biosensor that can quickly identify engineered PKSs variants with minimal structural disruptions. Using this biosensor, we screen a library of acyltransferase (AT)-exchanged PKS hybrids with randomly assigned domain boundaries, and we identify variants that maintain wild type production levels. We then probe each position in the AT linker region to determine how domain boundaries influence structural integrity and identify a set of optimized domain boundaries. Overall, we have successfully developed an experimentally validated, high-throughput method for making hybrid PKSs that produce novel molecules.

Fig. 1. Activity of misfolded protein biosensor and mCherry fusion proteins.

a GFP fluorescence of E. coli biosensor strains expressing PKSs with variable solubilities. pET = pET28a (empty vector control). b Fluorescence of mCherry tagged PKSs (left y-axis) and SDS-PAGE quantified abundance of the proteins relative to DEBSM6 amount (right y-axis). (c) mCherry fluorescence and (d) SDS-PAGE quantification of PKS proteins in different protein fractions of lysed cells, relative to “Total protein” for each replicate. Cells were induced with 250 µM IPTG in (a–d). e Fluorescence of ΔarsB::P_ibp GFP strain expressing mCherry tagged PKSs. f Same results as 1e with a simplified “solubility coefficient”: the ratio of the expressed protein (mCherry fluorescence) over activation of insolubility biosensor (GFP fluorescence). Data is presented as mean values of three biological replicates, dots are individual data points. In (a–c) fluorescence was measured using a microplate spectrophotometer, whereas in (e) fluorescence was measured using flow cytometry. Auto Fl = auto fluorescence, arb. units = arbitrary units. Source data are provided as a Source Data file.

Fig. 2. Predicted structure and sequence of DEBSM6 and KS-AT and post-AT linker.

a AlphaFold structure prediction of homodimeric DEBSM6 without TE, (b) a highlighted structure of the KS-AT linker in dark red, and (c) the post-AT linker in teal. The regions highlighted are the sequences selected for the junction library. d Alignment of the region selected for the junction library of DEBSM6 and EpoM4 KS-AT and (e) post-AT linker with secondary structure elements predicted by AlphaFold. Domain boundaries marked with blue boxes were predicted using an online tool, except the start of ψKR which was placed at the beginning of β5. Each junction position in KS-AT linker is sequentially named us1-102 and post-AT junctions called ds1-90. Highly conserved residues are marked with asterisks. DEBSM6 KS-AT linker sequence starts with HV to denote where the conserved GTNAH sequence is positioned. Gaps in alignments are marked in grey. For reference, the position of the first amino acid in the KS-AT sequence is I1908 for DEBSM6 and V1948 for EpoM4 and in the post-AT linker, the first position is A2301 and P2345, respectively.

Fig. 3. Creation and measurement of randomized junction library.

a Strategy for creating junction library: (A) Pooled libraries of oligos, that each has the junction (where green DEBSM6 sequence and blue EpoAT sequence meet) positioned at a different place in the linker region, amplifies EpoM4 AT with PCR and are inserted into (B) DEBSM6 mCherry with the native AT excised. The resulting library (C) consists of a randomized upstream and downstream junction, with 5256 possible combinations. b Fluorescence measurement of junction library in ΔarsB::P_ibp GFP strain using flow cytometry. Each dot represents one measured colony. Colored areas are estimations where differently soluble variants would appear. c,d Comparing junction positions between randomly picked colonies (blue sticks) with high solubility colonies (green sticks) in the KS-AT linker (c) and in the post-AT linker (d). Dotted line denotes selected linker region. Certain junction positions (e.g. ds62) were overrepresented due to how the library was designed: An alignment of DEBSM6-EpoM4 was used to decide junctions. Where there are gaps in the alignment (see position ds62 in Fig. 2e), the same start position of DEBSM6 was used for several end positions of EpoM4. This led to 8 unique variants all sharing ds62 as a junction. Source data are provided as a Source Data file. Arb. units = arbitrary units.

Fig. 4. Solubility and in vitro productivity of high solubility library variants.

a Solubility measurement using ΔarsB::P_ibp GFP biosensor strain of 40 high solubility colonies, a subset of which have their junction positions marked and were selected for (b) SDS-PAGE quantification of PKS abundance in different protein fractions. (c) In vitro reaction scheme with the synthetic starter 1 and methyl- or malonyl-CoA as substrate. DESBM6 AT natively accepts methylmalonyl-CoA while EpoM4 AT can accept both methyl- and malonyl-CoA. In vitro production of methyl TKL after 1 and 24 hours (d) and desmethyl TKL after 24 hours (e). DEBSM6 is the parental PKS, D1 is included as a reference to what is currently known as the optimal junctions for domain exchange. D0 was excluded due to it already been shown to be inactive. All strains were induced with 250 µM IPTG. Data is presented as mean values of three biological replicates, dots are individual data points. Arb. units = arbitrary units. Source data are provided as a Source Data file.

Fig. 5. Effect of KS-AT linker junction positions on solubility and product formation.

a Testing synergy between KS-AT and post-AT linker junctions. The solubility coefficient was measured using the ΔarsB::P_ibp GFP strain harboring DEBSM6 mCherry engineered with EpoM4 AT with different junction combinations. The color of the lines indicates the position of the post-AT junction, while the x-axis labels indicate the position of the KS-AT junction. b KS-AT junction effect on solubility coefficient measured using ΔarsB::P_ibp GFP strain harboring DEBSM6 mCherry engineered with EpoM4 AT or TiaM4 AT with downstream junction at ds25. Sample pairs tested for in vitro production in (e) and (f) are marked with an arrow. Data points from neighboring junction positions that give identical polypeptide sequence due to homology between the parental PKS sequence and exchanged AT sequences are repeated. For reference, the DEBSM6 mCherry solubility coefficient was 71.1 ± 0.5, and its expression was induced with 250 µM IPTG. c, d Solubility data from (b) visualized on predicted protein structure of DEBSM6 engineered with EpoM4 AT (c) or TiaM4 AT (d). e, f In vitro production of desmethyl TKL (e) and methyl TKL (f) of variant pairs of DEBSM6 with EpoM4 AT. Reactions were sampled at 1 hour (red bars) and 24 hours (green bars). Data is presented as mean values of three biological replicates, dots are individual data points. Arb. units = arbitrary units. Source data are provided as a Source Data file.