Dissecting how modular polyketide synthase ketoreductases interact with acyl carrier protein-attached substrates† †Electronic supplementary information (ESI) available: Complete experimental details and additional figures. See DOI: 10.1039/c7cc04625a

Interaction studies show that KR domains possess a generic binding site for ACP domains and provide evidence that the 5′-phosphopantetheine prosthetic group plays a key role in delivering acyl substrates to the active site in the correct orientation.

Electronic Supplementary Material (ESI) for ChemComm. This journal is © The Royal Society of Chemistry 2017

Materials, DNA isolation and manipulation
Restriction endonucleases and Taq ligase were obtained from New England BioLabs. Phusion DNA polymerase was purchased from Thermo Scientific. T5 exonuclease was purchased from Epicentre.
T4DNA ligase, alkaline phosphatase and protease inhibitors were purchased from Roche. All other chemicals were from Sigma-Aldrich.
All constructs were cloned from the MLSA1 or MLSB genes using either transfer polymerase chain reactions [Erijman et al., 2011;Erijman et al., 2014] or Gibson assembly [Gibson et al., 2009] and the primers described in Table S1. The products were treated with DpnI for 2 h at 37 °C and transformed into competent E. coli DH5α cells (Life Technologies) prior to conducting colony polymerase chain reactions using the method of Nybo [Nybo, 2012] to filter out false positive colonies. Plasmid DNA was isolated from an overnight culture using a Wizard Mini-prep set (Promega). Final confirmation of cloning steps was obtained by DNA sequencing (DNA Sequencing Facility, Department of Biochemistry, University of Cambridge).

mKRb
The sequence coding for mKRb in module 7 of the mlsB gene from Mycobacterium ulcerans (Uniprot: Q32YM8; residues 13784-13874; Table S2) was cloned into pVB and the resulting pVB-mKRb plasmid was transformed into competent E. coli Tuner (DE3) cells (Merck). The His 6 -GB1-mKRb fusion protein was expressed and the released mKRb product was purified using the same protocol described above for mKRa. All expression and purification steps were monitored by SDS-PAGE (NuPAGE) 4-12 % Bis-Tris gels (Life Technologies) stained with InstantBlue (Expedeon) ( Figure S1).

Protein expression and purification of apo ACP samples mACPa
The sequence coding for mACPa in module 5 of the MLSA1 gene from Mycobacterium ulcerans (Uniprot: Q6MZA4; residues 11087-11185; Table S2) was cloned into pVH, a modified pET28α vector in which the recognition sequence for thrombin had been replaced with that for tobacco etch virus (TEV) protease. The pVH-mACPa plasmid was transformed into competent E. coli Tuner (DE3) cells (Merck) and N-terminally His 6 -tagged mACPa protein was expressed by growing the cells at 37 °C in 1 L of LB medium, prepared according to standard protocols [Sambrook & Russell, 2001], with 30 μg/mL kanamycin (Sigma) for selection, to a 600 nm optical density of 0.8, followed by induction with 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG; Sigma) and incubation at 15 °C for 16 h.

mACPb
The amino acid sequence for mACPb coded in module 7 of the MLSB gene from Mycobacterium ulcerans (Uniprot: Q32YM8; residues 13784-13874; Table S2) is identical to that of the ACP domain (mACP 9 ) coded by the MLSA2 gene (Uniprot: Q6MZA5; residues 2050-2140) studied previously by Vance and coworkers [Vance et al., 2016]. As described in that paper, a pET28α vector was used to express N-terminally His 6 -tagged mACPb in E. coli Tuner (DE3) cells (Merck). The cells were grown at 37 °C in 1 L of LB medium, prepared according to standard protocols [Sambrook & Russell, 2001], with 30 μg/mL kanamycin (Sigma) for selection, to a 600 nm optical density of 0.8, followed by induction with 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG; Sigma) and incubation at 15 °C for 16 h.
The cells were harvested and mACPb was purified and concentrated as described above for mACPa, except His 6 -tag cleavage was performed using restriction grade thrombin (EMD Millipore). All expression and purification steps were monitored by SDS-PAGE (NuPAGE) 4-12 % Bis-Tris gels (Life Technologies) stained with InstantBlue (Expedeon) ( Figure S2) and the identity of the sample was confirmed by ESI-MS (PNAC facility, Department of Biochemistry, University of Cambridge; Table S3 and Figure S3).

Validation of KR domain secondary structure and reductase activity
Despite extensive trials, purified samples of mKRa and mKRb proved to be unsuitable for analysis by mass spectrometry using the Waters Q-TOF micro system in the Protein and Nucleic Acid (PNAC) Facility, Department of Biochemistry, University of Cambridge, due to aggregation during the injection step. To validate these samples, we confirmed that the enzyme domains possessed secondary structure using circular dichroism (CD) spectroscopy and displayed NADPH-dependent reductase activity against trans-1-decalone using a previously established assay [Siskos et al., 2005].
Runs acquired in triplicate were used to calculate the average spectra displayed in Figure S5, which indiacte that both mKRa and mKRb contain a significant proportion of α-helical secondary structure.  Table S4. These results confirm that the KR domain constructs prepared for this work possess kinetic parameters very similar to those obtained in previous studies [Bali & Weissman, 2006].
Overall, since both mKRa and mKRb contain appropriate secondary structure and show activities similar to those of previously studied equivalents, we concluded that our samples are likely to possess native tertiary structure.

Preparation of holo and acyl-loaded ACP samples
ACP constructs were prepared in the Ppant-attached holo form in vivo by co-expression. Two expression vectors, one coding for the ACP construct (pVH-mACPa or pET28α-mACPb) and the other coding for the broad specificity phosphopantetheinyl transferase Sfp (pET-Sfp) [Quadri et al., 1998], were co-transformed into E. coli Tuner(DE3) cells. Procedures for the expression and purification of the holo ACP species were as described for the apo species in Section 3 above. The modification state of the ACP domain was confirmed by ESI-MS (PNAC facility, Department of Biochemistry, University of Cambridge; Table S3 and Figure S3).
Loading reactions for acyl-ACP species were set up in vitro. An apo ACP sample (0.1 mM) was incubated at 27 °C for 2 h with Sfp (4.4 mM) and either acetoacetyl-CoA (2 mM) or β-hydroxybutyryl-CoA (2 mM) in phosphate buffer (45 mM Na 2 HPO 4 , 5 mM NaH 2 PO 4 , 150 mM NaCl, 0.01 % (v/v) NaN 3 , pH 7.5) supplemented with 10 mM MgCl 2 . To separate the loaded protein from Sfp and any excess substrate, the mixture was purified by size exclusion chromatography, as described in Section 3 above.
The identity of the eluted protein was confirmed by ESI-MS (PNAC facility, Department of Biochemistry, University of Cambridge; Table S3 and Figure S4). For acyl-ACP samples, ESI-MS analysis was carried out after conducting the ITC experiments described in Section 6 below. Although the samples were stored at 4 °C during this 24 h period, both of the acetoacetyl-ACP species proved to be mildly unstable: their mass spectra in panels A and B of Figure S3 show minor species that result from hydrolysis to the holo ACP form. However, the absence of signals from the starting material (apo ACP species) indicate that in both cases the loading reaction had run to completion. By contrast, no hydrolysis was observed for the two β-hydroxybutyryl-ACP species (panels C and D of Figure S3). Interestingly, the heat changes observed on titration of either apo mACPa or apo mACPb against their cognate KR domains in the absence of NADPH or NADP + were miniscule (data not shown); at ± 0.1 μcal s -1 per injection, these responses were little different from those observed in dilution control experiments (see Fig. 2 of the main text). The lack of a strong heat change signature in the absence of cofactor indicates that binding is weak or non-existent, or that the interaction between the injectant and its binding partner must be isothermic under the conditions of the experiment.

Isothermal titration calorimetry experiments
Furthermore, in the absence of cofactor titration of holo or acyl-ACP species, or of any of the prosthetic group fragments detailed in Scheme 2 of the main text, against either of the KR domains again produced thermograms that resembled dilution control experiments (data not shown). Finally, titration of NADPH or NADP + against either of the KR domains yielded similarly weak heat changes (data not shown). Since there is no doubt that the cofactor must bind to the enzyme so that ketoreduction can occur, we conclude that these interactions must be isothermic under the experimental conditions employed here.

Figure S1: Module organization for the mycolactone PKS system
Module organization for the three subunits of the mycolactone PKS system (MLSA1, MLSA2 and MLSB) [Stinear et al., 2004]. The structure of mycolactone is colour coded to match the subunits responsible for synthesizing each segment. A1-and B1-type KR domains are represented in white and magenta, respectively. DH domains that are predicted to be inactive are marked with diagonal black lines. Domain abbreviations: KS, ketosynthase; AT, acyltransferase; KR, ketoreductase; DH, dehydratase; ER,