An Iterative Module in the Azalomycin F Polyketide Synthase Contains a Switchable Enoylreductase Domain

Abstract Detailed analysis of the modular Type I polyketide synthase (PKS) involved in the biosynthesis of the marginolactone azalomycin F in mangrove Streptomyces sp. 211726 has shown that only nineteen extension modules are required to accomplish twenty cycles of polyketide chain elongation. Analysis of the products of a PKS mutant specifically inactivated in the dehydratase domain of extension‐module 1 showed that this module catalyzes two successive elongations with different outcomes. Strikingly, the enoylreductase domain of this module can apparently be “toggled” off and on : it functions in only the second of these two cycles. This novel mechanism expands our understanding of PKS assembly‐line catalysis and may explain examples of apparent non‐colinearity in other modular PKS systems.

(Epicentre, # MP5105). All procedures were carried out in accordance with the manufacturer's recommendations.
Procedure for gene disruption in vivo. The constructs used for gene disruption were introduced into Streptomyces sp. 211726 by conjugation using donor strain ET12567/pUZ8002 on SFM plates. After incubation at 28°C for 16 h, exconjugants were selected on SFM plates with 25 µg ml -1 apramycin and 25 µg ml -1 nalidixic acid. Single exconjugants were transferred to an SFM plate containing 50 µg ml -1 apramycin and 25 µg ml -1 nalidixic acid to confirm their antibiotic resistance. They were then patched onto SFM plates containing 50 µg ml -1 apramycin and onto SFM plates without antibiotic, respectively, to screen for the double crossover mutant. Candidate mutants with the correct phenotype (Apr S ) were further verified by PCR, sequencing or Southern blot.

Production, isolation and LC-ESI-HRMS analysis of AZL.
To obtain AZL production by Streptomyces sp. 211726 and mutants, 40 ml TSBY liquid medium was inoculated and grown as seed culture. After 2-3 days, 7 ml seed mycelium was inoculated into 700 ml SFM liquid medium and grown as described above. After 10 days, the supernatant was extracted with 700 ml ethyl acetate, and the mycelium was extracted with 500 ml methanol. The combined organic phase from both the culture supernatant and the mycelium (using both ethyl acetate and methanol extraction) was evaporated to dryness under reduced pressure using a Speed-Vac, then subjected to chromatography on Sephadex LH-20 (40-70 µm) and on a Phenomenex Synergi C18 column (250×10 mm, 4 µm) for separation and purification of AZL and derivatives.
On-line LC-ESI-HRMS analysis was carried out on a Thermo Electron LTQ-Orbitrap XL using positive-mode electrospray ionization. The LTQ-Orbitrap XL was coupled to Thermo Accela 600 fitted with a Phenomenex Luna C18 column (250×4.6 mm, 5 µm) at a flow rate at 1 ml min -1 . The gradient for separation of AZL from crude sample: 0 min 90% 0.1% formic acid in H 2 O (A) and 10% 0.1% formic acid in acetonitrile (B), 0-2 min 10% B to 80% B, 2-10 min 80% B to 95% B, 10-11 min 95% B, 11-12 min 95% B to 10% B, 12-15 min 10% B. The mass spectrometer was set to full scan (from 200 to 2000 m/z). To obtain enough of the AZL and derivatives to be characterized by NMR spectra, wild-type 211726 and ∆DH mutant were fermented on a larger scale. 1.5 l seed culture was inoculated to 40 l SFM liquid medium. After 5-day fermentation at 28˚C, same procedure was performed to obtain desired compounds. About 15 mg compound was obtained from 40 l fermentation.

Preparation of gene disruption constructs.
To disrupt the azl gene cluster, two homologous recombination fragments 2069 bp and 2001 bp flanking the ~120 kb azl cluster in the genome were amplified by PCR using primer pairs ∆azl-L1 and ∆azl-L2, ∆azl-R1 and ∆azl-R2, respectively. After digestion with appropriate restriction enzymes, the two fragments were cloned into Streptomyces-E. coli shuttle vector pYH7 [1] treated with NdeI and HindIII by three piece ligation to create pWHU2790. To verify the in-frame deletion in the construct and the mutant, a pair of primers ∆azl-CP1 and ∆azl-CP2 flanking the target gene were used for PCR and sequencing.
Site-directed mutation of the DH 1 domain. To inactivate the DH 1 domain by site-directed mutagenesis in vivo, a 11592 bp KpnI-digested fragment harboring the azlA gene from cosmid 3G11 was ligated to KpnI and FastAPtreated pYH7 [1] vector. Two fragments were amplified from the above recombinant plasmid using primer pairs ∆DH-L1 and ∆DH-L2, ∆DH-R1 and ∆DH-R2, and fused into one fragment by overlapping PCR, to obtain the sitedirected mutation donor. The recombinant plasmid was mutated by the donor via PCR-targeting [2] in BW25113/pIJ790 to generate pWHU2791, the final construct for site-directed mutation of DH 1 domain. To verify the site-directed mutation in construct and mutant, a pair of primers ∆DH-CP1 and ∆DH-CP2 flanking the target site were used for PCR and sequencing.

Characterization of AZL derivatives by NMR and LC-ESI-HRMS. The 1 H-,
and DEPT NMR spectra were collected on an Agilent 400MR DD2 NMR spectrometer. Chemical shifts were reported in ppm using tetramethylsilane as an internal standard, and NMR data processing was performed by using MestReNova software. LC-ESI-HRMS analysis was performed on a Thermo Electron LTQ-Orbitrap XL mass spectrometer using the method described above.

Preparation of protein expression constructs.
To generate the Azl4 expression construct, amplification of azl4 was accomplished with primers exAzl4-F and exAzl4-R. The PCR product was cloned into pET28a(+) to yield pWHU106. Similarly, the constructs pWHU107, pWHU86, pWHU2796 were generated by using primer pairs exAzl5-F and exAzl5-R for expression of Azl5, exAzlA-F and exAzlA-R for expression of AzlA, exACP1-F and exACP1-R for expression of ACP 1 domain, respectively. To generate the AzlA(∆DH 1 ) expression construct, amplification was accomplished by using the same expression primers of AzlA. The PCR product was cloned into pET28a(+) to yield pWHU2793.  The mass spectrometric data were processed and deconvoluted using the Bioworks software (Thermo Finnigan).     Figure S7. A detailed biochemical mechanism of module 1 processing in wild-type and ∆DH 1 mutant. In wild-type, 4-guanidinobutyryl-CoA, as the starter unit, is loaded onto loading domain (ACP L ) in AzlA by Azl5 (4-guanidinobutyryl-CoA:ACP acyltransferase), then transferred to the neighboring KS 1 domain of module 1. An extender unit, malonyl-CoA, is specifically selected by the AT 1 domain

Hydrolysis of thioester-bound products and LC-ESI
The first elongation by module 1 The second elongation by module 1 AzlA in wild-type AzlA in ∆DH1 mutant followed by a transthioesterification to generate the acyl-S-ACP. Subsequently, the KS 1 domain catalyzes the Claisen-condensation reaction between starter and extender unit resulting in release of CO 2 and formation of a β-keto ester intermediate. Following this, the nascent polyketide chain is reduced stepwise to yield hydroxyl and enoyl group by KR 1 and DH 1 domain, respectively, without further reduction by ER 1 due to domain switch-off in some way. Notably, after the first elongation, the polyketide chain is retrotransferred to KS 1 for an iterative extension rather than normally to KS 2 of the next module. During the second elongation by the same module 1, all steps are the same as first one, except the enoylreduction by ER 1 is switched on to form a fully reduced ketide unit. In contrast, the subsequent reduction is halted after KR 1 reaction in the DH 1 inactivation mutant, leaving one or two hydroxyl groups in the polyketide chain. To better monitor the change of chain on the giant PKS by LC-ESI-HRMS, an additional recombinant holo-ACP 1 domain dissociated from module 1 was added in the in vitro assay, which allow to compete the growing polyketide chain with the integral ACP 1 domain of module 1. The captured intermediates by the dissociated ACP 1 is highlighted in yellow.     Host for protein expression [5] Streptomyces Streptomyces sp.