Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2018 Jul 20.
Published in final edited form as: Cell Chem Biol. 2017 Jul 14;24(7):881–891.e4. doi: 10.1016/j.chembiol.2017.06.008

Formation of an Angular Aromatic Polyketide from a Linear Anthrene Precursor via Oxidative Rearrangement

Guixi Gao a, Xiangyang Liu a, Min Xu a, Yemin Wang a, Fei Zhang a, Lijun Xu a, Jin Lv a, Qingshan Long a, Qianjin Kang a, Hong-Yu Ou a, Ying Wang b, Jürgen Rohr c, Zixin Deng a, Ming Jiang a,*, Shuangjun Lin a,*, Meifeng Tao a,*,1
PMCID: PMC5584589  NIHMSID: NIHMS889197  PMID: 28712746

Summary

Bacterial aromatic polyketides are a group of natural products synthesized by polyketide synthases (PKSs) that show diverse structures and biological activities. They are structurally sub-classified into linear, angular, and discoid aromatic polyketides, the formation of which is commonly determined by the shaping and folding of the poly-β-keto intermediates under the concerted actions of the minimal PKSs, cyclases and ketoreductases. Murayaquinone, found in several streptomycetes, possesses an unusual tricyclic angular aromatic polyketide core containing a 9,10-phenanthraquinone. In this study, genes essential for murayaquinone biosynthesis were identified, and a linear anthraoxirene intermediate was discovered. A unique biosynthetic model for the angular aromatic polyketide formation was discovered and confirmed through in vivo and in vitro studies. Three oxidoreductases, MrqO3, MrqO6, and MrqO7, were identified to catalyze the conversion of the linear aromatic polyketide intermediate into the final angularly arranged framework, which exemplifies a novel strategy for the biosynthesis of angular aromatic polyketides.

TOC Image

Gao et al. discovered the formation of an angular 9,10-phenanthraquinone antibiotic from a linear tricyclic intermediate via oxidative rearrangement by three oxidoreductases, MrqO7, MrqO6, and MrqO3, proving a novel strategy for the biosynthesis of angular aromatic polyketides in nature.

graphic file with name nihms889197u1.jpg

Introduction

Polycyclic aromatic polyketides are a large group of natural products that show diverse chemical structures and a broad variety of biological activities such as anticancer, antibacterial, antifungal, and antiviral. Bacterial aromatic polyketides are generally synthesized by type II polyketide synthases (PKSs). A set of enzymes called minimal PKS, consisting of the ketosynthase heterodimer (β-keto synthase, KSα and chain length factor, KSβ) and a free standing acyl carrier protein (ACP), catalyze iterative Claisen condensations of ACP-tethered malonyl units to generate a highly active poly-β-ketoacyl thioester intermediate (McDaniel et al., 1994). Associated enzymes, including ketoreductases, cyclases, and aromatases, are required to direct the shaping and folding of the nascent poly-β-ketoacyl thioester intermediate and convert it into a polyphenol intermediate with linear, angular, or discoid initial framework (Hertweck et al., 2007). The initial frameworks may undergo oxidative rearrangements, which is often catalyzed by Baeyer-Villiger oxygenases, to afford modified skeletons, for example the transformation of dehydrorabelomycin to the heterocyclic jadomycin and gilvocarcin skeletons (Rix et al., 2005; Tibrewal et al., 2012). In this way an initial linear framework might also be converted into an angular one, e.g., the proposed transformation of a linear anthracycline precursor to the dioxabenzo[a]pyrene ring system of the chartreusin aglycone (Xu et al., 2005). In addition, numerous tailoring enzymes catalyze diverse post-modification reactions, such as hydroxylations (Shen and Hutchinson, 1994; Koskiniemi et al., 2007), epoxidations (Maier et al., 2014), alkylations (Zhang et al., 2007), halogenations (Zhu et al., 2013), and glycosylations (Trefzer et al., 2000; Garrido et al., 2006), to eventually form more complex and biologically active natural products.

Murayaquinone is an angular aromatic polyketide antibiotic initially identified from Streptomyces murayamaensis in 1986 (Sato et al., 1986). It was subsequently isolated from several other unidentified Streptomyces species (Chu et al., 1996). Murayaquinone shares its 9,10-phenanthraquinone core with piloquinone (Polonsky and Lederer, 1963) and haloquinone (Ewersmeyer-Wenk et al., 1981). Early feeding experiments using isotope-labeled precursors suggested a polyketide origin of murayaquinone, derived from ten acetate units via one-step decarboxylation. The unusual incorporation pattern of sodium [1,2-13C2]acetate and sodium [1-13C,18O2]acetate allowed to propose that murayaquinone was biosynthesized from a phenanthrene intermediate via oxidative rearrangement (Gould et al., 1997; Figure 1A).

Figure 1. Formation model of aromatic polyketide murayaquinone.

Figure 1

Open in a new tab

(A) Previously proposed formation of the angular murayaquinone.

(B) Confirmed formation of murayaquinone in this study.

In this work, we identified the murayaquinone biosynthetic gene cluster, and uncovered, through gene inactivation and enzyme studies, a novel biosynthetic route for the generation of angular aromatic polyketides (Figure 1B).

Results

Heterologous production of murayaquinone and its lactone derivatives

S. griseoruber Sgr29 was isolated from soil in the Shennongjia forest in the Hubei province of China. Its fermentation broth showed no inhibition against bacterial indicator strains, such as Staphylococcus aureus or Mycobacterium smegmatis. However, S. griseoruber Sgr29 was subjected to our library expression and analysis (LEXAS) project for activation of silent biosynthetic gene clusters of antimicrobial metabolites in a surrogate host, Streptomyces lividans SBT5 (Xu et al., 2016), leading to the identification of three BAC clones (3B4, 2E12, 6F3) that endowed the host with growth inhibitory activity against S. aureus and M. smegmatis. All three exconjugants, SBT5/3B4, SBT5/2E12, and SBT5/6F3, produced the same distinguished peaks (1, 2 and 3) in HPLC analyses (Figure 2A), suggesting that the three BAC clones shared the same biosynthetic gene cluster.

Figure 2. Production of murayaquinone and its derivatives by heterologous expression of three BAC clones containing the mrq gene cluster.

Figure 2

Open in a new tab

(A) Heterologous production of murayaquinone 1 and murayalactones 2 and 3 in S. lividans SBT5 containing BAC clones. Note that there was no production of 1–3 in the native strain S. griseoruber Sgr29. The pks deletion mutant (Δpks) and the vector control did not produce compounds 1–3. HPLC traces were recorded at 280 nm.

(B) Chemical structures of murayaquinone 1 and murayalactones 2–3.

A 10-L fermentation of S. lividans SBT5/3B4, referred to as the wild type strain (WT) in this study, allowed the purification and structural elucidation of the three compounds. Based on high-resolution mass spectrometry (HR-MS) data (Table S1) and 1H and 13C NMR data (Table S2), 1 and 2 were determined to be murayaquinone and murayalactone, respectively, that were previously identified from S. murayamaensis (Sato et al., 1986; Melville and Gould, 1994). The HR-MS and NMR data suggested 3 to be an isomer of 2 (Figure 2B; Table S1 and S2). Of critical importance were the assignments for C-8a and C-9a based on heteronuclear multiple bond correlation (HMBC). Correlations to C-8a (105.6 ppm) from H-5 (7.54 ppm) and H-7 (7.09 ppm) identified C-8a, which is linked to a carbonyl group. Correlation to C-9a (139.4 ppm) from H-4 (7.68 ppm) identified C-9a, which is attached to an oxygen atom (Table S2). The fact that compound 3 was found along with 1 and 2 indicated that it was biosynthesized in the same pathway. The structures of both 2 and 3 suggested these to be oxidative follow-up products of 1.

Identification of essential genes for the biosynthesis of murayaquinone

Sequencing of 3B4 yielded a circa 100 kb DNA region assumed to be responsible for the biosynthesis of murayaquinone, since it contained a minimal type II PKS gene set (mrqA, mrqB, and mrqC encoding KSα, KSβ, and ACP, respectively), whose deletion abolished the production of murayaquinone and its lactones (Figure 2A, Δpks). Bioinformatic analysis of the sequences flanking the minimal PKS genes revealed 58 open reading frames (ORFs) related to secondary metabolism (Figure 3A).

Figure 3. Determination of the boundaries of the mrq gene cluster by deletions.

Figure 3

Open in a new tab

(A) Murayaquinone biosynthetic gene cluster in BAC 3B4.

(B) HPLC analysis of S. lividans SBT5/3B4 and its deletion mutants. Murayaquinone 1 and lactones 2–3 were produced by S. lividans SBT5/3B4 and all mutants except for ΔRT. Note: ΔRT was the mutant in which genes mrqT2 to orf0 were deleted. Δright, mutant in which a 30-kb fragment (including orf25-orf28) near the right boundary of the mrq gene cluster was deleted. Δleft, mutant in which a 11.8-kb fragment (from orf-3) near the left boundary of the mrq gene cluster was deleted.

To determine the essential genes for the biosynthesis of murayaquinone among these ORFs, we first determined the boundaries of mrq gene cluster by gene knockouts. Mutant Aright (removing circa 30 kb including Δorf25-28) still produced 1–3 in amounts comparable to the wild-type strain (i.e.SBT5/3B4) (Figure 3B), while deletion of the adjacent gene mrqL resulted in a much higher production of 1 compared to the WT, and slightly lower production of 2 and 3 (Figure 4). Thus mrqL was determined as the right boundary of the gene cluster.

Figure 4. Metabolite profiles of the single gene deletion mutants of the mrq gene cluster.

Figure 4

Open in a new tab

(A) No metabolite was detected in mutants ΔmrqE, ΔmrqF, ΔmrqG, ΔmrqM, and ΔmrqO3.

(B) Accumulation of both angular compounds 1–3 and the anthraquinone-type compounds 4 and 5 in ΔmrqH, ΔmrqI, ΔmrqK, and ΔmrqO2.

(C) Accumulation of anthraquinones 4, 5 or anthrene 6 in ΔmrqO4, ΔmrqO5, ΔmrqO6, and ΔmrqO7.

(D) Production of 1–3 in ΔmrqD, ΔmrqO8, ΔmrqO1, ΔmrqJ, and ΔmrqL.

(E) Chemical structures of 4–6.

The determination of the left boundary of the mrq gene cluster was more complicated. Mutants Δleft (removing 11.8 kb from orf-3 to the insert-vector boundary), Δorf-2, Δorf-1, Δorf(1-22), Δorf23 and Δorf24 still produced 1–3 in the same amounts as SBT5/3B4. However, mutant ΔRT (removing 8 kb from mrqT2-orf0) abolished the production of 1–3, which was unexpected (Figure 3B). Therefore this region, which contains two transporter genes (mrqT1 and mrqT2) and two regulatory genes (mrqR2 and mrqR3), was called the regulation region. Inactivation of mrqO8 abolished the production of 1 and significantly reduced the production of 2 and 3 (Figure 4). Thus, the region between mrqO8 and mrqL was named the biosynthesis region. Therefore, the mrq gene cluster includes two regions that are separated by an insertion of about 28 kb DNA, with a transposase-encoding gene orf0 located at the left end of the insertion (Figure 3A).

In addition to the minimal PKS, the biosynthesis region encodes 3 ketoreductases (MrqF, MrqH, and MrqM), 3 cyclases (MrqD, MrqE, and MrqJ), 8 oxidoreductases (MrqO1-O8), 1 regulator (MrqR1), and 4 hypothetic proteins (MrqG, Mrql, MrqK, and MrqL) (Table 1). To interrogate the genes necessary for the biosynthesis of 1–3, single-gene deletion mutants were constructed, fermented, and analyzed by HPLC. The mutants were grouped roughly according to their abilities to produce 1–3 and to accumulate related compounds (Figure 4A–D). Mutants ΔmrqE, ΔmrqF, ΔmrqG, ΔmrqM, and ΔmrqO3 abolished the production of 1–3, and did not accumulate any metabolites related to 1, 2 or 3 (Figure 4A). Mutants ΔmrqH, Δmrqi, ΔmrqK, and ΔmrqO2 still produced 2 and 1 and significantly reduced the productio, but also accumulated either one or two anthraquinone-type of compounds, compounds 4 and 5, respectively (Figure 4B). ΔmrqO4 and ΔmrqO5 also accumulated 4 and 5, but abolished the production of 1–3. ΔmrqO7 abolished the production of 1–3, and accumulated another new metabolite, compound 6, while ΔmrqO6 accumulated trace amounts of 5 and 6 (Figure 4C). ΔmrqD and ΔmrqO8 produced only trace amount of 2 and 3. ΔmrqO1 and ΔmrqJ still produced 1–3 (Figure 4D). These findings confirmed all genes that partake in the production of 1–3 within the biosynthesis region, except for mrqJ and mrqO1.

Table 1.

Deduced function of the gene products of the mrq gene cluster.

Proteins (aa) Proposed function Homologues (Sequence ID in NCBI, Origin) Identity/Similarity (%)
Regulation region
MrqT2 (725) transporter WP_023423080.1, S. sp. GBA 94-10 57/69
MrqR3 (644) regulator WP_051823225.1, S. sp. NRRL S-1448 42/56
MrqT 1 (574) transporter WP_063760793.1, S. aureocirculatus 78/86
MrqR2 (177) regulator WP_051852929.1, S. aureocirculatus 73/86
Biosynthesis region
MrqO8 (471) oxygenase/dehydratase JadH, S. venezuelae 38/48
MrqR1 (266) regulator WP_020939953.1, S. collinus 52/68
MrqM (245) 3-oxoacyl-ACP reductase SsfK, S. sp. SF2575 51/67
MrqA (417) KSα PgaA, S. sp. PGA64 80/86
MrqB (414) KSβ ChaB, S. chartreusis 70/80
MrqC (87) ACP AGO50612.1, S. lusitanus 51/67
MrqO1 (308) oxidoreductase AHL46730.1, S. bottropensis 56/66
MrqD (319) second ring cyclase AlnR, S. sp. CM020 58/69
MrqE (317) cyclase PgaL, S. sp. PGA64 57/69
MrqF (262) polyketide C-9 ketoreductase PgaD, S. sp. PGA64 71/80
MrqO2 (172) flavin reductase AHL46683.1, S. bottropensis 32/51
MrqG (356) unknown WP_067167737.1, S. sp. ERV7 78/82
MrqO3 (489) FAD-binding oxidoreductase WP_030637315.1, S. flavovirens 49/64
MrqO4 (351) oxidase AHL46682.1, S. bottropensis 61/76
MrqH (256) ketoreductase MtmTIII, S. argillaceus 58/66
MrqI (153) unknown WP_049531873.1, Streptococcus pseudopneumoniae 31/44
MrqJ (177) cyclase AknE1, S. galilaeus 46/60
MrqK (154) unknown WP_067167164.1, S. sp. ERV7 75/84
MrqO5 (103) oxygenase ChaH, S. chartreusis 43/58
MrqO6 (532) FAD-binding monooxygenase BexE, Amycolatopsis orientalis subsp. vinearia 29/38
MrqO7 (287) NAD(P)-dependent oxidoreductase WP_040811703.1, Nocardia concava 53/63
MrqL (216) unknown WP_004941039.1, S. mobaraensis 46/61
Open in a new tab

Structural identification of the metabolites accumulated in the mutants

The new peaks 4–6 were only accumulated by a few of the deletion mutants. The compounds were purified and their structures determined through mass spectrometry and NMR spectroscopy. The structure of 4 was identified as mansoquinone by comparing the spectral data including HR-MS, 1H and 13C NMR data with literature data (Figure 4E; Tables S1 and S3; Abdelfattah, 2009). The HR-MS of 5 suggested its molecular formula to be C19H14O5, with a degree of unsaturation of 13, one more than that of 4. This indicated 5 to contain either one more double bond or another ring. The 1H and 13C NMR data of 5 (Table S3) were almost identical to those of 4. One obvious difference was that the methylene at C13-position (δH = 1.76 and δC = 17.1) of 4 was changed to a methine in 5 (δH = 4.78 and δC = 75.1) that is linked to an oxygen atom, and that 5 contained only one phenolic hydroxyl proton forming an intramolecular hydrogen bond. Furthermore, the HMBC spectrum showed a correlation of this methine proton (H-13) to the aromatic carbon C-1 (Table S3). Thus, 5 was identified as a derivative of 4 with C1 linking C13 though an oxygen bridge, thereby forming an additional pyran ring (Figure 4E).

The molecular formula of 6 was determined as C19H18O6 by HR-MS, revealing that metabolite 6 contains one O-atom more, but two H-atoms less than 4 (Table S1). The 13C-NMR spectrum of 6 (Table S3) showed only 10 aromatic carbons, two fewer than 4, and only 2 carbonyl carbons, one fewer than 4. There were 3 more carbons linked to oxygen, two of these are quaternary carbons (δC = 66.5 and 68.7) and one is a methine carbon (δH = 4.98 and δC = 69.2). The methyl group, which in 4 is attached to an aromatic ring (δH = 2.37 and δC = 20.1), is in 6 attached to a saturated carbon (δH = 1.66 and δC = 16.4). All these data combined led to structure 6, which was further supported by 2D NMR spectra (Figure 4E; Table S3).

Identification of genes responsible for the formation of 1 from 6

To check whether the compounds accumulated in the mutants are intermediates or shunt products in the murayaquinone biosynthetic pathway, we performed feeding experiments with compounds 4 and 6 to the Δpks mutant. We did not include 5 in these studies, because of its much different structure, which likely rendered 5 as a shunt product. When compound 4 was fed, neither 1 nor 2 nor 3 were detected by HPLC analysis (Figure S1), while 1–3 were produced when 6 was fed to the same mutant (Figure 5A). These findings demonstrated that 4 is a shunt product, while 6 is an actual biosynthetic intermediate of murayaquinone.

Figure 5. In vivo and in vitro characterization of the conversion of 6 to 1.

Figure 5

Open in a new tab

(A) Detection of in vivo conversions in four mutants. 6 was converted to 1–3 in Δpks, ΔmrqG, and ΔmrqM, but not in ΔmrqO3. The wild type strain (SBT5/3B4) producing 1–3 and two mutants producing 6 (ΔmrqO7 and ΔmrqO6) were used as controls to indicate the peaks.

(B) Biochemical conversion of 6 to 1 with different combinations of enzymes and cofactors. All reactions were conducted at 30°C for 60 min. The complete reaction sequence was achieved by a mixture of three enzymes (MrqO3, MrqO6 and MrqO7), three cofactors (NADPH, NADH and FAD), and the substrate 6. Other reactions were the same mixtures as for the complete reaction sequence except for the indicated omission, for example, −MrqO3 means minus MrqO3 from the complete reaction mixture. HPLC traces were recorded at 280 nm.

(C) LC-MS analysis of the enzymatic conversion of 6 by single proteins. Compound 6 (246 μM) was incubated with no enzyme, MrqO3 (2 μM), MrqO6 (2 μM), or MrqO7 (2 μM) in the presence of NADPH at 30°C for 30 min, and the samples were detected immediately. (i)–(iii) show the extracted ion chromatogram (EIC) of 365.3, 367.3, and 381.3, which are the m/z ([M+Na]+) values of the substrate (compound 6), the +2 Da intermediate, and the +16 Da intermediate, respectively.

The accumulation of 6 in ΔmrqO7 and ΔmrqO6 suggested that MrqO7 and MrqO6 played pivotal roles in the conversion of 6 into 1. To further determine whether other genes encode additional enzymes involved in this conversion cascade, compound 6 was fed to seven single-gene deletion mutants that abolished the production of 1–3. Compound 6 restored the production of 1–3 when fed to all of these mutants, for example the ΔmrqG and ΔmrqM mutants, except when fed to the ΔmrqO3 mutant (Figure 5A). This hinted that MrqO3 played a pivotal role in the conversion of 6 into 1, along with MrqO6 and MrqO7. Sequence analysis suggested MrqO3 to be a putative cholesterol oxidase, MrqO6 to be a putative flavin adenine dinucleotide (FAD)-dependent monooxygenase homologous to BexE, and MrqO7 to be a putative NAD(P)-dependent oxidoreductase. Note that BexE was proposed to catalyze a Baeyer-Villiger reaction in the BE-7585A biosynthesis (Jackson et al., 2016).

Biochemical characterization of MrqO3, MrqO6 and MrqO7 catalyzing conversion of 6 to 1

To uncover the mechanism of MrqO3, MrqO6 and MrqO7 that catalyze the conversion of 6 into 1, these three oxidoreductases were expressed in Escherichia coli BL21(DE3). MrqO3 and MrqO6 were purified as bright yellow proteins, revealing the presence of a flavin prosthetic group, while MrqO7 was purified as a colorless protein. LC-MS analysis of the supernatant of denatured MrqO3 and MrqO6 proved the presence of FAD (Figure S2). Thereafter, we designed an in vitro assay system containing MrqO3, MrqO6, and MrqO7 in the presence of NADPH, NADH and FAD, using 6 as a substrate. After the assay mixtures were incubated at 30°C for 60 min, 6 was completely converted to 1, whereas the control reaction with boiled enzymes did not convert 6. Whenever any one of these three enzymes was removed from the assay, 1 was not detected anymore, manifesting that the cooperation of all three enzymes was essential for the conversion of 6 to 1. The presence of either NADH or NADPH in the assay gave similar conversions of 6 to 1, and 1 was not generated in absence of these co-factors, suggesting that the enzymes show no preference for NADH or NADPH. However, the assay yielded similar conversions of 6 to 1 in absence of exogenous FAD (Figure 5B), likely due to FAD bound to two of the proteins. Based on these findings, a time-course study of the conversion of 6 to 1 catalyzed by the three enzymes was performed with 2 μM of each enzyme, 1 mM of NADPH, and 360 μM of 6. Under these conditions 6 was completely converted to 1 within 30 min. In addition, since MrqO3 is a homologue of cholesterol oxidase, we also monitored H2O2 production using Fluorimetric Hydrogen Peroxide Assay Kit. No H2O2 was detected in these reactions, indicating that H2O2 was not formed.

Subsequently, the sequence of events involving each enzyme was investigated. For this, the in vitro assay was changed, now using each enzyme individually in the presence of NADPH and 6. The reactions were carried out by incubation at 30°C for 30 min, and LC-MS was used to monitor the formation of possible products (Figure 5C). The reaction with MrqO3 did not show any conversion of substrate 6. The reaction with MrqO6 generated an obvious new ion peak at m/z 381.3 ([M+Na]+), 16 Da more than that of 6 (365.3, [M+Na]+), indicating one oxygen atom to be inserted into 6, resulting in structures 7 or 7′ in Figure 6. When MrqO7 was used in the assay, the LC-MS analysis revealed a new an ion peak at m/z 367.3 ([M+Na]+), 2 Da more than that of 6, indicating an addition of two hydrogen atoms, consistent with a reduced product, e.g., intermediate 8. These results indicated that both MrqO6 and MrqO7 could recognize 6 as substrate, while MrqO3 catalyzed the following last step of the conversion of 6 to 1.

Figure 6. Proposed biosynthesis of 1–3.

Figure 6

Open in a new tab

(A) Proposed biosynthetic pathway of murayaquinone 1 and its lactone derivatives 2 and 3.

(B) Proposed enzymatic mechanism of the oxidative rearrangement catalyzed by MrqO6.

(C) Proposed enzymatic mechanism of the conversion of 9 to 1 catalyzed by MrqO3.

To further clarify the catalytic order involving MrqO7 and MrqO6 during the conversion of 6 to 1, we carried out three parallel sequential reactions SR-I, SR-II, and SR-III. The sequential-reactions were all run in a two-step manner with the MrqO7-catalyzed reduction or the MrqO6-catalyzed Baeyer-Villiger reaction separately set as alternative first steps, while MrqO3 was always involved in the following second step. After the first-step reaction, the enzyme was removed by filtering, which was confirmed using the Bradford Protein Assay Kit. Subsequently, the second-step was initiated through addition of the other two enzymes. In SR-I, the MrqO7-catalyzed hydrogenation was set as the first-step reaction, while the MrqO6-catalyzed Baeyer-Villiger reaction was set as the first-step reaction in SR-II. The complete conversion of 6 to 1 catalyzed by all three enzymes was used as the positive control in SR-III. HPLC analysis showed that the production of 1 in SR-I was similar to that observed in the positive control, much higher than that in SR-II (Figure S3). These observations unambiguously verified that the reductive epoxide opening of 6 catalyzed by MrqO7 occurs prior to the MrqO6-catalyzed Baeyer-Villiger reaction.

Discussion

Bacterial aromatic polyketides are an important class of natural products because of their diverse chemical structures and bioactive properties. On the basis of the polyphenolic ring system, aromatic polyketides were classified into several different subclasses such as anthracyclines, angucyclines, tetracyclines (Zhan, 2009). These natural products possess either a linear or an angular shaped core structure, with the exception of resistomycin and resistoflavin, which have a discoid skeleton (Brockmann and Reschke, 1968; Brockmann et al., 1969; Ishida et al., 2006; Fritzsche et al., 2008). The formation of the linear or angular architecture of bacterial aromatic polyketides from poly-β-keto intermediates is proposed to result from the concerted actions of minimal PKSs and associated ketoreductases, cyclases, and aromatases (Hertweck et al., 2007). Surprisingly, several angular aromatic polyketides, such as the anticancer agent BE-7585A, the antibiotic murayaquinone, and chartreusin, displayed unexpected, atypical polyketide labeling patterns from 13C-acetate feeding experiments (Sasaki et al., 2010; Gould et al., 1997; Canham and Vining, 1976; Canham et al., 1977). Therefore, these angular aromatic polyketides were proposed to arise from oxidative rearrangements of linear polyphenolic intermediates, involving C-C bond cleavage and new C-C bond formation. Indeed, a linear anthracyclic metabolite, resomycin C, was accumulated upon inactivation of ChaZ, a Baeyer-Villigerase-like FAD-dependent oxygenase of the chartreusin biosynthesis. However, ChaZ was found unable to convert resomycin C (Xu et al., 2005). In the BE-7585A biosynthetic pathway, three oxygenases (BexE, BexI, and BexM) had been proposed to catalyze the oxidative rearrangement of an unidentified linear anthracyclinone intermediate to generate an angular angucyclinone intermediate in the biosynthesis of anticancer agent BE-7585A (Sasaki et al., 2010). Most recently, the crystal structure of BexE was resolved to 2.65 Å and used to perform docking stimulations to determine the key linear intermediate. However, unfortunately, the identification of BexE’s true substrate remains challenging (Jackson et al., 2016). In contrast, in this work on murayaquinone biosynthesis, we proved and confirmed the conversion of a linear polyphenolic intermediate to an angular aromatic polyketide.

(1) Characterization of the mrq gene cluster and the murayaquinone biosynthetic pathway

The entire biosynthetic gene cluster was cloned and expressed in the heterologous host S. lividans SBT5 utilizing the previously described LEXAS technology (Xu et al., 2016), because the natural strain S. griseoruber Sgr29 did not produce murayaquinone. The 28 kb mrq gene cluster consists of two parts, a regulation/efflux region and a biosynthesis region, separated by an insertion of a 28-kb unrelated fragment implied to have originated horizontally (evident from the presence of a transposase gene orf0 (Figure 3A)).

The murayaquinone biosynthesis pathway was updated after a bioinformatics analysis, genetic studies, and compound production from gene deletion mutants (Figure 6A). The generation of the first tricyclic aromatic intermediate (10) was proposed based on the typical biosynthetic logic of PKS II for the synthesis of some decaketides via C-9 ketoreduction and C7–C12 first ring cyclization such as nogalonic acid (Kantola et al., 2000; Hautala et al., 2003), tetracycline SF2575 (Pickens et al., 2009), and mensacarcin (Maier et al., 2015). Besides the minimal PKS, the sequences of ketoreductase MrqF, and of the two cyclases MrqE and MrqD were highly similar to SnoaD/SsfU/MsnO11, SnoaE/SsfY1/MsnC3, and SsfY2/MsnC2, respectively, that were responsible for the C-9 ketoreduction, C7–C12 first ring cyclization, and second/third ring cyclization, respectively.

Subsequent modifications of 10, by MrqH, MrqO8, Mrql, MrqK, MrqO5, MrqO1, and MrqO4, led to the production of key intermediate 6 (Figure 6A). MrqH, a ketoreductase homologue, was proposed to catalyze the reduction of the keto group for the butyryl side chain formation of 11. Since the distant keto group of 10 can possibly be reduced, less effectively, by ketoreductases from the host, traces of murayaquinone and the murayalactones can be detected as products of the ΔmrqH mutant. However, in absence of MrqH 10 reacts predominantly spontaneously, initially to form a pyrone ring by cyclization and dehydration, followed by other spontaneous reactions, an oxidation to form the anthraquinone, reduction of the pyrone lactone to a hemiacetal likely through a hydrogenase from the host, (McDaniel et al., 1994; Fu et al., 1994), to eventually produce 5 (Figure 4B).

MrqO8 is an FAD-binding motif-containing protein showing homology to JadH that catalyzes dehydration and oxidation in the jadomycin biosynthetic pathway (Chen et al., 2005; Chen et al., 2010), so it was speculated to participate in the formation of the first aromatic ring or the conversion of 11 to 12 by dehydration (Figure 6A). Its inactivation mutant still produced murayaquinone and murayalactones, but with drastically decreased production levels compared to the wild-type strain, probably due to spontaneous dehydration (Figure 4D).

MrqI and MrqK are NTF2-like superfamily small proteins with a SnoaL_2 and SnoaL_4 domain, respectively (Marchler-Bauer et al., 2016). The metabolic profiles of MrqI- and MrqK-inactivation mutants were very similar except that the compound production in Δmrqi was much lower. Both Δmrqi and ΔmrqK mutants accumulated 4 and 5 while decreasing production of 2 and 3 (Figure 4B). Taken together, MrqI and MrqK may take part in the reduction of 12 to 13, or reduce the keto group of 15 to form the hydroxyl of 6 (Figure 6A).

The formation of 6 from 13 also involves three oxidation steps catalyzed by MrqO5, MrqO1, and MrqO4 (Figure 6A). This was proposed based on bioinformatics analysis (Table 1) and metabolic profiling of their inactivation mutants (Figures 4C and 4D).

(2) Biochemical characterization of the conversion of the linear intermediate 6 to the angular product murayaquinone

The accumulation of anthrene type compounds (4, 5, 6) in several mutants indicates that a linear anthrene-type compound, not the previously proposed angular phenanthrene, may be the true substrate of the oxidative rearrangement to form the angular murayaquinone. Biotransformation of 6 to 1 in the PKS-inactivation mutant suggested 6 to be an actual intermediate of the biosynthetic pathway of 1. Accumulation of 6 in the mrqO6 and mrqO7 deletion mutants suggested that these two genes encoded enzymes for the 6 to 1 conversion.

The enzymatic conversion of 6 to 1 was then investigated. Since both the MrqO6- and MrqO7-inactivation mutants accumulated compound 6, this compound should generally be the substrate of either MrqO6 or MrqO7. MrqO6 is an FAD-dependent monooxygenase with 29% identity and 38% similarity to BexE that has been proposed to catalyzed a Beayer-Villiger reaction in the BE-7585A biosynthesis and MrqO7 is an NAD(P)-dependent oxidoreductase. Unfortunately, when 6 was incubated with MrqO6 or MrqO7, or both MrqO6 and MrqO7, 6 was consumed and many tiny peaks were detected by HPLC (Figure S4), indicating unstable products, consistent with the findings for BexE and BexF (Jackson et al., 2016), although MrqO7 shows no homology to BexF. Thus, as proposed for BexE and BexF, 6 might be converted to a very unstable compound through MrqO6 and/or MrqO7 that is readily degraded or oxidized to generate several other compounds. Therefore, we assumed that an/other enzyme/s may cooperate closely with MrqO6 and MrqO7 to convert the unstable product of MrqO6 and MrqO7 into a stable compound. But this/these enzyme/s may not accept 6 as the substrate. On the basis of this assumption, 6 was fed to the deletion mutants that did not produce 1–3, with the PKS-inactivation mutant as a positive control. This way, MrqO3, a predicted cholesterol oxidase homologue, was identified as a candidate cooperating with MrqO6 and MrqO7. Indeed, when MrqO3, MrqO6, and MrqO7 were incubated with 6 in the presence of FAD and NADPH, 6 was successfully converted to 1 within 30 min.

The order in which MrqO6 and MrqO7 act during the 6 to 1 conversion was determined by sequential enzymatic reactions. It was clearly demonstrated that the MrqO7-catalyzed hydrogenation on 6 is the first step, followed by the reactions of MrqO6 and MrqO3 to afford the final product 1. Some remaining questions were also addressed by these experiments: (1) MrqO6 may not interact with MrqO7 and no complex should be formed, which is different from the previous speculation (Jackson et al., 2016); (2) The products of MrqO7 and MrqO6 seem to be stable enough in the reaction mixture to allow the follow-up conversion steps to 1, but they readily decompose during the work-up process.

When LC-MS was used to follow the reactions catalyzed by MrqO6 and MrqO7, respectively, possible products can be detected and the type of the reactions can also be deduced from mass spectral analysis. The mass of the product 7/7′ from MrqO6 is 16 Da greater than 6, indicating a single oxygen atom to be inserted into 6, consistent with a Baeyer-Villiger reaction, like those proposed for ChaZ (Xu et al., 2005) or BexE (Jackson et al., 2016). The mass of product 8 from MrqO7 is by 2 Da greater than 6 and NAD(P)H is necessary for its formation, indicating that 2 protons were added to 6, and that the reaction catalyzed by MrqO7 is a reduction. On the basis of the nature of reductive reactions and the structure of the final product 1, we proposed that the hydride from NAD(P)H attacks C-16 to form a vicinal diol 8. The formation of 6 and 8 probably intents to disrupt the aromatic ring and enables the Baeyer-Villiger reaction. To some extent, this may explain why the aromatic resomycin C is not a substrate of ChaZ (Xu et al., 2005). As proposed above, the MrqO6-catalyzed Baeyer-Villiger reaction disrupts the C-C bond between C-14 and C-15 by insertion of an oxygen atom. Since C-6 cannot directly attack the carbonyl C-15 due to the spatial tension and hindrance, a thiol of a cysteine residue or a carboxyl group of MrqO6 may function as a nucleophile to attack C-15, generating an enzyme bound thioester or anhydride intermediate, as shown in Figure 6B. Thioesters are commonly used for an activation of a carboxyl carbon in PKSs and non-ribosomal peptide synthetases (NRPSs), the most similar example being KSs that use cysteine to load and activate the carboxyl carbons (Bisang et al., 1999; Robbins et al., 2016; Walsh, 2016). Anhydrides are also very commonly used for the carboxyl carbon activation, both chemically and enzymatically. For example, adenylation enzymes use acyl adenylation and some amidases use acyl adenylation or phosphorylation to activate carboxyl carbons (Begley et al., 2001; Gulick, 2009). It is very possible that a nucleophilic residue attacks C-15 to form a linear anhydride like 9″, and the electron-rich C-6 subsequently attacks C-15 of 9″, to form the new C-C bond of 9.

MrqO3 is a homologue of cholesterol oxidase. Cholesterol oxidases are a type of flavoenzymes that catalyze oxidation and isomerization of cholesterol in the first step of a metabolic pathway for utilizing cholesterol as a carbon and energy source (Murooka and Yamashita, 2001). Cholesterol oxidases catalyze the oxidation of cholesterol. During this reaction, the cofactor FAD is converted to FADH2 that in turn is spontaneously oxidized by aerial oxygen to generate H2O2 and to recycle FAD (Yue et al., 1999; Lario et al., 2003). In the assays catalyzed by the mentioned three enzymes, no H2O2 was detected using Fluorimetric Hydrogen Peroxide Assay Kit, suggesting that MrqO3 may catalyze oxidative and reductive reactions, accompanying the loss of 2 molecules of water, to eventually form the final product 1 (Figure 6C). First, FAD subtracts the hydride from 9 to generate FADH− and 9a. FADH− subsequently transfers the hydride to the carbonyl carbon C-15 to perform the reductive reaction to form 9b, thereby recycling FAD, distinct from the typical reactions catalyzed by cholesterol oxidases mentioned above. The loss of water of 9b is very likely catalyzed by MrqO3, but a spontaneous dehydration cannot be ruled out.

Significance

Angular aromatic polyketides are a subtype of aromatic bacterial polyketide natural products. Some of these are oxidatively rearranged from initially formed linear intermediates through an FAD-dependent Baeyer-Villiger oxygenase. However, only little biochemical evidence has been obtained to support this. In this study we demonstrated, in vivo and in vitro, that three oxidoreductases cooperate to catalyze the conversion of a linear tricyclic intermediate to the angular 9,10-phenanthraquinone. In the in vitro studies, the cholesterol oxidase homologue MrqO3 played an important role to convert an unstable intermediate to a chemically stable final product. Based on the sequence alignment of the three oxidoreductases, hypotheses for an oxidative rearrangement involving a Baeyer-Villiger reaction were proposed and interrogated. Overall, the studies revealed a novel strategy for the biosynthesis of angular aromatic polyketides.

STAR ★ METHODS

KEY RESOURCES TABLE

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Meifeng Tao (tao_meifeng@sjtu.edu.cn).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Bacterial strains

E. coli DH10B was used as a host for general cloning and construction of the genomic BAC library. E. coli BW25113/pIJ790 (Datsenko and Wanner, 2000; Gust et al., 2003) was used for λ-red mediated recombination to construct deletion mutant S. E. coli BT340 (i.e. DH5α/pCP20) was used for construction of in-frame deletion employing FLP-mediated site-specific recombination (Gust et al., 2003). E. coli ET 12567/pUB307 was used as a helper strain for E. coli-Streptomyces inter-genus conjugation (Flett et al., 1997). S. griseoruber Sgr29 was obtained from China Center for Type Culture Collection (CCTCC). S. lividans SBT5 (Bai et al., 2014) was used as a host for heterologous expression of the whole BAC library and BAC clones carrying the wild type mrq gene cluster and its deleted variant S. Staphylococcus aureus and Mycobacterium smegmatis mc2155 (Snapper et al., 1990) were used as indicators for bioassay.

Plasmids

pHL921 carrying an origin of transfer (oriT), an integrase gene (inf), and a site of integration (attPϕC31) was used as a BAC vector. pHL921-derived plasmids (BAC clones) can be mobilized into Streptomyces and integrate in the chromosomal attachment site (attBϕC31) (Xu et al., 2016). pJTU6722 carrying an FRT-eryB-FRT cassette, i.e. erythromycin-resistance gene ermB flanked by a FLP recombination target site (FRT) at either sides, was used for gene inactivation by λ-red-mediated recombination. pET-28a (Novagen) was used for protein expression.

Culture conditions

Lysogeny broth (LB) medium was used for E. coli growth. MS medium (Kieser et al., 2000) was used for Streptomyces growth, sporulation, and conjugation. No18 agar medium (glycerol 2%, cane molasses 1%, casamino acids 0.5%, peptone 0.1%, CaCO3 0.4%, agar 2%, pH 7.0) (Farnet et al., 2008) was used for fermentation. Streptomyces and E. coli BT340 were grown at 30°C.

METHODS DETAILS

Construction and screen of the genomic BAC library of S. griseoruber Sgr29

Large fragment genomic DNA was extracted from S. griseoruber Sgr29 and used to construct a genomic BAC library according to a previous protocol (Luo and Wing, 2003) in E.coli DH10B, yielding a library with 912 clones and an average insertion of ca. 100 kb. The library was stored in 96-well plates.

BAC clones carrying the mrq gene cluster were identified using the LEXAS technology (Xu et al., 2016). Briefly, the whole library was grown in LB broth (100 μL/well) for 6–8 h at 37°C in 96-well plates to OD600 about 0.6. The conjugation helper strain E. coli ET12567/pUB307 was grown in LB with kanamycin at final concentration of 50 μg/mL for 6 h at 37°C to OD600 0.6, harvested, washed, and resuspended in 1/10 volume of LB. ET12567/pUB307 was then added to each well (10 μL/well) of the 96-well plates and mixed. The conjugation recipient S. lividans SBT5 spores were flooded onto MS agar at ca.108 colony-forming units/9-cm Petri dish. Then the DH10B/BACs-ET12567/pUB307 mixtures in 96-well plates were transferred to the spores-coated Petri dishes by stamping with a 48-pin inoculator. The inoculated Petri dishes were incubated at 30°C for 12–16 h and flooded with 1 mL antibiotic solution of apramycin (to select for exconjugants) and trimethoprim (to select against E. coli cells). After incubation at 30°C for 7 d, the apramycin-resistant S. lividans SBT5 exconjugants were transferred to fresh MS plates containing apramycin and trimethoprim, incubated at 30°C for 7 d, the exconjugants were finally transferred to No18 agar plates for micro-fermentation. After 7 days of fermentation at 30°C, each plates of fer mented exconjugants was flooded with LB (supplemented with 0.5% agar) containing indicators (Staphylococcus aureus or Mycobacterium smegmatis mc2155) and incubated at 37°C for 12–20 h. Inhibitory zones could be observed.

Construction of large deletion mutants and single gene deletion mutants

Gene deletion mutants were constructed on the BAC clone 3B4 by λ-red mediated PCR targeting technology (Gust et al., 2003). The amplified products were purified and transformed into E. coli BW25113/pIJ790/3B4 competent cells by electroporation to replace a targeted fragment or single gene in 3B4 respectively, yielding replacement constructs. For construction of in-frame deletion, the replacement constructs were introduced to E. coli BT340 and cultured at 42°C to remove the FRT-eryB-FRT cassette by FLP-mediated excision with an 81 bp scar left. The resulted constructs were confirmed by PCR analyses. The constructs were introduced into S. lividans SBT5 individually by tri-parental mating, yielding different deletion mutants.

Purification of compounds 1–3 from S. lividans SBT5/3B4

S. lividans SBT5/3B4 was fermented (on 10 L No18 agar medium) and the culture was extracted with equal volume methanol for three times. The resulting crude extract was concentrated by evaporation, suspended in 5 L water, and adjusted to pH 7.0. The resulting water phase was extracted with ethyl acetate (EtOAc). The organic phase was collected and evaporated. The extract was re-dissolved and subjected to silica gel column chromatography, and eluted with a petroleum ether-EtOAc-CH3OH system. Fractions containing 2 were subjected to C18 reverse phase silica gel column chromatography eluted by CH3OH-H2O system, yielding 14 mg pure 2. Fractions containing 3 and 1 were subjected to silica gel column chromatography eluted with a CH2Cl2-CH3OH system to separate 3 and 1. Fractions containing 3 were collected, concentrated, dissolved in CH3OH, loaded on Sephadex LH-20 column, and eluted with CH3OH. 12 mg pure 3 was obtained. Fractions containing 1 were purified by semipreparative HPLC to yield 8 mg pure 1.

Purification of compounds 4 and 5 from S. lividans SBT5l3B4ΔmrqO4

Fermented culture of S. lividans SBT5/3B4ΔmrqO4 (10 L) were extracted with acetone. The resulting crude extract were concentrated by evaporation, suspended in 5 L water, and adjusted to pH 7.0. The water phase was extracted with ethyl acetate, and the resulting organic phase was dried by evaporation. The resultant residue was re-dissolved and subjected to silica gel column chromatography eluted with a petroleum ether-EtOAc-CH3OH system. Fractions containing compound 4 were subjected to a silica gel column chromatography eluted by a petroleum ether-CH2Cl2 system, yielding 12 mg compound 4. Fractions containing compound 5 were subjected to a C18 reverse phase silica gel column chromatography eluted by a CH3OH-H2O system to give fractions containing compound 5. After a silica gel column chromatography eluted by a petroleum ether-CH2Cl2 system, 12 mg pure compound 5 was obtained.

Purification of compound 6 from S. lividans SBT5l3B4ΔmrqO7

Fermented culture of S. lividans SBT5/3B4ΔmrqO7 (5 L) was extracted with acetone and the crude extract was concentrated, suspended in 3 L water, and adjusted to pH 7.0. The water phase was extracted with ethyl acetate, the organic phase was collected, dried by evaporation. The resultant residues were re-dissolved and subjected to silica gel column chromatography eluted with a CH2Cl2-CH3OH elution system. Compound 6 was eluted from the silica gel when the CH2Cl2:CH3OH ratio was of 100: 3. Fractions containing compound 6 were combined, dried by evaporation, and further purified by semipreparative chromatograghy at a flow rate of 2 mL/min on an Agilent ZORBAXSB-C18 column (5 μm, 9.4 mm × 250 mm) eluted by 75% MeOH for 25 min. The eluent should be evaporated as soon as possible since compound 6 was unstable in solution. 12 mg compound 6 was obtained.

In vivo conversion of 6 to 1 by feeding 6 to mutants during fermentation

S. lividans strains SBT5/3B4Δpks, SBT5/3B4ΔmrqE, SBT5/3B4ΔmrqE, SBT5/3B4ΔmrqG, SBT5/3B4ΔmrqM, SBT5/3B4ΔmrqO3, SBT5/3B4ΔmrqO4, SBT5/3B4ΔmrqO5 and the wild type SBT5/3B4 were fermented on No18 agar medium (in 9-cm Petri dishes) at 30°C Each dish was flooded with 30 mg compound dissolved in 1 mL methanol : water (1:1) after 3 days of incubation, and incubated for a further 3 days. The culture in each plate was extracted by ethyl acetate, dried by evaporation, dissolved in 0.5 mL methanol, and 20 μL sample was subjected to HPLC detection.

Feeding 4 to the Δpks mutant during fermentation

S. lividans strains SBT5/3B4Δpks and the wild type SBT5/3B4 were fermented in 50 mL No18 liquid medium (in 250 mL flask) at 30°C, 220 rpm. Each flask was added with 10 mg compound 4 dissolved in 1.5 mL methanol:CH2Cl2 (2:1) after 3 days of incubation, and incubated for a further 3 days. The culture in each flask was extracted by ethyl acetate, dried by evaporation, dissolved in 1 mL methanol, and 20 μL sample was subjected to HPLC detection.

Expression, purification and analysis of MrqO3, MrqO6 and MrqO7

mrqO3, mrqO6, and mrqO7 were amplified from BAC 3B4 using primer pairs proO3-F/proO3-R, proO6-F/proO6-R, and proO7-F/proO7-R, inserted into pET28a at either Ndel-Xhol or Ndel-Hindlll sites, so as to allow the expression of recombinant proteins with an N-terminal His(6) tag. MrqO6 and MrqO7 were produced in E. coli BL21(DE3) while MrqO3 was produced in E. coli strain BL21(DE3)/pGro7. The three proteins were expressed and purified in a similar manner. E. coli cultures (200 mL, in LB with kanamycin 50 μg/mL) were grown at 37°C to OD 600 ~0.6, induced by addition of 0.4 mM lPTG and followed by overnight growth (~18 h) at 16°C. Following incubation, cultures were harvested and the cell pellets were resuspended in 30 mL 50 mM Tris-HCl buffer (50 mM Tris, 150 mM NaCl, pH 7.5, glycerol 10%). Cells were lysed by repeated sonication, the lysate was centrifuged at 12,000 rpm for 25 min, and the resulting supernatant was added to a gravity-flow column containing 1 mL Ni2+ metal-chelate affinity resin (Bio-Rad). For the purification of MrqO3, the resin was washed with 10 mL 50 mM Tris-HCl supplemented with 30 mM imidazole to remove unspecifically bound protein. For MrqO6, the resin was washed with 10 mL 50 mM Tris-HCl supplemented with 40 mM imidazole. For MrqO7, the resin was washed with 10 mL 50 mM Tris-HCl supplemented with 50 mM imidazole. Then 5 mL 50 mM Tris-HCl supplemented with 500 mM imidazole were applied to the column to elute His-tagged proteins. PD-10 desalting columns were used to remove imidazole in the protein eluents. The desalted eluents were applied to Amicon Ultra-15 millipores (10 kDa, Merck) to concentrate the proteins. The concentrations of the proteins were determined by microplate reader using the Protein Bradford Assay Kit (Tiangen Biotech Company). The purified proteins were used for the in vitro enzyme activity assays.

Determination of the catalytic order of MrqO6 and MrqO7

Three sequential-reactions were conducted to determine the catalytic order of MrqO6 and MrqO7. These reactions were set up to 250 μL with 246 μM compound 6 and 1 mM NADPH and run in a two-step manner with the first step incubated for 45 min at 30°C the protein removed by Amicon Ultra-15 millipore (10 kD, Merck), and followed by the second step incubated for 1 h at 30°C In sequential-reaction I, MrqO7 (2 μM) was added for the first step reaction, and MrqO6 and MrqO3 (2 μM each) were added for the second step reaction. In sequential-reaction II, MrqO6 (2 μM) was added for the first step reaction, and MrqO7 and MrqO3 (2 μM each) were added for the second step reaction. In sequential-reaction III, no enzyme was added for the first step, and three enzymes MrqO7, MrqO6 and MrqO3 (2 μM each) were added for the second step reaction. After incubation, each reaction mixture was extracted with CH2Cl2, the organic phase was collected, air dried, dissolved with 20 μL MeOH, and detected by HPLC at 280 nm.

Detection of H2O2 in the in vitro conversion of 6 to 1 catalyzed by MrqO7, MrqO6 and MrqO3

H2O2 was detected using the Fluorimetric Hydrogen Peroxide Assay Kit (Catalog Number MAK166, Sigma) according to the product instruction. A liner standard curve was generated in the range of 0–10 μM H2O2 and R2 of the standard curve is 0.99776. The in vitro enzymatic assays were performed with 2 μM each enzyme, 1 mM NADPH, and 360 μM 6. The assay mixtures were incubated at 30°C for 20 min, then three samples of 10 μL, 20 μL and 30 μL were taken for the detection of H2O2. Meanwhile, the assay mixtures were extracted by CH2Cl2. The resultant organic phases were dried and resolved in 20 μL methanol for HPLC analysis, confirming the conversion of 6 to 1 in the in vitro assays.

QUANTIFICATION AND STATISTICAL ANALYSIS

See individual sections above and legends of pictures for details on the statistics used for analysis.

DATA AND SOFTWARE AVAILABILITY

Data Resources

The mrq gene cluster reported in this paper has been deposited in the NCBI under the accession number GenBank: KX817190.

Supplementary Material

supplement

Highlights.

  • Identification of the entire murayaquinone biosynthetic gene cluster

  • Anthrene type model compounds were identified from biosynthetic mutants

  • A linear 1(4H)-anthracenone epoxide is key intermediate for murayaquinone formation

  • Three oxidoreductases catalyze skeleton rearrangement of the tricyclic intermediate

Acknowledgments

We thank Songwang Hou, Tianshen Tao, and Colin P. Smith for gifts of bacterial strains. This work was supported by NSF (31370134, 31425001, 31300033, and 21661140002) of China, and the MOST (2012CB721000CB722010AA10A201; Chinese-Australia Joint Grant: 2016YFE0101000), the US National Institutes of Health (grant R01 GM 105977), and the Science and Technology Commission of Shanghai Municipality (15JC1400401).

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Supplemental Information

Document S1. Figures S1–S4, Tables S1–S3, and NMR spectra.

Author contributions

M.T., S.L, M.J., and Z.D designed research; G.G., X.L., M.X., Y.W., F.Z., L.X., J.L., and Q.L. performed research; M.T., S.L., M.J., J.R., Q.K., H-Y.O., and Y.W. analyzed data; M.T., S.L., M.J., and J.R., wrote paper.

References

  1. Abdelfattah MS. Mansoquinone: isolation and structure elucidation of new antibacterial aromatic polyketides from terrestrial Streptomyces Sp. Eg5. Nat Prod Res. 2009;23:212–218. doi: 10.1080/14786410801961550. [DOI] [PubMed] [Google Scholar]
  2. Bai T, Yu Y, Xu Z, Tao M. Construction of Streptomyces lividans SBT5 as an efficient heterologous expression host. J Huazhong Agric Univ. 2014;33:1–6. [Google Scholar]
  3. Begley TP, Kinsland C, Strauss E. The biosynthesis of coenzyme A in bacteria. Vitam Horm. 2001;61:157–171. doi: 10.1016/s0083-6729(01)61005-7. [DOI] [PubMed] [Google Scholar]
  4. Bisang C, Long PF, Corte J, Westcott J, Crosby J, Matharu A-L, Cox RJ, Simpson TJ, Staunton J, Leadlay PF. A chain initiation factor common to both modular and aromatic polyketide synthases. Nature. 1999;401:502–505. doi: 10.1038/46829. [DOI] [PubMed] [Google Scholar]
  5. Brockmann H, Meyer E, Schrempp K, Einers F, Reschke T. Resistomycin. IV. The constitution of resistomycins. Chem Ber. 1969;102:1224–1246. doi: 10.1002/cber.19691020415. [DOI] [PubMed] [Google Scholar]
  6. Brockmann H, Reschke T. The composition of resistomycin. Tetrahedron Lett. 1968 May;:3167–3170. doi: 10.1016/s0040-4039(00)89579-9. [DOI] [PubMed] [Google Scholar]
  7. Canham P, Vining LC. Pattern of acetate incorporation into the aglycone of chartreusin. Evidence from 13C nuclear magnetic resonance studies for a single-chain polyketide intermediate. J Chem Soc Chem Commun. 1976;169:319–320. [Google Scholar]
  8. Canham PL, Vining LC, Mclnnes AG, Walter JA, Wright JLC. Use of 13C in biosynthetic studies. Incorporation of 13C-labeled acetate into chartreusin by Streptomyces chartreusis. Can J Chem. 1977;55:2450–2457. [Google Scholar]
  9. Chen Y, Fan K, He Y, Xu X, Peng Y, Yu T, Jia C, Yang K. Characterization of JadH as an FAD- and NAD(P)H-dependent bifunctional hydroxylase/dehydrase in jadomycin biosynthesis. ChemBioChem. 2010;11:1055–1060. doi: 10.1002/cbic.201000178. [DOI] [PubMed] [Google Scholar]
  10. Chen YH, Wang CC, Greenwell L, Rix U, Hoffmeister D, Vining LC, Rohr J, Yang KQ. Functional analyses of oxygenases in jadomycin biosynthesis and identification of JadH as a bifunctional oxygenase/dehydrase. J Biol Chem. 2005;280:22508–22514. doi: 10.1074/jbc.M414229200. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Chu M, Mierzwa R, Truumees I, King A, Patel M, Berrie R, Hart A, Butkiewicz N, DasMahapatra B, Chan TM. Structure of Sch 68631: a new hepatitis C virus proteinase inhibitor from Streptomyces sp. Tetrahedron Lett. 1996;37:7229–7232. [Google Scholar]
  12. Datsenko KA, Wanner BL. One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc Natl Acad Sci U S A. 2000;97:6640–6645. doi: 10.1073/pnas.120163297. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Ewersmeyer-Wenk B, Zahner H, Krone B, Zeeck A. Metabolic products of microorganisms. 207. Haloquinone, a new antibiotic active against halobacteria. I. Isolation, characterization and biological properties. J Antibiot (Tokyo) 1981;34:1531–1537. doi: 10.7164/antibiotics.34.1531. [DOI] [PubMed] [Google Scholar]
  14. Farnet CM, Mcalpine JB, Bachmann BO, Staffa A, Zazopoulos E. System, knowledge repository and computer-readable medium for identifying a secondary metabolite from a microorganism. US20080010025[P] US. 2008
  15. Flett F, Mersinias V, Smith CP. High efficiency intergeneric conjugal transfer of plasmid DNA from Escherichia coli to methyl DNA-restricting streptomycetes. FEMS Microbiol Lett. 1997;155:223–229. doi: 10.1111/j.1574-6968.1997.tb13882.x. [DOI] [PubMed] [Google Scholar]
  16. Fritzsche K, Ishida K, Hertweck C. Orchestration of discoid polyketide cyclization in the resistomycin pathway. J Am Chem Soc. 2008;130:8307–8316. doi: 10.1021/ja800251m. [DOI] [PubMed] [Google Scholar]
  17. Fu H, Ebertkhosla S, Hopwood DA, Khosla C. Engineered biosynthesis of novel polyketides: dissection of the catalytic specificity of the act ketoreductase. J Am Chem Soc. 1994;116:4166–4170. [Google Scholar]
  18. Garrido LM, Lombó F, Baig I, Nur-e-Alam M, Furlan RL, Borda CC, Braña A, Méndez C, Salas JA, Rohr J. Insights in the glycosylation steps during biosynthesis of the antitumor anthracycline cosmomycin: characterization of two glycosyltransferase genes. Appl Microbiol Biotechnol. 2006;73:122–131. doi: 10.1007/s00253-006-0453-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Gould SJ, Melville CR, Chen J. The biosynthesis of murayaquinone, a rearranged polyketide. Tetrahedron. 1997;53:4561–4568. [Google Scholar]
  20. Gulick AM. Conformational dynamics in the Acyl-CoA synthetases, adenylation domains of non-ribosomal peptide synthetases, and firefly luciferase. ACS Chem Biol. 2009;4:811–827. doi: 10.1021/cb900156h. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Gust B, Challis GL, Fowler K, Kieser T, Chater KF. PCR-targeted Streptomyces gene replacement identifies a protein domain needed for biosynthesis of the sesquiterpene soil odor geosmin. Proc Natl Acad Sci U S A. 2003;100:1541–1546. doi: 10.1073/pnas.0337542100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Hautala A, Torkkell S, Raty K, Kunnari T, Kantola J, Mantsala P, Hakala J, Ylihonko K. Studies on a second and third ring cyclization in anthracycline biosynthesis. J Antibiot (Tokyo) 2003;56:143–153. doi: 10.7164/antibiotics.56.143. [DOI] [PubMed] [Google Scholar]
  23. Hertweck C, Luzhetskyy A, Rebets Y, Bechthold A. Type II polyketide synthases: gaining a deeper insight into enzymatic teamwork. Nat Prod Rep. 2007;24:162–190. doi: 10.1039/b507395m. [DOI] [PubMed] [Google Scholar]
  24. Ishida K, Maksimenka K, Fritzsche K, Scherlach K, Bringmann G, Hertweck C. The boat-shaped polyketide resistoflavin results from re-facial central hydroxylation of the discoid metabolite resistomycin. J Am Chem Soc. 2006;128:14619–14624. doi: 10.1021/ja064550u. [DOI] [PubMed] [Google Scholar]
  25. Jackson DR, Yu X, Wang G, Patel AB, Calveras J, Barajas JF, Sasaki E, Metsa-Ketela M, Liu HW, Rohr J, Tsai SC. Insights into complex oxidation during BE-7585A biosynthesis: structural determination and analysis of the polyketide monooxygenase BexE. ACS Chem Biol. 2016;11:1137–1147. doi: 10.1021/acschembio.5b00913. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Kantola J, Kunnari T, Kantola J, Mantsala P, Hakala J, Ylihonko K. Elucidation of anthracyclinone biosynthesis by stepwise cloning of genes for anthracyclines from three different Streptomyces spp. Microbiology. 2000;146:155–163. doi: 10.1099/00221287-146-1-155. [DOI] [PubMed] [Google Scholar]
  27. Kieser T, Bibb MJ, Buttner MJ, Chater KF, Hopwood DA. Practical Streptomyces Genetics. The John Innes Foundation; 2000. [Google Scholar]
  28. Koskiniemi H, Metsä-Ketelä M, Dobritzsch D, Kallio P, Korhonen H, Mäntsälä P, Schneider G, Niemi J. Crystal structures of two aromatic hydroxylases involved in the early tailoring steps of angucycline biosynthesis. J Mol Biol. 2007;372:633–648. doi: 10.1016/j.jmb.2007.06.087. [DOI] [PubMed] [Google Scholar]
  29. Lario PI, Sampson N, Vrielink A. Sub-atomic resolution crystal structure of cholesterol oxidase: what atomic resolution crystallography reveals about enzyme mechanism and the role of the FAD cofactor in redox activity. J Mol Biol. 2003;326:1635–1650. doi: 10.1016/s0022-2836(03)00054-8. [DOI] [PubMed] [Google Scholar]
  30. Luo M, Wing RA. An improved method for plant BAC library construction. Methods Mol Biol. 2003;236:3–20. doi: 10.1385/1-59259-413-1:3. [DOI] [PubMed] [Google Scholar]
  31. Maier S, Heitzler T, Asmus K, Brotz E, Hardter U, Hesselbach K, Paululat T, Bechthold A. Functional characterization of different ORFs including luciferase-like monooxygenase genes from the mensacarcin gene cluster. ChemBioChem. 2015;16:1175–1182. doi: 10.1002/cbic.201500048. [DOI] [PubMed] [Google Scholar]
  32. Maier S, Pfluger T, Loesgen S, Asmus K, Brotz E, Paululat T, Zeeck A, Andrade S, Bechthold A. Insights into the bioactivity of mensacarcin and epoxide formation by MsnO8. ChemBioChem. 2014;15:749–756. doi: 10.1002/cbic.201300704. [DOI] [PubMed] [Google Scholar]
  33. McDaniel R, Ebert-Khosla S, Fu H, Hopwood DA, Khosla C. Engineered biosynthesis of novel polyketides: influence of a downstream enzyme on the catalytic specificity of a minimal aromatic polyketide synthase. Proc Natl Acad Sci U S A. 1994;91:11542–11546. doi: 10.1073/pnas.91.24.11542. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Marchler-Bauer A, Bo Y, Han L, He J, Lanczycki CJ, Lu S, Chitsaz F, Derbyshire MK, Geer RC, Gonzales NR, Gwadz M, Hurwitz DI, Lu F, Marchler GH, Song JS, Thanki N, Wang Z, Yamashita RA, Zhang D, Zheng C, Geer LY, Bryant SH, Gwadz M. CDD/SPARCLE: functional classification of proteins via subfamily domain architectures. Nucleic Acids Res. 2016:gkw1129. doi: 10.1093/nar/gkw1129. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Melville CR, Gould SJ. Murayalactone, a dibenzo-α-pyrone from Streptomyces murayamaensis. J Nat Prod. 1994;57:597–601. doi: 10.1021/np50107a005. [DOI] [PubMed] [Google Scholar]
  36. Murooka Y, Yamashita M. Genetic and protein engineering of diagnostic enzymes, cholesterol oxidase and xylitol oxidase. J Biosci Bioeng. 2001;91:433–441. doi: 10.1263/jbb.91.433. [DOI] [PubMed] [Google Scholar]
  37. Pickens LB, Kim W, Wang P, Zhou H, Watanabe K, Gomi S, Tang Y. Biochemical analysis of the biosynthetic pathway of an anticancer tetracycline SF2575. J Am Chem Soc. 2009;131:17677–17689. doi: 10.1021/ja907852c. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Polonsky J, Lederer E. Piloquinone: a new phenanthrene-o-quinone isolated from the mycelium of Streptomyces pilosus. Nature. 1963;199:285–286. doi: 10.1038/199285a0. [DOI] [PubMed] [Google Scholar]
  39. Rix U, Wang C, Chen Y, Lipata FM, Remsing Rix LL, Greenwell LM, Vining LC, Yang K, Rohr J. The oxidative ring cleavage in jadomycin biosynthesis: a multistep oxygenation cascade in a biosynthetic black box. ChemBioChem. 2005;6:838–845. doi: 10.1002/cbic.200400395. [DOI] [PubMed] [Google Scholar]
  40. Robbins T, Liu YC, Cane DE, Khosla C. Structure and mechanism of assembly line polyketide synthases. Curr Opin Struct Biol. 2016;41:10–18. doi: 10.1016/j.sbi.2016.05.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Sasaki E, Ogasawara Y, Liu HW. A biosynthetic pathway for BE-7585A, a 2-thiosugar-containing angucycline-type natural product. J Am Chem Soc. 2010;132:7405–7417. doi: 10.1021/ja1014037. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Sato Y, Kohnert R, Gould SJ. Application of long range 1H/13C heteronuclear correlation spectroscopy (LR HETCOSY) to structure elucidation: the structure of murayaquinone. Tetrahedron Lett. 1986;27:143–146. [Google Scholar]
  43. Shen B, Hutchinson CR. Triple hydroxylation of tetracenomycin A2 to tetracenomycin C in Streptomyces glaucescens Overexpression of the tcmG gene in Streptomyces lividans and characterization of the tetracenomycin A2 oxygenase. J Biol Chem. 1994;269:30726–30733. [PubMed] [Google Scholar]
  44. Snapper SB, Melton RE, Mustafa S, Kieser T, Jacobs WR., Jr Isolation and characterization of efficient plasmid transformation mutants of Mycobacterium smegmatis. Mol Microbiol. 1990;4:1911–1919. doi: 10.1111/j.1365-2958.1990.tb02040.x. [DOI] [PubMed] [Google Scholar]
  45. Tibrewal N, Pahari P, Wang G, Kharel MK, Morris C, Downey T, Hou Y, Bugni TS, Rohr J. Baeyer-Villiger C-C bond cleavage reaction in gilvocarcin and jadomycin biosynthesis. J Am Chem Soc. 2012;134:18181–18184. doi: 10.1021/ja3081154. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Trefzer A, Hoffmeister D, Kunzel E, Stockert S, Weitnauer G, Westrich L, Rix U, Fuchser J, Bindseil KU, Rohr J, Bechthold A. Function of glycosyltransferase genes involved in urdamycin A biosynthesis. Chem Biol. 2000;7:133–142. doi: 10.1016/s1074-5521(00)00079-x. [DOI] [PubMed] [Google Scholar]
  47. Walsh CT. Insights into the chemical logic and enzymatic machinery of NRPS assembly lines. Nat Prod Rep. 2016;33:127–135. doi: 10.1039/c5np00035a. [DOI] [PubMed] [Google Scholar]
  48. Xu M, Wang Y, Zhao Z, Gao G, Huang SX, Kang Q, He X, Lin S, Pang X, Deng Z, Tao M. Functional genome mining for metabolites encoded by large gene clusters through heterologous expression of a whole-genome bacterial artificial chromosome library in Streptomyces spp. Appl Environ Microbiol. 2016;82:5795–5805. doi: 10.1128/AEM.01383-16. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Xu Z, Jakobi K, Welzel K, Hertweck C. Biosynthesis of the antitumor agent chartreusin involves the oxidative rearrangement of an anthracyclic polyketide. Chem Biol. 2005;12:579–588. doi: 10.1016/j.chembiol.2005.04.017. [DOI] [PubMed] [Google Scholar]
  50. Yue QK, Kass IJ, Sampson NS, Vrielink A. Crystal structure determination of cholesterol oxidase from Streptomyces and structural characterization of key active site mutants. Biochemistry. 1999;38:4277–4286. doi: 10.1021/bi982497j. [DOI] [PubMed] [Google Scholar]
  51. Zhan J. Biosynthesis of bacterial aromatic polyketides. Curr Top Med Chem. 2009;9:1958–1610. doi: 10.2174/156802609789941906. [DOI] [PubMed] [Google Scholar]
  52. Zhang W, Watanabe K, Wang CC, Tang Y. Investigation of early tailoring reactions in the oxytetracycline biosynthetic pathway. J Biol Chem. 2007;282:25717–25725. doi: 10.1074/jbc.M703437200. [DOI] [PubMed] [Google Scholar]
  53. Zhu T, Cheng X, Liu Y, Deng Z, You D. Deciphering and engineering of the final step halogenase for improved chlortetracycline biosynthesis in industrial Streptomyces aureofaciens. Metab Eng. 2013;19:69–78. doi: 10.1016/j.ymben.2013.06.003. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

supplement
NIHMS889197-supplement.pdf (2.3MB, pdf)

RESOURCES