Biosynthetic Baeyer–Villiger Chemistry Enables Access to Two

Jul 17, 2018 - ... two anthracene scaffolds typical of aromatic polyketides, were isolated from a culture of the deep-sea-derived Streptomyces olivace...
0 downloads 0 Views 1MB Size
Article Cite This: J. Nat. Prod. 2018, 81, 1570−1577

pubs.acs.org/jnp

Biosynthetic Baeyer−Villiger Chemistry Enables Access to Two Anthracene Scaffolds from a Single Gene Cluster in Deep-SeaDerived Streptomyces olivaceus SCSIO T05 Chunyan Zhang,†,‡ Changli Sun,† Hongbo Huang,† Chun Gui,†,‡ Liyan Wang,§ Qinglian Li,† and Jianhua Ju*,†,‡

Downloaded via UNIV OF SUSSEX on July 27, 2018 at 07:05:52 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.



CAS Key Laboratory of Tropical Marine Bio-resources and Ecology, Guangdong Key Laboratory of Marine Materia Medica, RNAM Center for Marine Microbiology, South China Sea Institute of Oceanology, Chinese Academy of Sciences, 164 West Xingang Road, Guangzhou, 510301, People’s Republic of China ‡ University of Chinese Academy of Sciences, 19 Yuquan Road, Beijing, 100049, People’s Republic of China § College of Life Sciences and Oceanology, Shenzhen Key Laboratory of Marine Bioresource and Eco-environmental Science, Shenzhen University, 3688 Nanhai Avenue, Shenzhen, 518060, People’s Republic of China S Supporting Information *

ABSTRACT: Four known compounds, rishirilide B (1), rishirilide C (2), lupinacidin A (3), and galvaquinone B (4), representing two anthracene scaffolds typical of aromatic polyketides, were isolated from a culture of the deep-seaderived Streptomyces olivaceus SCSIO T05. From the S. olivaceus producer was cloned and sequenced the rsd biosynthetic gene cluster (BGC) that drives rishirilide biosynthesis. The structural gene rsdK2 inactivation and heterologous expression of the rsd BGC confirmed the single rsd BGC encodes construction of 1−4 and, thus, accounts for two anthracene scaffolds. Precursor incubation experiments with 13C-labeled acetate revealed that a Baeyer−Villiger-type rearrangement plays a central role in construction of 1−4. Two luciferase monooxygenase components, along with a reductase component, are presumably involved in the Baeyer−Villiger-type rearrangement reaction enabling access to the two anthracene scaffold variants. Engineering of the rsd BGC unveiled three SARP family transcriptional regulators, enhancing anthracene production. Inactivation of rsdR4, a MarR family transcriptional regulator, failed to impact production of 1−4, although production of 3 was slightly improved; most importantly rsdR4 inactivation led to the new adduct 6 in high titer. Notably, inactivation of rsdH, a putative amidohydrolase, substantially improved the overall titers of 1−4 by more than 4-fold.

B

production called for the use of Streptomyces albus or S. lividans 1326 as heterologous hosts bearing the rsl cluster (within cosmid cos4).5 The success of this strategy suggested the potential for characterization of rishirilide biosynthesis and subsequent access to related analogues. Beyond the 2012 bioinformatics efforts, no further studies to characterize the rsl cluster and explore its presence in other microbial producers have been reported. During our efforts to identify new antitumor and antiinfective natural aromatic polyketides from marine actinomycetes,6−8 Streptomyces olivaceus SCSIO T05, isolated from a deep Indian Ocean sediment, was found to produce rishirilide B (1) and three analogues (2−4). In this paper, we report (i) the isolation of four known compounds from the deep-seaderived S. olivaceus SCSIO T05, rishirilide B (1), rishirilide C (2), lupinacidin A (3), and galvaquinone B (4), representative

acterially derived aromatic polyketides, especially those featuring a polycyclic phenolic skeleton, constitute an important group of natural products with diverse biological activities.1 Rishirilides A (Figure S1) and B (1) are aromatic polyketides originally isolated from Streptomyces rishiriensis OFR-1056 in 1984.2 Both agents were found to possess antithrombolytic activity by virtue of their ability to inhibit α2macroglobulin with IC50 values of 100 and 35 μg/mL, respectively. In addition, rishirilide B (1) has recently been shown to inhibit glutathione S-transferase with an IC50 of 26.9 μM, revealing potential applications in the prevention and treatment of fibrinolytic accentuation-induced thrombosis.3 Analogues of rishirilide B have yet to be reported in the literature. Consequently, there has been great interest in the prospect of engineering the rishirilide biosynthetic gene cluster (BGC) so as to enhance access to related metabolites.4 Initially identified in 2012 in Streptomyces bottropensis, the production of rishirilides in S. bottropensis has been unstable; reliable rsl expression and subsequent production of rishirilides A and B © 2018 American Chemical Society and American Society of Pharmacognosy

Received: January 25, 2018 Published: July 17, 2018 1570

DOI: 10.1021/acs.jnatprod.8b00077 J. Nat. Prod. 2018, 81, 1570−1577

Journal of Natural Products

Article

of two anthracene scaffold variants (aromatic polyketidebased); (ii) identification and analysis of the entire rsd BGC as it is housed in S. olivaceus SCSIO T05; (iii) elucidation of the rsd BGC responsible for the biosynthesis of these two anthracene scaffold variants using gene inactivation and heterologous expression methodologies; (iv) a unique Baeyer−Villiger-type rearrangement involved in the rsdencoded biosynthesis as identified by means of precursor incubation experiments with 13C-labeled acetate; (v) bioinformatic analyses and targeted gene disruptions enabling identification and characterization of the two luciferase monooxygenase components, along with a reductase component, involved in the rsd-encoded Baeyer−Villiger rearrangement; (vi) characterization of six regulators involved in the biosynthesis of rishirilides and new analogue 6; and (vii) enhanced production of 1−4 by specific mutagenesis of rsdH, an amidohydrolase protein.

Figure 1. HMBC correlations identified for compounds 2 and 6.

reported to modulate epigenetic activities at 1.0 μM and to exert moderate toxicities against non-small-cell lung cancer cell lines Calu-3 and H2887.10 Until now, the BGCs for 3 and 4 have not been reported. Identification of a Single Gene Cluster Responsible for the Biosynthesis of the Full Set of Compounds 1−4. The wild-type producing strain S. olivaceus SCSIO T05 is a highly reliable producer of the rishirilides. Yet, the rishirilide BGC (rsd cluster) housed within the S. olivaceus SCSIO T05 genome has eluded characterization. This contrasts with the previously characterized rsl BGC from S. bottropensis, which produces rishirilides A and B but in a highly unpredictable fashion in the S. bottropensis microbe.5 Predicated on this observation we sought to identify the rsd biosynthetic gene cluster using whole genome scanning and annotation methods. Using a combination of PacBio RS and Illumina HiSeq 2500 technologies, we acquired the complete 8.46 Mb genome sequence. We applied online antiSMASH13 software to analyze the S. olivaceus SCSIO T05 genome, which revealed a type II PKS gene cluster designated as the rishirilide B gene cluster with an 80% similarity to the rsl BGC, containing 28 open reading frames and spanning 31 kb. The genetic organization of the putative rsd gene cluster is described in Figure 2, and the nucleotide sequences have been deposited in GenBank with accession number MF437311. Genes are color-coded on the basis of their proposed functions, as is summarized in Table 1. The rsd BGC shows high similarity to the reported rsl cluster; both clusters display the same organization of genes (Table 1 and Figure 2).5 Consequently, it appears that the rishirilide BGCs from different producers are likely conserved. For instance, RsdK1, RsdK2, and RsdK3 (from S. olivaceus) are respectively homologous to RslK1, RslK2, and RslK3 (from S. bottropensis) originally reported by Bechthold and co-workers; both sets of machinery likely participate in rishirilide chain extension reactions.5 To verify that the rsd cluster is responsible for the production of rishirilides, we inactivated rsdK2 (ketosynthase β) using PCR-targeting methods14 in the wild-type S. olivaceus SCSIO T05. Unexpectedly, not only were biosynthetic products 1 and 2 absent from ΔrsdK2 mutant fermentations but compounds 3 and 4 also were (Figure 3, trace ii). These four compounds differ from each other with respect to their side chain linkage points. The 3-methylbutyl moiety in 1−3 is attached to C-4, whereas in compound 4 the 4-methylpentanoyl is tethered to C-3, indicating that 1−4 share the same ancestral anthracene scaffold. Thus, it became apparent that the lone rsd BGC drives the production of rishirilide B (1), rishirilide C (2), lupinacidin A (3), and galvaquinone B (4), representing two anthracene scaffold variants. Heterologous Expression of the rsd Gene Cluster in S. lividans Yields 1 and 4. To further confirm that the rsd BGC accounts for two anthracene scaffold variants, we expressed the rsd BGC in the heterologous host S. lividans TK64. Cosmid 01-



RESULTS AND DISCUSSION Isolation of Compounds 1−4 from Marine-Derived S. olivaceus SCSIO T05. The strain SCSIO T05 was isolated from an Indian Ocean deep-sea-derived sediment and identified as S. olivaceus based on morphology and 16S rDNA sequence analysis. The strain yielded an array of secondary metabolites when fermented using modified RA medium and analyzed by HPLC-DAD-UV. Silica gel and semipreparative HPLC column chromatography of the 12 L fermentation extract yielded compounds 1−4 as major compounds. The structures of compounds 1, 3, and 4 were identified as rishirilide B (1),2 lupinacidin A (3),9 and galvaquinone B (4),10 respectively, by comparisons of their HRESIMS and 1H and 13C NMR spectroscopic data to previously reported data. The molecular formula of compound 2 was determined as C21H24O7 by its negative HRESIMS data, with one more hydroxy group than rishirilide B (1). The hydroxy group was placed at C-9 by comparisons of its 13C NMR chemical shifts of C-8, C-8a, C-9, C-9a, and C-10 with those in rishirilide B (1), which was also supported by the HMBC correlation from H-8 to C-9 (Figures S6−S9). Therefore, the structure of compound 2 was elucidated as shown and named rishirilide C (Figure 1). Compound 2 was identified as a known compound (CAS no. 1217602-55-5) in SciFinder, although no literature citations were noted.11 Compounds 3 and 4 are anthraquinones, a class of polyketides extensively studied and well-known for their diverse biological activities.12 Lupinacidin A (3) inhibits, in a dose-dependent fashion, in vitro invasion of colon 26-L5 cells with an IC50 of 0.07 μg/mL.9 Alternatively, galvaquinone B (4) has been 1571

DOI: 10.1021/acs.jnatprod.8b00077 J. Nat. Prod. 2018, 81, 1570−1577

Journal of Natural Products

Article

Figure 2. (A) Genetic organization of the rsd gene cluster from S. olivaceus SCSIO T05; cosmids harboring the entire candidate gene cluster are indicated by solid lines. (B) Genetic organization proposed by Bechthold and co-workers for the rsl gene cluster from Streptomyces bottropensis.6

Table 1. Deduced Function of ORFs in the rsd Biosynthetic Gene Cluster ORF

sizea

proposed function

ID/SIb

protein homologue and origin

Orf(−3) Orf(−2) Orf(−1) RsdC1 RsdK1 RsdK2 RsdK3 RsdA RsdK4 RsdT1 RsdT2 RsdT3 RsdO1 RsdO2 RsdP RsdR1 RsdC2 RsdO3 RsdO4 RsdO5 RsdC3 RsdR2 RsdR3 RsdO6 RsdR4 RsdT4 RsdO7 RsdO8 RsdO9 RsdO10 RsdH Orf(+1) Orf(+2) Orf(+3)

433 547 413 315 89 408 422 360 346 321 249 311 354 168 375 270 301 238 100 361 164 273 1080 347 153 512 306 325 558 283 427 179 431 237

unknown function major facilitator superfamily protein transcriptional regulatory protein aromatase acyl carrier protein ketosynthase (β) ketosynthase (α) acyl transferase 3-oxoacyl-ACP synthase III ABC-transporter (substrate binding) ABC-transporter (ATP-binding) ABC-transporter transmembrane luciferase-like monooxygenase flavin reductase phosphotransferase SARP family regulator Second ring cyclase 3-oxoacyl-ACP reductase Anthrone monooxygenase NADH:flavin oxidoreductase cyclase SARP regulatory protein LAL-family regulator luciferase-like monooxygenase MarR family transcriptional regulator drug resistance transporter putative NADPH quinone reductase putative NADPH:quinone oxidoreductase FAD-dependent oxidoreductase C9-keto reductase amidohydrolase cupin citrate/H+ symporter transcriptional regulator

82/89 92/95 74/83 83/87 85/94 89/92 93/95 78/85 89/93 86/93 90/93 87/89 94/97 87/92 85/90 79/83 88/91 91/95 89/96 84/90 85/87 87/89 85/89 93/96 91/96 89/93 87/90 88/93 88/92 91/96 81/86 90/92 90/96 96/99

hypothetical protein (AKQ13284.1); Streptomyces canus major facilitator superfamily MFS_1 (AEN13008.1); Streptomyces sp. SirexAA-E transcriptional regulatory protein (EDY42647.2); Streptomyces sp. SPB74 RslC1(AHL46707.1); Streptomyces bottropensis RslK1(AHL46708.1); Streptomyces bottropensis RslK2(AHL46709.1); Streptomyces bottropensis RslK3(AHL46710.1); Streptomyces bottropensis RslA(AHL46711.1); Streptomyces bottropensis RslK4(AHL46712.1); Streptomyces bottropensis RslT1(AHL46713.1); Streptomyces bottropensis RslT2(AHL46714.1); Streptomyces bottropensis RslT3(AHL46715.1); Streptomyces bottropensis RslO1(AHL46716.1); Streptomyces bottropensis RslO2(AHL46717.1); Streptomyces bottropensis RslP(AHL46718.1); Streptomyces bottropensis RslR1(AHL46719.1); Streptomyces bottropensis RslC2(AHL46720.1); Streptomyces bottropensis RslO3(AHL46721.1); Streptomyces bottropensis RslO4(AHL46722.1); Streptomyces bottropensis RslO5(AHL46723.1); Streptomyces bottropensis RslC3(AHL46724.1); Streptomyces bottropensis RslR2(AHL46725.1); Streptomyces bottropensis RslR3(AHL46726.1); Streptomyces bottropensis RslO6(AHL46727.1); Streptomyces bottropensis RslR4(AHL46728.1); Streptomyces bottropensis RslT4(AHL46729.1); Streptomyces bottropensis RslO7(AHL46730.1); Streptomyces bottropensis RslO8(AHL46731.1); Streptomyces bottropensis RslO9(AHL46732.1); Streptomyces bottropensis RslO10(AHL46733.1); Streptomyces bottropensis RslH(AHL46734.1); Streptomyces bottropensis cupin (WP_004923763.1); Streptomyces griseoflavus citrate/H+ symporter (EDY63061.2); Streptomyces pristinaespiralis ATCC 25486 transcriptional regulator (EDY63059.1); Streptomyces pristinaespiralis ATCC 25486

a

Amino acids. bIdentity/similarity.

strain fermentation, validating that the single rsd gene cluster drives the biosynthesis of the two anthracene scaffold variants. 13 C-Labeling Experiments Unveil an Oxidative Baeyer−Villiger-Type Rearrangement Reaction en Route to 1−4. We hypothesized that an oxidative rearrangement likely occurs during rishirilide and lupinacidin A biosynthesis. To study this putative oxidative rearrangement cascade, we carried out experiments with 13C-labeled acetate to investigate isotopic incorporation patterns. We introduced [1-13C], [2-13C], and [1,2-13C2] acetate to growing cultures of

10B, containing all 28 ORFs, was selected (Figure 2) for use and modified by replacing the kanamycin resistance gene within the SuperCos 1 vector with a pSET152AB-derived fragment.7 The resulting cosmid was then transferred into S. lividans TK64 by conjugation and stably integrated via attB/ attp-site-specific recombination, to yield the S. lividans TK64/ 01-10B strain.15 This engineered strain was then fermented using the previously employed conditions, and its extracts were analyzed by HPLC-DAD-UV. Products 1 and 4 were clearly observed (Figure 3, trace xi) in the extracts of the mutant 1572

DOI: 10.1021/acs.jnatprod.8b00077 J. Nat. Prod. 2018, 81, 1570−1577

Journal of Natural Products

Article

S. olivaceus SCSIO T05, and the resulting 13C-labeled products were purified and analyzed by 13C NMR spectroscopy. Fermentation doping with [1-13C] acetate gave rise to isotopically enriched 4, in which 13C incorporation at C-2, C-4, C-5, C-6, C-8, C-9a, C-10a, and C-12 was clearly evident upon analysis of the NMR data (Figures 4 and S35). Doping of the fermentation with [2-13C] acetate gave rise to 13C NMR signal enhancements at C-1, C-3, C-4a, C-5a, C-7, C-9, C-10, C-11, and C-13 (Figures 4 and S36). To determine the orientation of acetate units during assembly, [1,2-13C2] acetate was utilized. Notably, 13C−13C spin couplings in 4 were observed for C-1/C-2, C-3/C-4, C-4a/C-5, C-5a/C-6, C-7/C8, C-9/C-9a, C-10/C-10a, and C-12/C-13 pairing (Figures 4 and S37). Importantly, the acetate incorporation pattern observed for 4 proved typical for the anthracene scaffold. We envision that anthracene core biosynthesis probably starts with a 4-methylpentanoyl moiety and subsequent assemblage of eight acetate units catalyzed by a type II PKS. In comparing the 13 C-labeling/enrichment results obtained for 3 with those determined for 4, it is notable that 13C−13C spin couplings were observed for C-1/C-2, C-4a/C-5, C-5a/C-6, C-7/C-8, C9/C-9a, C10/C-10a, and C-4/C-12 (Figures 4 and S34). This suggests that a subsequent oxidative C−C bond cleavage event occurs between C-4 and C-4a followed by rearrangement and recyclization (Figure 4). It is noteworthy that similar oxidative C−C bond cleavage reactions of polyketide-derived tetracyclic substrates play key roles in the biosynthesis of mithramycin,16 gilvocarcin, and jadomycin.17 These data also suggest that an oxidative Baeyer−Villiger-type rearrangement serves as a biosynthetic branch point giving rise to the two different anthracene scaffold variants. Proposed Model for the Biosynthesis of 1−4 and Identification of Genes Putatively Involved in the Oxidative Baeyer−Villiger-Type Rearrangement. Assembly of the aromatic backbone of 1−4 (Figure 5) likely starts with the synthesis of ACP-bound intermediate 7.5 Labeling results observed in the 13C-acetate experiments suggest the assembly of eight acetate units and a subsequent decarboxylation step to generate intermediate 8 (Figure 5). This invokes catalysis by the minimal PKS (RsdK1, RsdK2, and

Figure 3. HPLC analysis of fermentation broths. (i) Wild-type S. olivaceus SCSIO T05; (ii) ΔrsdK2 mutant strain; (iii) Δorf(−1) mutant strain; (iv) Δorf(+3) mutant strain; (v) ΔrsdR1 mutant strain; (vi) ΔrsdR2 mutant strain; (vii) ΔrsdR3 mutant strain; (viii) ΔrsdR4 mutant strain; (ix) ΔrsdH mutant strain; (x) wild-type S. lividans TK64; (xi) S. lividans TK64/01-10B strain; (xii) ΔrsdO1 mutant strain; (xiii) ΔrsdO2 mutant strain; (xiv) ΔrsdO6 mutant strain. Extracts used for metabolite analyses contained the contents of both fermentation media and cells to ensure comprehensive capturing of metabolites generated for each strain. Compound X was identified as xiamycin, an indolosesquiterpene compound unrelated to the rsd pathway.

Figure 4. Suggested oxidative ring opening and follow-up reactions leading to compounds 1, 3, and 4 based on 13C-labeling experiments. 1573

DOI: 10.1021/acs.jnatprod.8b00077 J. Nat. Prod. 2018, 81, 1570−1577

Journal of Natural Products

Article

Figure 5. Proposed biosynthetic pathway for compounds 1−4. Compounds 7−11 are proposed intermediates.

an active site.23,24 Luciferases are the most extensively studied representatives of class C flavoprotein monooxygenases; these enzymes generally use a reduced flavin mononucleotide (FMN) as a coenzyme. The reduced FMN is likely generated by a reductase component within the system.25 Both RsdO1 and RsdO6 contain a single-domain eightstranded β/α barrel motif, analyzed by Bioinformatic Toolkit,26 suggesting that RsdO1 and RsdO6 might serve as the two monooxygenase components and that RsdO2 may function as the reductase responsible for generating reduced FMN. Production of 1−4 was blocked in the ΔrsdO1 mutant strain (Figure 3, trace xii). To confirm that the deletion of rsdO1 was the sole reason for the loss of 1−4, a 1065-bp fragment, the complete rsdO1 gene was reintroduced into the ΔrsdO1 mutant strain. Importantly, complementation of the ΔrsdO1 mutant strain led to restoration of 1, 3, and 4 biosynthesis (Figure S28, trace iii). The loss of 1−3 production by the ΔrsdO1 mutant strain suggests that RsdO1 may play a critical role as the α subunit needed for the monooxygenase system to accomplish O-atom insertion into the C4−C4a bond. In deleting the rsdO6 gene from the rsd cluster, we observed little change in the titers for 1−4 (Figure 3, trace xiv). This finding, using the ΔrsdO6 mutant strain, revealed that RsdO6 is most likely a β subunit devoid of active sites. Significantly, HPLC analyses of extracts from the ΔrsdO2 mutant strain indicated that loss of RsdO2 activity led to the accumulation of product 4 and a new shunt product, 5 (Figure 3, trace xiii). Compound 5 was identified as galvaquinone A (Figures S14 and S15) on the basis of comparisons of HRESIMS and 1H

RsdK3), RsdO10, and three cyclases (RsdC1, RsdC2, and RsdC3), with high homology to rsl-encoded enzymes RslK1, RslK2, RslK3, RslO10, RslC1, RslC2, and RslC3, respectively.5 RsdO10 encodes a C9-keto reductase catalyzing reduction of the 9-keto group.18 RsdO4 codes for an anthrone monooxygenase, with sequence similarity to AknX, responsible for converting emodinanthrone to emodin anthrone in the aklavinone pathway.19 We envision that 8 then serves as the substrate for the cryptic oxidative Baeyer−Villiger-type rearrangement. Following O-atom insertion into the C4−C4a bond, C−C bond cleavage takes place, followed by rearrangement and recyclization to render the two anthracene scaffold variants observed. A similar oxidative process has been proposed in BE-7585A biosynthesis.20 Sequence analysis (Figure S16) revealed that RsdO1 and RsdO6 encode proteins homologous to the luciferase-like monooxygenases (LLMs). Notably, these enzymes are highly homologous to RslO1 and RslO6 from the S. bottropensis system, but no relevant enzymatic functions were suggested.5 Historically important is that the LLM Hgc3 is required for Oatom insertion into the C5−C6 bond of hygrocin; this realization showcases the role of LLMs as Baeyer−Villiger monooxygenases (BVMOs).21 Notably, bacterial luciferases have been classified as type II BVMOs since 1997.22 Luciferases from bacterial species usually consist of two subunits, α and β, with molecular masses of approximately 40 and 35 kDa, respectively.23 The two subunits are homologous in sequence and fold into a single-domain eightstranded β/α barrel motif, which exhibits a TIM (triosephosphate isomerase)-barrel fold. But only the α subunit contains 1574

DOI: 10.1021/acs.jnatprod.8b00077 J. Nat. Prod. 2018, 81, 1570−1577

Journal of Natural Products

Article

and 13C NMR spectroscopic data with previously reported data.12 Both nonrearranged compounds 4 and 5 were produced by the ΔrsdO2 mutant strain, suggesting that flavin reductase RsdO2 is necessary for FMN reduction. These results suggest that two monooxygenase components, RsdO1 and RsdO6, along with a reductase component, RsdO2, catalyze cryptic O-atom insertion into the C4−C4a bond of putative intermediate 8 (Figure 5), followed by C−C bond cleavage, rearrangement, and recyclization, to afford the key intermediate 11. Analyses of rsdO5, rsdO7, and rsdO8 by BLASTP failed to aid us in assigning functions to their products. However, RsdO7 and rsdO8 were found to have sequence similarities to NADPH-dependent oxidoreductases and, thus, may be involved in redox chemistries leading to compounds 1−4. Characterization of the Regulatory Network for 1−4 Production and Discovery of RsdH, a Gene Enabling Improved Titers. To investigate the regulation of 1−4 production, we inactivated the six regulatory genes in this gene cluster. RsdR1 and RsdR2 are highly similar to RslR1 and RslR2, SARP family transcriptional regulators, which are generally reported to be pathway-specific activators in many biosynthetic gene clusters.5,27 RsdR3 is highly homologous to RslR3, a LAL family regulator that can directly and indirectly control the expression of multiple genes and play a positive role in the production of secondary metabolites.28 Not surprisingly, separate deletions of rsdR1, rsdR2, and rsdR3 abolished production of 1−4 in those mutant strains bearing the regulatory deficiencies (Figure 3, traces v−vii). Complementation of the ΔrsdR1 mutant strain led to restoration of 1, 2, and 4 biosynthesis. Complementation experiments focusing on rsdR2 and rsdR3 both led to restored production of 1−4 (Figure S28, traces v, vii, and ix). These data revealed rsdR1−R3 as critical positive regulators in rishirilide biosynthesis. Additionally, we had noted rsdR4 as a potentially essential regulatory element in the rsd BGC. RsdR4 is highly homologous to RslR4,5 a MarR family transcriptional regulator; such homologues have a variety of bioactivities, including catabolism of aromatic compounds. By binding to the intergenic region between the marR gene and a divergently oriented gene, MarR homologues can positively or negatively regulate the transcription of both genes.29 In deleting rsdR4 from the rsd cluster we observed a slight increase in 3 but otherwise little change to 1−4 production. Striking in the ΔrsdR4 mutant strain however was the presence of new analogue 6 (Figure 3, trace viii), whose yield is extremely low in the wild-type strain. In addition to regulatory genes rsdR1− R4, orf(−1) and orf(+3) were identified as excellent regulatory candidates since they display similarities to established Streptomycete transcriptional regulators. Inactivation of orf(−1) exerted no influence upon 1−4 biosynthesis (Figure 3, trace iii), revealing its unlikely involvement in 1−4 biosynthesis. Alternatively, inactivation of orf(+3) exerted little to no influence upon production of 1 and 4 but did appear to slightly impair 2 and 3 production (Figure 3, trace iv), thereby implying a very subtle regulatory role, possibly by virtue of transcriptional regulation (Table 1). Finally, we also sought to inactivate rsdH in the rsd cluster inspired, in part, by BLAST program sequence analyses, showing that RsdH has 62% similarity to an amidohydrolase from Streptomyces sp. NRRL F-4335 (WP_052874117.1). As a putative member of this large metal-dependent hydrolase superfamily, we envisioned that RsdH might also serve as a

form of regulatory agent in setting 1−4 concentrations either inside or outside of the producing microbe. Interestingly, we found that inactivation of rsdH gave rise to a mutant strain (ΔrsdH) capable of significantly improved rishirilide production (Figure 3, trace ix). Importantly, the ΔrsdH mutant totally produced more than 4-fold as much 1−4 as the wild-type producer. In revisiting the results obtained with the ΔrsdR4 mutant, we sought to elucidate the structure of 6 (Figure 1). A large-scale (12 L) fermentation of the ΔrsdR4 mutant was followed by broth extraction, extract purification, and acquisition of a full set of 1D and 2D NMR data (Table S2, Figures S38−42). Compound 6 was isolated (titer = 8 mg from a 12 L fermentation) as a minor product with a molecular formula of C18H16O5 as determined by HRESIMS analyses, one C2H4 unit less than 3. The 1H and 13C NMR spectroscopic data for 6 were almost identical to those of 3, barring resonances representative of the lupinacidin A (3) 3-methylbutyl side chain. A propane moiety was attached at C-17, which was confirmed by HMBC correlations of H2-12 with C-3, C-4a, C13, and C-14, combined with COSY correlations of H2-12/H213/H3-14. Thus, the structure of 6, aptly named lupinacidin D and generated via rsdR4 inactivation, was determined (Figure 1). Compound 6 was also detected in the ΔrsdH mutant. During the biosynthesis of rishirilides, lupinacidin A, and galvaquinone B, RsdK4 is presumably responsible for extending the isobutyryl CoA unit to the 4-methyl-3-oxopentanoyl CoA unit (Figure 5). Compound 6 is a product of the gene cluster, using acetyl CoA as the starter unit. We envision that RsdR4 might contribute to the substrate selection of RsdK4.



CONCLUSIONS In summary, four compounds representing two anthracene scaffold variants, rishirilide B (1), rishirilide C (2), lupinacidin A (3), and galvaquinone B (4), were isolated from deep-seaderived S. olivaceus SCSIO T05. We have identified and characterized the rsd gene cluster by genome sequencing and bioinformatics analysis and determined that this lone BGC accounts for both anthracene scaffold variants. Gene inactivation, heterologous expression, and 13C-labeling experiments played key roles in providing these insights. Additionally, a Baeyer−Villiger-type rearrangement chemistry has been identified, on the basis of 13C-labeling and gene disruption experiments, to play a key role in the biosynthesis of 1−3. In addition, inactivation of a MarR family transcriptional regulator gene (rsdR4) within the rsd BGC enabled production of a new rishirilide congener lupinacidin D (6), while inactivation of an amidohydrolase gene rsdH was found to significantly improve total titers of 1−4 by more than 4-fold. Taken together, these efforts provide an excellent springboard for the further study and development of 1−4 and related congeners via the application of rsd cluster engineering efforts.



EXPERIMENTAL SECTION

General Experimental Procedures. Optical rotations were determined with an MCP 500 (Anton Paar) polarimeter. UV spectra were recorded on a UV-2600 spectrometer (Shimadzu). IR spectra were obtained on an IR Affinity-1 spectrometer (Shimadzu). NMR spectra were acquired with an Avance 500 spectrometer (Bruker) at 500 MHz for 1H nucleus and 125 MHz for 13C nucleus and an Avance 700 spectrometer (Bruker) at 700 MHz for 1H nucleus and 175 MHz for 13C nucleus. Carbon signals and the residual proton signals attributable to CD3OD (δC 49.0 and δH 4.87), CDCl3 (δC 1575

DOI: 10.1021/acs.jnatprod.8b00077 J. Nat. Prod. 2018, 81, 1570−1577

Journal of Natural Products

Article

77.16 and δH 7.26), and DMSO-d6 (δC 39.52 and δH 2.50) were used for calibration. High-resolution mass data were determined using a Maxis quadrupole-time-of-flight mass spectrometer (Bruker). Column chromatography (CC) was performed using silica gel (100−200 mesh, Yantai Jiangyou Silica Gel Development Co., Ltd.). Reversedphase medium-pressure preparative liquid chromatography (RPMPLC) was carried out using a Cheetah MP200 (Agela Technologies) column filled with ODS (40−63 μm, YMC). Preparative HPLC was performed using an LC3000 solvent delivery module equipped with a smartline UV detector 2550 (Knauer) and a YMC-Pack ODS-A column (250 × 20 mm, 5 μm). DL2000 DNA marker (Takara Bio Inc.) and easyTaq DNA polymerase (TransGen Biotech Co., Ltd.) were used to carry out all PCR verification procedures (see Supporting Information). Bacterial Strains and Plasmids. Strain SCSIO T05 was isolated from a sediment sample obtained at a depth 4617 m from the Indian Ocean (S 1°25.536′, E 94°20.184′). It was identified as Streptomyces olivaceus SCSIO T05 on the basis of 16S rDNA gene sequence analysis by comparison with others in the GenBank database. The 16S rDNA gene sequence has been deposited in GenBank under accession number MF429815. E. coli XL 1-Blue MR (Stratagene) was used as the host strain for construction of the S. olivaceus SCSIO T05 genomic cosmid library. E. coli BW25113/pIJ790 was used as the host for Red/ET-mediated recombination.18 E. coli ET12567/pUZ8002 was used to transfer DNA into the Streptomyces. Plasmid pIJ773 was used to amplify the aac (3)IV apramycin resistance cassette. E. coli strains with plasmids were grown on agar or in liquid Luria−Bertani medium at 28 or 37 °C. SCSIO T05 wild-type and mutant strains were grown at 28 °C, on ISP-4 medium supplemented with 0.1% peptone, 0.05% yeast extract, and 3% sea salt. Media were supplemented with apramycin (50 μg/mL), chloramphenicol (25 μg/mL), or kanamycin (50 μg/mL). SuperCos1 (Stratagene) was used for the construction of the S. olivaceus SCSIO T05 genomic cosmid library. Production, Isolation, and Identification of Rishirilide B (1), Rishirilide C (2), Lupinacidin A (3), and Galvaquinone B (4). Strain SCSIO T05 was incubated on modified ISP-4 agar plates at 28 °C. The spores were inoculated into 50 mL of modified RA medium including 2% soluble starch, 1% glucose, 1% maltose, 1% maltose extract, 0.5% corn flour, 3% crude sea salt (from Guangdong Province Salt Industry Group, Guangdong, China), and 0.2% CaCO3, pH 7.2, and incubated at 28 °C on rotary shakers (200 rpm) for 2 days. The resultant seed cultures were transferred to 1 L Erlenmeyer flasks containing 150 mL of RA medium and grown under the same conditions for 8 days. About 12 L of the growth culture was centrifuged at 4000g for 10 min to generate the supernatant and mycelium. The supernatant was extracted with equal volumes of butanone (3×); the mycelium cake was extracted with 1 L of acetone three times. The two extracts were combined after HPLC analysis and subjected to silica gel CC using gradient elution with a CHCl3/ MeOH mixture (100:0, 98:2, 96:4, 94:6, 92:8, 90:10, 85:15, 80:20, 70:30, and 50:50) to give 10 fractions (AFr.1−AFr.10). AFr.1 and AFr.2 were combined and subjected to silica gel CC again using gradient elution with a petroleum ether/EtOAc mixture (100:0, 95:5, 90:10, 85:15, 80:20, and 70:30) to give six fractions (BFr.1−BFr.6). BFr.2−BFr.6 were purified by preparative HPLC with an ODS column, eluting with 95% solvent B (A: H2O; B: CH3CN) over the course of 30 min at a flow rate of 2.5 mL/min (using detection at 254 nm), to afford lupinacidin A (50.8 mg) and galvaquinone B (21.6 mg) with retention times of 29.7 and 31.5 min, respectively. AFr.9 and AFr.10 were combined and subjected to silica gel CC again using gradient elution with a CHCl3/MeOH mixture (100:0, 95:5, 90:10, 85:15, 80:20, and 70:30) to give six fractions (CFr.1−CFr.6). CFr.4− CFr.6 were purified by preparative HPLC with an ODS column, eluted with a linear gradient of 10% to 90% solvent B (A: H2O/HOAc 100/0.1; B: CH3CN/HOAc 100/0.1) in over 18 min and then a linear gradient of 90% to 100% solvent B from 18 to 25 min at a flow rate of 2.5 mL/min (using detection at 254 nm), to afford rishirilides B (51.6 mg) and C (18.8 mg) with retention times of 19.1 and 21.6 min, respectively.

Rishirilide C (2): yellow solid; [α]22D +99.8 (c 1.34, EtOH); UV (EtOH) λmax (log ε) 223 (4.05) 269 (4.09) 302 (3.40) 395 (3.39) nm; IR (ATR) νmax 3422, 1699, 1684, 1601, 1576 cm−1; 1H and 13C NMR spectroscopic data, Table S1; (−)-HRESIMS m/z 387.1461 [M − H]− (calcd for C21H23O7, 387.1449). Lupinacidin D (6): orange solid; [α]25D +5.2 (c 1.00, MeOH); UV (MeOH) λmax (log ε) 212 (3.76) 236 (3.47) 254 (3.54) 276 (3.54) 334 (3.00) 444 (3.28) nm; IR (ATR) νmax 3356, 1647, 1628, 1558, 1541, 1506, 1456, 1016 cm−1; 1H and 13C NMR spectroscopic data, Table S2; (−)-HRESIMS m/z 311.0925 [M − H]− (calcd for C18H15O5, 311.0925). Whole Genome Scanning and Bioinformatics Analysis. The S. olivaceus SCSIO T05 genomic DNA used for scanning was isolated according to a slightly modified protocol.30 Whole genome scanning and annotation of S. olivaceus SCSIO T05 were accomplished using a combination of PacBio RS and Illumina HiSeq 2500 technologies at Shanghai Majorbio Biopharm Technology Co., Ltd. Secondary metabolite biosynthetic gene clusters were detected and analyzed using online antiSMASH software (http://antismash. secondarymetabolites.org/). ORFs were analyzed using online FramePlot 4.0beta software (http://nocardia.nih.go.jp/fp4/), and their functional predictions were accomplished with an online BLAST program (http://blast.ncbi.nlm.nih.gov/). The rsd gene cluster was deposited in GenBank under the accession number MF437311. Genomic Library Construction and Screening. The S. olivaceus SCSIO T05 genomic cosmid library was constructed using SuperCos 1 according to the manufacturer’s instructions. About 2500 clones were picked and placed into 96-well plates and stored at −80 °C. Three pairs of primers associated with the orfs rsdA, rsdO3, and rsdO8 (Table S3, Supporting Information) were designed and used to screen the genomic cosmid library using PCR methods. Construction of Gene Inactivation Mutants. Gene deletions were performed following the REDIRECT protocol. The apramycin resistance gene oriT/aac(3)IV fragment obtained by PCR (all primers used are listed in the Supporting Information) was used to replace the target genes in the cosmid 01-10B or 13-8E. The constructed mutant cosmids were introduced into nonmethylating E. coli ET12567/ pUZ8002 and then transferred into S. olivaceus SCSIO T05. Because the strain was only sensitive to kanamycin, exconjugants were grown on solid kanamycin containing ISP-4 medium to select for the chromosomal integration of the inactivation constructs. To ensure loss of the target gene from the chromosome, exconjugants were replica-plated once onto antibiotic-free ISP-4 plates. Single colonies were again replica-plated onto kanamycin-containing ISP-4 plates and antibiotic-free ISP-4 plates. Kanamycin-sensitive clones were evaluated by PCR to ensure the proper generation of desired mutant clones. Complementation of ΔrsdO1, ΔrsdR1, ΔrsdR2, and ΔrsdR3 Mutants. Using the genomic DNA of S. olivaceus SCSIO T05 as the template, a DNA fragment carrying the complete rsdO1, rsdR1, rsdR2, or rsdR3 gene was amplified by PCR (all primers used are listed in the Supporting Information). The PCR product was then digested by NdeI/SpeI. The NdeI/SpeI digested fragment was cloned into the integrative vector pL646AKE to give the complementation plasmid. The constructed complementation plasmid was introduced into nonmethylating E. coli ET12567/pUZ8002 and then transferred into ΔrsdO1, ΔrsdR1, ΔrsdR2, or ΔrsdR3 mutants, respectively. This resulted in kanamycin-resistant exconjugants. The correct exconjugants were confirmed by PCR analysis. Metabolite Analyses of Wild-Type, ΔrsdK2, ΔrsdO1, ΔrsdO2, ΔrsdO6, Δorf(−1), ΔrsdR1, ΔrsdR2, ΔrsdR3, ΔrsdR4, Δorf(+3), and ΔrsdH Mutants. Strains were grown on ISP-4 agar medium at 28 °C to obtain sporulation and then used to inoculate 250 mL flasks containing 50 mL of modified RA medium. Fermentations were carried out at 28 °C on rotary shakers (200 rpm) for 7 days. Each of the fermentations, containing the contents of both media and cells, was extracted with 100 mL of butanone, and then the solvents were removed under reduced pressure. The extracts were dissolved into 1 mL of MeOH and then subjected to HPLC analysis with UV detection at 254 nm. 1576

DOI: 10.1021/acs.jnatprod.8b00077 J. Nat. Prod. 2018, 81, 1570−1577

Journal of Natural Products

Article

Large-Scale Fermentation and Isolation of Compounds 5 and 6. To obtain compound 5, large-scale fermentation (12 L) of the ΔrsdO2 mutant was performed. To obtain new analogue 6, large-scale fermentation (12 L) of the ΔrsdR4 mutant was carried out. Both fermentation procedures were the same as that employed for the wildtype producer. The extract was subjected to silica gel CC using gradient elution with a CHCl3/MeOH mixture (100:0, 98:2, 96:4, 94:6, 92:8, 90:10, 85:15, 80:20, 70:30, and 50:50) to give 10 fractions (AFr.1−AFr.10). ΔrsdO2/AFr.1−AFr.3 containing compound 5 were purified by preparative HPLC with an ODS column, eluted with 95% solvent B (A: H2O; B: CH3CN) over a period of 30 min at a flow rate of 2.5 mL/min (using detection at 254 nm), to afford compound 5 (12 mg). ΔrsdR4/AFr.1 containing compound 6 was purified by preparative HPLC with an ODS column, eluted with 95% solvent B (A: H2O; B: CH3CN) over a period of 30 min at a flow rate of 2.5 mL/min (using detection at 254 nm), to afford compound 6 (8 mg).



(7) Zhang, Y.; Huang, H.; Chen, Q.; Luo, M.; Sun, A.; Song, Y.; Ma, J.; Ju, J. Org. Lett. 2013, 15, 3254−3257. (8) Gui, C.; Mo, X.; Gu, Y. C.; Ju, J. Org. Lett. 2017, 19, 5617−5620. (9) Igarashi, Y.; Trujillo, M. E.; Martínez-Molina, E.; Yanase, S.; Miyanaga, S.; Obata, T.; Furumai, T. Bioorg. Med. Chem. Lett. 2007, 17, 3702−3705. (10) Hu, Y.; Martinez, E. D.; MacMillan, J. B. J. Nat. Prod. 2012, 75, 1759−1764. (11) http://www.cas.org/products/scifinder. (12) Griffiths, S.; Mesarich, C. H.; Saccomanno, B.; Vaisberg, A.; De Wit, P. J.; Cox, R.; Collemare, J. Proc. Natl. Acad. Sci. U. S. A. 2016, 113, 6851−6856. (13) Medema, M. H.; Blin, K.; Cimermancic, P.; de Jager, V.; Zakrzewski, P.; Fischbach, M. A.; Weber, T.; Takano, E.; Breitling, R. Nucleic Acids Res. 2011, 39, 339−346. (14) Gust, B.; Challis, G. L.; Fowler, K.; Kieser, T.; Chater, K. F. Proc. Natl. Acad. Sci. U. S. A. 2003, 100, 1541−1546. (15) Li, Q.; Song, Y.; Qin, X.; Zhang, X.; Sun, A.; Ju, J. J. Nat. Prod. 2015, 78, 944−948. (16) Prado, L.; Fernández, E.; Weiβbach, U.; Blanco, G.; Quirós, L. M.; Braña, A. F.; Méndez, C.; Rohr, J.; Salas, J. A. Chem. Biol. 1999, 6, 19−30. (17) Tibrewal, N.; Pahari, P.; Wang, G.; Kharel, M. K.; Morris, C.; Downey, T.; Rohr, J. J. Am. Chem. Soc. 2012, 134, 18181−18184. (18) Hertweck, C.; Luzhetskyy, A.; Rebets, Y.; Bechthold, A. Nat. Prod. Rep. 2007, 24, 162−190. (19) Chung, J. Y.; Fujii, I.; Harada, S.; Sankawa, U.; Ebizuka, Y. J. Bacteriol. 2002, 184, 6115−6122. (20) Sasaki, E.; Ogasawara, Y.; Liu, H. W. J. Am. Chem. Soc. 2010, 132, 7405−7417. (21) Li, S.; Wang, H.; Li, Y.; Deng, J.; Lu, C.; Shen, Y.; Shen, Y. ChemBioChem 2014, 15, 94−102. (22) Villa, R.; Willetts, A. J. Mol. Catal. B: Enzym. 1997, 2, 193−197. (23) Fisher, A. J.; Raushel, F. M.; Baldwin, T. O.; Rayment, I. Biochemistry 1995, 34, 6581−6586. (24) Baldwin, T. O.; Christopher, J. A.; Raushel, F. M.; Sinclair, J. F.; Ziegler, M. M.; Fisher, A. J.; Rayment, I. Curr. Opin. Struct. Biol. 1995, 5, 798−809. (25) Van Berkel, W. J. H.; Kamerbeek, N. M.; Fraaije, M. W. J. Biotechnol. 2006, 124, 670−689. (26) Zimmermann, L.; Stephens, A.; Nam, S. Z.; Rau, D.; Kübler, J.; Lozajic, M.; Gabler, F.; Söding, J.; Lupas, A. N.; Alva, V. J. Mol. Biol. 2018, 430, 2237−2243. (27) Yang, K.; Qi, L. H.; Zhang, M.; Hou, X. F.; Pan, H. X.; Tang, G. L.; Yuan, H. Curr. Microbiol. 2015, 71, 458−464. (28) Zhu, Z.; Li, H.; Yu, P.; Guo, Y.; Luo, S.; Chen, Z.; Mao, X.; Guan, W.; Li, Y. Appl. Microbiol. Biotechnol. 2017, 101, 1547−1557. (29) Zhang, Q.; Chen, Q.; Zhuang, S.; Chen, Z.; Wen, Y.; Li, J. Appl. Environ. Microbiol. 2015, 81, 3753−3765. (30) Kieser, T.; Bibb, M. J.; Buttner, M. J.; Chater, K. F.; Hopwood, D. A. Practical Streptomyces Genetics; John Innes Foundation: Norwich, UK, 2000.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jnatprod.8b00077.



Tables of NMR spectroscopic data of compounds mentioned in the main text and primers used in this study; figures of NMR spectra of compounds 1−5; phylogenetic analyses for RsdO1 and RsdO6; gene inactivations; gene complementation experiments 13Clabeling analyses of compounds 1, 3, and 4; NMR spectra of compound 6 (PDF)

AUTHOR INFORMATION

Corresponding Author

*Tel/Fax: +86-20-89023028. E-mail: [email protected]. ORCID

Hongbo Huang: 0000-0002-5235-739X Jianhua Ju: 0000-0001-7712-8027 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This study was supported in part by the National Natural Science Foundation of China (41706169, 81425022, U1501223, U1706206), Natural Science Foundation of Guangdong Province (2016A030312014), and the Program of Chinese Academy of Sciences (XDA11030403). We are also very grateful to Ms. Xiao, Ms. Sun, Ms. Zhang, and Mr. Li in the analytical facility at SCSIO for recording spectroscopic data.



REFERENCES

(1) Kim, J.; Yi, G. S. BMC Microbiol. 2012, 12, 169. (2) Iwaka, H.; Nakayama, Y.; Takahashi, M.; Uetsuki, S.; Kido, M.; Fukuyama, Y. J. Antibiot. 1984, 37, 1091−1093. (3) Komagata, D.; Sawa, R.; Kinoshita, N.; Chiaki Imada, C.; Sawa, T.; Naganawa, H.; Hamada, M.; Okami, Y.; Takeuchi, T. J. Antibiot. 1992, 45, 1681−1683. (4) Garcia, I.; Vior, N. M.; González-Sabín, J.; Braña, A. F.; Rohr, J.; Moris, F.; Méndez, C.; Salas, J. A. Chem. Biol. 2013, 20, 1022−1032. (5) Yan, X.; Probst, K.; Linnenbrink, A.; Arnold, M.; Paululat, T.; Zeeck, A.; Bechthold, A. ChemBioChem 2012, 13, 224−230. (6) Huang, H.; Yang, T.; Ren, X.; Liu, J.; Song, Y.; Sun, A.; Ma, J.; Zhang, Y.; Huang, C.; Zhang, C.; Ju, J. J. Nat. Prod. 2012, 75, 202− 208. 1577

DOI: 10.1021/acs.jnatprod.8b00077 J. Nat. Prod. 2018, 81, 1570−1577