Chimeric natural products derived from medermycin and the nature-inspired construction of their polycyclic skeletons

Medermycin, produced by Streptomyces species, represents a family of antibiotics with significant activity against Gram-positive pathogens. The biosynthesis of this family of natural products has been studied, and new skeletons related to medermycin have rarely been reported until recently. Herein, we report eight chimeric medermycin-type natural products with unusual polycyclic skeletons. The formation of these compounds features some key nonenzymatic steps, which inspired us to construct complex polycyclic skeletons via three efficient one-step reactions under mild conditions. This strategy was further developed to efficiently synthesize analogues for biological activity studies. The synthetic compounds, chimedermycins L and M, and sekgranaticin B, show potent antibacterial activity against Staphylococcus aureus, methicillin-resistant Staphylococcus aureus, and methicillin-resistant Staphylococcus epidermidis. This work paves the way for understanding the nonenzymatic formation of complex natural products and using it to synthesize natural product derivatives.

anthranilic acid, and 4-hydroxybenzaldehyde, under mild conditions 18 . Aeruginaldehyde and different aminopyrrolnitrin derivatives undergo an abiotic Pictet−Spengler reaction to generate pyonitrins A−D 19 . Dibohemamines A−C are formed via the spontaneous dimerization of bohemamines and formaldehyde 17 . In addition to their fascinating structures, these compounds formed through nonenzymatic reactions also exhibit potent biological activities [15][16][17] .
Mangrove-associated microorganisms, especially actinomycetes, have been demonstrated to be an important source of antibiotics 20 . To discover new compounds with antibacterial activity against drugresistant pathogens, we developed a library of Streptomyces sp. strains isolated from mangrove samples. This library was utilized to screen for antibacterial activity. The results showed that the metabolites of strain OUCMDZ-4982 exhibit significant antibacterial activity against Staphylococcus aureus (SA) and methicillin-resistant Staphylococcus aureus (MRSA). To obtain the active compounds from this strain, we investigated its secondary metabolites.

Isolation and structural elucidation
Strain OUCMDZ-4982 was isolated from a mud sample collected from Hat Chao Mai National Park, Thailand. It was identified as a Streptomyces species by analysis of the 16 S rRNA gene sequence. The preliminary study on its major metabolites resulted in the isolation of medermycin, a known antibiotic with strong activity against Grampositive pathogens [21][22][23] . Production-scale fermentation of OUCMDZ-4982 was carried out using rice-based solid medium to provide sufficient material for the identification of new metabolites. Normal-phase fractionation of the extract yielded nine fractions. The MS/MS-based molecular networking ( Supplementary Fig. 1) indicated that OUCMDZ-4982 can produce higher molecular weight medermycin derivatives. Purification of compounds from the active fraction led to chimedermycins A−H (1−8) and medermycin (9).
Chimedermycin A (1) was determined to have a molecular formula of C 40 H 43 NO 13 based on the high-resolution electrospray ionization mass spectrometry (HRESIMS) peak at m/z 746.2798 [M + H] + . Analysis of its 1 H and 13 C NMR data and comparison with those of medermycin (9) suggested that compound 1 contains a pyranonaphthoquinone moiety similar to medermycin [21][22][23] . The other part of compound 1 was determined to be 6-deoxy-dihydrokalafungin (DDHK) 24,25 by analysis of the remaining signals in the 1 H and 13 C NMR spectra. The connection between the medermycin moiety and DDHK moiety was confirmed by the heteronuclear multiple bond correlations (HMBCs) of H-4 to C-11′′, H-5 to C-10′′, and H-9′′ to C-14 ( Supplementary Fig. 2). The relative configurations of chimedermycin A (1) were determined by nuclear Overhauser effect spectroscopy (NOESY) correlations ( Supplementary  Fig. 2), coupling constant analysis (Supplementary Table 1), and 13 C NMR calculations [26][27][28] (Supplementary Fig. 3, Supplementary Data 1-4). The absolute configuration of medermycin (9) has been confirmed by total synthesis 29 . Moreover, the stereochemistry of the angolosamine ring in the natural medermycin-type derivatives isolated from Streptomyces species seems to be constant thus far because the genes related to this angolosamine ring are highly conserved (Fig. 2   Therefore, the absolute configuration of compound 1 was speculated based on the shared biosynthetic origin with medermycin (9) (Figs. 2, 3a). This deduction was verified by electronic circular dichroism (ECD) calculations 34 and the ECD exciton chirality method 35 . The ECD spectrum of 1 showed a negative Cotton effect at 344 nm and a positive Cotton effect at 297 nm, which was indicative of negative chirality between the two chromophores ( Supplementary Fig. 3). The structure of compound 1 was further confirmed by its semisynthesis from 9 and 10 ( Fig. 3b).
Comparison of the NMR spectra of chimedermycins B-D (2-4) with those of compound 1 revealed that compounds 2-4 were methyl ester derivatives of 1 at two different carboxyl groups (Fig. 1). Their structures were further confirmed by the chemical transformations from 2 to 1, from 1 to 4, and from 3 to 4. Analysis of the 1D and 2D NMR data of chimedermycin E (5) revealed that it has a chimeric skeleton derived from medermycin and dehydroxy-GTRI-02 36 . The NOESY correlations ( Supplementary Fig. 2) and ECD Cotton effects (Supplementary Fig. 4) indicated that compound 5 has the same configuration as compound 1. Medermycin (9) and dehydroxy-GTRI-02 are thus the biosynthetic precursors of chimedermycin E (5).
Chimedermycins F-H (6-8) also contain the same moiety derived from medermycin (9) as compound 1. Their polycyclic skeletons were determined by NMR data analysis ( Supplementary Fig. 2, Supplementary Table 3). Compound 7 is the methyl ester derivative of 6, which was verified by the chemical transformation from 6 to 7. The structures of compounds 6 and 8 were further confirmed by semisyntheses from 9 ( Fig. 3b). Full details on the structural assignments of chimedermycins A-H (1-8) can be found in the Supplementary Information.

Proposed biosynthesis of medermycin and the chimedermycins
To investigate the formation of the chimeric skeletons, the whole genome of Streptomyces sp. OUCMDZ-4982 was sequenced, and the medermycin biosynthetic gene cluster (med cluster) in this strain was compared with the reported cluster from Streptomyces sp. AM-7161 37,38 . The med cluster in Streptomyces sp. OUCMDZ-4982 contains 27 open reading frames (ORFs), which are highly homologous with those of Streptomyces sp. AM-7161 (Fig. 2a). The product of the missing gene med-21 is likely associated with pathway regulation (putative kinase) and med-24 encodes a putative phosphopantetheinyl transferase 39 . Furthermore, no additional genes involved in the biosynthesis of the chimeric frameworks were found in the med cluster. Therefore, we speculated that the formation of compounds 1-8 may be driven by nonenzymatic reactions. The biosynthetic pathway for medermycin (9) has been thoroughly studied ( Fig. 2) 37,[39][40][41] . It consists of polyketide formation and attachment of the sugar moiety. First, the key bicyclic intermediate of octaketide origin forms, which further undergoes ketoreduction, cyclization, and dehydration to form the chiral intermediate 4-dihydro-9-hydroxy-1-methyl-10-oxo-3-H-naphtho-[2,3-c]pyran-3-(S)-acetic acid ((S)-DNPA). Subsequent enoyl reduction of (S)-DNPA affords 6-deoxy-dihydrokalafungin (DDHK), which undergoes a series of redox reactions to yield dihydrokalafungin (DHK) and  kalafungin (KAL). Finally, angolosamine, produced via the deoxyhexose pathway, is attached to KAL to generate medermycin (Fig. 2). Based on the above facts, we proposed a nonenzymatic process for the formation of the chimeric natural products 1-8. (+)-DDHK, derived from (S)-DNPA (10), can react with medermycin (9) to form chimedermycin A (1). It seems likely that (+)-DDHK and medermycin (9) undergo Michael addition followed by another nucleophilic addition (Fig. 3a). Medermycin (9) is hydrolysed to generate demethylmenoxymycin B (11), which can react with butane-2,3-dione (12) to form the key intermediate d by Michael addition followed by ketol condensation. Then, intermediate d undergoes hemiketalization to yield chimedermycin F (6) (Fig. 3a). 4-Hydroxyphenylpyruvic acid (13) and medermycin (9) are considered two biosynthetic precursors of chimedermycin H (8). 4-Hydroxyphenylpyruvic acid (13) can be easily oxidized into quinone methide by molecular oxygen. Electrophilic attack of quinone methide at the 8-position of medermycin (9) followed by dehydro-aromatization results in the formation of intermediate e, which undergoes hemiketalization followed by oxidative decarboxylation to generate chimedermycin H (8) (Fig. 3a). These small molecules, which react with medermycin to form chimaeras, are derived from different pathways. DDHK is a key intermediate in the biosynthesis of medermycin. Butane-2,3-dione may be produced by Streptomyces sp. OUCMDZ-4982. It has been reported that various bacteria can produce butane-2,3-dione. In all these producers, the key enzyme acetohydroxyacid synthase (AHAS; acetolactate synthase) is involved in the biosynthesis of acetolactate from pyruvate. Acetolactate is unstable and can be easily converted to butane-2,3-dione through nonenzymatic decarboxylation in the presence of oxygen 42 . AHAS is a heterotetramer that is composed of two subunits, a catalytic one and a regulatory one, which are encoded by the genes ilvB (Gen-Bank accession no. AY785370.1) and ilvN (GenBank accession no. AF175526.1), respectively, in a Streptomyces cinnamonensis strain 43,44 .
We searched these genes in the genome of Streptomyces sp. OUCMDZ-4982, and found two genes with high homology to ilvB (91% identity) and ilvN (90% identity). So, we speculated that Streptomyces sp. OUCMDZ-4982 has the ability to produce butane-2,3-dione. 4-Hydroxyphenylpyruvic acid is an important intermediate during the formation of tyrosine in the primary metabolic process 45 .

Semisynthesis of compounds 1, 6, 8 and their analogues
By analyzing the structural characteristics and possible formation pathways of compounds 1, 6, and 8, we think that their skeletons may be synthesized by simple chemical reactions. To investigate the formation of chimedermycin A (1), a reaction between medermycin (9) and the model compound naphthalene-1,8-diol (14) was designed (Fig. 4a). At the outset, inspired by the living environment of microorganisms, this reaction was biomimetically carried out in DMSO/ phosphate-buffered saline (PBS; 0.01 M, pH 7.2-7.4) (1:10) at room temperature (rt) (Fig. 4a). Surprisingly, compound 15, which we named chimedermycin I, formed almost quantitatively within 1 h (Fig. 4b). Furthermore, we investigated the influence of different solvent systems on this reaction (Fig. 4) and found that it hardly occurs in DMSO or DMSO/MeCN (1:10), and only a small amount of product can be observed in DMSO/H 2 O (1:10). Although aqueous sodium bicarbonate (pH 7.5) can promote the reaction, this condition is not better than that of the DMSO/PBS system (Fig. 4b).

Antimicrobial activity
The antimicrobial activities of compounds 1-9, 15, and 21-27 were evaluated against two Gram-negative bacteria (Pseudomonas aeruginosa and Escherichia coli), three Gram-positive bacteria (SA, MRSA, and MRSE), and two pathogenic fungi (Candida albicans and Candida glabrata). Given the antibiotic resistance of MRSA and MRSE, ciprofloxacin and vancomycin were chosen as positive controls for antibacterial activity, as these two antibiotics can significantly inhibit methicillin-resistant Staphylococcus species. Ketoconazole was used as a positive control for the pathogenic fungi. The antibacterial activity results are shown in Table 1 as minimal inhibitory concentrations (MICs) and minimal bactericidal concentrations (MBCs). These compounds were found to have no activity against Gram-negative bacteria or fungi. Interestingly, the known compounds 9 and 26 exhibited stronger activities against the Gram-positive bacteria SA, MRSA, and MRSE than the positive controls. Compounds 23, 24, and 27 showed activities roughly equivalent to vancomycin and slightly stronger than ciprofloxacin (Table 1) (Fig. 7), indicating that the antibacterial effect of compound 27 is much stronger than that of compounds 23 and 24. The strategy of combining fermentation and nature-inspired synthesis described in this report can provide enough material for further biological activity studies.

Discussion
In conclusion, we obtained eight chimeric medermycin-type natural products (1-8) from a marine-derived Streptomyces species. These compounds contain three types of complex polycyclic skeletons that are formed through spontaneous nonenzymatic reactions. Based on the possible mechanisms of these reactions, we investigated the syntheses of the above mentioned natural products. Astonishingly, these complex structural skeletons can be constructed through three simple one-step reactions, which occur smoothly under near physiological conditions. Using this nature-inspired strategy, we accomplished the semisyntheses of chimedermycins A (1), F (6), and H (8) from medermycin (9). Moreover, seven chimeric derivatives, chimedermycins I-N (15 and 21-25) and sekgranaticin B (27), were efficiently synthesized. During the construction of these chimeric skeletons, some common natural products, such as 4-hydroxycoumarin (18) and pyoluteorin (20), were involved. This synthetic method provides an effective strategy for the reutilization of simple and common structures and complex and new skeletons for drug development. Three synthetic compounds (23, 24, and 27) exhibited potent antibacterial activity against Gram-positive bacteria. Hence, obtaining more analogues for further biological studies using this method will be part of our future efforts.
There are a growing number of natural products that are produced by key nonenzymatic steps. Spontaneous reactions, together with enzymes for biosynthesis, play important roles in the formation of complex skeletons. The in-depth investigation of the nonenzymatic formation of natural products can not only help us to further understand the formation mechanism of some complex skeletons but also inspire us to synthesize more compounds for biological activity research. The nonenzymatic formation mechanism of some natural products has been elucidated, but there are still some challenges in this process. Currently, the discovery of nonenzymatic steps remains very accidental. In the future, we need to explore more rational research methods to further understand nonenzymatic biosynthesis, such as utilizing small molecule probes with chemical activity to explore natural products with highly reactive functional groups or structural fragments [50][51][52] . These compounds may have the potential to form different skeletons with other natural products through spontaneous reactions. On the one hand, it is highly necessary to analyze and verify the formation mechanisms of complex natural products based on chemical principles. On the other hand, analysis of biosynthetic genes can also help us to speculate whether these processes are nonenzymatic. In addition, from an evolutionary perspective, metabolites with high chemical reactivity may have some important ecological functions, but this needs to be confirmed by further systematic study.

Collection and phylogenetic analysis of strain OUCMDZ-4982
The actinobacterium Streptomyces sp. OUCMDZ-4982 was isolated from a mud sample collected from Hat Chao Mai National Park, Thailand. The sample (2 g) was dried over 24 h in an incubator at 35°C. The dried sample was diluted to 10 −3 g/mL, 100 μL of which was dispersed across a solid-phase agar plate (Modified Emerson agar media: 1 g of glucose, 1 g of yeast powder, 4 g of peptone, 2.5 g of NaCl, 18 g of agar, and 1 L of seawater) and incubated at 28°C

Genome sequencing and analysis
The whole genome sequencing of OUCMDZ-4982 was performed using Illumina HiSeq/NovaSeq PE150 platform at Allwegene Technology Company Limited (Beijing, China). All good quality paired reads were assembled using the SPAdes software (v3.13.0) into several scaffolds. The genome was annotated with Prokka software (v1.13.7) and the function of genes was predicted by universal database (NR, SwissProt, GO, KEGG, COG, InterProScan, CAZymes, Pfam, and CARD  53 . The tested strains were cultivated in LB agar plates (5 g of yeast extract, 10 g of tryptone, 5 g of NaCl, 15 g of agar powder, and 1 L of water) for bacteria and in YPD agar plates (10 g of yeast extract, 20 g of peptone, 20 g of glucose, 15 g of agar powder, and 1 L of water) for fungi at 37°C. Ciprofloxacin lactate and ketoconazole were used as positive controls for bacteria and fungi, respectively. Compounds 1-9, 15, 21-27, and positive controls were dissolved in MeOH at the concentration of 100 μg/mL. A 10 μL quantity of test solution was absorbed by a paper disk (5 mm diameter) and placed on the assay plates. After 24 h incubation, zones of inhibition were observed. The MICs were determined by the broth microdilution method 54 . Broth dilution used LB growth medium containing increasing concentrations (a twofold dilution series) of the antimicrobial compounds, which was inoculated with 5 × 10 5 CFU/mL of pathogenic cells. The MICs were considered to be the lowest concentration that caused total inhibition of bacterial growth. MICs were also determined in triplicate for each strain.

Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The 16 S rRNA gene sequence of Streptomyces sp. OUCMDZ-4982 has been deposited in the GenBank database under accession number of MW193757. The data that support the findings of this study are available within the paper and its Supplementary Information file. Additional data are available from the corresponding author upon request. Source data are provided with this paper.