Introduction

Actinomycetes have been a useful source of bioactive natural products [1]. For instance, the immunosuppressant FK506 and the antibiotic vancomycin were discovered from this class of microorganisms. However, the discovery rate of natural products with novel structures is decreasing [2], and some previously discovered compounds are often found repeatedly, although novel antibiotics are still needed because of the emergence of multidrug-resistant pathogens. In contrast, the recent advances in next-generation DNA sequencers have resulted in extensive genomic information on actinomycetes and showed that most of these microorganisms have at least 30 secondary metabolite biosynthesis gene clusters [3]. This number is much higher than the numbers of natural products that have been isolated from respective strains, suggesting that most of such biosynthetic gene clusters are not active under usual laboratory conditions. These silent gene clusters are expected to be a useful source of novel natural products.

Genome mining is an approach to discover novel natural products based on genetic information [3]. The activation of silent gene clusters is one of the most important parts of genome mining. Therefore, many methods have been developed to activate silent biosynthetic gene clusters [3,4,5,6]. One approach is to cultivate a certain microorganism with another microorganism. This co-culture method often results in the production of natural products that are not produced in pure culture. Co-cultivation of Streptomyces with a mycolic acid-containing bacterium, for example, is called combined-culture, and various compounds have been isolated using this method [4, 7,8,9,10,11,12].

Recently, we analyzed the biosynthetic pathway of cremeomycin and discovered an unprecedented nitrous acid biosynthetic pathway composed of two enzymes, CreE and CreD [13]. Nitrous acid produced by this pathway is involved in the formation of the diazo group of cremeomycin (Fig. 1a). In this pathway, first CreE converts l-aspartic acid to nitrosuccinic acid. Sequentially, CreD converts nitrosuccinic acid to nitrous acid and fumaric acid. The nitrous acid reacts with the amino group of 3-amino-2-hydroxy-4-methoxybenzoic acid to form the diazo group of cremeomycin. We call this the ANS (l-aspartate-nitrosuccinate) pathway in this manuscript. In silico analysis of genome sequences in the public databases showed that operons consisting of creE and creD homologs are present in a wide variety of actinomycetes [13]. In addition, many of these operons are surrounded by genes apparently related to secondary metabolism, suggesting that the ANS pathway is involved in the biosynthesis of various secondary metabolites. This idea was supported by a study on the fosfazinomycin biosynthetic pathway [14]. FtzM and FtzL, encoded by the fosfazinomycin biosynthetic gene cluster, have the same function as CreE and CreD and are presumably involved in the biosynthesis of the N–N bond of fosfazinomycin. However, we have still limited knowledge of the role of the ANS pathway in secondary metabolism.

Fig. 1
figure 1

Cremeomycin biosynthetic pathway (a) and putative biosynthetic pathway for novel desferrioxamine derivatives (b)

To further understand this role, we carried out genome mining of the ANS pathway. We selected a putative secondary metabolite biosynthetic gene cluster containing creE and creD homologs in the genome of Streptomyces davawensis JCM 4913 as a target (Fig. 2 and Table 1). By comparing the metabolic profiles of the wild-type and ΔcreE (BN159_4422) strains, we identified three novel desferrioxamine derivatives having either of two unusual five-membered ring structures (Fig. 1b). This research expands our knowledge on the role of the ANS pathway for the biosynthesis of secondary metabolites.

Fig. 2
figure 2

Biosynthetic gene cluster containing creE and creD homologs in S. davawensis

Table 1 Predicted functions of the genes located around the creE and creD homologs

Materials and methods

Strains and chemicals

S. davawensis JCM 4913 was obtained from the Japan Collection of Microorganisms (JCM, Ibaraki, Japan). T. pulmonis TP-B0596 was isolated from the soil sample from Toyama Prefecture, Japan, previously [15]. Escherichia coli JM109 was used for DNA manipulation. E. coli ET12567 harboring pUZ8002 was used for conjugation. TSB medium was prepared by dissolving 3% tryptic soy broth in water and autoclaving prior to incubation. A-3M medium was prepared by dissolving 0.5% glucose, 2% soluble starch, 2% glycerol, 1.5% Pharmamedia®, 0.3% yeast extract, and 1% HP-20 (Mitsubishi Chemical Corporation, Tokyo, Japan), adjusted to pH 7.0, and autoclaving prior to incubation. Bennett maltose medium was prepared by dissolving 0.1% yeast extract, 0.07% bonito extract, 0.038% meat extract, 0.2% NZ amine type A, and 1% maltose, adjusting to pH 7.3, and autoclaving prior to incubation. Restriction enzymes and DNA polymerase were purchased from Takara Bio Inc. (Shiga, Japan). In the polymerase chain reaction (PCR), the genomic DNA of S. davawensis was used as a template unless otherwise noted.

Production and purification of BN159_4422 (CreE homolog) and BN159_4421 (CreD homolog)

BN159_4422 was amplified by PCR using primers 5′-CATATGACCGGCAGCAGGACCAA-3′ (an NdeI site is underlined, and the start codon is italicized) and 5′-AAGCTTTACGATCACCCCGGCGCCCT-3′ (a HindIII site is underlined, and the stop codon was removed for the histidine-tag fusion at the C-terminus of the recombinant protein). The obtained DNA fragment was cloned into NdeI and HindIII sites of pET26b, resulting in pET26-BN159_4422. BN159_4421 was amplified by PCR using primers 5′-CATATGAGCGGGCGAGGCGACAC-3′ (an NdeI site is underlined, and the start codon is italicized) and 5′-AAGCTTTCAGCGCAGGGCGCGGTCCA-3′ (a HindIII site is underlined, and the stop codon is italicized). The obtained DNA fragment was cloned into the NdeI and HindIII sites of pColdI, resulting in pColdI-BN159_4421.

E. coli BL21(DE3) strains harboring pET26-BN159_4422 and pColdI-BN159_4421 were individually cultured in 100 mL of TB medium with 50 µg/mL kanamycin at 37 °C until OD600 reached 0.6. After incubation at 18 °C (for the strain harboring pET26-BN159_4422) and 15 °C (for the strain harboring pColdI-BN159_4421) for 15 min, 0.05 mM isopropyl β-d-l-thiogalactopyranoside (IPTG) with 2 mg/L riboflavin and 0.02 mM IPTG, respectively, was added into the culture broths. After 20 h, the cells were harvested using centrifugation and resuspended in lysis buffer I (20 mM Tris-HCl pH 9.0, 20% glycerol, 200 mM NaCl, 10 mM β-mercaptoethanol) and lysis buffer II (20 mM HEPES pH 8.0, 20% glycerol, 500 mM NaCl), respectively. Then lysozyme (final ~0.5 mg/mL) was added to each solution. After each solution was incubated on ice for 1 h, the cells were lysed using sonication [0.05 mM flavin adenine dinucleotide (FAD) was added to the sample of the strain harboring pET26-BN159_4422 prior to sonication]. After cell debris was removed by centrifugation, the recombinant BN159_4422 and BN159_4421 proteins were purified by Ni2+ affinity chromatography using His60 Ni Superflow Resin (Takara Bio Inc.; 1 mL). The former and latter proteins were eluted with elution buffer I (20 mM Tris-HCl pH 9.0, 20% glycerol, 200 mM NaCl, 10 mM β-mercaptoethanol, 200 mM imidazole) and elution buffer II (20 mM HEPES pH 8.0, 20% glycerol, 500 mM NaCl, 200 mM imidazole), respectively, desalted with the same lysis buffer and concentrated using an Amicon Ultra centrifugal filter with a 10,000-molecular-mass cutoff (Merck Millipore, Darmstadt, Germany).

In vitro assay of CreE and CreD

The reaction mixture containing 10 µM CreE, 10 µM CreD, 5 mM l-aspartic acid, 2 mM NADPH, and 100 mM Tris-HCl (pH 7.5) was incubated at 30 °C for 1 h. The reaction was quenched by adding formic acid (final concentration 2%). Production of nitrous acid was examined with Saltzman reagent as described previously. [16]

Construction of the ∆BN159_4422 (creE homolog) strain

The BN159_4422 disruptant was constructed by substituting the core region of BN159_4422 with an apramycin resistance gene (aac(3)IV) by homologous recombination. A downstream fragment (~2000 bp) of BN159_4422 was amplified by PCR using primers 5′-AAGCGGCCGCTGCTGTCTGGTGGTCATGTC-3′ (a NotI site is underlined) and 5´-TTGATATCATCCGTACGCTGCTCAACTC-3′ (an EcoRV site is underlined). The obtained DNA fragment was cloned into NotI and EcoRV sites of pKGLP2 [17], resulting in pKGLP2-BN159_4422-down. Meanwhile, aac(3)IV was amplified by PCR using primers 5′-AATCTAGAAGCAAAAGGGGATGATAAGTTTATC-3′ (an XbaI site is underlined) and 5′-TTGCGGCCGCAGAATAGGAACTTCGGAATAGG-3′ (an NotI site is underlined) and an acc(3)IV-containing DNA fragment as a template. The obtained DNA fragment was cloned into XbaI and NotI sites of pKGLP2-BN159_4422-down, resulting in pKGLP2-BN159_4422-down/apra. An upstream fragment (~2000 bp) of BN159_4422 was amplified by PCR using primers 5′-AACTGCAGCATCAACTCGTGCCGACGGT-3′ (a PstI site is underlined) and 5′-TTTCTAGAACAGCCGCCTGATGTGACGT-3′ (an XbaI site is underlined). The obtained DNA fragment was cloned into PstI and XbaI sites of pKGLP2-BN159_4422-down/apra, resulting in pKGLP2-∆BN159_4422. To construct a ∆BN159_4422 strain, this plasmid was transferred into S. davawensis using a conjugation method described below. The desired recombination was confirmed by PCR using primer sets, 5′-GCCAGTCCTTCCACCGCGTC-3′ (primer 1 in Figure S2A) plus 5′-GGCGACAGCCCTGGGTCAAC-3′ (primer 2) and 5′-CCCATCTTCGAGGGGCCGGA-3′ (primer 3) plus 5′-CTCCTTCGGGGTGCCGTTCC-3′ (primer 4).

Construction of pTYM19ep-BN159_4422-4421

A DNA fragment harboring BN159_4422 and BN159_4421 was amplified by PCR using primers 5′-AACATATGACCACCAGACAGCACAC-3′ (an NdeI site is underlined, and the start codon of BN159_4422 is italicized) and 5′-TTAAGCTTTCATGCCGCCGACCGGGTGCG-3′ (a HindIII site is underlined, and the stop codon of BN159_4421 is italicized). The obtained DNA fragment was cloned into NdeI and HindIII sites of pTYM19ep [18], resulting in pTYM19ep-BN159_4422-4421, in which the BN159_4422- 4421 operon is located under the control of ermE* promoter.

Construction of the ∆BN159_5485 (desD homolog) strain

The BN159_5485 disruptant was constructed by substituting the core region of BN159_5485 with aac(3)IV by homologous recombination. An upstream fragment (~2000 bp) of BN159_5485 was amplified by PCR using primers 5′-GCGGCCGCGCGCGATGAGCAGTTGATACGGGTCGT-3′ and 5′-AACCCGATAGCGGGGCTGTA-3′. For the amplification of a downstream fragment (~2000 bp), primers 5′-GACGGATTCGGCGAGGCTCA-3′ and 5′-GATTACGAATTCGATGACGCCGTCGTGGCGGGCAC-3′ were used. The aac(3)IV gene was amplified by PCR using primers 5′-CCCCGCTATCGGGTTGAATAGGAACTTCGGAATAG-3′ and 5′-CTCGCCGAATCCGTCAGCAAAAGGGGATGATAAGT-3′ and an acc(3)IV-containing DNA fragment as a template. The amplified fragments were cloned together into an EcoRV site of pKGLP2 using In-fusion (Takara Bio Inc.), resulting in pKGLP2-∆BN159_5485. To construct a ∆BN159_5485 strain, this plasmid was transferred into S. davawensis using the conjugation method described below. The desired recombination was confirmed by PCR using a primer set, 5′-ATCGACCGCGCCAACCTCTA-3′ (primer 5 in Figure S2C) plus 5′-TCGGTGAAGTACTCGGCGAC-3′ (primer 6).

Conjugational transfer

E. coli strain ET12567/pUZ8002 harboring a plasmid for gene disruption was inoculated into 100 mL of Luria-Bertani (LB) medium and incubated at 37˚C until OD600 reached 0.4–0.6. The cells were harvested and washed twice with LB medium and resuspended in 5 mL of LB medium. S. davawensis spores in 100 µL of 20% glycerol were suspended in 0.5 mL of TSB medium and incubated at 50˚C for 10 min. The spores and E. coli cells were mixed and inoculated on MS agar plate containing 10 mM MgCl2 and 60 mM CaCl2. After incubation at 30 °C for 18 h, the plate was overlaid with an antibiotics solution (nalidixic acid [0.75 mg/mL] and apramycin [0.75 mg/mL] in 1 mL of sterile water). After incubation at 30 °C for 1 week, several apramycin-resistant S. davawensis colonies were obtained, in which the plasmid was integrated into the chromosome through single-crossover recombination. For the second recombination, the strains were repeatedly incubated on MS agar with apramycin (50 mg/L) for few days. Double crossover mutants were selected by blue-white selection using gusA on pKGLP2 and X-gluc (5-bromo-4-chloro-3-indolyl β-d-glucuronide cyclohexylammonium salt). The spores harvested from MS agar were inoculated on Bennett maltose agar containing 50 mg/L apramycin and 0.3% X-gluc. White colonies were selected as candidates for a double crossover mutant. The desired recombination was confirmed by PCR using primers described above.

Analysis of metabolites of S. davawensis in combined culture

The wild-type and mutant S. davawensis strains were inoculated into TSB medium and incubated at 30 °C for 3 days. In parallel, T. pulmonis was inoculated into TSB medium and incubated at 30 °C for 2 days. Portions of S. davawensis pre-culture (3 mL) and T. pulmonis pre-culture (1 mL) were together inoculated into 100 mL of A-3M medium in a K-1 flask. After incubation at 30 °C for 2 days, 5 mL of culture broth was harvested and incubated with 0.1 g of Amberlite FPX66 resin (Organo, Tokyo, Japan) at room temperature for 2 h. The resin and cells were together harvested using centrifugation and washed with distilled water. Then the metabolites were extracted from the resin and cells with methanol. The methanol extract was evaporated to dryness and dissolved in 200 μL of dimethyl sulfoxide (DMSO) for LC-ESIMS analysis with a 1100 series spectrometer (Agilent Technologies, Santa Clara, CA) coupled to a High-Capacity Trap Plus system (Bruker Daltonics, Billerica, MA) equipped with a Cosmosil π NAP Packed column (2.0ID × 150 mm; Nacalai Tesque, Kyoto Japan). Compounds were eluted with a linear gradient of water and acetonitrile containing 0.1% formic acid. The flow rate was 0.4 mL/min. The gradient elution profile started with 2% acetonitrile and kept 2% for 2 min. The acetonitrile concentration was gradually increased to 50% in 28 min, to 100% in 2 min, and kept 100% for 2 min. The column was re-equilibrated with 2% acetonitrile for 3 min before the next injection. Absorbance at 210 nm was monitored.

Isolation and structural elucidation of compounds

For isolation of desferrioxamine derivatives, the wild-type S. davawensis strain was co-cultured with T. pulmonis in multiple flasks as described above. After the addition of Amberlite FPX66 resin (0.02 g/mL culture), the culture broth was incubated at room temperature for 2 h for the adsorption of compounds. The cells and FPX66 resin were harvested by filtration, and the compounds were extracted with methanol. The solvent was evaporated to dryness, and the residual materials were dissolved in methanol and subjected to medium-pressure liquid chromatography (MPLC; Purif-Compact A, Shoko Scientific, Tokyo, Japan) with a silica gel column (SIZE60; Shoko Scientific). Absorbed compounds were eluted with a linear gradient of chloroform/methanol from 100/0 to 0/100 for 20 min. The fraction containing the target compounds was evaporated to dryness. The residual materials were dissolved in methanol and subjected to high-performance liquid chromatography (HPLC; Shimadzu, Tokyo, Japan) equipped with Cosmosil π NAP column (10ID × 250 mm, Nakalai Tesque) and Cosmosil 5C18 AR-II column (10ID × 250 mm, Nakalai Tesque). In both cases, the target compounds were eluted in a linear gradient of water and methanol containing 0.1 % formic acid. As a result, compounds 1 (1 mg from 6 L of fermentation broth), 2 (11 mg from 9 L), and 3 (4.8 mg from 6 L) were isolated. The 1H NMR, 13C NMR, COSY, HMBC, HMQC, and TOCSY spectra were recorded in DMSO-d6 on JNM-A500 NMR System (JEOL, Tokyo, Japan).

Siderophore assay

Desferrioxamine derivatives (1, 2, and 3) were suspended in deionized water, and an equal volume of 5 mM FeCl3 was added to the solution. Then, these samples were analyzed with LC-MS. The analytical conditions were the same as those in the analysis of S. davawensis metabolites.

Results and discussion

In silico analysis of a putative secondary metabolite biosynthesis gene cluster containing creE and creD homologs in S. davawensis

By BLAST search using CreE and CreD as queries, creE (BN159_4422) and creD (BN159_4421) homolog genes were discovered in the genome of S. davawensis JCM4913 [19] (Table 1 and Fig. 2). BN159_4422 and BN159_4421 showed 56.3% and 65.9% identity to CreE and CreD, respectively, of S. cremeus. The putative gene cluster containing BN159_4422 and BN159_4421 possesses several other genes that seem to be involved in secondary metabolite biosynthesis (Table 1 and Fig. 2). For example, BN159_4424 encodes a TrpE(G)-like protein, which is involved in anthranilate biosynthesis. Interestingly, S. davawensis has another copy of trpE (BN159_6401) and BN159_6401 is highly homologous to TrpE proteins of other Streptomyces species (e.g., SGR_5465 of Streptomyces griseus, 88.3% identity; SAV_6171 of Streptomyces avermitilis, 95.9%; SCO2043 of Streptomyces coelicolor A3(2), 92.7%). In contrast, BN159_4424 shows very poor homologies to these enzymes (around 10% identity). This fact suggests that BN159_6401 and BN159_4424 should be involved in tryptophan biosynthesis and secondary metabolism, respectively. Furthermore, this gene cluster encodes two putative ligases (BN159_4426 and BN159_4430) and two putative acyl carrier proteins (BN159_4427 and BN159_4431). The presence of these genes suggests that this gene cluster should be responsible for the biosynthesis of a secondary metabolite(s). However, it was very difficult to predict their chemical structures by bioinformatic analysis.

In vitro analysis of CreE and CreD homologs (BN159_4422 and BN159_4421) from S. davawensis

To confirm the enzymatic functions of the CreE and CreD homologs of S. davawensis, we produced C-terminally His-tagged BN159_4422 and N-terminally His-tagged BN159_4421 in E. coli and purified them by Ni2+ affinity chromatography (Figure S1A). The recombinant enzymes were used for the nitrous acid formation assay described previously. [16] As expected, these recombinant proteins produced nitrous acid from l-aspartic acid in the presence of NADPH and FAD (Figure S1B). This result clearly shows that CreE (BN159_4422) and CreD (BN159_4421) homologs from S. davawensis have the same functions as CreE and CreD from S. cremeus.

Natural products produced by the creE and creD homolog-containing gene cluster

To identify the natural products produced by the gene cluster containing the creE and creD homologs, we constructed a BN159_4422 disruptant (ΔBN159_4422) by substituting the core region of BN159_4422 with an apramycin resistance gene (Figure S2A and B) and compared its metabolic profile with that of the wild-type strain. Although we cultivated these two strains in various conditions, we could not observe any significant differences in their metabolic profiles (Fig. 3a, b). Thus, we assumed that this gene cluster is silent under usual laboratory conditions, and therefore we decided to use combined-culture to awaken the gene cluster. The two S. davawensis strains were individually co-cultured with T. pulmonis in A-3M medium. Comparison of their metabolic profiles showed that three compounds (1, 2, and 3) were detected only in the wild-type strain (Fig. 3d, e). These compounds were not produced when S. davawensis and T. pulmonis were cultured alone (Fig. 3a, c), indicating that the compounds were produced by S. davawensis and that its production was activated by co-cultivation with T. pulmonis. When the BN159_4422-4421 operon was introduced into the ΔBN159_4422 strain using the chromosome integrative vector pTYM19ep, the production of 1, 2, and 3 was restored (Figure S3). Taken together, these results clearly show that the creE and creD homologs are responsible for the biosynthesis of 1, 2, and 3.

Fig. 3
figure 3

LC-MS analysis of metabolites produced by the wild-type and mutant S. davawensis strains in pure culture and combined-culture with T. pulmonis. BN159_4422 and BN159_5485 encode CreE and DesD homologs, respectively. Sd and Tp indicate S. davawensis and T. pulmonis, respectively. The peak marked with an asterisk was concluded not to be related with the CreE and CreD homologs because it was observed occasionally

Isolation and structural elucidation of compounds 1, 2, and 3

The culture was scaled up for the purification of compounds 1, 2, and 3. These three compounds were purified by silica gel chromatography using MPLC and reverse-phase chromatography using HPLC. High-resolution mass spectrometry (HR-MS) showed that the molecular formulae of sodium adducts of compounds 1, 2, and 3 were C32H54N6O11Na (m/z 721.37482; calcd. for m/z 721.37428), C33H56N6O11Na (m/z 735.39047; calcd. for m/z 735.38993), and C33H56N6O9Na (m/z 703.40065; calcd. for m/z 703.40010), respectively. The compounds were further analyzed by 1D and 2D NMR (Table 2 and Fig. 4). Comparison of 1H and 13C NMR spectra with several desferrioxamine derivatives suggested that these compounds are desferrioxamine B derivatives with different modifications (Table 2) [20,21,22]. These structures were further supported by MS/MS analysis (Figure S4) [22]. The fragment ions observed were consistent with the cleavage at peptide bonds. Further analysis of unassigned NMR signals clearly showed that these compounds are novel desferrioxamine derivatives with either of two unusual N-containing five-membered ring structures (Fig. 4). The terminal amine of desferrioxamine B (and its analog) is apparently modified to produce these heterocyclic structures (a pyrrole ring in compound 3 and more oxidized one in compounds 1 and 2). In the structures of 1 and 2, the terminal heterocyclic structure is common and the difference is originated from the desferrioxamine chain length; 1 has a one-carbon shorter chain than 2. Meanwhile, in the structures of 2 and 3, the desferrioxamine chain length is common, but their terminal heterocyclic structures are different. According to their structures, we named these compounds desferrioxamine I2a (1), I1a (2), and I1b (3).

Table 2 1H and 13C NMR data for 1, 2, and 3
Fig. 4
figure 4

Structures of 1, 2, and 3. Key correlations of 1H-1H COSY and 1H-13C HMBC are depicted as illustrated in the lower part

Analysis of the putative desferrioxamine biosynthetic gene cluster

Although involvement of the creE homolog in the biosynthesis of these desferrioxamine derivatives was obvious, no genes related to desferrioxamine biosynthesis are encoded by the gene cluster containing the creE and creD homologs. Instead, a putative desferrioxamine biosynthetic gene cluster is present at a different locus in the genome (BN159_5485-5490, Figure S5 and Table S1) [23, 24]. To examine whether this biosynthetic gene cluster is related to the biosynthesis of 1, 2, and 3, a BN159_5485 (desD homolog) disruptant (ΔBN159_5485) was constructed (Figure S2C and D). As expected, the ΔBN159_5485 strain did not produce 1, 2, or 3 in the combined-culture with T. pulmonis (Fig. 3f). Interestingly, production of another compound (4) was also abolished in the disruptant (Fig. 3f). By MS/MS analysis, this compound was identified to be desferrioxamine B (4) (Figure S4) [25]. Taken together, we concluded that the BN159_5485-5490 operon is responsible for the biosynthesis of desferrioxamine derivatives including compounds 1, 2, 3, and 4 in S. davawensis. We speculate that desferrioxamine B (4) should be a biosynthetic intermediate of 2 and 3. We also speculate that a desferrioxamine B analog (5), which has a shorter carbon chain by one methylene than desferrioxamine B (4) (Fig. 1), should be a biosynthetic intermediate of 1; a trace amount of 5 was also detected in the culture broth of the wild-type S. davawensis strain (both pure culture and combined-culture with T. pulmonis) (Figure S4). It should be noted that production of 4 and 5 was also observed in the pure culture of both the S. davawensis wild-type and ΔBN159_4422 strains. This result indicates that the BN159_5485-5490 operon responsible for the biosynthesis of 4 and 5 is expressed under the normal culture condition and that the gene cluster containing the creE and creD homologs is awakened by combined culture, which results in the conversion of 4 and 5 into the novel desferrioxamine derivatives 1, 2, and 3.

Siderophore assay of 1, 2, and 3

According to their structures, 1, 2, and 3 were expected to have siderophore activity. To confirm this, these compounds were incubated with Fe3+ ion and analyzed by LC-MS (Figure S6). As a result, these compounds gained 53 Da, which corresponds to one Fe3+ ion. In addition, a significant change in the UV spectra was observed: new λmax appeared around 430 nm. These results clearly show that compounds 1, 2, and 3 act as a siderophore.

Conclusion

In this study, we discovered three novel desferrioxamine siderophores produced by S. davawensis under combined-culture with T. pulmonis through genome mining of the ANS pathway. The compounds are probably biosynthesized by modifying desferrioxamine B (4) or its analog (5), which are produced by the function of the BN159_5485-5490 operon (Fig. 1b). However, because the terminal heterocyclic structures in 1, 2, and 3 have not been reported so far, it is difficult to predict how these structures are biosynthesized. Because BN159_4422 and BN159_4421 were shown to produce nitrous acid in vitro and these genes are required for the biosynthesis of 1, 2, and 3 in vivo, nitrous acid probably plays an important role in the biosynthesis of these terminal heterocyclic structures. It is our future challenge to reveal the biosynthetic pathway for these novel desferrioxamine derivatives including the role of nitrous acid in this pathway. However, involvement of the ANS pathway in the biosynthesis of the unique terminal heterocyclic structures of these novel desferrioxamine derivatives provided an important example for the diverse usage of the ANS pathway for the biosynthesis of various secondary metabolites in actinomycetes. This study also indicates that the ANS pathway could be a useful source of information for the discovery of novel natural products as well as novel secondary metabolite gene clusters.