Escherichia coli DH5α and Morganella morganii SBK3 were cultivated in Lysogeny Broth (LB), containing 1% (w/v) bacto tryptone, 1% (w/v) NaCl and 0.5% (w/v) yeast extract, at 37°C. A. nidulans strains were cultivated in Minimal Medium (MM) [AspA (70 mM NaNO₃, 11.2 mM KH₂PO₄, 7 mM KCl, pH 5.5), 1% (w/v) glucose, 0.1% (v/v) Hutner's trace elements] or in London medium (LM) [1% (w/v) glucose, 7 mM KCl, 2 mM MgSO₄, 11.2 mM KH₂PO₄, 0.1% (v/v) Hutner's trace elements (Hill and Kafer, 2001) without iron sulfate, 10 μM FeCl₃, 10 mM NaNO₃, pH = 6.5] with 0.0001% (w/v) 4-aminobenzoic acid at 30 or 37°C. Two percentage (w/v) agar was added for cultivation in petri dishes. For induction of the dba gene cluster, 0.1 μg/ml bleomycin (VWR, 203408-10) or 1 μg/ml phleomycin (InvivoGen, ant-ph-5p) were added. All Aspergillus strains were constructed in AGB552 (Bayram et al., 2012) background. Streptomyces strains were cultivated in GYM medium (4 g/l glucose, 4 g/l yeast extract, 10 g/l malt extract, pH = 7.2). For solid GYM medium, 2% (w/v) agar and 2 g/l CaCO₃ were added. For co-cultivation of Streptomyces strains with A. nidulans, 93 ml London medium were mixed with 32 ml GYM medium and inoculated with each 1.25^*10⁷ spores of Streptomyces spp. and A. nidulans AGB552. The cultures were grown for 2 days at 120 rpm and 30°C. S. mobaraenis (DSM 40847) was provided by Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH (DSMZ). Streptomyces coelicolor and Streptomyces cellulosae subsp. griseorubiginosus were provided by Axel Zeeck. M. morganii SBK3 was provided by Günther Winkelmann.

For the construction of plasmids used for gene deletions in A. nidulans, pME4319 served as vector, which contains a recyclable phleomycin resistance cassette (Phle resistance gene from Streptoalloteichus hindustanus (ble) (Drocourt et al., 1990; Punt and van den Hondel, 1992; Hartmann et al., 2010; Liu et al., 2021). Sequences of all used primers in this study are listed in Supplementary Table 1. Approximately 1,500 bp of the 5' and 3' flanking regions of the gene of interest were inserted into the SwaI and PmlI sites of pME4319, respectively, by using the GeneArt™ Seamless Cloning and Assembly Enzyme Mix or the GeneArt™ Seamless Cloning and Assembly Kit (Thermo Fisher Scientific). The 5' and 3' flanking regions were amplified from A. nidulans genomic DNA with the primer pairs given in Supplementary Table 2. For transformation of A. nidulans, the deletion cassettes were excised from the corresponding plasmids with PmeI. Supplementary Table 3 includes all plasmids and Supplementary Table 4 all A. nidulans strains used in this study. E. coli and A. nidulans were transformed using the heat-shock method (Inoue et al., 1990) or the polyethylene glycol-mediated protoplast fusion method (Punt and van den Hondel, 1992), respectively. 5 μg/ml phleomycin was added to the medium to select for positive A. nidulans clones. The recyclable marker cassette in A. nidulans was removed by cultivation of the fungus on Minimal Medium containing 0.5% (w/v) glucose and 0.5% (w/v) xylose. Thereby, the xylose inducible promoter of the cassette activates expression of the ß-recombinase, whose gene product recognizes and cleaves off the marker cassette at its two flanking β-six recognition sites, leaving one β-six site (Hartmann et al., 2010; Liu et al., 2021).

Fungal mycelia were harvested, frozen in liquid nitrogen and ground with a table mill. RNA was isolated according to the instruction of the RNeasy® Plant Miniprep Kit (Qiagen). Approximately 0.8 μg of RNA was used for cDNA synthesis according to manufacturer's instructions of the QuantiTect® Reverse Transcription Kit (Qiagen). RT-PCR was performed with the MESA GREEN qPCR MasterMix Plus for SYBR Assay (Eurogentec) in a CFX Connect Real-Time System (BioRad). One microliter of 1:10 diluted cDNA was used per reaction. Two biological replicates with each three technical replicates were used. Primers are listed in Supplementary Table 5. Data were recorded with the CFX Manager software (BioRad). Expression of the housekeeping genes h2A, gpdA, and rps15 were used for normalization.

For SM extraction from liquid cultures, mycelia were removed by filtration. The aqueous filtrates were adjusted to pH 5 and extracted with equivalent volumes of ethyl acetate. The organic solvent was evaporated, and the dried metabolites were stored at −20°C. For LC-MS analysis, the metabolites were reconstituted in acetonitrile/H₂O (1:1) and analyzed with a Q Exactive™ Focus orbitrap mass spectrometer coupled to an UltiMate™ 3000 HPLC (Thermo Fisher Scientific) equipped with a DAD-3000 Diode Array Detector (Thermo Fisher Scientific) and a Corona™ Veo™ RS Charged Aerosol Detector (Thermo Fisher Scientific). Five microliter of each sample was injected on a HPLC column [Acclaim™ 120, C18, 5 μm, 120 Å, 4.6 × 100 mm (Thermo Fisher Scientific)] applying a linear acetonitrile/0.1% (v/v) formic acid in H₂O/0.1% (v/v) formic acid gradient [from 5 to 95% (v/v) acetonitrile/0.1 formic acid in 20 min, plus additional 10 min with 95% (v/v) acetonitrile/0.1 formic acid] with a flow rate of 0.8 ml/min at 30°C. The measurements were performed in a mass range of 70–1,050 m/z in positive and negative mode. Data acquisition and analysis was performed with Thermo Scientific Xcalibur™ 4.1 (Thermo Fisher Scientific) and with FreeStyle™ 1.6 (Thermo Fisher Scientific) software.

For SM extraction of the fungal biomass, the mycelia were washed with water and dried. The mycelia were covered with 100 ml acetone and placed in a Sonorex™ Digital 10 P ultrasonic bath (Bandelin) for 30 min at 10% performance. The mycelia were removed by filtration and anhydrous magnesium sulfate was added to the organic filtrate to bind remaining water. After another filtering the organic solvent was evaporated as described before. For the HR-ESI-MS analysis, the metabolites were reconstituted in acetonitrile/H₂O (1:1) and 2 μl was injected into an Agilent 1200 Infinity Series HPLC (Agilent Technologies, Böblingen, Germany) utilizing a C18 Acquity® UPLC BEH column (2.1 × 50 mm, 1.7 μm; Waters, Milford, MA/USA) connected to a maXis® ESI-TOF-MS (Bruker). HPLC parameters were set as follows: solvent A: H₂O + 0.1% formic acid, solvent B: acetonitrile + 0.1% formic acid; gradient: 5% B (0.5 min), 5–100% (19.5 min), 100% (5 min), flowrate 0.6 ml/min, and DAD detection 190–600 nm.

NMR spectra were recorded with an Avance III 700 spectrometer with a 5 mm TCI cryoprobe (¹H 700 MHz, ¹³C 175 MHz) and an Avance III 500 spectrometer (¹H 500 MHz, ¹³C 125 MHz; both Bruker, Billerica, MA/USA). NMR data were referenced to selected chemical shifts of acetone-d₆ (¹H: 2.05 ppm, ¹³C: 29.32 ppm), pyridine-d₅ (¹H: 7.22 ppm, ¹³C: 123.87 ppm) and DMSO-d₆ (¹H: 7.27 ppm, ¹³C: 77.00 ppm), respectively. Optical rotations were taken with a MCP 150 polarimeter (Anton Paar, Graz, Austria); UV spectra were taken with a UV-2450 UV/VIS spectrophotometer (Shimadzu, Kyoto, Japan); ECD spectra were collected with a JD 815 spectrophotometer (Jasco, Pfungstadt, Germany).

ESI-MS spectra were recorded with an UltiMate® 3000 Series uHPLC (Thermo Fisher Scientific, Waltman, MA/USA) utilizing a C18 Acquity® UPLC BEH column (2.1 × 50 mm, 1.7 μm; Waters, Milford, MA/USA) connected to an amaZon® speed ESI Iontrap MS (Bruker). HPLC parameters were set as follows: solvent A: H₂O + 0.1% formic acid, solvent B: acetonitrile + 0.1% formic acid; gradient: 5% B (0.5 min), 5–100% (19.5 min), 100% (5 min), flowrate 0.6 mL/min, and DAD detection 190–600 nm. HR-ESI-MS spectra were obtained with an Agilent 1200 Infinity Series HPLC (Agilent Technologies, Böblingen, Germany; conditions same as for ESI MS spectra) connected to a maXis® ESI-TOF-MS (Bruker).

Anhydrosepedonin (1) and antibiotic C (2) were isolated from crude extracts of A. nidulans strains: AGB535 (ΔdbaF) was used for the isolation of the mixture of anhydrosepedonin (1) and antibiotic C (2) as well as the isolation of pure antibiotic C (2), whereas anhydrosepedonin (1) was purified from AGB1430 (ΔdbaB). In total 5 L liquid LM supplemented with 10 μM FeCl₃, 0.0001% (w/v) 4-aminobenzoic acid and 1 μg/ml phleomycin (InvivoGen, ant-ph-5p) was inoculated with 1 × 10⁶ spores/ml of either AGB535 or AGB1430 and incubated on a shaker for 2 days at 30°C. Mycelia were removed by filtration, the aqueous filtrates were adjusted to pH 5 and extracted with equivalent volumes of ethyl acetate. The organic solvent was evaporated, and the crude extracts were stored at −20°C.

98 mg crude extract was dissolved in 3 mL MeOH/DMSO (9:1) and subjected to preparative reverse phase HPLC (Pure C-850 FlashPrep, Büchi Labortechnik, Flawil, Switzerland). As stationary phase, Synergi Polar-RP column (250 × 50 mm, 10 μm, Phenomenex, Torrance, CA, USA) was used. The mobile phase consisted of deionized water (solvent A) and acetonitrile (solvent B). Purification of the crude extract was performed by using an isocratic elution of 5% aqueous ACN with 0.05% formic acid at a flow rate of 50 mL/min for 5 min, 5–100% solvent B in 60 min and finally isocratic elution at 100% solvent B for 10 min to afford the mixture of compound 1 and 2 (_tR: 37.7–38.1 min, 12.1 mg).

220 mg crude extract of AGB1430 (ΔdbaB) was dissolved in 5 mL MeOH and subjected to preparative reverse phase HPLC (PLC 2250, Gilson, Middleton, WI, USA). As stationary phase, Synergi Polar-RP column (250 × 50 mm, 10 μm, Phenomenex, Torrance, CA, USA) was used. The mobile phase consisted of deionized water (solvent A) and acetonitrile (solvent B). Purification of the crude extract was performed by using an isocratic elution of 15% aqueous ACN with 0.05% formic acid at a flow rate of 50 mL/min for 5 min, 15–50% solvent B in 50 min, 50–100% solvent B in 10 min, and finally isocratic elution at 100% solvent B for 10 min to afford compound 1 (t_R: 46.0–46.5 min, 12.7 mg).

148.6 mg crude extract of AGB535 (ΔdbaF) was dissolved in 6 mL MeOH, divided into 4 equal parts and subjected to preparative reverse phase HPLC (Pure C-850 FlashPrep, Büchi Labortechnik, Flawil, Switzerland). As stationary phase, Luna C18(2) column (250 × 21.2 mm, 5 μm, Phenomenex, Torrance, CA, USA) was used. The mobile phase consisted of deionized water (solvent A) and acetonitrile (solvent B). Purification of the crude extract was performed by using an isocratic elution of 15% aqueous ACN with 0.05% formic acid at a flow rate of 20 mL/min for 5 min, 15–45% solvent B in 32 min, 50–100% solvent B in 3 min, and finally isocratic elution at 100% solvent B for 10 min to afford compound 2 (t_R: 32.0–32.3 min, 3.6 mg).

Ten liter liquid MM was inoculated with 10⁹ spores of strain AGB527 and grown at 37°C for 36 h. Mycelia were removed by filtering with Miracloth and the pH of the culture filtrate was adjusted to 5. The culture filtrate was extracted twice with an equivalent amount of ethyl acetate and dried to yield the crude extract. The crude extract was separated by preparative HPLC (Instrumentelle Analytik Goebel GmbH: HPLC pump RAININ Dynamax SD-1, HPLC detector RAININ Dynamyx UV-1, column Nucleodur 100-5 C18 ec, 250 × 20 mm, solvent system: A = H₂O, B = acetonitrile) under gradient conditions (20% B to 100% B in 20 min). Detection was carried out at 230 nm. The isolation yielded 7.7 mg 7-hydroxy-3,7-dimethyl-6H-2-benzopyran-6,8(7H)-dione that we named azanidulone. ¹H-NMR spectra were recorded on a Varian Mercury-Vx 300 (300 MHz) spectrometer. ¹³C-NMR spectra were recorded on a Varian Inova-500 spectrometer (125.7 MHz). ESI-MS data were acquired using a Finnigan LC-Q mass spectrometer.

Four liter liquid LM with 1 μg/ml phleomycin was inoculated with 10⁹ spores of strain AGB1418 and grown at 30°C for 2 days. Mycelia were removed and the pH of the culture filtrate was adjusted to 5. The culture filtrate was extracted twice with an equivalent amount of ethyl acetate and dried to yield 515.9 mg crude extract. The crude extract was dissolved in 10 mL MeOH and subjected to preparative reverse phase HPLC (PLC 2020, Gilson, Middleton, WI, USA). As stationary phase, VP Nucleodur 100-5 C18 ec column (250 × 20 mm, 7 μm, Macherey-Nagel, Düren, Germany) was used. The mobile phase consisted of deionized water (solvent A) and acetonitrile (solvent B). Purification of the crude extract was performed by using a linear gradient elution of 10–30% aqueous ACN with 0.05% formic acid at a flow rate of 20 mL/min for 35 min, 30–50% solvent B in 10 min, 50–100% solvent B in 5 min, and finally isocratic elution at 100% solvent B for 10 min to afford compound 5 (t_R: 29–29.8 min, 3 mg).

The bioassay for detection of siderophores with keto-hydroxy bidentate ligands was performed as described previously (Thieken and Winkelmann, 1993). Plates with solid LB medium (containing agar agar, SERVA, 9002-18-0) with or without 600 μM 2,2'-dipyridyl were inoculated with fresh M. morganii SBK3 over-night cultures. SMs were extracted from 125 ml liquid cultures of A. nidulans AGB552 and ΔdbaI as described above and dissolved in 100 μl methanol. Filter discs were coated with 10 μl metabolite extracts diluted 1:1 with methanol and placed on the agar. Ten microliter pure methanol served as control. Plates were cultivated for 3 days at 37°C in darkness. Growth and inhibition zones were detected and measured.

Bioactivity assays were performed to analyze the direct growth inhibiting effect of the SM extracts from A. nidulans on Streptomyces. 2 × 10⁶ spores of S. mobaraensis, S. coelicolor, and S. cellulosae were distributed on whole agar plates containing 40 ml GYM medium. The SM extract of fungal reference strain AGB552 and ΔdbaI was dissolved in 80 μl methanol and 10 μl were spotted on a filter disk which was incubated with the inoculated plates for 2 days at 30°C. To test the direct effect of anhydrosepedonin and antibiotic C these metabolites were isolated from the fungal extracts by preparative HPLC. The anhydrosepedonin/antibiotic C mixture as well as separately isolated anhydrosepedonin (1) and antibiotic C (2) were dissolved in methanol to a final concentration of 2 mg/ml. 12.5 μl of the extract was spotted on filter disks, which were incubated with the Streptomyces inoculated GYM plates for 2 days at 30°C. Methanol served as control.

Minimum inhibitory concentrations (MIC) were determined as described previously (Becker et al., 2021). The following fungal and bacterial organisms were tested: Bacteria: Bacillus subtilis (DSM10), Staphylococcus aureus (DSM346), Acinetobacter baumanii (DSM30008), Chromobacterium violaceum (DSM30191), E. coli (DSM1116), Pseudomonas aeruginosa (PA14); mycobacteria: Mycolicibacterium smegmatis (ATCC700084); fungi: Candida albicans (DSM1665), Schizosaccharomyces pombe (DSM70572), Mucor hiemalis (DSM2656), Pichia anomala (DSM6766), Rhodotorula glutinis (DSM10134). The cytotoxicity assay against mouse fibroblast cell line L929, human cervical cancer cell line KB 3.1, human epidermoid carcinoma cell line A431, human lung carcinoma cell line A549, human prostate carcinoma cell line PC-3, human ovarian adenocarcinoma cell line SKOV-3 as well as human breast adenocarcinoma cell line MCF-7 was performed as described by (Becker et al., 2021).

To unravel the chemical response of A. nidulans toward different Streptomyces species, A. nidulans AGB552 was co-cultivated with S. mobaraensis, S. coelicolor or S. cellulosae on solid and in liquid medium at 30°C. The three Streptomyces strains are producers of different kinds of antibiotics (Höfs et al., 2000; Bentley et al., 2002; Komaki and Harayama, 2006).

First, solid GYM medium was inoculated with A. nidulans and one of the Streptomyces strains. After 6 days of growth an inhibition zone was visible between the colonies of S. mobaraensis and A. nidulans (Figure 1). No inhibition zone was observed when S. coelicolor or S. cellulosae were co-cultivated with A. nidulans. S. mobaraensis is a known producer of the glycopeptide antibiotic bleomycin, which is active against Aspergilli (Austin et al., 1990; Moore et al., 2003).

Next, SMs from liquid co-cultivation experiments were extracted to analyze whether A. nidulans produces specific molecules as a chemical response to S. mobaraensis, which might induce bacterial inhibition. After 2 days of cultivation, mycelia were removed and the metabolites in the filtrate were extracted with ethyl acetate and analyzed in LC-MS/MS equipped with a diode array detector (DAD). The HPLC chromatogram at 366 nm revealed two peaks in the sample extracted from the S. mobaraensis co-culture that were not present in the other co-cultures (Figure 2).

S. mobaraensis is a producer of the glycopeptide antibiotic bleomycin, which shows antifungal activity against Aspergilli and other ascomycetes (Austin et al., 1990; Moore et al., 2003). A growth test of A. nidulans AGB552 on medium supplemented with different concentrations of bleomycin revealed that A. nidulans grows on low concentrations of 0.1 μg/ml bleomycin (Supplementary Figure 1). To unravel the bacterial signal for fungal production of the tropolones, A. nidulans AGB552 was cultivated in liquid medium supplemented with 0.1 μg/ml bleomycin. After metabolite extraction, the chromatogram revealed the production of several metabolites that were not present in extracts from medium without bleomycin (Figure 3A). Phleomycin is another glycopeptide antibiotic produced by Streptomyces and closely related to bleomycin. According to the literature, A. nidulans cannot grow at phleomycin concentrations of c ≥ 10 μg/ml (Punt and van den Hondel, 1992). A growth test of A. nidulans AGB552 on medium supplemented with phleomycin revealed fungal growth at 1 μg/ml phleomycin (Supplementary Figure 1A). Cultivation of A. nidulans AGB552 in liquid medium supplemented with 1 μg/ml phleomycin and subsequent metabolite analysis showed the production of metabolites 1 and 2 as well as the two new peaks 4 and 5 (Figure 3B) that were also detected in cultures induced by bleomycin (Figure 3A; Supplementary Table 6).

Metabolites 1 and 2 were isolated by preparative HPLC and elucidated as the tropolones anhydrosepedonin (1) and antibiotic C (2) (Figure 4A) by NMR, HRESIMS and UV/VIS data. In addition, DHMBA (3) was identified by HRESIMS and UV/VIS spectra in small amounts from the S. mobaraensis co-culture. DHMBA (3) was identified earlier as PKS-derived metabolite in A. nidulans (Ahuja et al., 2012; Gerke et al., 2012). Anhydrosepedonin (1) has been isolated earlier from the fungus Sepedonium chrysospermum (Wright et al., 1970) and antibiotic C (2) from A. nidulans strain IMI238850 (McDonald et al., 1983), but no full set of NMR data has been published. The corresponding biosynthetic gene clusters for both compounds have not been identified, either. We conclude that the three identified compounds are produced by A. nidulans in co-culture with S. mobaraensis.

Compound 4 was isolated by preparative HPLC and identified as the azaphilone 7-hydroxy-3,7-dimethyl-6H-2-benzopyran-6,8(7H)-dione that we named azanidulone (Figure 4). It has been isolated before from the ascomycete Daldinia concentrica (Buchanan et al., 1995) and the ascomycete Nigrospora sp. YE3033 (Zhang et al., 2016). A 7S configuration was assigned due to a negative CE observed at ca. 355 nm in the CD spectrum of 4 (Supplementary Figure 7) (Clark et al., 2008).

Taken together, the glycopeptide antibiotics bleomycin and phleomycin, which are naturally produced by Streptomyces, induce the production of the tropolones anhydrosepedonin (1) and antibiotic C (2) as well as of DHMBA (3), azanidulone (4) and 5 in A. nidulans.

Compound 5 was isolated by preparative HPLC as a red oil. Its UV/VIS spectrum with maxima at 292 and 389 nm indicated an extensive π-system. The molecular formula C₁₇H₁₆O₇ was obtained from the peak at m/z 333.0972 in the HRESIMS spectrum, specifying 10 units of unsaturation. The ¹H and ¹³C NMR spectra of 5 in DMSO-d₆ contained two sets of resonances with a ratio of about 4:3, signifying spontaneous interconversion. Structure elucidation is described with data of the main isomer as follows: The proton and HSQC spectra of 5 indicated the presence of three methyls (δ_C/δ_H 29.5/2.19, 23.1/1.25, 19.8/2.28), one methylene (δ_C/δ_H 46.3/3.83 and 3.76) and three olefinic methines (δ_C/δ_H 124.0/5.92, 121.2/6.89, 100.1/6.34). Additionally, the carbon NMR spectrum showed signals for two ketones (δ_C 204.4, 198.5) and eight additional (δ_C 165.1, 163.2, 160.8, 147.8, 126.3, 100.0, 98.4, 78.9) carbon atoms devoid bound protons. Since 5 has a ratio of H/C atoms of <1, the crowding of HMBC correlations from just a few protons complicated the structure elucidation process. Consequently, an ADEQUTE NMR spectrum was measured for the distinction between ²J_{C, H} and ³J_{C, H} correlations. Foremost, the assignment of two-bond correlations from 3-H to C−4, from 10–H to C−9 and C−11, from 13–H to C−12 and C−14, from 15–H₃ to C−14 and from 17–H₃ to C−7 facilitated the assignment. Subsequently, observed ³J correlations from the HMBC NMR spectrum assembled the planar structure on 5 (Figure 4B). The largest difference in chemical shift between the isomers was observed for methyl group C−17 (δ_C 14.2 vs. 23.1), suggesting that the isomers differ in this region of the molecule. Since all HMBC correlations are analogous for both isomers, a stereo chemical difference was expected. The stereochemistry of C−7 is immutable and can be deduced from the absolute configuration of the biosynthetically related azanidulone (4) as 7S, but the neighboring hemiketal C−8 is metabolically unstable and prone to epimerize. Since a ROESY correlation between 8–OH and 17–H₃ was observed for the minor isomer, 7S, 8S and 7S,8R absolute configurations were assigned for the minor and for the major epimer, respectively.

Compound 5 is a novel substance with a novel tricyclic scaffold that we named tripyrnidone. It showed no bioactivity in our test assays against bacteria or fungi and no cytotoxicity against the tested cell lines.

DHMBA (3) was shown to be the direct product of the non-reducing PKS DbaI in A. nidulans (Ahuja et al., 2012; Gerke et al., 2012). We deleted dbaI in AGB552 background and analyzed the SM profile with LC-MS/MS with and without the addition of bleomycin and phleomycin (Figure 3). The deletion of dbaI abolished the production of DHMBA (3) as well as of the SMs 1, 2, 4, and 5. It can be concluded that the tropolones as well as the compounds 4 and 5 are produced by the PKS DbaI and are most likely biosynthesized from DHMBA by other enzymes.

In fungi, biosynthetic genes for a specific SM are usually clustered in the genome (Keller, 2019; Rokas et al., 2020). The gene encoding the DHMBA-producing PKS DbaI is embedded within the dba gene cluster on chromosome II of A. nidulans (Gerke et al., 2012) (Figure 5A). The cluster spans 9 genes, dbaA-dbaI, and encodes two putative transcription factors DbaA and DbaG. Overexpression of dbaA upregulates all genes of the cluster, which is usually silenced under standard laboratory conditions (Gerke et al., 2012). To understand the influence of glycopeptide antibiotics on dba gene cluster expression, the relative gene expressions of the dba cluster genes were analyzed with quantitative real-time polymerase chain reaction (qPCR). The neighbored genes AN7890-AN7895 and AN7904 were included in the analysis. The genes AN7893-7895 were shown earlier to be co-regulated with the dba cluster under certain conditions (Nahlik et al., 2010; Gerke et al., 2012; Bohnert et al., 2014; Oiartzabal-Arano et al., 2015). The A. nidulans reference strain AGB552 (rf) and the two transcription factor deletion strains ΔdbaA and ΔdbaG were cultivated with and without phleomycin. The relative gene expression pattern showed that all 9 dba genes as well as AN7893-AN7895 were upregulated in the reference strain when phleomycin was added (Figure 5B). In contrast, expression of AN7893-dbaI was not increased in phleomycin-containing medium when dbaA was deleted, proving that DbaA is the specific transcriptional activator for the gene cluster. The putative transcription factor DbaG is not involved in dba gene regulation but showed a repressive function on expression of the uncharacterized AN7891 gene, which encodes a β-1,4-endoglucanase, in medium without phleomycin. In summary, the presence of 1 μg/ml phleomycin induces the expression of the dba genes plus AN7893-AN7895. To test what is the lowest phleomycin concentration for the gene cluster induction, we performed qPCR with RNA extracted from cultures containing 0.01 to 1 μg/ml phleomycin. The analysis revealed expression of dbaI at 0.5 and 1 μg/ml phleomycin, but lower concentrations are rather insufficient (Supplementary Figure 1B).

Compounds 1–5 are polyketides. The structures as well as the abolished production in the PKS deletion strain ΔdbaI suggest that the compounds 1, 2, 4, and 5 are synthesized from the DbaI-product DHMBA (3) (Gerke et al., 2012) by other co-regulated enzymes. Therefore, all genes from AN7893-dbaI were deleted in A. nidulans AGB552 background. A SM analysis with LC-MS/MS after cultivation in medium with phleomycin revealed that without the transcriptional activator DbaA and without the PKS DbaI the compounds 1–5 were not produced (Figure 6A; Supplementary Figure 12). Deletion of the transcription factor gene dbaG as well as AN7895, dbaE and dbaF had no impact on compound production. In the strain without the transporter gene dbaD, the compounds 2 and 4 were not detected in the medium, but 5, which is released to the medium independently of DbaD. 1 was only detected in very low amounts (Supplementary Figure 12A). An additional intracellular extraction of SMs from mycelia of ΔdbaD did not reveal the presence of any of the metabolites (Supplementary Figure 13). For the synthesis of anhydrosepedonin (1) the non-heme Fe(II) dioxygenase AN7893 as well as the FAD-dependent monooxygenase DbaH are needed in addition to the PKS DbaI. Interestingly, the production of 1 increased when dbaB was deleted, indicating that DbaB converts 1 into another compound. Antibiotic C (2) was not detected from strains lacking AN7893, AN7894, DbaB, DbaC, DbaH, and DbaI. To produce azanidulone (4) only the PKS DbaI and the FAD-dependent monooxygenase DbaH are necessary. Tripyrnidone (5) is not detected from strains without DbaB, DbaH and DbaI. Double deletion of dbaB and AN7893 abolished 1 and 2, but 3–5 were still produced (Figure 6B).

Starting from DHMBA as precursor, we propose the following biosynthesis for compounds 1, 2, and 4 (Figure 7). DHMBA is oxidized by the FAD-dependent monooxygenase DbaH at position C-3. A ring closure follows to form the azaphilone skeleton and a subsequent spontaneous release of a water molecule leads to the formation of azanidulone (4). For the biosynthesis of the tropolones 1 and 2, the carbon atom of the methyl group at position C-3 is incorporated by the non-heme Fe(II) dioxygenase AN7893 yielding sepedonin (6) with the 7-membered tropolone ring. A spontaneous water molecule release forms the tropolone anhydrosepedonin (1), which is further oxidized by the FAD-dependent monooxygenase DbaB to the tropolone antibiotic C (2). In summary, the PKS DbaI, DbaH and AN7893 are responsible for the formation of tropolones. Due to its key role in tropolone ring formation, AN7893 was designated TroA (tropolone-forming A).

Since the putative intermediate sepedonin (6) was not identified from the HPLC chromatogram at 366 nm, we searched for the presence of its mass in the LC-MS/MS data of the reference and the deletion strains. The extracted ion chromatogram at m/z = 225.0757 [M + H]⁺ revealed a peak with the UV/VIS spectrum of sepedonin (Supplementary Figure 2) (Supka, 1981). As expected, sepedonin (6) is not produced in the deletion strains of troA, dbaA, dbaH, and dbaI. In contrast, the production of sepedonin (6) is enhanced in ΔdbaB as well as in the two YCII-domain encoding deletion strains ΔAN7894 and ΔdbaC. The function of YCII domains was not identified yet, but they seem to have a regulatory role on SM production.

For the biosynthesis of tripyrnidone (5), the oxidized DHMBA precursor presumably reacts with triacetic acid lactone (TAL). To date, TAL has not been identified from A. nidulans. We checked our LC-MS/MS data whether the reference strain AGB552 cultivated in bleomycin- or phleomycin-containing medium produces TAL. The extracted ion chromatogram at m/z = 127.0390 [M + H]⁺ revealed a peak at a retention time of 5.4 min (Figure 8). The peak was identified as TAL (7) by comparison with commercial TAL (Sigma-Aldrich). The TAL-producing PKS has not yet been identified in A. nidulans.

Theoretically, three tropolone molecules can form a complex with one Fe³⁺ ion. The iron binding activity of anhydrosepedonin (1) was shown earlier (Supka, 1981). To test whether the dba cluster produced tropolones can function as alternative siderophores, a bioassay for detection of siderophores with keto-hydroxy bidentate ligands was used (Thieken and Winkelmann, 1993). M. morganii SBK3 is a bacterium which uses siderophores with keto-hydroxy bidentate ligands for growth. Filter discs with A. nidulans SM extracts from reference strain AGB552 and ΔdbaI were put on LB agar plates supplemented with the iron chelator 2,2'-dipyridyl and inoculated with M. morganii. LB plates without 2,2'-dipyridyl were used as control. If the extract contains siderophores with keto-hydroxy bidentate ligands, a Morganella growth zone can be detected. The bioassay shows that M. morganii grows on the whole LB plate without 2,2'-dipyridyl but an inhibition zone of 1.6 ± 0.1 cm (p ≤ 0.01, n = 4) is visible for the extract of the reference strain, which is not present for the ΔdbaI extract (Figure 9). On plates where all free iron is bound by 2,2'-dipyridyl, M. morganii is able to grow in 3.9 ± 0.5 cm (p ≤ 0.001, n = 4) distance around the reference strain extract, whereas no growth is detected around the ΔdbaI extract. However, an inhibition zone is detected 2.6 ± 0.4 cm (p ≤ 0.001, n = 4) around the reference strain extract, surrounded by the growth zone. No bacterial growth can be observed close to the extract from the PKS deletion strain ΔdbaI neither to the methanol control. These results show that the tropolones produced by the A. nidulans AGB552 upon induction of the dba/troA gene cluster by phleomycin serve as siderophores and provide the bacteria with iron at low SM concentrations. At high concentrations, the SMs produced by the dba/troA cluster have an inhibitory effect on growth of the Gram-negative M. morganii.

Confrontation experiments of A. nidulans and Streptomyces species on solid medium showed an inhibition zone (Figure 1). SM extraction revealed that A. nidulans produces the tropolones anhydrosepedonin and antibiotic C from the dba/troA cluster in response to either co-cultivation with S. mobaraensis or by media supplementation with the glycopeptides bleomycin and phleomycin. Hence, it was tested whether the SM extract from an A. nidulans culture supplemented with phleomycin would inhibit growth of Streptomyces species. Filter disks containing the extract of the reference strain AGB552 and the PKS deletion strain ΔdbaI were incubated with spores of different Streptomyces strains. The tested Streptomyces strains show a zone of inhibition around the filter disk containing the SM extract of the Aspergillus reference strain, whereas no inhibition zone was detected around the control filter disk containing methanol (Figure 10A). For the extract of the PKS deletion strain, only a minor or no inhibition zone was detected. These results revealed that the SM extract of the reference strain has a dba-cluster dependent antibacterial activity on Streptomyces strains.

To test whether the antibacterial activity is based on anhydrosepedonin (1) and/or antibiotic C (2), which are components of the reference strain extract, these metabolites were isolated from A. nidulans (Supplementary Figure 14). A filter disk was coated with either the pure compounds or a mixture of both (Figures 10B,C). Both purified compounds as well as the mixture inhibited bacterial growth.

Furthermore, anhydrosepedonin (1) and antibiotic C (2) were subjected to a broad range bioactivity test pipeline. The bacteria B. subtilis (DSM10), S. aureus (DSM346), A. baumanii (DSM30008), C. violaceum (DSM30191), E. coli (DSM1116) and P. aeruginosa (PA14), the mycobacterium M. smegmatis (ATCC700084) and the fungi C. albicans (DSM1665), S. pombe (DSM70572), M. hiemalis (DSM2656), P. anomala (DSM6766) and R. glutinis (DSM10134) were used. Additionally, the mouse fibroblast cell line L929, human cervical cancer cell line KB 3.1, human epidermoid carcinoma cell line A431, human lung carcinoma cell line A549, human prostate carcinoma cell line PC-3, human ovarian adenocarcinoma cell line SKOV-3 and human breast adenocarcinoma cell line MCF-7 were used. Antibiotic C (2) showed a broad, weak activity against all the tested organisms, except for P. aeruginosa, and showed strong cytotoxic activity against the tested cell lines with the highest activity against the cell line from human breast adenocarcinoma with an IC₅₀ of 0.041 μg ml⁻¹. Comparison showed lower activities for the tested anhydrosepedonin (1) (Table 2).

Taken together, the soil fungus A. nidulans produces the SMs anhydrosepedonin (1) and antibiotic C (2) from the dba/troA gene cluster, which have an antibiotic activity against Streptomycetes and a range of different other microorganisms.