Actinoalloteichus hymeniacidonis HPA 177T was purchased from DSMZ, Germany. Bacillus subtilis ATCC6051 was obtained from the research group of Professor Ákos T. Kovács, Department of Biotechnology and Biomedicine, Technical University of Denmark.

To identify, annotate and analyze the secondary metabolite biosynthetic gene clusters, NCBI BLAST (Boratyn et al., 2013) and AntiSMASH 5.0 (Blin et al., 2019) were used in the genome mining process of A. hymeniacidonis (CP014859.1) and other PTM producers (Supplementary Table S2).

The following GenBank releases are sources of ftd-like gene clusters used in comparative analyses: Streptomyces sp. SPB78, GenBank accession NZ_ACEU01000453; S. griseus subsp. griseus NBRC 13350, GenBank access AP009493.1; Lysobacter enzymogenes, GenBank access EF028635.2.

NMR spectra were recorded on a Bruker Advance III 800 MHz spectrometer. Silica gel chromatography was performed on silica gel 60 (Merck, 0.04–0.063 mm, 230–400 mesh ASTM) and Sephadex LH20 (Pharmacia). Ikarugamycin and equisetin were ordered from Sigma-Aldrich. Solvents of analytical grades were ordered from VWR.

To obtain sufficient amounts of metabolites for chemical characterization, A. hymeniacidonis was cultured in a 6 L liquid fermentation at 28°C, 200 rpm for 7 days. Slant of spores of A. hymeniacidonis was inoculated in a 500 ml flask containing 150 ml medium soluble starch 4 g/L, KNO₃ 2 g/L, NaCl 1 g/L, MgSO₄·7H₂O 0.5 g/L, CaCO₃ 0.02 g/L, yeast extract 1 g/L. The culture broth was separated into supernatant and mycelia parts, respectively. Both were extracted by ethyl acetate and the organic phases were combined. Evaporation of the organic solvent yielded 1.8 g of crude extract that was partially purified in flash chromatography by silica gel chromatography using gradient solutions of dichloromethane and methanol into 10 fractions F1-10. The F7 fraction containing PTMs was further purified by semi-preparative HPLC using an XBridge RP₁₈ HPLC Column 10 × 250 mm, 5 μm, a flow rate of 4 ml/min, 40.0°C. Using a 28 min multi-step method and acetonitrile and water as mobile phases the following method was applied in semi-preparative HPLC: at 0–5 min 10–50% acetonitrile, at 5–7 min 50–60% acetonitrile, at 7–15 min 60–80% acetonitrile, at 15–18 min 80–100% acetonitrile, and acetonitrile was maintained at 100% for another 5 min and followed by re-equilibration to 10% acetonitrile until 28 min. Pure compound 1 (0.8 mg) was obtained and analyzed by NMR spectroscopy.

A UHPLC–DAD–QTOF method was set up for the screening, with an injection volume of 1 μl extract. The separation was performed on a Dionex Ultimate 3000 UHPLC system (Thermo Scientific, Dionex, Sunnyvale, CA, United States) equipped with a 100 × 2.1 mm, 2.6 μm, Kinetex C₁₈ column, held at a temperature of 40°C, and using a linear gradient system composed of A: water, and B: acetonitrile. The flow rate was 0.4 ml min⁻¹.

Time-of-flight detection was performed using a maXis 3G QTOF orthogonal mass spectrometer (Bruker Daltonics, Bremen, Germany) operated at a resolving power of ∼50,000 full width at half maximum FWHM. The instrument was equipped with an orthogonal electrospray ionization source, and mass spectra were recorded in the range m/z 100–1,000 as centroid spectra, with five scans per second. For calibration, 1 μl of 10 mmol sodium formate was injected at the beginning of each chromatographic run, using the divert valve 0.3–0.4 min. Data files were calibrated post-run on the average spectrum from this time segment, using the Bruker HPC high-precision calibration algorithm.

For ESI⁺ the capillary voltage was maintained at 4,200 V, in the spray chamber, the gas flow to the nebulizer was set to 2.4 bar, the drying temperature was 220°C, and the drying gas flow was 12.0 L min⁻¹. Transfer optics ion-funnel energies, quadrupole energy were tuned on HT-2 toxin to minimize fragmentation. For ESI⁻ the settings were the same, except that the capillary voltage was maintained at −2,500 V. Ion-cooler settings were: transfer time 50 µs, radiofrequency RF 55 V peak-to-peak Vpp, and pre-pulse storage time 5 µs.

The iron II detecting agent ferrozine was used to test the iron-reducing activity of three tetramates, xanthobaccin A, equisetin, and ikarugamycin. A reaction solution comprised of 10 µl test tetramate (1 mg/ml), 10 µl ammonium iron III citrate C₆H₈FeNO₇ (5 mg/ml), and 20 µl aqueous ferrozine (1% wt/vol). FeSO₄ mixed with aqueous ferrozine (1% wt/vol) was used as a positive control. Tetramate mixed with ammonium iron III citrate was used as a negative control. All components were dissolved in ammonium chloride buffer 1 M, pH 4.5. After 5 min reaction under darkness, the reaction mixtures were analyzed by HPLC-HRMS.

An agar diffusion assay was carried out to test the antimicrobial activity of tetramates against Bacillus subtilis ATCC 6051. Whatman Antibiotic assay discs of 6 mm were loaded with 20 µg pure tetramates with/without vitamin C (10 µg). The growth medium for B. subtilis was 3 g meat extract, 5 g (Bacto)-peptone, 5 g glucose, 1 L tap water, pH 7.3–7.5, 18 g agar. The test plates were prepared by pouring 14 ml of L-agar as a base layer; after solidifying, this was overlaid with 4 ml of the inoculated seed layer. Roximycin was used as a positive antibiotic control. Pure methanol and vitamin C were used as negative controls. The plates were incubated at 37°C for 24 h, and antimicrobial activity was recorded as clear zones (in mm) of inhibition surrounding the disk. The test sample was considered active when the zone of inhibition was greater than 6 mm. The MIC assay was done by the broth dilution method according to the NCCLS (National Committee for Clinical Laboratory Standards, 1997).

Actinoalloteichus hymeniacidonis HPA 177^T is a Gram-positive, rare actinomycete isolated from the marine sponge Hymeniacidon perlevis (Zhang et al., 2006). Our genome mining revealed that it harbored a putative PTM BGC (Figure 2), and the core biosynthetic PKS-NRPS protein, which exhibited 67 and 61% similarity to HSAF and frontalamide synthetase proteins, respectively. The individual genes from the PTM gene cluster are listed in Supplementary Table S1 and the proposed biosynthetic pathway is shown in Supplementary Figure S2. The biosynthesis was proposed to be carried out by a hybrid iterative PKS-NRPS, and a single set of the functional domains KS-AT-DH-KR-ACP that iteratively incorporate six malonyl-CoA to form two polyene chains, which were further condensed with the two free amine groups of L-ornithine via the NRPS activity. This resulted in a tetramate-polyene intermediate, which was then cyclized via reduction by the tailoring oxidative enzymes to form the PTM skeleton. The potential biosynthetic gene cluster and the homologs of the A. hymeniacidonis core PKS-NRPS protein (TL08_RS13440) identified in the NCBI database and their phylogenetic relationships are depicted in Supplementary Figure S1 and Supplementary Table S2, respectively.

To test whether the tetramate gene cluster identified through genome mining is functional in A. hymeniacidonis, we conducted a small-scale fermentation (10 ml) and confirmed the production of possible PTMs supported by LC-MS analyses (Supplementary Figure S3). To characterize the active compounds, A. hymeniacidonis was cultivated in a 6 L scale to yield a crude extract subjected to separation by chromatography on silica gel and Sephadex LH-20 columns, yielding pure compound 1 (0.8 mg).

Compound 1 was isolated as a major component. HRESIMS data of 1 ([M + H]⁺ 511.2788, calculated for 511.2803, Δ 2.7 ppm) suggested a molecular formula of C₂₉H₃₈N₂O₆ and implied that it might be xanthobaccin A through AntiBase search (Laatsch, 2012). Four olefinic protons (H-8, δ 5.93; H-9, δ 5.74; H-23, δ 6.63; H-24, δ 6.97) corresponding to two double bonds were observed. A trans and a cis configuration for the two double bonds was deduced by the coupling constants (16.0 and 10.8 Hz between H23/H24 and H8/H9, respectively). In the HMBC spectrum, NH (δ 7.87, t, 5.6 Hz), H-8, and H-9 showed correlations to C-7 (δ 166.1). The ¹³C NMR spectrum indicated the presence of a tetramate unit by the presence of signals for C-1 (δ 196.2), C-27 (δ 178.5), C-2 (δ 61.4), and C-26 (δ 99.4). The location of a carbonyl group at C-20 (δ 207.9) was established by HMBC correlations between H-12, H-21 and H-22 and C-20. The aliphatic parts of the molecule were confirmed by both COSY and HMBC correlations. The selected HMBC correlations can be seen in Supplementary Figure S4. Both ¹H NMR and ¹³C NMR spectra were identical to xanthobaccin A reported in the literature (Hashidoko et al., 1999). Xanthobaccin A was firstly reported from the Gram-negative bacterium Stenotrophomonas sp. SB-K88 living in the rhizosphere, which exhibited high activity against the plant fungal pathogen Pythium ultimum (Hashidoko et al., 1999). Here, for the first time, we demonstrated that xanthobaccin A is also produced by a Gram-positive bacterium associated with a sponge. Together with the discovery of the cytotoxic ikarugamycins and clifednamide A from a sponge-associated Streptomyces sp. (Dhaneesha et al., 2019), it indicates that sponge endosymbionts might be the true producers for those PTMs reported from sponges, such as the antinematode geodin A (Capon et al., 1999).

Xanthobaccin A was reported to be the principal active metabolite of A. hymeniacidonis. The compound exhibited a minimum inhibitory concentration (MIC) of 1 μg/ml against Pythium ultimum (Hashidoko et al., 1999). To compare the antibacterial activity of xanthobaccin A to other related tetramates, equesetin and ikarugamycin, were tested against Bacillus subtilis. All three tetramates had antimicrobial activity against B. subtilis with MIC values of 10, 0.62, 2.5 μg/ml for xanthobaccin A, equestin and ikarugamycin, respectively.

Next, we tested the ion-chelating activity of all three tetramates by adding ferric citrate to the solutions of tetramates, followed by HPLC analyses. However, it was hard to observe the corresponding stoichiometry under acidic conditions. Thus, commercially available tetramates equisetin and ikarugamycin were further analyzed for their iron complex under a neutral HPLC condition.

Through extracted ion chromatography, both Fe(III)(equisetin)₃ (m/z 1,171.5973 [M + H]⁺, calc. 1,171.5993 for C₆₆H₉₀N₃O₁₂Fe, Δ 1.7 ppm) and Fe(III)(equisetin)₂ (m/z 798.3720 [M + H]⁺, calc. C₄₄H₅₉N₂O₈Fe 798.374, Δ 2.6 ppm) could be observed (Figure 3). It is not surprising since there are two different enolic tautomers of equisetin which lead to two or three complex structures. A higher electron density on the amide carbonyl compared to the carbonyl on the C-4 position could lead to the observation of a dominant stoichiometry under a neutral pH condition.

Different from equisetin, the relative larger tetramate ikarugamycin could form Fe(III)(ikarugamycin)₂ (m/z 1,008.4909 [M + H]⁺, calc. C₅₈H₇₃FeN₄O₈ 1,008.4897, Δ 1.2 ppm) as the dominant iron complex (Figure 3). During the submission of the manuscript, another study (Yu et al., 2021) reported that HSAF could act as an iron-chelator with a similar chelating pattern. Also, the HSAF-mutant was less susceptible to oxidative stress (Yu et al., 2021).

We hypothesized that there is a similar scenario compared to the copper bis(tenuzonate) structure. Revisiting the former investigation of tenuazonic acid-Fe (III) complex revealed a similar observation, where ions derived from a loss of fragment radicals were detected (Lebrun et al., 1985). In the same investigation, a reduction process of Fe(III)(TA)3 to Fe(II)(TA)3 was proposed in the report (Lebrun et al., 1985). This led us to our hypothesis that the Fenton chemistry follows the complexation of Fe(III) with tetramates.

To further address the function of tetramates, we studied the previous evidence from HSAF in Candida albicans IBT656, where a transcriptomics analysis was carried out. RNA-seq of PTM-treated Candida albicans revealed that HSAF triggered apoptosis via ROS-dependent pathway (Ding et al., 2016). However, the exact mechanism remains unknown.

Nevertheless, this evidence and the iron-chelating activity pointed to a possible link to Fenton chemistry, which describes the oxidative degradation of organic matter by hydrogen peroxide (H₂O₂) in the role of Fe²⁺ under acidic conditions, first discovered by H. J. Fenton in 1894. H₂O₂ is naturally produced by living organisms, from bacteria, algae to human phagocytic cells. Although H₂O₂ has limited reactivity, in the presence of Fe²⁺, it can initiate a very strong reaction to produce high active hydroxyl radicals.

To clarify the iron-reducing activities, we carried out an iron-reducing experiment with ferrozine. As ferrozine forms a pink complex with iron II, which can be analyzed by a UV spectrometer, we determined the iron III reducing activity of the three tetramates using the ferrozine method. All tetramates showed positive results in the ferrozine test. Upon incubation of xanthobaccin A with ammonium iron III citrate and ferrozine, a ferrozine-iron II complex was detected by HPLC-HRMS, which showed a characteristic UV maximum absorption at 562 nm and formula of C₄₀H₂₈FeN₈O₁₂S₄ ([M + H]⁺ m/z 995.0203, Δ-2.5ppm) (Supplementary Figure S5). This suggested the reduction of iron III to iron II by xanthobaccin A. It is likely that the complexation of xanthobaccin with iron II triggers Fenton chemistry (Figure 4) and produces reactive hydroxyl radicals as depicted in Figure 4. As expected, adding the radical scavenger vitamin C reduced the antibacterial effects of tetramates. After adding vitamin C, the inhibition zones were reduced from 23 to 18 mm for equisetin, and 7 to 6 mm for xanthobaccin A, respectively.