Streptomyces tanashiensis BYF-112 was isolated from the workers of O. formosanus as described in our previous work and was preserved on YMS (1.0 g L^–1 KNO3, 1.5 g L^–1 yeast extract, 10 g L^–1 soluble starch, and 20 g^–1 agar) slants at 4°C until use (Long et al., 2022). From a 7-day old YMS solid culture plate, a conidial spore suspension was prepared by adding 15 ml of distilled water containing 0.2% Tween 80 (w/v) under agitation by glass rod. To study the effect of iron, aliquots (2 ml) of spore suspension were added into 100 ml of YMS production media supplemented with deionized water dissolved FeSO₄•7H₂O at the concentrations of 50 mg/l. Equal amount of deionized water was added to the control group. The filter of each fermentation broth was extracted three times with equal amount of ethyl acetate (EtOAc) after fermenting at 28°C in a shaker rotating at 180 rpm for 7 days. The EtOAc extracts were fast analyzed by thin-layer chromatography (TLC).

Solution (250 μg/ml) of yellow pigments (YP) and the green pigments (GP) were dissolved in DMSO and deionized water respectively, and stored in the dark. The pigments stability under heat, light and metal ions were performed by the methods described in detail previously with slight modification (Zhu et al., 2020; Wu et al., 2022). Heat stability: the tested tubes were placed into 101-3A electric thermostatic drying oven (Changzhou, Jiangsu, China) at 40, 60, 80, 100, and 120°C for 2 h. The samples were then removed and quickly cooled to room temperature in an ice bath for further analysis. Stability under condition with light and dark: the effect of light on pigment stability was determined by after exposure to indoor incandescent light and stored at dark place for 1–7 days. Stability under condition with metal ions: the effect of metal ions (Ca²⁺, Zn²⁺, Mg²⁺, Na⁺, and K⁺, 1 mmol/l) on the pigment was determined in the dark for 2 days.

The retention rate of YP and GY were calculated at the maximum absorbance 407 and 700 nm respectively, using UV–Visible spectrophotometer (UNICO UV-2800) according to following formula (Wu et al., 2022).

Retention rate of YP/GP = A1/A0 × 100%.

A1/A0: the absorbance of YP/GP after/before treatment.

The fermentation procedure was described as the previously reference with slight modifications (Zhang et al., 2020). The fresh spores of S. tanashiensis BYF-112 cultivated on YMS agar plates at 28°C for 3–5 days were inoculated into 250 ml Erlenmeyer flask containing 100 ml YMS liquid culture medium (10 g soluble starch, 1.5 g yeast extract, 1.0 g KNO₃, 1,000 ml deionized water). After 3 days of incubation at 28°C on rotary shakers at 180 rpm, 30 ml of cultured liquid was transferred as a seed into Erlenmeyer flasks (1,000 ml), each containing 400 ml YMS broth (or YMS broth supplemented with FeSO₄•7H₂O). The cultures were continuously shaken at 180 rpm for 7 days at 28°C to produce pigments.

The yellow culture broth (16 l) of BYF-112 grown on a YMS liquid medium was filtered by gauze to obtain the supernatant, which was then extracted with ethyl acetate (EtOAc, 3 × 16 l) at room temperature. The EtOAc phase was evaporated in vacuo to afford crude extract (6.6 g). The crude extract was subjected to a silica gel column eluting with a stepwise gradient of CH₂Cl₂/MeOH (100, 0–100:16, v/v) to give five fractions (Fr1–Fr5). Compound 6 (15 mg) was crystallized from the CH₂Cl₂ solution from the Fr2 (CH₂Cl₂/MeOH, 100:1, v/v). Fr2 was repeatedly chromatographed over a Sephadex LH-20 column (MeOH) to yield metabolite 5 (6.0 mg). Fr3 (CH₂Cl₂/MeOH, 100:2, v/v) was further fractionated on a silica gel column, eluting with (CH₂Cl₂/MeOH, 100:0, 100:1 and 100:2, v/v) to give three subfractions (R1–R3). Subfractions R2 and R3 were loaded onto a Sephadex LH-20 column eluting with MeOH to give compounds 1 (40.2 mg) and 3 (5.2 mg), respectively. Compound 2 (30.1 mg) was crystallized from the methanol-dichloromethane binary solvent solution of Fr4 (CH₂Cl₂/MeOH, 100:4). The remaining part of Fr4 was further purified by a Sephadex LH-20 eluting with MeOH to yield compounds 7 (8.8 mg) and 8 (7.9 mg). Fr5 (CH₂Cl₂/MeOH, 100:16, v/v) was also loaded onto a Sephadex LH-20 column to yield compound 4 (8.6 mg).

The green culture broth (8 l) of BYF112 grown on a YMS liquid media supplemented with FeSO₄•7H₂O was centrifuged to afford the liquid phase at 6,000 rpm for 10 min. The liquid phase was passed through a macroporous resin column D-101 (Shangxi Lebobiochem Co., Ltd., China), rinsed with deionized water until the color of the effluent turned colorless, and then eluted with methanol. The organic fraction was evaporated in vacuo to afford crude extract (2.6 g). The crude extract was subjected to a silica gel column eluting with a stepwise gradient of CH₂Cl₂/MeOH (100:4, 100:8, and 100:16, v/v) to give two fractions (Fr1–Fr2). Fractions Fr1 and Fr2 were further purified by a Sephadex LH-20 eluting with MeOH to yield compounds 9 (20.1 mg) and 10 (10.9 mg), respectively.

Structural identifications of the metabolites were made on the basis of the spectroscopic analysis. NMR spectra were recorded on an Agilent II DD2 instrument operating at 600 MHz for ¹H and 150 MHz for ¹³C, while 2D spectra (COSY, HMQC, HMBC, and DEPT) were obtained using standard Agilent software. Chemical shifts were given in parts per million (δ) downfield from the TMS internal standard. A Mariner Mass 5,304 instrument was used to measure the HR-ESI-MS spectra.

The metabolites 1–4 were evaluated for their cytotoxic activities against human malignant melanoma cell line (A375), human ovarian cancer cell line (SKOV-3), human breast cancer cell line (MDA-MB231), human gastric cancer cell line (MGC-803) and human normal cell line (L-02) with the MTT method (Wang et al., 2022). The cell lines were maintained in Dulbecco’s modified eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS) along with 1% penicillin and streptomycin. Cells were grown at 37°C in an atmosphere with 5% CO₂ and then seeded in 96-well plates and permitted to grow for 20 h. Then tested compounds at preset concentrations were added. The 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT; 20 μl, 5 mg/ml) was added into each well after a 48 h exposure period. After an additional 4 h further incubation, the medium was replaced by DMSO (150 μl), then shaking the plates for 15 min. The absorbance at 490 nm were recorded. The percentage of cell viability was calculated using the following formula: Cell viability (%) = the OD of tested group/the OD of CK × 100%. Finally, IC₅₀ values were determined by EXCEL. The azithromycin was used as positive control.

The Staphylococcus aureus (ATCC6538), methicillin-resistant S. aureus (MRSA, SA2), Escherichia coli (ATCC8739) and Micrococcus tetragenus (ATCC35098) donated by Professor Ting Xue from Anhui Agricultural university were used as a target pathogen for the examination of antibacterial activity of the metabolites. The disc diffusion (Özogul et al., 2022) and minimum inhibitory concentrations (MICs; Cui et al., 2022) methods were performed to evaluate the antibacterial activities of the new natural metabolites 1, 2, and 4. Filter paper disks (6 mm) with metabolite dissolved in DMSO in a concentration (30 μg/filter paper) were added to the culture medium, and the plates were incubated at 37°C for 12 h. Filter papers with DMSO and levofloxacin were set as negative and positive controls, respectively. All experiments were performed in triplicate, and data were shown as mean values standard deviation. And the MICs of the compounds were determined by the method as we previously described (Cui et al., 2022).

Effect of Iron on the metabolic profile of S. tanashiensis BYF-112 was displayed in Supplementary Figure S1A. The result showed that the color of the fermentation broth exhibited a great difference between YMS and YMS culture broth treated with iron, which presented yellow and green, respectively. Furthermore, TLC profiles indicated that two green spots were observed in iron treated extract, and only yellow spots were seen in the extracts of iron free cultures (Supplementary Figure S1B). These results showed that iron had a great influence on the metabolite profiles of the strain BYF-112.

Previous reports showed that the color stability of natural pigment was affected by external factors such as heat, light, and metal ions (Zhang et al., 2006; Zhu et al., 2020). Therefore, the stability of YP and GP under different environmental conditions (heat, light, and metal ions) were evaluated in this study.

As shown in Figures 1A,B, both YP and GP presented strong heat stability with the residual rate of 89.44–98.22% under exposure to 40–100°C for 2 h at the concentration of 250 μg/ml. The result indicated that microbial pigments might be applied in the food industry due to their superior thermal stability, compared to other natural pigments (Vendruscolo et al., 2013; Teran Hilares et al., 2018).

The color stability of YP and GP stored in light and dark was conducted under being treated for 1–7 days at 25°C (Figures 1C,D). The results showed that the YP retention rate (86.23%) in the indoor incandescent light decreased slightly, compared to that of in the dark for 7 days (99.44%), while the GP remained stable under light and dark for 7d with the retention rate of 91.57 and 91.42%, respectively. Hence, both pigments can be preserved in light to some extent.

The effect of metal ions (Ca²⁺, Zn²⁺, Mg²⁺, Na⁺, K⁺, 1 mmol/l) on the pigment was determined in the dark for 2 days. The results showed that both YP (Figure 1E) and GP (Figure 1F) were stable with the retention rate of more than 92% under the treatments of different metal ions.

Four new natural compounds 1–4, along with six known metabolites 5–10 were isolated from the culture of S. tanashiensis BYF-112. Chemical structures of compounds 1–10 are shown in Figure 2.

Compound 1 was obtained as a yellow brown powder, and its molecular formula C₁₃H₁₀N₂O₃ was deduced from HR-ESI-MS data (m/z [M + H]⁺ 243.0757, which was quite in accordance with ¹H and ¹³C NMR data (Supplementary Figures S2–S4). ¹H NMR and ¹³C NMR data were as following: ¹H NMR (DMSO-d₆, 600 MHz) δ_H 4.55 (2H, d, J = 5.6 Hz), 5.30 (1H, m), 6.33 (H, s), 6.33 (H, s), 6.74 (2H, brs), 7.38 (1H, d, J = 8.3 Hz), 7.43 (1H, d, J = 8.3 Hz), 7.60 (1H, brs); ¹³C NMR (DMSO-d₆, 150 MHz) δ_C 62.5 (CH₂), 98.8 (CH), 103.6 (CH), 115.9 (CH), 125.7 (CH), 127.5 (CH), 133.8 (C), 140.2 (C), 141.0 (C), 147.6 (C), 148.5 (C), 149.2 (C), 180.6 (C). The data were almost no different from 2-amino-8-hydroxymethyl-3H-phenoxazin-3-one, a semisynthetic derivative of the pigment aminophenoxazinone described in the literature (Pasceri et al., 2013). However, to our knowledge, the compound was first discovered as a new natural product.

Compound 2 was purified as a yellow brown powder, and its molecular formula C₁₅H₁₂N₂O₄ was deduced from HR-ESI-MS ion peak at m/z [M + H]⁺ 285.0860, which was consistent with ¹H and ¹³C NMR data (Table 1; Supplementary Figures S5–S11). The ¹H NMR and ¹³C NMR spectra of 2 were resembled to those of 1, except an acetyl unit signal (δ_C 170.1, 20.6; δ_H 2.09, s, 3H) at position 12 in 1 appeared to be COCH₃, which corresponded to the increase in molecular of compound 2 by 42 amu compared to 1. Further structure elucidation of 2 was established by the 1D and 2D NMR spectra. The ¹H NMR data exhibited the protons resonances of a trisubstituted aromatic ring proton signals at δ_H 7.49, (1H, d, J = 8.4 Hz, H-6), δ_H 7.44, (1H, d, J = 9.5 Hz, H-7) and δ_H 7.69, (1H, s H-9), another two aromatic protons at δ_H 6.34 (1H, s, H-1), 6.36 (1H, s, H-4), one oxygenated methylene proton peak at δ_H 5.16 (2H, s, H-11), one methyl protons of acetyl group resonances line at δ_H (2.09, 3H, s, H-14) and one α-amino proton signal at δ_H 6.81 (2H, d, J = 6.7 Hz, NH₂-2). The observation agreed with its ¹³C NMR spectra, which displayed as anticipated one methyl carbon (δ_C 20.6, CH₃, C-14), one methylene carbon (δ_C 64.6, CH₂, C-11), one carbonyl carbon (δ_C 180.1, CO, C-2), and 11 aromatic carbon resonance lines (δ_C 98.4, 142.4, 103.4, 148.8, 141.4, 115.9, 128.3, 133.3, 127.2, 133.4, and 148.4). These substructures were pieced together by the 2D NMR experiments including its ¹H–¹H COSY spectrum highlighting the two coupling sequences (11-H to 7-H and 9-H), and its HMBC spectrum (Supplementary Figure S12) indicating the key correlations of H-1 to C-3 and C-4a, and of H-4 to C-2 and C-10a, and of H-6 to C-5a, C-8 and C-9a, and of H-7 to C-11 and C-5a, and of H-9 to C-5a and C-11, and of H-8 to C-8 and C-9, and of 2-NH₂ to C-1 and C-3. Thus, the structure of metabolite 2 was determined as a new aminophenoxazinone derivative and named exfoliazone B.

Compound 3 was purified as a red yellow powder, and its molecular formula was evidenced to be C₂₂H₁₉N₃O₅ from the quasimolecular ion peak at m/z [M + Na]⁺ 428.1217 in its HR-ESI-MS spectrum, which was consistent with ¹H and ¹³C NMR data (Table 1; Supplementary Figures S13–S19). The ¹H NMR (DMSO-d₆, 600 MHz) spectrum of 3 exhibited five aromatic protons signals (δ_H 6.04, 6.36, 7.45, 7.38 and 7.58) and one imino proton resonance (δ_H 7.60) and one oxygenated methylene signal (δ_H 4.54 and 5.33), and the ¹³C NMR (DMSO-d₆, 125 MHz) spectrum displayed 11 aromatic carbon resonance lines (δ_C 96.3, 145.7, 103.3, 149.3, 140.7, 115.6, 127.2, 140.0, 125.39, 133.5, 149.9), one carbonyl carbon signal (δ_C 179.9) and one methylene carbon (δ_C 63.2). These data showed almost no difference with exfoliazone described in the literature (Zhang et al., 2015). Apart from these signals of exfoliazone skeleton, there were six aromatic carbon (δ_C 128.5, 121.1, 126.5, 147.9, 115.9, and 123.5), two aromatic proton (δ_H 7.68 and 6.80), one carbonyl carbon (δ_C 169.1), one methyl carbon (δ_C 23.6), one methyl proton (δ_H 2.05), one methylene carbon (δ_C 45.2) one methylene proton (δ_H 4.33), one imino proton (δ_H 7.68) and one hydroxyl proton (δ_H 9.73) as shown in the Table 1. These unassigned data indicated that a subgroup N-(2-hydroxy-5-methyl-phenyl)-acetamide were existed in 3, which corresponded to be increase in molecular weight of 3 by 164 amu compared to that of exfoliazone. These hypotheses were verified by the key HMBC (Fig. S12) correlations in compound 3. Thus, the metabolite 3 was identified as a new natural product, named exfoliazone C.

Compound 4 was obtained as a yellow red powder, and its molecular formula C₂₁H₂₂N₂O₉ was deduced from HR-ESI-MS (m/z [M + H]⁺ 447.1378, [M + Na]⁺469.1205 calcd for C₂₁H₂₃N₂O₉, 447.1404 and C₂₁H₂₂N₂O₉Na, 469.1223), which was consistent with the ¹H and ¹³C NMR data (Table 2; Supplementary Figures S20–S26). The ¹H NMR (DMSO-d₆, 600 MHz) spectrum of 4 exhibited five aromatic protons signals δ_H 6.45 (1H, s, H-1), 6.44 (1H, s, H-4), 7.54 (1H, d, J = 8.4 Hz, H-6), 7.52 (1H, d, J = 9.5 Hz, H-7) and 7.76 (1H, s, H-9; one methylene proton δ_H 5.18 (2H, s, H-11); one methyl proton δ_H 2.08 (3H, s, H-14) and one imino proton δ_H 6.96 (1H, d, J = 6.7 Hz, NH-8′); and the ¹³C NMR (DMSO-d₆, 125 MHz) spectrum displayed 11 aromatic carbon resonance lines δ_C 99.6 (CH, C-1), 145.1 (C, C-2), 103.4 (CH, C-4), 148.9 (C, C-4a), 141.7 (C, C-5a), 116.0 (CH, C-6), 129.3 (CH, C-7), 133.5 (C, C-8), 127.5 (CH, C-9), 133.3 (C, C-9a), 148.4(C, C-10a), one carbonyl carbon signal 179.6 (CO, C-2), one methylene carbon signal 64.6(CH₂, C-11)and one methyl carbon signal 20.6 (CH₃, C-14). These proton and carbon signals of substructures were very similar to that of metabolite 1 suggesting that 4 might be a derivative of 1.The residual proton signals δ_H 4.49 (1H, m, H-1′), 3.37 (1H, m, H-3′), 3.15 (1H, m, H-4′), 4.98 (1H, d, J = 5.2 Hz, OH-4′), 3.30 (1H, m, H-5′), 5.06 (1H, d, J = 4.3 Hz, OH-5′), 3.34 (1H, m, H-6′), 5.23 (1H, d, J = 5.3 Hz, OH-6′), 3.70 (1H, m, H-7), 3.46 (1H, m, H-7′), 4.50 (1H, d, J = 5.1 Hz, OH-7′) and carbon signals δ_C 83.3 (C-1′), 77.0 (C-3′), 70.0 (C-4′), 78.1 (C-5′), 72.2 (C-6′), 60.8 (C-7′) indicated that a hexose were existed in 4. These substructures were pieced together by the 2D NMR experiments including its ¹H-¹H COSY spectrum highlighting the two coupling sequences (11-H to 7-H and 9-H), and its HMBC spectrum (Supplementary Figure S12) indicating the key correlations of H-1 to C-2 and C-3, and H-4 to C-3, C-2 and C-10a, and H-6 to C-5a and C-8, and H-7 to C-11 and C-5a, and H-9 to C-11 and C-5a, and H-11 to C-7 and C-9, and H-14 to C-1, and H-8′ to C-1, C-3, C-1′ and C-6′. Thus, the structure of metabolite 3 was determined as a new glycosylated derivative of 1 and named exfoliazoneglycoside.

The known compounds 5–10 were identified as exfoliazone (5) (Zhang et al., 2015), N-(2-Hydroxy-5-methylphenyl acetamide (6) (Suzuki et al., 2006a), 3-acetylamino-4-hydroxybenzaldehyde (7) (Suzuki et al., 2006b), 3-acetylamino-4-hydroxybenzonic acid (8) (He et al., 2018), viridomycin A (9) and viridomycin F (10) (Omura et al., 1999), by comparing the NMR and MS data with those reported in the literature. It was reported that the known compounds N-(2-Hydroxy-5-methylphenyl) acetamide (6), 3s-acetylamino-4-hydroxybenzaldehyde (7) and 3-acetylamino-4-hydroxybenzonic acid (8) were derived from the precursor 3-amino-4-hydroxybenzoic acid (3,4-AHBA; Gould et al., 1996). The remaining aminophenoxazinones were presumably synthesized from the precursor 3,4-AHBA in the gene cluster, which contained a key tyrosinase-like copper-containing monooxygenase responsible for the C-nitrosation (Noguchi et al., 2010). Streptomyces has been reported as pigments producer (Chatragadda et al., 2019b; Zhu et al., 2020). Our results enriched the pigment molecules diversity produced by the Streptomyces.

In vitro toxicity of the compounds 1–4 was evaluated against the cell lines using the MTT method (Table 3). The results showed that the metabolites 2–4 showed no cytotoxic activities against the human normal liver cell line L-02 with the IC₅₀ values of more than 100 μM. These safety performance of 2–4 might be an advantage for applying in the food industry. Furthermore, the metabolite 2 showed selective cytotoxic activities toward human ovarian cancer cell line SKOV-3 and gastric cancer cell line MGC-803 with the IC₅₀ values of 75.78 and 69.88 μM, respectively, which were lower than those of positive azithromycin with the IC₅₀ values of 0.41 and 1.57. The compound 1 exhibited inhibitory activity against the tested human cancer cell lines (human malignant melanoma cell line A375, SKOV-3, breast cancer cell line MDA-231 and MGC-803) with the IC₅₀ values of 22.33–67.70 μM. However, it is low toxicity against the normal cell L-02 with the IC₅₀ value of 72.01 μM would be a limitation in using as a food pigment.

The disc diameters of zone of inhibition (ZOI) values of metabolites (1, 2 and 4) against bacterial pathogens were presented in Table 4; Supplementary Figure S27. The results showed that metabolite 1 possessed strong antibacterial activity against methicillin-resistant S. aureus (MRSA) with the ZOI of 15.3 mm, which was equal to that of referenced levofloxacin (ZOI = 15.2 mm). In addition, the metabolite 1 showed moderate antibacterial activities against Gram-positive S. aureus (ZOI = 17.2 mm) and M. tetragenus (ZOI = 15.mm), which were weaker than those of referenced levofloxacin with the ZOI values of ZOI = 28.9 and 24.8 mm, respectively. And in the MICs test (Table 4), the compound 1 exhibited potential antibacterial activities against M. tetragenus, S. aureus and MRSA in the MIC tests with the MIC values of 12.5, 12.50, and 25.0 μg/ml, which were comparable to those of positive gentamycin sulfate with the MIC values of 3.13, 3.13, and 6.25 μg/ml, respectively. However, both compounds 2 and 4 did not showed antibacterial activities against the test strains. By comparing the structure of compounds 1 and 2, the oxhydryl substituent at C-11 in 2 could reduce antibacterial activities against tested bacteria. It indicated that the 11-position hydroxyl group might play a very important role in inhibiting the growth of the Gram-positive bacteria.