Secondary metabolism is commonly associated with morphological development in microorganism. In fact, Actinobacteria and Myxobacteria, both of which possess relatively complex morphology, produce a number of secondary metabolites that include biomedically and industrially useful chemicals.1, 2, 3 Thus, complex life cycles may be indicative of microorganisms with active secondary metabolism. In this context, Thermosporothrix hazakensis SK20-1T, a thermophilic bacterium isolated from ripe compost produced by a field-scale composter,4 attracted our attention because it develops aerial mycelia, which bud to form multiple exospores per mother cell.5 This morphological differentiation is similar to that observed in Streptomyces species, which belong to Actinobacteria, known producers of a variety of secondary metabolites. We therefore postulated that SK20-1T might yield novel metabolites. Indeed, we previously identified new acyloins in fermentation broth from SK20-1T cells.6 In the present study, we identified two new secondary metabolites derived from T. hazakensis SK20-1T.

The SK20-1T strain was grown on agar medium containing 0.1% Bacto Yeast Extract (Becton, Dickinson and Company, Sparks, NV, USA), 0.2% Bacto Tryptone, 0.1% NaCl, 0.1% MgSO4·7H2O and 2% agar at 50 °C for 7 days and then cultured in Difco ISP1 medium (Becton, Dickinson and Company) in a 500-ml Sakaguchi flask for 3 days to generate a seed culture. The seed culture was then transferred to a 5-l jar fermenter (Bioneer C500, B.E. Marubishi, Tokyo, Japan) containing 3 l of fermentation medium (1.0% soluble starch, 0.4% Bacto Yeast Extract and 0.2% Bacto Peptone), and SK20-1T was cultured for 7 days at 55 °C while stirring at 300 r.p.m. The secondary metabolites produced by the bacteria were analyzed by liquid chromatography–mass spectrometry (Shimadzu UFLC/AB SCIEX TripleTOF 5600 System, Tokyo, Japan) using a C18 column (Capcell Pak, 2.0 × 50 mm, Shiseido, Tokyo, Japan) and a solvent gradient of 10–90% CH3CN (containing 0.1% formic acid) over 30 min (flow rate 0.4 ml min−1) at various time points over the course of the 7 days to observe metabolite production over time. Two chromatographic peaks, each thought to be a natural product unrecorded in the Dictionary of Natural Products on DVD ver. 22:2 (CRC Press) based on an investigation using the molecular formulae calculated by ESI–HRMS, were selected for further purification. After 7 days of fermentation, a crude extract was prepared by extracting the culture broth with an equal volume of water-saturated butanol. The crude extract (1.0 g) was fractionated on a Diaion HP-20 flash chromatography column (Nippon Rensui, Tokyo, Japan) using different concentrations of MeOH (20, 40, 60, 80 and 100% MeOH in water, 100 ml each) as the elution solvent. The 100% MeOH fraction was injected into a preparative HPLC system (JASCO, Tokyo, Japan) equipped with a C18 column (PEGASIL ODS column, 20 × 250 mm, Senshu Scientific, Tokyo, Japan) using 60% MeOH containing 0.1% trifluoroacetic acid (TFA) as the eluent at a flow rate of 8 ml min−1 to yield compounds 1 (5 mg) and 2 (1 mg).

Compound 1 was isolated as a pale-yellow solid with the molecular formula C13H8N2O3S, indicating 11 double-bond equivalents. The UV/visible spectrum of compound 1 displayed maxima at 352 nm and 278 nm, which suggested a chromophore including an indole moiety. The 1H NMR spectrum (600 MHz, DMSO-d6) supported the presence of the indole moiety based on the appearance of typical chemical shifts, including a broad doublet signal (δH 12.32) resulting from an exchangeable NH-proton, a downfield-shifted doublet signal (δH 9.11), and four aromatic proton signals (δH 8.27, 7.55, 7.26 (2H)). 1H-1H COSY, HSQC and HMBC analyses of 1 revealed correlations consistent with a three-substituted indole structure (Figure 1). The product ion at m/z 116 in the ESI–MS spectrum, which corresponds to C8H6N+, also supported the presence of an indole group (Figure 1). In addition, the product ion at m/z 144, which corresponds to C9H6NO+, revealed the presence of a 3-carbonyl-indole group. The remaining singlet (δH 8.77) gave HMBC cross signals with three unassigned quaternary carbons (δC 170.1, 162.5 and 148.9). Along with these correlations, the presence of remaining one sulfur and two oxygen atoms in 1 suggested the presence of a thiazole ring connected to carboxylic acid. The existence of a carboxylic acid group in 1 was also suggested by the significant shift in the HPLC retention time due to an ion-pair effect with the mobile phase containing 0.1% TFA in comparison with the mobile phase without TFA. The position of the group is most likely C-4 on the thiazole ring of 1 because the thiazole skeleton is usually generated from cysteine as described below. Although the unassigned quaternary carbon (δC 177.1) with a weak HMBC correlation from H′-2 (δH 9.11) in the indole moiety did not show any correlation with the thiazole moiety in the HMBC experiment, the overall structure of 1 was established as 2-(1′H-indole-3′-carbonyl)-thiazole-4-carboxylic acid by comparing the NMR spectral data, mass fragmentation and UV/visible spectrum of 1 to those of the previously reported 2-(1′H-indole-3′-carbonyl)-thiazole-4-carboxylic acid methyl ester (ITE) isolated from porcine lung.7

Figure 1
figure 1

Structures of 1, 2 and 2-(1′H-indole-3′-carbonyl)-thiazole-4-carboxylic acid methyl ester (ITE). 1H-1H COSY (bold) and HMBC (arrow) correlations and specific mass fragmentations of 1 are also shown. The correlation denoted by a dashed arrow was observed weakly in the HMBC spectrum of 1.

Full size image

Compound 1: UV-visible (MeOH) λmax, nm (log ɛ): 272 (3.95), 278 (3.95), 352 (3.91); 1H NMR and 13C NMR (Table 1, Supplementary Figure S1 and S2); ESI–HRMS: m/z 273.0328 [M+H]+; calculated for C13H9N2O3S, 273.0329.

Table 1 1H (600 MHz) and 13C (150 MHz) NMR spectral data for 1 and 2 in dimethyl sulfoxide-d6
Full size table

Compound 2 was also isolated as a pale-yellow solid. The UV/visible spectrum of 2 showed maxima at 346 nm and 277 nm, similar to those of 1. The molecular formula was determined to be C12H8N2OS based on ESI–HRMS. The molecular weight of 2 is 44 Da smaller than that of 1, suggesting that the structures of 1 and 2 likely differ by the presence of a carboxyl group. In addition, the product ions at m/z 116 and m/z 144 in the ESI–MS spectrum also confirmed the presence of a 3-carbonyl-indole moiety. Based on the UV/visible and MS spectra, 2 was identified as the decarboxylated form of 1. This structure was fully supported by the 1H and two-dimensional NMR spectral data. In contrast to the 1H NMR spectrum of 1, two doublets (δH 8.13 and 8.11) were observed in the 1H NMR spectrum of 2 in agreement with the decarboxylation of 1. Furthermore, these two proton signals correlated with the quaternary carbon (δC 170.2) of the thiazole ring in the HMBC spectrum of 2. Based on these spectral analyses, the complete structure of 2 was defined as 2-(1′H-indole-3′-carbonyl)-thiazole.

Compound 2: pale-yellow solid; UV-visible (MeOH) λmax, nm (log ɛ): 271 (3.86), 277 (3.86), 346 (3.75); 1H NMR and 13C NMR (Table 1, Supplementary Figure S3 and S4); ESI–HRMS: m/z 229.0429 [M+H]+; calculated for C12H9N2OS, 229.0430.

Neither 1 nor 2 exhibited antimicrobial activities against Candida albicans NBRC1594 at concentrations as high as 100 μM and Micrococcus luteus ATCC9341 at concentrations as high as 10 μM. Similarly, cytotoxicity tests against human ovarian carcinoma SKOV3 cells, mesothelioma Meso-1 cells and T lymphoma Jurkat cells revealed that high concentrations of 1 induced slight cytotoxicity against only the Jurkat cell line (approximately 30% inhibition at 50 μM, Supplementary Figure S5).

Natural small molecules with indole and thiazole moieties have been isolated from various biological sources, including animals,7 plants,8 bacteria,9, 10 fungi11 and marine sponges.12 However, small molecules such as 1 and 2 that contain a 3-carbonyl-indole moiety have not been explored thoroughly. In particular, the 2-(1′H-indole-3′-carbonyl)-thiazole carbon skeleton is rarely encountered among natural products reported in the literature; according to the Dictionary of Natural Products on DVD ver. 22:2, the only other naturally derived 2-(1′H-indole-3′-carbonyl)-thiazole is ITE, which was isolated from porcine lung.7 Symbiotic or enteric bacteria are often suspected to be the biosynthetic source of secondary metabolites isolated from animals.13, 14 Therefore, we also suspect that a bacterial endosymbiont in porcine lungs might be responsible for producing ITE or a precursor of ITE. In fact, we detected ITE, as well as 1 and 2, in the SK20-1T culture (Supplementary Figure S6). Intriguingly, 2 was discovered in cultures of a myxobacterial strain 706, which was recently isolated from compost in Germany.15

The thiazole moieties found in natural products are usually generated through the oxidation of a thiazoline ring formed by heterocyclization between the sulfhydryl group of cysteine and the preceding carbonyl group.16 Therefore, 1 and 2 are presumably generated through the oxidation of the thiazoline formed by the cyclization of an indole-3-glyoxylamide intermediate synthesized by condensation between cysteine and indole-3-glyoxylic acid (Supplementary Figure S7). The elucidation of the biosynthesis of 1 and 2, which will likely lead to the identification of novel metabolic pathways in T. hazakensis SK20-1T, will be the aim of our next study.