Introduction

Cancer stem cells (CSCs) are one of the major causes of tumor recurrence because of their high malignancy and resistance to conventional anti-cancer agents [1, 2]. We previously established a new “CSC spheres” culture system using medium for CSCs in a three-dimensional (3D) culture; the culture had elevated CSC markers (Sox2, Nanog, Oct4) and resistance to anticancer agents such as camptothecin and paclitaxel [3]. Our recent screening from microbial metabolites resulted in the isolation of a novel compound, named streptospherin A (1; Fig. 1) from Streptomyces sp. KUSC-240, which was an inhibitor of CSC sphere formation and growth [4]. Structural analysis of 1 revealed a characteristic skeleton with pentasubstituted benzene and tetrahydropyran moieties. These findings led us to further isolation and structural determination of derivatives from Streptomyces sp. KUSC-240. Because of the presence of the unique pentasubstituted benzene moiety in the left C-2’–C-12’ fragment of 1, LC-MS analysis in precursor ion scan mode (negative mass of 179.1 as product ion mass) was performed for the EtOAc extract of Streptomyces sp. KUSC-240 to efficiently find other derivatives.

Fig. 1
figure 1

Structures of streptospherins A–F (1–6)

Full size image

In this study, we found several derivatives containing the left fragment of 1 and isolated five novel compounds, named streptospherins B–F (2–6). We describe the isolation, structural determination, stereochemical analysis, and inhibitory activities in CSC sphere formation of 2–6. We also describe a plausible biosynthesis pathway of 3–5 by bioinformatic analysis of the whole genome sequence of Streptomyces sp. KUSC-240. Our results indicate that streptospherins might be potential therapeutic agents for targeting CSCs.

Materials and methods

Isolation of streptospherins B–E (2–5)

Streptomyces sp. KUSC-240, isolated from a soil sample of Gunma, Japan [4], was cultured in GY production medium [glucose 5.0%, yeast extract 0.4%, and CaCO3 0.25%] at 30 °C for 4 days. After fermentation, the culture medium (3 L) was extracted with the same volume of acetone, filtered and concentrated in vacuo. The remaining culture broth was adjusted to pH 7.0 and extracted with EtOAc twice (3 L × 2) to give a brown residue (742.5 mg). The extract was subjected to silica gel column chromatography (Silica gel 60 N, spherical neutral, particle size 40–50 μm; Kanto Chemical Co. Inc., Tokyo, Japan) using a CHCl3–MeOH stepwise system to give two fractions A and B. Fraction A (CHCl3:MeOH = 10:1; 141.8 mg) was subjected to silica gel column chromatography using a n-hexane–EtOAc stepwise system. The n-hexane:EtOAc = 1:9–0:1 fraction (28.0 mg) was purified by preparative HPLC using a COSMOSIL Cholester column (Nacalai Tesque, Inc., Kyoto, Japan; ϕ20 × 250 mm; 19 mL min−1) with 37% MeCN in H2O. Further purification was performed by preparative TLC (PLC silica gel 60 F254, 0.5 mm; Merck KGaA, Darmstadt, Germany) with EtOAc to give streptospherin B (2; 1.4 mg). Fraction B (CHCl3:MeOH = 100:1; 209.2 mg) was subjected to silica gel column chromatography with n-hexane:EtOAc = 3:1. The streptospherins-rich fraction (65.8 mg) was purified by preparative HPLC using a COSMOSIL Cholester column (ϕ20 × 250 mm; 19 mL min−1) with 57% MeCN in H2O to give three fractions B1–B3. Fraction B1 (1.9 mg) was further purified by preparative TLC with CHCl3:MeOH = 50:1 to give streptospherin C (3; 1.2 mg). Fraction B2 (27.6 mg) was further purified by preparative TLC with n-hexane:EtOAc = 2:1 to give streptospherin D (4; 19.8 mg). Fraction B3 (3.5 mg) was further purified by preparative TLC with n-hexane:EtOAc = 2:1 to give streptospherin E (5; 2.3 mg).

Isolation of streptospherin F (6)

Streptomyces sp. KUSC-240 was cultured in A-16 production medium [glucose 2.0%, pharmamedia 1.0%, and CaCO3 0.5%] at 28 °C for 7 days. After fermentation, the culture medium (3 L) was extracted with the same volume of acetone, filtered and concentrated in vacuo. The remaining culture broth was adjusted to pH 9.0 and extracted with EtOAc twice (3 L × 2) to give a brown residue (516.9 mg). The extract was subjected to silica gel column chromatography using a CHCl3–MeOH stepwise system. The CHCl3:MeOH = 50:1 fraction (45.0 mg) was subjected to silica gel column chromatography using a n-hexane–EtOAc stepwise system. The n-hexane:EtOAc = 3:2 fraction (19.1 mg) was purified by preparative HPLC using a COSMOSIL Cholester column (ϕ10 × 250 mm; 5 mL min−1) with 42% MeCN in H2O. Further purification was performed by preparative TLC with n-hexane:EtOAc = 2:3 to give streptospherin F (6; 0.7 mg).

Streptospherin B (2): white powder; [α]20D + 47.4 (c = 0.30, MeOH); UV λmax (MeOH) nm (logε): 220 (3.89), 240 (sh), 289 (3.83), 333 (3.41); IR (KBr) cm–1: 3341, 2958, 2930, 2871, 1619, 1464, 1421, 1381, 1338, 1307, 1295, 1274, 1234, 1213, 1164, 1065, 1039, 1028, 931, 879, 828; ESI–MS m/z: 437.2558 (calcd for C24H37O7–: 437.2545). 1H and 13C NMR data are summarized in Tables 1 and 2.

Table 1 1H NMR data (600 MHz) for streptospherins B–F (2–6) in CDCl3
Full size table
Table 2 13C NMR data (150 MHz) for streptospherins B–F (2–6) in CDCl3
Full size table

Streptospherin C (3): white powder; [α]20D + 25.3 (c = 0.10, MeOH); UV λmax (MeOH) nm (logε): 222 (4.01), 240 (sh), 285 (3.99), 331 (3.72); IR (KBr) cm–1: 3317, 2961, 2952, 1618, 1446, 1420, 1380, 1314, 1232, 1166, 1073, 1039, 980, 906, 803; ESI–MS m/z: 391.2507 (calcd for C23H35O5–: 391.2490). 1H and 13C NMR data are summarized in Tables 1 and 2.

Streptospherin D (4): white powder; [α]20D + 18.7 (c = 1.0, MeOH); UV λmax (MeOH) nm (logε): 221 (3.98), 240 (sh), 284 (3.97), 331 (3.72); IR (KBr) cm–1: 3366, 2958, 2932, 2871, 1619, 1464, 1419, 1380, 1336, 1318, 1307, 1272, 1236, 1216, 1163, 1050, 879, 822; ESI–MS m/z: 405.2642 (calcd for C24H37O5–: 405.2646). 1H and 13C NMR data are summarized in Tables 1, 2, and S1.

Streptospherin E (5): white powder; [α]20D + 6.7 (c = 0.20, MeOH); UV λmax (MeOH) nm (logε): 219 (3.95), 240 (sh), 288 (3.81), 333 (3.40); IR (KBr) cm–1: 3357, 2957, 2928, 2870, 1620, 1455, 1418, 1381, 1337, 1309, 1275, 1239, 1201, 1162, 1035, 994, 832; ESI–MS m/z: 419.2796 (calcd for C25H39O5–: 419.2803). 1H and 13C NMR data are summarized in Tables 1 and 2.

Streptospherin F (6): white powder; [α]20D + 12.2 (c = 0.036, MeOH); UV λmax (MeOH) nm (logε): 221 (3.91), 240 (sh), 291 (3.93), 334 (3.54); IR (KBr) cm–1: 3325, 2956, 2919, 2871, 2848, 1622, 1540, 1465, 1419, 1381, 1310, 1273, 1236, 1217, 1164, 1015, 969, 877; ESI–MS m/z: 361.1979 (calcd for C19H30O5Na+: 361.1985). 1H and 13C NMR data are summarized in Tables 1 and 2.

Synthesis of MTPA ester 7

To a solution of 4 (10.0 mg, 24.6 μmol) in 200 μL CH2Cl2 was added (S)-α-methoxy-α-(trifluoromethyl)phenylacetic acid (MTPA; 17.3 mg, 73.8 μmol), 3-ethylcarbodiimide hydrochloride (EDCI-HCl; 14.1 mg, 73.8 μmol), and 4-dimethylaminopyridine (DMAP; cat.) for 5 h at room temperature. After the solution was dried in vacuo, the residue was subjected to preparative HPLC using a COSMOSIL Cholester column (ϕ10 × 250 mm; 5 mL min−1) with 75% MeCN in H2O to yield 7a (0.95 mg, 1.5 μmol, 6%) as a white powder.

To a solution of 4 (5.0 mg, 12.3 μmol) in 200 μL CH2Cl2 was added (R)-α-methoxy-α-(trifluoromethyl)phenylacetic acid (MTPA; 5.8 mg, 24.6 μmol), 3-ethylcarbodiimide hydrochloride (EDCI-HCl; 4.7 mg, 24.6 μmol), and 4-dimethylaminopyridine (DMAP; cat.) for 1 h at room temperature. After the solution was dried in vacuo, the residue was subjected to preparative HPLC using a COSMOSIL Cholester column (ϕ10 × 250 mm; 5 mL min−1) with 75% MeCN in H2O to yield 7b (1.48 mg, 2.4 μmol, 19%) as a white powder.

ESI–MS m/z: 621.3059 for 7a and 7b (calcd for C34H44F3O7–: 621.3045). 1H NMR data are summarized in Table S2.

Cell culture

HT29 (human colorectal adenocarcinoma) cells were purchased from the American Type Culture Collection. HT29 cells were maintained in Roswell Park Memorial Institute (RPMI)-1640 medium (Nissui Pharmaceutical Co. Ltd., Tokyo, Japan) supplemented with 10% heat-inactivated fetal bovine serum at 37 °C in a humidified atmosphere containing 5% CO2.

Sphere formation assay

Cells were resuspended in 3D Tumorsphere Medium XF (PromoCell GmbH, Heidelberg, Germany) and cultured on poly(2-hydroxyethyl methacrylate) [poly-HEMA (Sigma-Aldrich/Merck KGaA, Darmstadt, Germany)]-coated 6-well plates. A half volume of fresh medium was added every 3 days. After 7 days of culture, spheres were collected and dispersed with Accutase (Nacalai Tesque, Inc.), resuspended in the XF medium, and reseeded on poly-HEMA-coated 96-well plates. Simultaneously, the test compound was added. After 3 days of treatment, spheres were observed under a microscope. Cell growth was measured by a WST-8 assay using Cell Counting Kit-8 (Dojindo Laboratories, Kumamoto, Japan).

Results and discussion

Streptospherins B–F (2–6) were obtained as a white powder with high purity (Fig. S1). According to high-resolution ESI–MS, 1H NMR, and 13C NMR data, the molecular formula were found to be C24H38O7 (2; m/z: 437.2558, calcd for C24H37O7–: 437.2545), C23H36O5 (3; m/z: 391.2507, calcd for C23H35O5–: 391.2490), C24H38O5 (4; m/z: 405.2642, calcd for C24H37O5–: 405.2646), C25H40O5 (5; m/z: 419.2796, calcd for C25H39O5–: 419.2803), and C19H30O5 (6; m/z: 361.1979, calcd for C19H30O5Na+: 361.1985) (Fig. S2). The UV absorption spectra of 2–6 were identical to that of 1, which suggested that the presence of similar partial structure. 1H, 13C, and HMQC NMR analyses revealed the presence of the same pentasubstituted benzene moiety (C-1’ to C-12’) as in 1.

The full planar structure of streptospherin D (4), which had the highest yield among these compounds 2–6, was elucidated from 1D and 2D NMR spectra. The significant differences between 4 and 1 were the presence in spectrum of 4 of downfield shift protons (δH: 5.42 for H-3) and carbons (δC: 127.5 for C-3 and 136.2 for C-4) and the absence of a hemiacetal moiety (δC: 100.1 for C-3 in 1). These data indicated that the tetrahydropyran ring in 1 was altered to an alkenyl group in 4. The 1H-1H COSY spectrum of 4 revealed the proton-proton correlations from H-1 to H-3 and from H-5 to H-11 (Fig. S20). The linkage from C-12 to C-3/C-4/C-5 was determined by HMBC correlations from H-12 to C-3/C-4/C-5 (Fig. S21). NOESY correlations from H-3 to H-5 indicated that the C-3–C-4 double bond is in the E configuration (Fig. S22). From these structural analyses, the planar structure of 4 was elucidated (Fig. 2a).

Fig. 2
figure 2

Structural elucidation of streptospherin D (4). a COSY, key HMBC, and NOESY correlations of 4. b ΔδH values of MTPA ester of 4. c Acetonide formation of 1,3-diol at C-5 and C-7 positions of 4. d Newman projection at the C-8 position of 4. e Plausible absolute configuration of 4

Full size image

The absolute configuration at C-5 in 4 was determined by the modified Mosher’s method [5]. (R) or (S)-MTPA acid [α-methoxy-α-(trifluoromethyl)phenylacetic acid] was condensed with the allylic hydroxy group at C-4 of 4 to produce 7a and 7b (Scheme S1a). Analysis of the Δδ values between 7a and 7b revealed the C-5 to be in the R configuration (Figs. 2b, S35, S36, and Table S2). The hydroxy groups at C-5 and C-7 of 4 were converted to a 1,3-dioxane ring to obtain the acetonide 8 (Scheme S1b). 13C chemical shifts of the acetonide methyl groups at C-14 and C-15 were observed at δC 24.7 and 25.0 ppm, respectively, indicating that the six-membered acetonide ring in 8 was in a boat conformation (Fig. S38, Table S3). The acetonide ring formed by an anti-1,3-diol takes a boat conformation, therefore, the hydroxy groups at C-5 and C-7 of 4 were determined as anti and C-7 was revealed to be in the R configuration (Fig. 2c) [6]. The absolute configuration at C-8 was proposed by the J-based configuration analysis (JBCA) method [7]. The large coupling constant between H-7 and H-8 (J = 9.5 Hz) in CD3OD indicated an anti relationship of H-7 and H-8 (Fig. S17, Table S1). In addition, the absence of a cross signal between H-6 and H-9 in NOESY spectrum (Fig. S22) supported a threo relationship of the C-8 propyl group with the C-7 hydroxy group (Fig. 2d), indicating the absolute configuration of 4 as (5R,7R,8S) (Fig. 2e). We are working on the total synthesis of compound 4 to unambiguously confirm its absolute configuration in our laboratory.

The 1D and 2D NMR spectra of streptospherins C (3) and E (5) were almost identical to those of 4, with the difference that the alkyl chain at position C-8 was ethyl (3) or isobutyl (5). The optical rotation of 4 ([α]20D + 18.7) is close to that of 3 ([α]20D + 25.3) and 5 ([α]20D + 6.7), and the ECD spectrum of 4 is much similar to those of 3 and 5 (Fig. S3), confirming the same absolute configurations at the chiral center of 3 and 5 as in 4. Streptospherin F (6) was indicated to be as a shortened fragment of 4, with the mass decreased by C5H8 and similar 1D and 2D NMR spectra. A partial structure of streptospherin B (2) was determined via the similarity of 1D and 2D NMR spectra with those of 3–5. The addition of hydroxy groups at C-2 and C-12 was confirmed by the mass spectral increment of two oxygen and downfield shift protons observed at δH 4.42 ppm for H-2 and 3.98/4.28 ppm for H-12. NOESY correlations from H-2 to H-5 indicated that the C-3–C-4 double bond of 2 is in the Z configuration (Fig. S9), which differs from the E configuration in 3–5. These structural analyses led us to the elucidation of the planar structure of 2; however, the absolute configuration at the C-2 position was not determined because of the limited available amount of 2. The absolute configurations of the chiral centers of 2 and 6 are presumed to be as shown in Fig. 1 given that 3–5 are produced by the same producing strain.

To reveal the putative biosynthesis pathway of the streptospherins, we extracted the genomic DNA of Streptomyces sp. KUSC-240, sequenced the genome, and analyzed the sequence assembly by using antiSMASH version 7.1.0 [8]. As a result, we found a core type I PKS gene cluster for 3–5. On the basis of the predicted PKS modules of this cluster, we propose the putative biosynthetic pathway of 3–5 in Fig. 3.

Fig. 3
figure 3

Putative biosynthesis pathway of streptospherins C–E (3–5)

Full size image

We previously demonstrated that 3D-cultured HT29 cell spheres in 3D Tumorsphere Medium XF (XF medium) exhibited an abundance of CSC markers and resistance to high dose of anticancer agent such as 10 μM camptothecin [3]. Here, we observed the effect of 2–6 on HT29 cell spheres in XF medium under a microscope and found that these compounds significantly inhibited CSCs sphere formation (Fig. 4a). Cell growth analysis using a WST-8 assay also showed that 2–6 inhibited the cell growth of spheres in XF medium with IC50 values of 10.5–86.3 μM (Fig. 4b). These results indicate that streptospherins targeted CSCs and inhibited their growth moderately, compared to lenoremycin that we previously reported [3]. The cell growth inhibitory activities of 3–5 were greater than those of 2 and 6, which indicated that the C-1–C-4–C-12 moiety without oxygen modification is important for targeting the CSCs. Compound 1 had lower activity than 3–5, which suggests that streptospherins with linear structure in the right fragment have potential for therapeutic targeting of CSCs, but further structure–function relationship analyses are needed.

Fig. 4
figure 4

Inhibitory activity of streptospherins A–F (1–6) toward cancer stem cells (CSCs). a CSC sphere formation observed under a microscope (2–6). b IC50 calculation from WST-8 assay (2–6) or an ATP assay (1; ref. [4])

Full size image

In summary, we identified streptospherins B–F (2–6) as inhibitors of CSC sphere formation; these compounds, produced by Streptomyces sp. KUSC-240, have a common substructure. We and others have previously reported that polyether ionophore compounds, such as lenoremycin and salinomycin, exhibit the inhibitory activity against CSCs [3, 9]. We speculate that the mechanisms of action of streptospherins differ from those of these compounds because of the differences in their core chemical structure. To further investigate the mechanisms of action of streptospherins, detailed bioassays, along with additional fermentation, isolation, and total synthesis efforts, are currently underway in our laboratory.