Geldanamycin (GDM) is an ansamycin produced by Streptomyces hygroscopicus. In our study of secondary metabolites from S. hygroscopicus 17997 (a GDM producing strain deposited at China Pharmaceutical Culture Collection, with an accession number CPCC 200120), we identified a natural GDM analog, 19-S-methylgeldanamycin (1, Figure 1). Compound 1 retained potent cytotoxicity against cancer cells, and meanwhile exhibited increased water solubility and photostability compared with GDM.1

Figure 1
figure 1

Chemical structures of some natural products carrying methylthio-group. A full color version of this figure is available at The Journal of Antibiotics journal online.

As a methylthio-derivative of GDM, 1’s synthetic mechanism in S. hygroscopicus 17997 remains unclear. The biosynthetic pathway of GDM does not contain any methylthiolation reaction.2, 3, 4 Besides, bioinformatics analysis of GDM biosynthetic gene clusters revealed no candidate gene encoding radical-SAM enzyme for methylthiolation reaction.5 So, 19-methylthiolation of GDM is not an essential and/or extended modification reaction in GDM biosynthesis.

An examination of natural ansamycins reveals quite a few compounds with methylthio- group (such as trierixin, quinotrierixin, awamycin and 3-methylthiorifamycin SV, Figure 1) linking to an aromatic carbon of the molecules.6, 7, 8, 9 Besides, urdamycin E as an angucycline antibiotic contains also a methylthio- group (Figure 1).10

Among these compounds, urdamycin E and 3-methylthiorifamycin SV were known to be closely related to L-methionine (Met) for their production.8, 10 Rohr10 proposed a mechanism for chemical conversion of urdamycin A to E, which involved a Michael addition of –SMe (derived biogenetically from Met) to urdamycin A. The mechanism may be applicable to methylthio-containing quinone antibiotics such as awamycin.10 In fact, methanethiol can react chemically with GDM (or naphthomycin A), yielding 1 (or 30-methylthionaphthomycin A by addition-elimination).11, 12 Therefore, the methylthio- group in 1 should come from methanethiol, and methanethiol should be derived from Met as a catabolite by S. hygroscopicus 17997. Other natural ansamycins carrying methylthio-groups (Figure 1) must have the same biological origin and synthetic mechanism.

To demonstrate the above deduction, we supplemented Met into ISP2 medium (1.0% malt extract, 0.4% yeast extract, 0.4% glucose, 1.5% agar, 0.3% Met) to culture S. hygroscopicus 17997 at 28 °C for 5 days. We found that the production of 1 increased to a level of ∼20 mg l−1 (Compound 1 was only MS detectable if Met was not supplemented). Besides, a strong and repulsive smell filled the room for culturing S. hygroscopicus 17997. GC–MS analysis confirmed that the bad smell originated from methanethiol diffused out of the headspace gas of these Met-plus ISP2 medium plates with Streptomyces hygroscopicus 17997 (Figure 2). And methanethiol as a catabolite of Met by S. hygroscopicus 17997 reacted chemically with GDM, which yielded 1 (Figure 3 and Supplementary Figure S1).

Figure 2
figure 2

GC–MS of headspace gas of ISP2 medium (with and without supplementing Met) plates cultured with Streptomyces hygroscopicus 17997 at 28 °C for 5 days. The GC–MS was performed on an Agilent QP2010 with a column Rxi-5MS (30 m × 0.25 mm) under the following parameters. For GC, column oven temperature 35 °C, injection temperature 200 °C, flow control mode linear velocity (36.0 cm s−1), helium as carrier gas at flow rate 1.0 ml min−1 and pressure 47.6 kPa, split ratio 100. For MS, ion source temperature 200 °C, interface temperature 250 °C, acquisition mode scan (285 per second), time 1.0–5.0 min, m/z 45–100. Samples (10 μl) were injected manually in split mode at 100:1, with injector temperature 200 °C. GC (upper): a methanethiol peak at 1.58 min appeared only from the headspace gas of culture medium with Met (3.0 g l−1). The peak was proved to be methanethiol by MS. MS (lower): left, MS spectrum of the peak at 1.58 min; right, standard MS spectrum of methanethiol from US National Institute of Standards and Technology. The two MS spectra were identical (m/z 48, M+, molecular ion; m/z 47, M+-H; m/z 45, M+-H-2H). A full color version of this figure is available at The Journal of Antibiotics journal online.

Figure 3
figure 3

Silical gel TLC for 19-S-methylgeldanamycin production. Two ISP2 plates (one with 3.0 g l−1 L-Met, the other without L-Met as control) inoculated with Streptomyces hygroscopicus 17997 were incubated at 28 °C for 3 days for mycelia growth. The third ISP2 plate was prepared without inoculating Streptomyces hygroscopicus 17997, but added Geldanamycin (GDM) (purity ⩾95%, prepared by the author’s laboratory) to a final concentration of 50 mg l−1. Upon removing the lids, the third ISP2 plate was placed upside down over the former one ISP2 plate with L-Met, and then incubated at 28 °C for 0, 10, 20 and 30 h to detect 19-S-methylgeldanamycin production. At each time, 1/4 agar culture of each plate was cut off for ethyl acetate extraction. The organic extract mixed with a small volume of 10% FeCl3 for 30 min at room temperature, to oxidize hydroquinones to quinones, before silica gel TLC with a mobile phase of EtOAc/CH2Cl2/hexane/methanol (9:6:6:1, v/v), in which GDM had a Rf of 0.55 and 19-S-methylgeldanamycin 0.38. The significant increase of 19-S-methylgeldanamycin production indicated that 19-S-methylgeldanamycin came from GDM reacting chemically with methanethiol. Left: 19-S-methylgeldanamycin production in the two ISP2 plates with Streptomyces hygroscopicus 17997. Right: 19-S-methylgeldanamycin production in the ISP2 plate without Streptomyces hygroscopicus 17997 (cell-free plate). S stands for 19-S-methylgeldanamycin standard (purity ⩾93%, prepared by the author’s laboratory). A full color version of this figure is available at The Journal of Antibiotics journal online.

When we supplemented L-ethionine to ISP2 medium plates inoculated with S. hygroscopicus 17997, we detected the production of an expected red compound with m/z 643, corresponding to the sodium adduct ion of 19-S-ethylgeldanamycin (C31H44N2O9SNa; see Supplementary Information for its silica gel thin layer chromatography (TLC) and MS–MS confirmation; Supplementary Figures S2 and S3). These S. hygroscopicus 17997 cultures also gave off a strong smell of thiols (ethanethiol). This result provided an additional evidence that the methylthio- group in 1 originated from Met, which was catabolized by S. hygroscopicus 17997 to methanethiol as substrate for the production of 1.

To further confirm the above synthetic mechanism, we supplemented Met (3.0 g l−1) in culturing S. hygroscopicus N02Z-0421 (a herbimycin producer isolated by China NCPC New Drug Research and Development Co. Ltd.) to obtain methylthio- derivative of herbimycin, which is a close analog of GDM.13 An expected red compound was detected in the secondary metabolites of S. hygroscopicus N02Z-0421 by silica gel TLC (Figure 4). The red compound was then purified by a procedure of ethyl acetate extraction, silica gel chromatography and reversed-phase HPLC, and 10 mg pure preparation of the red compound (purity >98%, calculated by area% of HPLC at 250 nm; Supplementary Figure S4) were obtained from 1.2 l fermentation supernatant of S. hygroscopicus N02Z-0421. (See Supplementary Information for detailed descriptions of fermentation of S. hygroscopicus N02Z-0421 and purification of the red compound.)

Figure 4
figure 4

Silica gel TLC (developed with the same mobile phase as that of Figure 3) of ethyl acetate extract of fermentation supernatant of Streptomyces hygroscopicus N02Z-0421. Herbimycin A showed a Rf value of 0.60, and its methylthiolated derivative 0.55. A full color version of this figure is available at The Journal of Antibiotics journal online.

HR-ESI(+)-MS of the red compound revealed a principal peak at m/z 689.25554 ([M+Na]+) and a minor (34S isotope) peak at m/z 691.24797 (Supplementary Information, Supplementary Figure S5). The molecular formula of the red compound was thus established as C32H46N2O9S2 (exact mass 689.25369 for [M+Na]+), which is S2C2H4 (or two SCH2) more than herbimycin A (C30H42N2O9).

The NMR spectra of the red compound (Supplementary Figures S6–S11) were highly similar to those of herbimycin A except a few differences.14, 15 In the 13C-NMR, two additional carbon signals of -SCH3 at δC 15.1 and δC 16.3 showed up; in the 1H-NMR, two additional hydrogen signals (δH 2.52, δH 2.60) from two -SCH3 appeared, whereas the hydrogen signals of C-17 and C-19 in herbimycin A disappeared. Long-range correlations from methyl proton at δH 4.28 (H-15) and -SCH3 proton at δ 2.52 to aromatic carbon C-17 suggested that this -SCH3 (δH 2.52) group was connected to C-17. The other -SCH3 group was connected to C-19, as indicated by its methyl proton at δH 2.60 exhibiting long-range correlation to the aromatic carbon C-19. Therefore, the chemical structure of the red compound was 17,19-dimethylthioherbimycin A (2). Its NMR chemical shifts were assigned completely by HSQC, COSY and HMBC (Table 1 and Figure 5).

Table 1 NMR data of 17,19-dimethylthioherbimycin A
Figure 5
figure 5

NMR correlations for 17,19-dimethylthioherbimycin A. A full color version of this figure is available at The Journal of Antibiotics journal online.

A comparison of the benzoquinone moiety of GDM and herbimycin A shows that C-17 of GDM links to a methoxyl group, while C-17 of herbimycin A links to a hydrogen atom that can be easily substituted by other groups. Accordingly, GDM can be mono-methylthiolated at C-19, while herbimycin A di-methylthiolated at C-17,19. But it is interesting to note that both trierixin and quinotrierixin (Figure 1) are mono-methylthiolated derivatives of ansatrienin/mycotrienin,16, 17 although ansatrienin contains two aromatic carbons for methylthiolation like herbimycin A.

Ansamitocin is a benzenic ansamycin produced by Actinosynnema pretiosum ATCC 31565.18 When supplementing Met to culture medium, we detected no methylthio-derivative of ansamitocin from A. pretiosum ATCC 31565. This result suggests that only antibiotics with quinone moiety (such as benzoquinone and naphthoquinone ansamycins) are able to react chemically with methanethiol to form methylthio-derivatives.

Microorganisms such as Streptomyces can catabolize Met to methanethiol. Ashraf19 reported that L-methioninases, ubiquitous in all organisms except mammals, catalyzed the α, γ-elimination of L-Met to α-ketobutyrate, methanethiol and ammonia. A blast search of genome sequences of Streptomyces in NCBI indicated that L-methioninases existed in various species of Streptomyces including Streptomyces hygroscopicus (most of these L-methioninases were annotated as either methionine gamma-lyase or cystathionine gamma-synthases). When we added some Met to ISP2 medium to culture Streptomyces lividans TK24, a model streptomycete strain, we could smell the appearance of methanethiol (if GDM was also added, 1 could be detected by silica gel TLC).

S. hygroscopicus 17997 produced low level (about 3 mg l−1) of 1 in a culture medium consisted of 2% starch, 0.5% glucose, 0.5% cottonseed meal, 1% cornsteep liquor, 0.5% yeast powder and 0.2% CaCO3. One possible explanation for low level production of 1 may be that this culture medium with rich organic nitrogen resources, could provide more sulfur-containing amino acids (Met and cysteine) than needed for normal growth of S. hygroscopicus 17997. Therefore, a part of the surplus Met was catabolized to methanethiol, which then reacted chemically with GDM to form 1.

Thiol-containing compounds such as ethanethiol, glutathione (GSH), N-acetyl-L-cysteine, and so on, can also react chemically with benzoquinone or naphthoquinone ansamycins, producing many semisynthetic or natural thioansamycins.12, 20, 21 Recently, Yang et al.22 reported two thionaphthomycins (naphthomycins M and N). Naphthomycin M contains a thioglycolic acid group connected to C-30 of the naphthomycin skeleton via sulfur, and naphthomycin N is a dimer-like molecule, with a 2-aminoethanethiol group as bridge to connect two naphthomycin monomers. But the (bio)synthetic mechanism(s) of naphthomycins M and N remains unclear.22

Ōmura23 and Shibata24 reported chemical modification of herbimycin A, and obtained more than a dozen herbimycin A derivatives with various modifications at C-17 or C-19 of the benzoquinone moiety of herbimycin A. Among them, derivatives with a methylpiperazino group or bromine substituent at C-19 showed high antitumor activity.23, 24 Our herbimycin A derivative 2 contains two identical substituents (methylthio-) at the benzoquinone moiety of herbimycin A, one at C-17 and the other at C-19 of the molecule. A preliminary cytotoxicity assay of 2 against HepG2 cancer cells was conducted by us using MTT assay.25 Compound 2 showed potent cytotoxic activity against HepG2 cancer cells, with an IC50 of 18.7 μM (for herbimycin A, 14.0 μM), almost the same to 1 with an IC50 of 19.0 μM.1