Cryptic sulfur incorporation in thioangucycline biosynthesis

Abstract

Sulfur incorporation into natural products is a critical area of biosynthetic studies. Recently, a subset of sulfur-containing angucyclines has been discovered, and yet, the sulfur incorporation step is poorly understood. In this work, a series of thioether-bridged angucyclines were discovered, and a cryptic epoxide Michael acceptor intermediate was revealed en route to thioangucyclines (TACs) A and B. However, systematic gene deletion of the biosynthetic gene cluster (BGC) by CRISPR/Cas9 could not identify any gene responsible for the conversion of the epoxide intermediate to TACs. Instead, a series of in vitro and in vivo experiments conclusively showed that the conversion is the result of two non-enzymatic steps, possibly mediated by endogenous hydrogen sulfide. Therefore, the TACs are proposed to derive from a detoxification process. These results are expected to contribute to the study of both angucyclines and the utilization of inorganic sulfur in natural product biosynthesis.

Graphical Abstract

The cryptic incorporation of inorganic sulfur into thioangucycline biosynthesis was confirmed to be non-enzymatic by heterologous expression, systematic gene deletion, and in vitro experiments. A stable epoxide-containing angucycline was identified as the final natural product and precursor to the thioether-bridged angucyclines, likely as part of a hydrogen sulfide-mediated detoxification process.

Introduction

Angucyclines represent the largest group of polycyclic aromatic polyketide natural products and possess an intriguing architecture and wide-ranging bioactivities.^[1] Sulfur-containing angucyclines, however, are rare, with only six known examples, including BE-7585A,^[2] urdamycin E,^[3] grecocycline B,^[4] gephysulfuromycin,^[5] grisemycin,^[6] and the donghaesulfins^[7] (Figure 1).

Figure 1.

Structures of the six sulfur-containing angucyclines characterized to date and the origin of the sulfur, if known.

The biosynthetic pathway of the characteristic tetracyclic benz[α]anthracene scaffold of angucyclines has been extensively studied. The core scaffold is biosynthesized by type II polyketide synthases (PKSs) via consecutive decarboxylative condensations of a short acyl-CoA starter and nine malonyl-CoA extenders, and additional structural diversity observed within this family of natural products results from the myriad of tailoring enzymes.^[1] However, a detailed investigation into the sulfidation tailoring step is still lacking, and thus, the origin of the sulfur and its mechanism of incorporation into the angular tetracyclic backbone remains largely elusive. In the biosynthesis of urdamycin E, a methanethiol moiety enzymatically derived from methionine was proposed to undergo a non-enzymatic Michael addition to the C-5 site of the benz[α]anthraquinone.^[3] A similar non-enzymatic mechanism was proposed in which a thiosugar moiety acts as the nucleophile for C-5 attack in the biosynthesis of BE-7585A.^[2] In contrast, the C-7 sulfhydryl installation onto grecocycline B was proposed to be catalyzed by a thioesterase (Figure 1); however, no evidence was provided in support of this mechanism.^[4] The mechanism of sulfur incorporation between C3 and C12a in gephysulfuromycin,^[5] at C-6 in grisemycin^[6] and the dimerized donghaesulfins^[7] are unknown.

During our ongoing effort to isolate natural products as small molecule probes and drug leads from the microbial strain collection at The Scripps Research Institute, we discovered a series of sulfur-bridged angucycline dimers, thioangucyclines A-E (Figure 2A), from Streptomyces sp. CB00072.^[8] The combination of a genome-sequenced producer and several isolated thioangucycline (TAC) congeners provides an outstanding opportunity to investigate sulfur incorporation in angucycline biosynthesis.

Figure 2.

Native and heterologous expression of the tac BGC resulted in identification of the thioangucyclines and key intermediates. A) The thioangucyclines (TACs) and related angucycline monomers isolated from S. sp. CB00072 and S. albus recombinant strains SB4051 and SB4061 in this study. Compounds 1 and 2 were isolated from S. sp. CB00072; compounds 6, 8-O-methylrabelomycin (10),^[9] tetrangomycin (11),^[10] X-14881 E (12)^[11], 13,^[12] 14,^[13] and oleaceran (15)^[14] were isolated from strain SB4051; and the remaining compounds were isolated from SB4061. The crystal structure of 9 is shown for reference. See Figure S1 for other structures. B) HPLC profiles of TACs and congeners produced by the S. sp. CB00072 wild-type and S. albus SB4051 recombinant strains in medium F. (i) S. sp. CB00072 wild-type, (ii) S. albus SB4051, (iii) S. albus SB4058 (ΔPKS, negative control).

Herein, we report the (i) identification of the tac BGC and its expression in a heterologous host for TAC production, (ii) characterization of the unusual angucycline epoxide as the nascent final product encoded by the tac BGC, and (iii) sulfur incorporation into an angucycline epoxide via a non-enzymatic Michael addition process generates TACs as products of H₂S-detoxification. These findings provide new insights into sulfur incorporation in natural product biosynthesis and cellular thiol-mediated detoxification of bioactive natural products.

Results and Discussion

Discovery of TACs and congeners from Streptomyces sp. CB00072

During our ongoing natural product discovery program at Natural Products Discovery Center (NPDC) at Scripps Research, a group of sulfur-bridged angucycline dimers (TACs) were identified and isolated from S. sp. CB00072. Briefly, four standard media applied in our previous study (Table S1)^[13] were utilized to ferment S. sp. CB00072, and medium F showed the richest production profile based on HPLC analysis (Figure S2). Large-scale fermentation (10 L) of S. sp. CB00072 in shake flasks led to the isolation of new compounds TAC-A (1) and TAC-B (2) (Figure 2A), each with approximate yields of 0.1 mg L^-1. The structures of 1 and 2 were elucidated by a combination of NMR (1D and 2D) and HRMS spectra (Figure S3–S4); the NMR data was summarized in Tables S11 and S12. The structures were similar to the previously described donghaesulfins,^[7] and the sequenced genome of S. sp. CB00072 provides an opportunity to study the biosynthetic origin of the thioethers.

Production of TACs by expressing the tac BGC in Streptomyces albus J1074 and improvement of TAC titer by thiosulfate supplementation

The genome of S. sp. CB00072 has previously been reported, and further analysis revealed 38 putative BGCs annotated by antiSMASH 4.0 and summarized in Table S2.^[15] One BGC, encoding a type II polyketide synthase (PKS, BGC 9, Table S2), showed high similarity with the BGC of the angucycline kiamycin from Streptomyces sp. W007 (Figure S23).^[16] While likely encoding the newly isolated TACs, no sulfur-containing angucyclines had been previously isolated from S. sp. W007. Upon closer comparison of the two BGCs, a putative resistance gene, tacR (homologous to urdJ2 from urdamycin biosynthesis),^[17] was identified outside the original boundaries of the tac BGC based on the kiamycin BGC (Figure S23). Therefore, the boundaries of the tac BGC were expanded beyond that of the kiamycin BGC.

Though low production of the TACs could be seen in the native producer, successful heterologous expression of the tac BGC would enable both a clean metabolic background in the new host and more facile genetic manipulations. Towards this goal, the predicted tac BGC (34 kb, Figure 3) was cloned into pStreptoBAC V using the Cas9-assisted targeting of chromosome segments (CATCH) strategy to yield pBS4063 (Table S3).^[18] Transformation of this construct into E. coli yielded the cloning host SB4031 (Table S5), in which genetic manipulation was facilitated by well-developed E.coli CRISPR/Cas9 tools.^[19] From E. coli, pBS4063 was conjugated into the model organism S. albus J1074 and integrated into the genome for stable heterologous expression.^[20] The resultant strain, SB4051 (Table S6), demonstrated similar titers of TACs in medium F as the native producing strain (Figure 2B, panels i vs ii), so the poor overall yield in both hosts remained a challenge for continuing biosynthetic studies.

Figure 3.

Genetic organization of the 34 kb thioangucycline (tac) biosynthetic gene cluster reveals no obvious candidates for sulfur incorporation. Orf: open reading frame.

Given the many TAC congeners lacking sulfur detected in the fermentation profile of the heterologous host (Figure 2A), the sulfur source(s) were expected to be a major bottleneck in the biosynthesis of TACs under the original conditions. Considering that sulfur in natural products is often derived from amino acids,^[21] but also rarely can be derived from inorganic sulfur,^{[2b, 22]} the fermentation media was supplemented with a panel of sulfur sources. While l-methionine and sodium sulfite showed no improvement in titers, sodium thiosulfate, and to a lesser extent, l-cysteine and sodium sulfide, showed improved yields of TACs (Figure S24). Sodium thiosulfate (1 mM) was able to increase the yield of 1 8.5-fold, though the new titer was still less than 1 mg L⁻¹ (Figure S25). The new medium, designated F-S, was utilized for additional experiments to improve the yield of TACs (Figure 4).

Figure 4.

Titer comparison of TAC-A (1) by HPLC analysis. (i) SB4051 fermented in F-S medium showed increased titer of 1; (ii) SB4051 fermented in F medium yielded less 1 than in F-S medium; iii) SB4061 fermented in F-S medium showed a very high yield of 1; (iv) SB4061 fermented in F medium. See Figure S24 for metabolite profiles of S. albus SB4051 supplemented with other sulfur sources.

Unveiling of an angucycline epoxide as the nascent final product encoded by the tac BGC by systematic inactivation of genes

With the TAC titer improvements resulting primarily from inorganic sulfur sources, the mechanism of the sulfur incorporation at C-6 and other transformations remain unclear. If these moieties resulted from enzymatic transformations, the corresponding gene(s) would be expected to be located within the tac BGC. To assign these functions, systematic gene deletion was performed, and the resulting metabolite profiles from the S. albus heterologous host were examined.

Though efficient CRISPR/Cas9 genetic manipulation tools have recently been developed for Streptomyces,^[23] access to a bacterial artificial chromosome (BAC) containing the tac BGC in E. coli preempted the use of those tools in favor of the more rapid E. coli tools.^[19] In addition to the shortened growth cycles of E. coli relative to Streptomyces, genetic manipulation in E. coli has the advantage of any off-target effects outside of the BGC not being transferred into the heterologous Streptomyces host. Systematic in-frame deletion of most open reading frames (ORFs) within the tac BGC via CRISPR/Cas9 confirmed the BGC boundaries and the roles of several biosynthetic genes. Mutants Δorf(−1) (SB4052) and Δorf(+1) (SB4069) did not demonstrate significantly different HPLC profiles relative to the control strain SB4051. Accordingly, these two ORFs were designated to be outside the BGC boundaries (Figure S26). All other ORFs within the boundaries, except for tacH-M (PKS genes), tacF (O-methyltransferase), and tacE/R (transporters), were deleted individually (mutants’ genotypes were confirmed by sequencing, Figure S27). Each of the constructed mutants was fermented in F-S media and the extracts of the fermentation broth were analyzed by HPLC.

A primary purpose of the BGC manipulations was to increase the yield of TACs and their congeners, and yet, the deletion of the four putative BGC-specific regulators (tacB/D/U/X) in the heterologous host all failed to improve the yield of TACs (Figure S28). Fortunately, deletion of tacP, a gene encoding a malonyl-CoA decarboxylase homologue, resulted in a mutant strain, SB4061, with significantly higher titers of 1 (30 mg L⁻¹) than the wild-type strain (Figure 4). Though the role of TacP in TAC biosynthesis is unclear, the loss of a malonyl-CoA decarboxylase would be expected to boost the intracellular levels of malonyl-CoA required for biosynthesis of the angucycline polyketide core. The higher titers of SB4061 enabled isolation and characterization of additional TACs (3-5) and known monomers 7-9 (Figure 2A),^{[13, 24]} and their structures were elucidated by a combination of NMR (1D and 2D) and HRMS spectra (Figures S5–S7) (NMR data see Tables S11 and S12).

In addition to higher titers, the newly-constructed mutants were able to address unanswered biosynthetic questions from the pathway. Among the mutants, SB4053 (ΔtacA), SB4060 (ΔtacO), SB4063 (ΔtacS), and SB4064 (ΔtacT) were each found capable of producing 6 as the major compound while abolishing the production of 1 and 2 (Figure S29). The two ketone reductases TacA and TacO were implicated in the reduction of the ketones on the C-ring of 6 (Figures S30–S31). Cumulatively, these results indicate that TacO is responsible for the reduction of the ketone at C-7 while TacA reduces the ketone at C-12. Initially, two putative monooxygenases, TacS and TacT, could not be assigned functions in TAC biosynthesis due to low identity and similarity to other known proteins (Table S8). Deletion of either tacS or tacT abolished production of 9 and the dimerized TACs 1 and 2, instead accumulating a late-stage shunt product lacking the epoxide (Figure S29). Therefore, TacS and TacT must be involved in the installation of the epoxide moiety of 9 (Figure S30).

Of the remaining six deletion mutants SB4055 (ΔtacC), SB4057 (ΔtacG), SB4059 (ΔtacN), SB4062 (ΔtacQ), SB4066 (ΔtacV), and SB4067 (ΔtacW), none resulted in a metabolite profile consistent with being responsible for the installation of sulfur or acting after TacO (Figure S32). The HPLC profiles indicated TacG and TacN were involved in earlier stages of angucycline biosynthesis, while deletion of tacC, tacQ, tacV, and tacW did not abolish the production of TACs. In the example of grecocycline B, a putative thioesterase GreTh (sequence not available) was proposed, without evidence, to install the thiol group,^[4] while the deletion of tacW (encoding a putative thioesterase, Table S8) did not abolish the production of the sulfur-containing TACs. Taken together, no gene in tac BGC could be proven to be responsible for installing sulfur into the angucycline scaffold, and the epoxide compound 9 was assigned as the nascent final product of the tac BGC. The relative configuration of 9 was confirmed by X-ray diffraction (CCDC No. 1969840, Figure 2A and Table S9). As the hydroxyl group at C-3 takes α orientation,^[1] the absolute configuration of C-3 was determined as R when referenced to a known angucyclinone whose absolute stereochemistry was confirmed by X-ray diffraction analysis.^[6]

Without a candidate from the BGC for conversion of 9 to 1, other enzymatic transformations needed to be eliminated. A fermentation time course of the overproducing strain SB4061 was conducted (Figure 5). By the second day, production of 9 had already peaked. Over the next three days, there was a significant increase in the titer of 1 and a concomitant decrease in the titer of 9 during the same time period (as the peak for 2 overlapped with another peak, the order of formation of 1 and 2 was obfuscated). From four to seven days, the amount of 3 increased slowly. These results suggest that the biosynthesis of 9 is quite rapid, while conversion of 9 to 1 (and further to 3) occurs much more slowly, consistent with a non-enzymatic conversion.

Figure 5.

HPLC profiles of the time course for the angucycline epoxide and TAC production by the S. albus SB4061 overproducing strain fermented in F-S medium show a slow accumulation of TACs after rapid production of 9. The peak for 2 overlaps with another unrelated compound.

Non-enzymatic sulfur incorporation into TACs with H₂S as a direct sulfur source

Based on the hypothesis that the conversion of 9 to 1 is non-enzymatic, a second look at sulfur sources was necessary. With thiosulfate as the most impactful of these sources, it was necessary to test if it acted as a direct or indirect sulfur donor for TAC biosynthesis. Towards this goal, 9 was incubated with thiosulfate in vitro, and a thiosulfate-9 adduct 16 was isolated (Figures 6 and S21). Notably, however, 16 did not spontaneously dimerize to any of the TACs. To mimic the addition of an aromatic sulfide, such as would be expected during a dimerization reaction, 9 was incubated with thiophenol. As expected, the nucleophilicity of thiophenol enabled near total conversion of 9 to the expected product 17 (Figures 6 and S22).

Figure 6.

Sulfur incorporation into TACs is non-enzymatic with KSH as a direct sulfur source. A) The angucycline epoxide product 9 can readily undergo nucleophilic addition with thiol nucleophiles as exemplified by thiophenol, Na₂S₂O₃, KSH, or Na₂S₂O₃ in the presence of selected endogenous rhodanese-like proteins. Conditions: a) 3 eq thiophenol, 50 mM Tris-HCl pH 7.5, 25 °C, 3 h, >95%; b) 3 eq Na₂S₂O₃, 50 mM phosphate pH 7.5, 25 °C, 5 h, >95%; c) 3 eq KSH, 50 mM phosphate pH 7.5, 25 °C, 5 h; d) 2 mM Na₂S₂O₃, 0.8 mM rhodanese, 50 mM Tris-HCl pH 7.5, 28 °C, 6 h; e) 10 mM MBBr, 25 °C, 5 h. B) HPLC analysis of the angucycline epoxide product reactions with selected thiol nucleophiles: (i) thiophenol, (ii) sodium thiosulfate, (iii) KSH, and (iv-vi) Na₂S₂O₃ in the presence of the three selected rhodanese-like proteins of J-MoeZ, Rhd-1, or Rhd-2.

Based on these results, it was expected that hydrogen sulfide, a known gasotransmitter and cytoprotectant in bacteria,^[25] reacts with 9 to yield the intermediate pre-TAC (18) before reacting with a second molecule of 9 to yield a TAC dimer (Figure 6A). To confirm these steps, isolated 9 was treated with excess KSH, and the formation of TAC-A (1) and TAC-B (2) was observed in situ (Figure 6B). In addition to 1 and 2, other new peaks could be seen by HPLC, but there was no obvious peak for intermediate 18. To confirm the existence of 18, the reaction mixture was treated with monobromobimane (mBBr), a widely used thiol-specific reagent that generates a product with strong UV absorbance.^{[2b, 26]} As predicted, the free thiol of 18 was trapped by mBBr and the resultant product 19 was confirmed by HRMS (Figures 6B and S34).

As thiosulfate was observed as the most impactful primary sulfur source to improve yield of TAC-A during fermentation (Figure S24), and the possibility of thiosulfate acting as the direct sulfur donor had been excluded through the isolation of the thiosulfate adduct 16 (Figure 6), it was important to demonstrate the in vitro conversion of 9 to TACs using thiosulfate. The conversion of thiosulfate to H₂S has been demonstrated in cell-free E. coli extract,^[27] and the reduction of thiosulfate to H₂S also has been well established and reviewed.^[28] Recent work has also shown rhodanese homologues are involved in the sulfur incorporation in BE-7585A and thioplatensimycin/thioplatencin.^{[2b, 22b]} Multiple rhodanese gene homologues were identified within the genomes of the native host S. sp. CB00072 and the heterologous host S. albus J1074 (Table S10), so three homologues were selected to mediate the conversion of 9 to TACs using thiosulfate (Figure S34). The three rhodanese enzymes, termed J-MoeZ (a MoeZ homologue),^{[2b, 22b]} Rhd-1 (a RhdA homologue),^[29] and Rhd-2 (a GlpE homologue),^[30] were produced in E. coli and purified (Figure S35). When 9 was incubated with thiosulfate in the absence of any enzymes, only the thiosulfate adduct 16 was observed, but upon addition of a rhodanese homologue, conversion of 9 to 1 and 2 was observed (~15%, Figure 6B).

After confirming the ability of H₂S to convert the enzymatic product 9 into 1 and 2, it is now possible to propose the entire late-stage biosynthetic pathway (Figure 7). Following the enzymatic synthesis of 9, endogenous HS⁻ can attack the electrophilic C-6 position via a 1,6-Michael addition, leading to the epoxide opening. The thermodynamically unstable ring-opened product re-aromatizes, losing water and yielding pre-TAC (18). The highly reactive 18 attacks another molecule of 9 in a similar manner as thiophenol to yield 1.

Figure 7.

Biosynthesis of the angucycline epoxide 9 and its detoxification by cellular thiols yields TACs such as 1.

TACs as detoxification products of bioactive angucycline natural products and cellular thiols

Within bacteria and other organisms, the high reactivity of sulfur has often enabled non-enzymatic detoxification and signal transduction through small molecule thiols such as glutathione, mycothiol, and other thiol species derived from methionine and cysteine.^[31] Over the last ten years, H₂S has been implicated in signaling and cytoprotection at low levels and toxicity at higher concentrations.^[32] The cytoprotective effects have been proposed to result from its antioxidative properties, and bacterial production of H₂S has been correlated with improved resistance towards antibiotics that trigger oxidative stress.^{[32a, 32b]} Bacterial H₂S can be derived from either organic sulfur sources (e.g., cysteine and methionine catabolism) or inorganic sulfur sources (e.g., sulfate and thiosulfate reduction); however, the primary sulfur source is often species-specific and depends on sulfur availability.^{[28b, 32b, 33]}

The non-enzymatic production of 1 and 2 is proposed to result from detoxification of 9 by H₂S. Consequently, all three compounds were tested for their antibacterial activities against Staphylococcus aureus ATCC 25923, Bacillus subtilis ATCC 23857 and Mycobacterium smegmatis ATCC 607 in Müller-Hinton (MH) broth in 96-well plates with tetracycline as a positive control and the expectation that 9 was the true final bioactive natural product. Gratifyingly, the results showed 9 to be at least four- to eight-fold more active than the TACs when tested against the three Gram-positive bacteria (Table 1). As expected, addition of 1 mM sodium thiosulfate (termed as MH-S broth) resulted in a significant decrease in the activity of 9 against the same strains (Table 1). LC-MS analysis of the cultured cells with 9 in each medium showed formation of 1 and 2 from 9, though dimerization occurred at a much greater rate in the presence of thiosulfate (Figure S36). As neither 1 nor 2 was active against the three strains and the formation of thiosulfate adduct 16 was not detected in the assay (Figure S37), the loss of activity for 9 can be attributed to the conversion to TACs, with even endogenous levels of thiols resulting in some detoxification of 9.

Table 1.

Antibacterial assays for 1, 2, and 9.

	antibacterial activity MIC (μM)
	MH				MH-S[a]
	9	1[b]	2[b]	Tet[c]	9	1[b]	2[b]	Tet[c]
B. subtilis	20	ND	ND	5	40	ND	ND	5
S. aureus	40	ND	ND	2.5	80	ND	ND	2.5
M. smegmatis	20	ND	ND	2.5	40	ND	ND	2.5

^[a]MH-S, MH broth with 1 mM Na₂S₂O₃.

^[b]ND, MIC > 160 μM (maximum tested concentration).

^[c]Tet, tetracycline.

Conclusion

In this study, the discovery of a group of new sulfur-bridged angucycline dimers has been reported, and their biosynthetic origins have been explored. Using a thorough and systematic approach, a combination of heterologous expression, media optimization, and extensive genetic manipulation was utilized to improve the yield of TACs by ~300-fold, thus enabling the study of their biosynthesis and highlighting the key role of the epoxide-containing 9. Based on the results of in vivo and in vitro assays, H₂S, either exogenous or endogenous, was elucidated to non-enzymatically attack the backbone of 9. The resultant reactive intermediate 18 was demonstrated to initiate dimerization to yield 1 and other thioangucyclines. The final installation of sulfur onto an aromatic ring via a cryptic epoxide intermediate provides new insights into understanding sulfur incorporation in other natural product biosynthesis, especially the growing collection other sulfur-containing angucyclines.

Cellular thiols are found to play increasing roles in bacterial physiology and defense mechanisms against antibiotics and oxidative stress. Given the reduced cytotoxicity of the dimerized compounds relative to 9, in combination with the established roles of H₂S in signaling,^[34] antibiotic resistance,^[32b] and toxicity,^[35] we propose that TACs are the result of an innate detoxification mechanism. Therefore, the study of TACs and their precursors is critical for increasing our knowledge of cryptic sulfur incorporation into natural products and the implications thereof.