The Isolation of Pyrroloformamide Congeners and Characterization of Their Biosynthetic Gene Cluster

Abstract

Dithiolopyrrolones are microbial natural products containing a disulfide or thiosulfonate bridge embedded in a unique bicyclic structure. By interfering with zinc ion homeostasis in living cells, they show strong antibacterial activity against a variety of bacterial pathogens, as well as potent cytotoxicity against human cancer cells. In the current study, two new dithiolopyrrolones, pyrroloformamide C (3) and pyrroloformamide D (4), were isolated from Streptomyces sp. CB02980, together with the known pyrroloformamides 1 and 2. The biosynthetic gene cluster for pyrroloformamides was identified from S. sp. CB02980, which shared high sequence similarity with those of dithiolopyrrolones, including holomycin and thiolutin. Gene replacement of pyfE, which encodes a non-ribosomal peptide synthetase (NRPS), abolished the production of 1–4. Overexpression of pyfN, a type II thioesterase gene, increased the production of 1 and 2. Genome neighborhood network analysis of the characterized and orphan gene clusters of dithiolopyrrolones revealed a unified mechanism for their biosynthesis, involving an iterative-acting NRPS and a set of conserved tailoring enzymes for the bicyclic core formation.

Graphical Abstract

Sulfur, the 10^th most abundant element in the Universe, is essential to all life forms found on Earth, and has been proposed to be metabolized by ancient microbes on Mars.¹ Dithiolopyrrolones, such as holomycin,² thiolutin,³ xenorhabdins⁴ and thiomarinols,^5-7 are sulfur-containing natural products characterized by the presence of a disulfide or thiosulfonate bridge embedded in a unique bicyclic scaffold (Figure 1A). The dithiolopyrrolones have been identified from diverse bacteria species, including soil-dwelling Streptomyces, and Gram-negative bacteria, including Xenohabdus, Alteromonas, Yersinia and Photobacterium, which suggests that dithiolopyrrolones may play important chemical ecological roles for the producing organisms.^4,5,8,9

Figure 1.

Structures of selected dithiolopyrrolone natural products. (A) The structures of N-acylpyrrothine or N-methyl-N-acylpyrrothine dithiolopyrrolones. (B) The structures of pyrroloformamides, the N-formylated dithiolopyrrolones from Streptomyces sp. CB02980, in which 1 and 2 were previously reported.

Both holomycin and thiolutin are broad-spectrum antibiotics active against Gram-positive and Gram-negative pathogens.^2-7,10 They were also shown to be cytotoxic against human non-small-cell lung cancer cell line H460, and LCC6, a highly metastatic MDA-MB-435 derivative cell line, as well as HeLa cells.^11,12 The mode of action of holomycin and thiolutin in Escherichia coli, fungi and human cells was recently revealed to be metal-chelation.^12,13 The reduced holomycin was able to chelate Zn²⁺ ion of the essential E. coli class II fructose bisphosphate aldolase and the metallo-β-lactamases responsible for clinical carbapenem resistance.¹³ The reduced thiolutin was also a Zn²⁺ ion chelator for bacterial metalloproteinase thermolysin in vitro, as well as AB1/MPN/Mov34 (JAMM) domain-containing metalloprotease Rpn11 and other JAMM metalloproteases in fungi and mammals.¹² The inhibition of Rpn11, a de-ubiquinating enzyme of the 19S proteasome, or other JAMM proteases, may disrupt protein or genome homeostasis of cancer cells, which are critically dependent on the ubiquitin-proteasome systems.¹⁴ The elucidation of these fundamental mechanisms for the observed inhibitory activity of holomycin and thiolutin is significant and would be instrumental for the future drug development of this family of metal-chelating natural products. In addition, the potent antibiotic thiomarinol consists of two distinct moieties–the clinically-used topical antibiotic mupirocin and a dithiolopyrrolone group joined together by a fatty acyl amide linker. The hybrid showed superior antibacterial activity than its constituents, which suggested a natural hybridity for new antibiotic development. For example, Thomas and co-workers have prepared dozens of thiomarinol analogues through mutasynthesis, while Li and co-workers characterized a unique TmlU/HolE enzyme pair as a potential tool to engineer new analogues.^15,16

Pyrroloformamides 1 and 2 (previously named vD 844 and vD 846) are unique among the reported dithiolopyrrolones, due to the modification of their amino terminal by N-formylation (Figure 1B).^17,18 The additional N-methyl group on 1 results in the presence of two rotamers at ambient temperature. Both 1 and 2 were active against methicillin-resistant Staphylococcus aureus, E. coli and K. pneumoniae. Pyrroloformamide (1) acted as a cytokinesis modulator to interrupt cell cycle, and was cytotoxic against metastatic prostate cancer cells with an IC₅₀ of 1.7 μM. No new analogues of pyrroloformamides have been reported since their initial discovery in 1969.^{17 -19} In addition, to our knowledge, the biosynthetic gene cluster (BGC) of pyrroloformamides has not been reported, despite recent strides in understanding the biosynthesis, heterologous expression and mutasynthesis of holomycin, thiolutin and thiomarinols.^20-25

We are interested in discovering new natural product drug leads from un- or under-explored environmental niches, and recently identified disulfide-containing guangnanmycins through a genome mining approach.²⁶ During our continuing search for new natural products from actinomycetes, we discovered that S. sp. CB02980, isolated from Yandang Mountain in eastern China, could produce pyrroloformamides 1 and 2, along with two new congeners pyrroloformamide C (3) and pyrroloformamide D (4) (Figure 1B). In this study, we report the isolation and structure characterization of these two new pyrroloformamides. In addition, the identification and preliminary characterization of the BGC for pyrroloformamides in S. sp. CB02980 revealed a unified mechanism for dithiolopyrrolone biosynthesis using an iterative-acting non-ribosomal peptide synthetase (NRPS) and highly homologous tailoring enzymes.

RESULTS AND DISCUSSION

Fermentation , Isolation and Structure Elucidation of Pyrroloformamide C (3) and D (4).

During activity-based metabolite profiling of actinomycete strains, the fermentation broth of S. sp. CB02980 showed strong inhibitory activity against S. aureus ATCC 29213 and E. coli. Further fermentation of S. sp. CB02980 and fractionation led to the isolation of pyrroloformamide (1) (Table S1 and Figure S1-S8). Since 1 shares the common disulfide-containing bicyclic core with other dithiolopyrrolones, it was postulated that the addition of L-cysteine to a culture of S. sp. CB02980 may increase the yield of 1. The yield of 1 reached 32.6 ± 6.2 mg/L, about a 3-fold increase over the original medium, when 10 mM of L-cysteine was added to the growth medium (Figures S9 and S10). This is consistent with a previous report that addition of L-cysteine may increase the titers of thiolutin and the related dithiolopyrrolone antibiotics.²⁷ A large-scale fermentation using this optimized medium led to the isolation of compounds 1–3 with the addition of macroporous resins, while 4 was isolated from the fermentation culture without resins. Compounds 2–4 are close analogues of pyrroloformamide (1), in which 2 was the previously isolated vD 846 (renamed as pyrroformamide B in this study) (Table S2 and Figures S11-S16), while pyrroloformamide C (3) and pyrroloformamide D (4) are new congeners (Figure 2).

Figure 2.

Structure elucidation of 3 and 4. (A) HMBC () correlations of 3 and 4. (B) The X-ray crystallographic study of 3. (C) The chemical shifts of holothin part of thiomarinol A, thiomarinol B, 1, and 4. (D) Semisynthesis of 4 from 1.

Pyrroloformamide C (3) was obtained as light-yellow needle crystals. Its molecular formula C₈H₈N₂O₂S₂, was established by HR-ESI-MS, differing from the molecular formula of 1 by one methylene unit (Figure S18). The ¹H and ¹³C-NMR data indicated that 3 existed as a mixture of two rotamers 3a and 3b, similar to compound 1 (Table S3). The difference was the presence of a methylene singlet in pyrroloformamide C (3) [δ_H 4.40 (2H, s), δ_C 30.4 (CH₂, C-5′) for 3a and δ_H 4.28 (2H, s), δ_C 30.3 (CH₂, C-5′) for 3b], which indicated the presence of a methylene group embedded in two sulfur atoms. The above chemical shifts are similar to those observed previously in other 1,3-dithiolane containing natural products, including bis(methylthiomethyl) disulfide and a dithiolopyrrolone derivative, holomycin A.²⁸ The assignment was further supported by the HMBC correlations from H-5 to C-5′ and H-5′ to C-3, C-5 for 3a, and H-5 to C-5′ and H-5′ to C-3, C-5 for 3b (Figure 2A). The complete structure of 3a and 3b was assigned based on their ¹H and ¹³C-NMR, HSQC and HMBC spectra (Figures S19-S23). A single-crystal X-ray crystallographic analysis was applied to further confirm the structure of 3, in which only rotamer 3a was observed (Figures 2B and S24). This was consistent with the presence of a major conformer 3a (appr. 8:2) in the ¹H and the other NMR spectra, suggesting that some steric effects imposed by the 3-dithiine moiety in 3 might restrict the dynamic equilibrium between 3a and 3b. In contrast, both rotamers (appr. 6:4) were present in the crystal of compound 1.¹⁸

Compound 4 was obtained as yellow oil. The molecular formula C₇H₆N₂O₄S₂, was established by HR-ESI-MS analysis, differing from the molecular formula of 1 by two more oxygen atoms (Figure S26). The ¹H, ¹³C NMR, HSQC and HMBC spectra of 4 were similar to those of 1, with the presence of two set of signals (appr. 5:3), suggesting the formation of two rotamers 4a and 4b (Figures S28-S31). The only difference was the replacement of the disulfide moiety in 1 by a thiosulfonate unit in 4, supported by the comparison of chemical shifts among 1a, 4a, as well as thiomarinol A and B. Compared to compound 1a, the downfield-shifted C-3 (δ_C 141.5), upfield-shifted C-4 (δ_C 121.6) and H-5 (δ_H 6.8) in 4a suggested that the position of the sulfone in 4a was different from those of thiomarinol B,²⁹ as well as other thiosulfonate analogues oxo-holomycin, thiolutin dioxide and xenorxides (Figures 1 and 2C).^6,30-32 This assignment was also confirmed by the preparation of 4 through the treatment of 1 using 30% H₂O₂ in aqueous acetone solution, with an isolated yield of 60% (Figure 2D).²⁸ The semisynthetic compound 4 was identical to that of the natural product 4 (Figure S34).

Antibacterial Activities and Cytotoxicity of Compounds 3 and 4.

The newly isolated pyrroloformamide congeners 3 and 4 were first evaluated for their in vitro antibacterial activities against S. aureus ATCC 29213 and methicillin-resistant S. aureus (MRSA), as well as E. coli and Klebsiella pneumoniae, using 1, 2 and linezolid as controls (Table 1). Compared to 1 and 2, compound 4 had attenuated antibacterial activities with minimal inhibitory concentration (MIC) ranging from 8–64 μg/mL, while compound 3 showed no antibacterial activity against any of the four strains tested. Compounds 3 and 4 were not cytotoxic to non-small cell lung cancer cell line A549 or a human epithelial colorectal adenocarcinoma (Caco-2).³³ In contrast, 1 and 2 had MICs of 0.5 μg/mL or 1–4 μg/mL against the pathogens tested, respectively. Compound 1 showed strong cytotoxicity against A549 and Caco-2 cells. Since only the reduced holomycin or thiolutin could chelate the essential Zn²⁺ ion from Zn²⁺-dependent enzymes, the loss of antibacterial activity and cytotoxicity of 3 was likely due to its inability to affect zinc ion homeostasis in living cells. In contrast, compound 4 still had moderate antibacterial activity against S. aureus and E. coli, which was consistent with the reported antibacterial activity of many thiosulfonate- or thiosulfinate-containing natural products, such as pseudoallicin and allicin found in garlic,³⁴ as well as leinamycin from Streptomyces atroolivaceus, whose unique redox-active and catalytic properties have been previously documented.³⁵

Table 1.

The antibacterial activity (MICs, μg/mL) and cytotoxicity (IC₅₀, μM) of pyrroloformamides.

Bioactivity	Bacteria/Cancer cells	Compounds
		Linezolid	1	2	3	4
Antibacterial activities	S. aureus ATCC 29213	1	0.5	4	> 64	8
	MRSA	1	0.5	2	> 64	8
	E. coli	> 64	0.5	1	> 64	32
	K. pneumoniae	> 64	0.5	1	> 64	64
Cytotoxicity	A549	-	1.34±0.01	-	> 64	> 64
	Caco-2	-	0.57±0.01	-	> 64	> 64

^-,not tested.

Pyrroloformamide Biosynthesis

The isolation of pyrroloformamides 1–4 in S. sp. CB02980 indicated an excellent opportunity to study their biosynthesis and the evolutionary relationships with other dithiolopyrrolone natural products. Shotgun genome sequencing of S. sp. CB02980 and bioinformatic analysis led to an assumed candidate pyrroloformamide BGC with high sequence similarity with those of holomycin and thiolutin (Figure 3 and Table S7).³⁶ The pyrroloformamide gene cluster contained essential genes for the dithiolopyrrolone core biosynthesis, including the multi-domain containing NRPS PyfE for the biosynthesis of the core Cys-Cys dipeptide.

Figure 3.

(A) Comparison of the pyrroloformamide gene cluster in S. sp. CB02980, with other similar gene clusters. (B) Predicted functions of ORFs in the pyrroloformamide-type gene cluster from S. sp. CB02980. (C) HPLC analysis of pyrroloformamides production in S. sp. CB02980 wild-type strain and mutants. (D) GNN analysis of 31 orphan dithiolopyrrolone gene clusters and 4 characterized dithiolopyrrolone gene clusters (pyrroloformamide, holomycin, thiolutin and thiomarinol). : orphan holomycin-type; : orphan pyrroloformamide-type; : holomycin; : thiolutin; : thiomarinols; : pyrroloformamide in S. sp. CB02980. (E) The unified dithiolopyrrolone biosynthetic pathways.

To study if pyfE was involved in the biosynthesis of pyrroloformamides in S. sp. CB02980, pyfE was replaced with a mutant copy in which pyfE was disrupted by the thiostrepton-resistance gene with a kasOp* promoter.³⁷ The gene replacement of pyfE completely abolished the production of compounds 1–4 in S. sp. CB02980 (Figure 3C).

There are two standalone thioesterase (TE) PyfC and PyfN genes in the pyf gene cluster, which shared 58% and 53% sequence identity with HlmC and HlmK from the holomycin BGC, respectively. Since co-expression of hlmK with thiolutin and aureothricin BGC in S. albus led to the increase of their production, we hypothesized that overexpression of its homologous gene pyfN can also improve the production titer of pyrroloformamides. PyfN shared the same S92G mutation with HlmK and may also act as a proofreading TE for pyrroloformamide biosynthesis.³⁸ A pyfN overexpression plasmid containing the constitutive promoter ermE*, was thus constructed in the integrative vector pSET152 and introduced into S. sp. CB02980 by conjugation to obtain a pyfN overexpression mutant. The production of pyrroloformamides 1 and 2 increased more than 2-fold in the mutant strain with one additional copy of pyfN. These data confirmed that the pyf gene cluster is responsible for pyrroloformamides biosynthesis in S. sp. CB02980.

To study the evolutionary differences among the pyf BGC and other dithiolopyrrolone BGCs, a genome neighborhood network (GNN) analysis was conducted (Figure 3D). Using PyfE as a query sequence, the survey of the NCBI GenBank resulted in identifying a total of 115 proteins with sequence identity > 43%. Thirty-one putative dithiolopyrrolone BGCs were then identified by using a 95% cutoff to de-duplicate BGCs containing PyfE homologs of extremely high similarity, followed by removal of 36 BGCs lacking essential gene for dithiolopyrrolone biosynthesis (Tables S8-S38). Although a set of highly conserved dithiolopyrrolone core biosynthetic genes is present in almost all of those BGCs, it seemed that two distinct types of dithiolopyrrolone BGCs were present, which were tentatively named as holomycin-type or pyrroloformamide-type BGCs. For example, the Cys-specific adenylation domain in the NRPSs, such as PyfE and HolE, share highly conserved sequence motifs with the Cys-specific adenylation domain of LnmI, which loads an L-Cys to its cognate PCP in leinamycin biosynthesis (Figure S39).^39,40 A standalone thioesterase presumably responsible for proof-reading the peptide intermediate, as well as a thioredoxin-disulfide reductase and acyl-CoA dehydrogenase for the formation of the ene-thiol core motif, are also present. In contrast, compared to the holomycin-type BGCs, the pyrroloformamide-type BGCs contain a set of unique biosynthetic genes, such as RidA family proteins and two methyltransferases, while they lack the representative acetyl- or acyl-transferases, such as HlmA (Figure S40). Interestingly, the homologous protein of HlmI is only present in 14 out of 31 orphan dithiolopyrrolone gene clusters. Since HlmI is a thioredoxin reductase responsible for the formation of disulfide bond in dithiolopyrrolones, the lack of its homologue in some BGCs suggests an alternative mechanism for the disulfide bond formation and resistance for these putative natural products^{22, 41}

Based on the above analysis, the dithiolopyrrolone core biosynthesis in pyrroloformamides may be similar to holomycin and thiolutin. However, pyrroloformamide biosynthesis may diverge significantly when holothin is off-loaded from the NRPS assembly line (Figure 3E).²⁰ Since either holothin or its biosynthetic intermediates are heavily modified during pyrroloformamide biosynthesis, a detailed enzymatic understanding of the tailoring of the Cys-Cys dipeptide intermediate remains obscure. In addition, the mechanism responsible for N-formylation in pyrroloformamides could not be deduced based on the current analysis of the biosynthetic pathway. Interestingly, two orphan pyrroloformamide-type gene clusters from our in-house strains, S. sp. CB02009 and S. sp. CB02261, have also identified. They share high sequence similarity and similar gene organization with pyf BGC in S. sp. CB002980 (Figure 3A). Therefore, the identification and preliminary characterization of these pyrroloformamide BGCs provides us an opportunity to study their biosynthesis, and to generate additional dithiolopyrrolones by manipulating enzymes from these biosynthetic machineries.

Experimental Section

General Experimental Procedures.

IR spectra were recorded on Nicolet iS50 FT-IR (Thermo Scientific). HRMS spectra were recorded on an LTQ-ORBITRAP-ETD instrument. NMR spectra were acquired using a Bruker 500 or 400 MHz spectrometer. Chemical shifts were reported in ppm relative to CD₃CN (δ = 1.94 ppm), CD₃OD (δ = 3.31 ppm) or DMSO-d₆ (δ = 2.50 ppm) for ¹H-NMR and CD₃CN (δ = 1.32 and 118.26 ppm), CD₃OD (δ = 49.00 ppm) or DMSO-d₆ (δ = 39.60 ppm) for ¹³C NMR spectroscopy. Column chromatography (CC) was carried out on silica gel (100–200 mesh and 300–400 mesh, Yantai Jiangyou Silica Gel Development Co., Ltd., Yantai, China). Semi-preparative reversed phase-high-performance liquid chromatography (RP-HPLC) was performed using a Waters 1525 Binary HPLC Pump equipped with a Waters 2489 UV/Visible Detector and using a Welch Ultimate AQ-C18 column (250 × 10 mm, 5 μm). Crystal data were acquired on a Rigaku APEX-II XtaLAB PRO MM007HF diffractometer using Cu Kα radiation at 100 K. The structure was analyzed by a SHELXS-97 software and refined by means of full-matrix least-squares.

Bacterial Strains.

The S. sp. CB02980 wild-type and mutant strains were grown on G1 agar plates (2% soluble starch, 0.1% KNO₃, 0.05% K₂HPO₄, 0.05% MgSO₄·7H₂O, 0.05% NaCl, 0.001% FeSO₄·7H₂O, 2% agar) at 28 °C for 7-9 days to obtain spores.

Fermentation.

The spores of S. sp. CB02980 strains were inoculated into 250 mL Erlenmeyer flasks containing 50 mL tryptic soy broth (TSB) medium (1.7% tryptone, 0.3% soya peptone, 0.5% NaCl, 0.25% K₂HPO4, 0.25% dextrose, pH 7.3 ± 0.2) at 30 °C on a rotary shaker at 220 rpm for 24 h. Then 10% (v/v) seed cultures were transferred into 50 mL production medium (2% soluble starch, 2% corn flour, 0.05% KH₂PO₄, 0.025% MgSO₄, trace element solution (v/v=1/1000) (0.1% ZnSO₄•7H₂O, 0.1% FeSO₄•7H₂O, 0.1% MnCl₂•4H₂O, 0.1% CaCl₂) in a 250 mL flask. The pH of production medium was adjusted to 7.1, followed by the addition of 0.5% (w/v) CaCO₃ and 6.0% (v/v) wet macroporous resin DA201-H (Jiangsu Su Qing Water Treatment Engineering Group Co., Ltd., Jiangyin, China). Before inoculation, the production medium may be supplemented with different concentration sterile cysteine. The S. sp. CB02980 strains were then cultured for 8 days at 30 °C on a rotary shaker at 230 rpm. For large-scale fermentation, the seed cultures (50 mL) were aseptically transferred to 2 L Erlenmeyer flasks containing 500 mL of production medium.

Extraction and Isolation.

After fermentation, the resins from 12-L culture of S. sp. CB02980 wild-type strain were filtered and dried in air. They were extracted with MeOH and dried in vacuum, further extracted by ethyl acetate (EtOAc) to afford the crude extracts (10 g). The crude extracts were subjected to silica gel chromatography (CC) and eluted with a gradient of petroleum ether (PE)/EtOAc (100:0, 9:1, 8:2, 7:3, 6:4, 1:1, 4:6, 3:7, 2:8, 1:9, 0:100, v/v) to give 7 fractions (Fr. A1–Fr. A7). Fr. A7 was purified by silica gel CC, using dicholoromethane/MeOH (100:0, 99:1, 98:2, 97:3, 96:4, 95:5, 94:6, 93:7, 92:8, 91:9, 9:1, 1:1, 0:100, v/v) to obtain compound 1 (250 mg) and compound 2 (284 mg). Fr. A6 was further purified by Sephadex LH-20 chromatography and then silica gel CC to obtain compound 3 (24 mg).

Compound 4 was obtained from the culture without the addition of resins in the medium. After fermentation, the culture broth (4 L) was centrifuged (4 000 rpm, 15 min) to obtain the supernatant, which was extracted with EtOAc (3 × 4 L) and dried over anhydrous Na₂SO₄. The crude extracts (188 mg) were obtained after concentration under reduced pressure, which were subjected to semipreparative RP-C18 HPLC (Welch Ultimate AQ 5 μm, 250 × 10 mm) with a flow rate of 2 mL/min and a gradient elution of CH₃CN /H₂O in 16 min (10% to 95% for 8 min, followed by 95% for 2 min, and 95% to 10% for 2 min, followed by 10% for 4 min), to afford compounds 4 (t_R = 11.5 min, 69 mg).

X-ray Single-Crystal Data of 3.

The crystals of 3 were obtained by crystallization of 3 from a solution of CH₃CN/H₂O (v/v, 1:1). Monoclinic crystals, a = 3.9579 (15) Å, b = 9.5320 (3) Å, c = 13.2614 (5) Å, α= 70.820(3)°, β = 85.006(3)°, γ =79.423(3)°, Z = 2, μ = 5.006 mm⁻¹, F (000) = 238, and T = 100 K; Crystal dimensions: 0.3 × 0.01 × 0.01 mm³, Volume = 464.32 (3) ų, 1898 reflections measured, 1809 independent reflections (R_int = 0.0420), the final R indices [I > 2σ(I)] R₁ = 0.0388, wR₂ = 0.1107, R indices (all data) R₁ = 0.0400. The goodness of fit on F² was 1.0377. The crystal data of 3 have been deposited with the Cambridge Crystallographic Data Center (deposition no. CCDC 1900736).

Preparation of Semisynthetic 4 from Pyrroloformamide (1).

Compound 1 (20 mg) was dissolved in acetone/water solution (4 mL of acetone and 2 mL of water), and then 2 drops of sodium hydrogen carbonate aqueous solution and 30% aqueous H₂O₂ (14 μL) was added. The whole reaction mixture was stirred at room temperature for 5-10 min and concentrated under reduced pressure. The resulting residue was subjected to semi-preparative RP-C18 HPLC (Welch Ultimate AQ 5 μm, 250 × 10 mm) separation to afford semisynthetic 4 (t_R = 11.5 min, 12 mg) with a yield of 60%.

Pyrroloformamide A (1): yellow needle crystal. UV (CH₃CN) λmax 231.3, 364.4 nm. ¹H NMR (500 MHz, CD₃CN) and ¹³C NMR (126 MHz, CD₃CN) see Table S1. HRESIMS m/z 214.9947 [M + H]⁺ (calcd for C₇H₇O₂N₂S₂, 214.9949).

Pyrroloformamide B (2): yellow powder. UV (CH₃CN) λmax 245.5, 297.7, 386.1 nm. ¹H NMR (500 MHz, DMSO-d₆) and ¹³C NMR (126 MHz, DMSO-d₆) see Table S2. ESIMS m/z 200.96 [M + H]⁺ (calcd for C₆H₅N₂O₂S₂, 200.98).

Pyrroloformamide C (3): light yellow needle crystal. UV (CH₃CN): λmax 357 nm (Figure S17). IR ν_max (film) 3136, 3037, 2999, 2937, 2814, 1686, 1667, 1587, 1342, 1303, 1125, 799, 775, 715 cm⁻¹. HRESIMS m/z 229.0095 [M + H]⁺ (calcd for C₈H₉O₂N₂S₂, 229.0105).

Rotamer 3a: ¹H NMR (500 MHz, DMSO-d₆) δ 6.55 (1H, s, H-5), 4.40 (2H, s, H-5’), 8.23 (1H, s, H-6), 3.07 (3H, s, H-7), 10.53 (1H, s, H-8); ¹³C NMR (126 MHz, DMSO-d₆) δ 163.6 (s, C-1), 127.4 (s, C-2), 125.1 (s, C-3), 130.1 (s, C-4), 108.2 (s, C-5), 30.4 (s, C-5’), 163.0 (s, C-6), 30.4 (s, C-7). Rotamer 3b: ¹H NMR (500 MHz, DMSO-d₆) δ 6.51 (1H, s, H-5), 4.28 (2H, s, H-5’), 8.17 (H, s, H-6), 3.25 (3H, s, H-7), 10.37 (H, s, H-8); ¹³C NMR (126 MHz, DMSO-d₆) δ 163.8 (s, C-1), 125.1 (s, C-2), 130.7 (s, C-3), 130.2 (s, C-4), 108.1 (s, C-5), 30.3 (s, C-5’), 162.7 (s, C-6), 34.3 (s, C-7).

Pyrroloformamide D (4): yellow oil. UV (CH₃CN): λmax 296.5, 364.4 nm (Figure S25). IR ν_max (film) 3435, 3074, 2922, 2850, 1739, 1667, 1620, 1311, 1173, 1099, 549 cm-1. HRESIMS m/z 244.9694 [M − H]⁻ (calcd for C₇H₆O₄N₂S₂, 244.9691).

Rotamer 4a: ¹H NMR (500 MHz, CD₃CN) δ 6.80 (1H, s, H-5), 8.84 (1H, s, H-6), 3.15 (3H, s, H-7); ¹³C NMR (126 MHz, CD₃CN) δ 165.0 (s, C-1), 127.2 (s, C-2), 141.5 (s, C-3), 121.6 (s, C-4), 107.7 (s, C-5), 161.9 (s, C-6), 29.3 (s, C-7).

Rotamer 4b: ¹H NMR (500 MHz, CD₃CN) δ 6.63 (1H, s, H-5), 8.10 (1H, s, H-6), 3.46 (3H, s, H-7); ¹³C NMR (126 MHz, CD₃CN) δ 165.3 (s, C-1), 124.1 (s, C-2), 141.7 (s, C-3), 126.5 (s, C-4), 108.0 (s, C-5), 163.6 (s, C-6), 33.7 (s, C-7).

Gene Replacement of pyfE in S. sp. CB02980.

A pOJ260-based plasmid pXY2001 was constructed to generate the ΔpyfE gene replacement mutant in S. sp. CB02980 via a double crossover homologous recombination. To inactivate pyfE, a 2,052-bp fragment of pyfE gene was replaced with thiostrepton resistance gene with a kasOp* promoter using the In-Fusion cloning kit (TSINGKE, China), and the mutated pyfE gene was cloned into pOJ260 between the HindIII and XbaI restriction sites. This plasmid was introduced into S. sp. CB02980 by conjugation and selected for thiostrepton resistance and apramycin-sensitive phenotype to isolate the desired double-crossover mutant strain S. sp. YX2001. The genotype of the mutant was confirmed by PCR and DNA sequencing. The PCR primers are shown in Table S5.

Overexpression of pyfN Gene in S. sp. CB02980.

A pSET152-based plasmid pXY2002, in which the expression of pfyN is under the control of the constitutive promoter ermE*, was constructed. A 900-bp pyfN gene was PCR-amplified by high-fidelity DNA polymerase from S. sp. CB02980 using the primers listed in Table S5 and the resultant PCR fragment was sub-cloned into the pSET152 between BamHI and XbaI restriction sites under the control of a constitutive promoter ermEp*. This plasmid pXY2002 was introduced into S. sp. CB02980 by conjugation and selected for apramycin resistant conjugants to afford S. sp. YX2002.

Genome Neighborhood Network Analysis of Dithiolopyrrolone BGCs in GenBank.

Using PyfE as the query sequence, its homologous proteins were searched in the GenBank database on March 6, 2019. A total of 115 proteins with sequence identity over 43% were collected, and their lengths range between 1000 to 1300 amino acids. Protein homologues below 43% were unlikely involved in dithiolopyrrolone biosynthesis due to the lack of key dithiolopyrrolone biosynthetic homologue genes, such as hlmB, hlmC, hlmD and hlmF. The 115 PyfE homologues were winnowed into 65 representative sequences by filtering them to a maximum identity of 95% using CD-HIT.⁴² A total of 34 proteins were further removed from the 65 representative sequences, whose BGCs also lack the essential hlmB, hlmC, hlmD and hlmF homologue genes for dithiolopyrrolone biosynthesis. All the proteins in the remaining 31 putative dithiolopyrrolone BGCs were used in an all versus all BLAST using BLAST+ and an E value limit of 10.⁴³ Cytoscape v3.4 was used for GNN visualization and analysis.⁴⁴

Phylogenetic Analysis of the Adenylation Domain in NRPSs.

The amino acid sequences of adenylation domains in NRPSs of dithiolopyrrolone BGCs were aligned using ClustalW in MEGA 7.⁴⁵ A maximum likelihood phylogenetic tree was generated using the Jones–Taylor–Thornton model of amino acid substitution and 100 bootstrap replications. The adenylation domain of LnmI in leinamycin BGC was used as an outer group.

Biological Assays.

The cytotoxicity of compounds 1, 3, 4 against the human cancer cell lines A549 and Caco-2 were evaluated with the CCK8 assay.³³ The antibacterial activity of 1-4 was tested against S. aureus ATCC 29213 and MRSA, as well as Gram-negative pathogen E. coli and K. pneumoniae, in accordance with the previously reported method.⁴⁵ All the tested compounds were dissolved in DMSO.

Availability of data and materials.

The shotgun genome sequence of S. sp. CB02980 (PRJNA524931) was submitted to GenBank and may be accessed.