|Home | About | Journals | Submit | Contact Us | Français|
Chemical examination of an extract from an Erythrobacter sp. isolated from mangrove sediments yielded erythrazoles A (1) and B (2). The erythrazoles are of mixed biosynthetic origin containing a tetrasubstituted benzothiazole, an appended diterpene side chain and a glycine unit. Erythrazole B is cytotoxic to a panel of non-small cell lung cancer (NSCLC) cell lines, with IC50 values of 1.5, 2.5 and 6.8 μM against H1325, H2122 and HCC366 respectively.
Marine bacteria have proven to be a valuable resource of biologically active natural products.1 A predominant focus of these studies has been on marine-derived actinomycetes, leading to a number of biologically and structurally interesting natural products, such as abyssomycin C2, salinosporamide A3, marinomycin4 and ammosamide A.5 However other bacteria, such as species of Bacillus, Pseudomonas and Burkholderia have proven to be an additional source of biologically and chemically interesting natural products.6 As part of our efforts to isolate marine-derived bacteria from mangrove sediments, we isolated bacterial strain SNB-035 that by 16S rRNA analysis was identified as closely resembling Erythrobacter sp. Erythrobacter are Gram-negative bacteria that are known producers of carotenoids, but to our knowledge no other natural products have been reported for this class of bacteria.7 Our studies on this strain reveal that Erythrobacter species are prolific producers of natural products.
Through high-throughput screening efforts, we identified the extract from Erythrobacter strain SNB-035 to possess activity in the Locus Derepression Assay (LDR), which identifies molecules that modulate epigenetic regulation, such as DNA methylation and histone acetylation.8 Ultimately, bioassay guided fractionation led to two other compounds with the ability to modulate epigenetics, however it allowed us to identify erythrazoles A and B (1-2). Structurally, 1 and 2 possess a benzothiazole moiety, which is rare in natural products. Furthermore, 2 arises from four biosynthetic pathways; NRPS, terpene, shikimate and polyketide. Although combinations of two of these pathways is common, natural prouducts containing all four pathways are seldom observed. Subsequent biological evaluation in a variety of cancer related screening efforts revealed 2 to have low μM cytotoxicity against the non-small cell lung cancer (NSCLC) cell lines H1395, H2122 and HCC366.
Marine bacterium SNB-035 was isolated from a sediment sample collected from Trinity Bay (Galveston, TX) and isolated on an acidified Gauze media. Analysis of the strain by 16S rRNA revealed 98% identity to Erythrobacter citreus. A large-scale fermentation (30 L) by shake fermentation was carried out to obtain sufficient material for full chemical and biological analysis of the new compounds. The excreted metabolites were collected using XAD-7 resin and the resulting crude extract purified by a combination of solvent/solvent extraction (n-hexane, DCM, ethyl acetate and methanol/H2O) and reversed phase chromatography to give fractions enriched in terpenoid metabolites. Final purification by gradient reversed phase HPLC gave erythrazole A (1, 1.0 mg) and B (2, 0.6 mg).
Erythrazole A (1) was isolated as a light yellow glass, [α]D +3.3 (c 0.12 MeOH); UV (MeOH) λmax(log ε) 210 (4.1), 254 (4.1), 332 (3.8). High-resolution ESI-MS (HRESIMS) analysis of 1 gave an [M + H]+ at m/z 587.2792 consistent with a molecular formula of C31H42N2O7S (calcd for C31H43N2O7S, 587.2791) and 12 degrees of unsaturation. Analysis of the 13C and HSQC NMR spectra of 2 revealed five methyl carbons, nine methylene carbons, five methine carbons, and twelve sp2 quaternary carbons. A combination of the four methyl singlets from δH 1.09 to 1.83 ppm and the number of sp2 carbons from δC 121 to 138 ppm were highly indicative of a terpenoid chain (Table 1). This was confirmed by the analysis of the 1H-1H COSY and gHMBC spectra of 1 (Figure 1). HMBC correlations from the H17 methyl doublet (δH 1.09) to C1 (δC 182.8), C2 (δC 40.4) and C3 (δC 34.2), as well as COSY correlations from H2/H3, H3/H4 and H4/H5 established C1-C5. COSY correlations from H7/H8, H8/H9, H11/H12 and H12/H13 and key HMBC correlations from H18 (δH 1.50) to C5 (δC 40.3), C6 (δC 135.6) and C7 (δC 125.5); from H19 (δH 1.51) to C9 (δC 40.5), C10 (δC 135.8) and C11 (δC 125.0); from H20 (δH 1.83) to C13 (δC 40.7) C14 (δC 137.9) and C15 (δC 121.7) provided assignment of a fragment terminating in a carboxylic acid. The three double bonds in 1 were all assigned to have the E configuration based on the chemical shift of vinyl methyl groups, C18 (δC 15.6), C19 (δC 16.1), and C20 (δC 16.5).9 The downfield chemical shift of the methylene doublet of H16 (δH 3.60) and HMBC correlations to the C14/C15 olefin and to three additional sp2 carbons, indicate the terpene chain is attached to an aromatic ring.
The remaining structural assignment of 1 required significant interpretation of NMR data, as we still had to account for C11H9N2O5S and eight degrees of unsaturation. The UV spectrum of 1 showed a λmax at 254 and 332 nm, which indicated a heteroaromatic fused ring system. Based on the UV, remaining molecular formula, and chemical shift values, we deduced 1 to contain a benzothiazole. As there is only one aromatic proton in 1, we relied on key HMBC correlations to assign the tetrasubtituted benzothiazole moiety, including from H18 (δH 3.60) on the terpene side-chain to C8′ (δC 145.5), C9′ (δC 120.0) and C10′ (δC 131.8); from H15 (δH 5.24) to C9′; and from H6′ (δH 7..47) to C5′ (δC 147.4), C7′(δC 150.1), C8′ and C10′.
The 13C chemical shifts of C7′ and C8′ at δC 150.1 and 145.5, respectively, suggested the ortho-dioxygen substituted pattern. An HMBC correlation from the methyl singlet H11′ (δH 3.98) to C7′ indicated the –OMe was placed at C7′. The 1H chemical shift of the H11′ protons ruled out the possibility of a thioether or a methylamine at C7′. The presence of an –OH at C8′ was confirmed by treatment of 1 with CH2N2 in Et2O and obtaining a product with three new methyl singlets at 3.60, 3.63 and 3.87 ppm, with the signal at 3.87 representing the newly formed –OMe at C8′.10 Moreover, utilizing the 13C chemical shifts of C4′(δC 161.2), C5′ and C10′, we could assign the regiochemistry of the benzothiazole ring such that the nitrogen atom is attached to C5′, the sulfur atom to C10′, and both the N atom and S atom are attached to C4′. This is based on comparison to literature values of previously reported benzothiazoles with similar substitution patterns (Figure S1).11 In all examples, the sulfur bearing carbon (C10′) is shifted upfield to around ~135 ppm, whereas the nitrogen bearing carbon (C5′) is downfield ~150 ppm. This data matches our assignment of the benzothiazole ring. To complete the structural assignment of 1, we deduced the presence of a terminal glycine based on HMBC correlations from the H2′ methylene pair (δH 4.03) to C1′ (δC 175.3) and to C3′ (δC 161.6). Recording the NMR data of 1 in DMSO-d6 Table S2) indicated the presence of an NH proton at δH 8.17 (t, J = 4.0 Hz), which showed a COSY correlation with H2′ at δH 3.52 (d, J = 4.0 Hz), and weak HMBC correlation with C4′(δC 158.1) on the benzothiazole ring. As a result, the assignment of the entire backbone of 1 was completed. Erythrazole A (1) represents the first example of a benzothiazole containing diterpene. As discussed below, benzothiazole containing natural products are quite rare.
To determine the absolute configuration at C2 in 1, we turned to Kusumi’s phenylglycine methyl ester (PGME) method.12 Derivitization with S-3 or R-3 gave the diamides 4a and 4b (Figure 2). To our surprise the 1H NMR spectrum (CDCl3) of 4a revealed a 1.0:0.6 ratio of diastereomers (based on integration of the H17 methyl doublet), indicating that erythrazole A is a mixture of enantiomers. The 1.0:0.6 ratio was verified by resolving 1 via chiral HPLC (Chiracel OD). Unfortunately, due to overlapping 1H NMR signals of the diagnostic protons in 4a and 4b, we are unable to assign the absolute configuration with confidence.
Erythrazole B (2) was isolated as a light yellow glass, UV (MeOH) λmax(log e) 210 (4.09), 254 (3.99), 334 (3.7). The molecular ions identified in negative ESI-MS at m/z 611 [M − H]− and positive ESI-MS at m/z 635 [M + Na]+ allowed the deduction of its molecular weight of 612 Da, 26 Da more than 2. High-resolution ESI-MS gave an [M + H]+ at m/z 613.2951 consistent with a molecular formula of C33H44N2O7S (calcd for C33H45N2O7S, 613.2947) and 13 degrees of unsaturation. Based on this molecular formula and the similarity of the UV, 1H and 13C NMR of 1 and 2 (Table S1), we could deduce that the benzothiazole moiety was intact and that 2 differed from 1 only in the length and substitution of the terpene side chain. This was confirmed by the analysis of the 1H-1H COSY and gHMBC spectra of 2 to establish an unusual 22 carbon terpene fragment terminating in a carboxylic acid. A few key HMBC correlations for the terpene portion of 2 were from C1 (δC 180.5) to H2 (δH 2.95) and from C3 (δC 121.2), C4 (δC 137.3) and C5 (δC 32.4) to the methyl singlet H19 (δH 1.69). COSY correlations from H5/H6 and H6/H7 established the modified portion of the terpene chain. The four double bonds in 2 were assigned as 3Z, 8E, 12E, and 16E, respectively, based on the chemical shift of the vinyl methyls, C19 (δC 23.5), C20 (δC 15.9), C21 (δC 15.9), and C22 (δC 16.5).9 This is an unusual derivitazation of a terpene. Previous examples of C12 or C7 terpenes (eg. mycophenolic acid) have been observed, with the unusual chain length arising from an oxidative cleavage of a C15 to a C12 or C7 terpene.13 Due to the location and geometry of the C3 olefin in 2, we believe 2 arises from a two-carbon homologation of 1, presumably via acetate addition and subsequent elimination of H2O to give the olefin at C3.
Benzothiazoles are a prevelant heterocyclic moiety in pharmaceuticals, such as in the diabetic drug zopolrestat14 and the fatty acid oxidation inhibitor CVT-3501.15 However in natural products, benzothiazoles are relatively rare, with only a few examples in the literature, including firefly luciferin, which was isolated in the late 1940’s.16,17 Microbially derived benzothiazole containing natural products include rifamycins P and Q18 and thiazinotrienomycin F and G.19 There are few biosynthetic investigations of benzothiazoles, but it has been suggested that the motif arises from condensation of cysteine with a benzoquinone.20 The presence of high levels of coenzyme Q and other ubiquinone analogs in the fermentation broth would suggest that 1 and 2 are from a shunt pathway of ubiquinone biosynthesis.
A plausible biosynthetic sequence beginning with prenylation of 4-hydroxybenzoic acid (5) with geranylgeranyl pyrophosphate to generate intermediate 6, followed by a series of well precedented steps found in ubiquinone biosynthesis (decarboxylation, hydroxylation, methylation, oxidation) to generate quinone 7. Condensation of cysteine with 7 followed by an oxidative cyclization would provide benzothiazine 8, followed by a decarboxylation, addition of H2O, and rearrangement to give the benzothiazole containing aldehyde 9 (Figure 3). Subsequent oxidation of the aldehyde to the carboxylic acid and condensation with glycine would provide the entire benzothiazole moiety 10. Efforts are currently underway to study the biosynthesis of the erythrazoles using labeled precursors and use of genome sequencing to elucidate the biosynthetic gene cluster.
Although the erythrazoles were in the active fraction of the LDR assay, when tested as the pure compounds, they showed no activity up to 20 μM concentration. However, when these molecules were examined for cytotoxic activity against a panel of cancer cell lines, we found that 2 had an IC50 of 1.5 μM against the NSCLC cell line H1395, 2.5 μM against H2122 and 6.8 μM against HCC366. 1 had no cytotoxicity up to 20 μM against these three NSCLC cell lines. The difference in bioactivity for 1 and 2 is surprising as they only differ in two carbons. We are further exploring the biological activity of these molecules to understand this difference.
The authors thank Elisabeth Martinez (University of Texas Southwestern Medical Center, Department of Pharmacology) for the LDR assay, Bruce Posner and Shuguang Wei (University of Texas Southwestern Medical Center, Biochemistry) for cytotoxicity assays and Aaron Legako (University of Texas Southwestern Medical Center, MacMillan lab) for scale-up fermentation. We acknowledge the following grants for funding this project: NIH R01 CA149833, P01 CA095471 and the Welch Foundation I-1689. JBM is a Chilton/Bell Foundation Endowed Scholar.
Supporting Information Available General procedures, bioassay protocols, chemical derivatization, data tables and NMR spectra.