Streptomyces diastaticus NBU2966

Liu, Yang, Ding, Lijian, Deng, Yueting, Wang, Xiao, Cui, Wei & He, Shan, 2022, Feature-based molecular networking-guided discovery of siderophores from a marine mesophotic zone Axinellida sponge-associated actinomycete Streptomyces diastaticus NBU 2966, Phytochemistry (113078) 196, pp. 1-11 : 2-8

2.1. MS/MS analysis of Streptomyces diastaticus NBU2966 based on molecular networking

The LC – MS/MS analysis was performed for the EtOAc extract of Streptomyces diastaticus NBU2966 cultivated with rice medium. As the initial step (preprocessing) of digitized data analysis, 388 mass features were extracted through MZmine2 from the entire dataset, then the extracted feature table and MS/MS spectral.mgf files were both uploaded to the feature-based molecular networking workflow via GNPS platform. All of mass spectral nodes (mass range from 89 to 937 m /z) were organized into a network consisting of 41 molecular families. To maximize annotation coverage over the entire dataset, we employed MolNetEnhancer, a developed recent workflow that integrated metabolome mining and annotation tools (NAP, DEREPLICATOR, VARQUEST, MS2LDA) into molecular networks (Ernst et al., 2019). Detailed NAP (Network Annotation Propagation) will assign a consensus candidate structure for each node, then cluster or molecular family (with more than 2 nodes) were automatically classified using ClassyFire. The consensus classifications with maximum frequency per cluster/network were retrieved to appoint putative chemical class annotation ( Moreno-Ulloa et al., 2020), which were level 3 annotations in MSI. It was revealed that a total of 8 structural annotations at the superclass level, and the organoheterocyclic superclass (compounds containing a ring with at least one carbon atom and one non-carbon atom) as the main class, occupied 39.38% in the molecular network as shown in Fig. 1a .

The MS /MS spectral library search was performed and resulted in 22 hits matched to reference MS/MS spectra in GNPS, which hits were defined as MSI (2007 Metabolomics Standards Initiative) level 2 annotations ( Sumner et al., 2007). Furthermore, during manual inspection of these annotations, two interesting heterocyclic molecules were observed in molecular families A and B, corresponding separately to ( R)-2-(2-Hydroxyphenyl)-4-hydroxymethyl-4,5-dihydrothiazole, and pyochelin methyl ester ( Fig. 1b ). As we see, both of them were featuring the thiazoline ring and were analogs of the pyochelin, a well-known phenolate-thiazoline type siderophore (involved in iron chelation), and reports about undescribed natural products of pyochelin-family siderophores are very rare. Within these two molecular families, most of the remaining nodes could not be identified through GNPS matching. Since these nodes may have potentially undescribed siderophores and may exhibit interesting biological properties, the large-scale isolation of EtOAc extract of the strain NBU2966 was performed so as to isolate these potentially undescribed siderophore analogs. As a result of this purification protocol, two undescribed phenolate/thiazoline-type pyochelin family members [thiazostatin C (6) and methyl thiazostatin B (7)] and four known compounds (1–4) were isolated and identified ( Fig. 1c ), it should be noted here that pulicatin J (5) was also purified in this isolation procedure according to the similar UV spectrum with those of aerugine/pyochelin-class compounds (characteristic UV absorption maxima located at 204, 251, and 316 nm), even though it was not detected in this molecular network of the crude extract, besides, two phenol/oxazoline-type compounds (8 and 9) and one undescribed phenol/pyochelin/oxazoline-type compound 10 were isolated .

Pulicatin J (5) was isolated as a white powder. A molecular formula of C 14 H 16 N 2 O 4 S with eight indices of hydrogen deficiency (IHD), was given by HRESIMS (m/z 309.0912 [M + H] +; calcd for C 14 H 17 N 2 O 4 S, 309.0909). The UV spectrum of 5 possessed close similarity with those of aerugine/pyochelin-class compounds (characteristic UV absorption maxima located at 204, 251, and 316 nm), indicated the existence of the 2-(2-hydroxyphenyl)-2-thiazoline substructure. Although, it was repeatedly isolated as a single peak by HPLC, the result of analyzing of NMR spectra of 5 suggested a pair of epimers due to two methyl signals [1.21, d (7.2); 1.19, d (7.2)] with a ratio of 1: 1 in the 1 H NMR spectrum ( Fig. S1 , Table 1) and the presence of two distinct sets of signals in the 13 C NMR spectrum ( Fig. S2 , Table 2). Structure elucidation was directly carried out on the mixture since two sets of the constituents in 5 could not be separated. For simplifying, one set of signals was chosen to clarify the gross structure. Examination of the 13 C NMR and HSQC spectra revealed 14 carbon signals, attributable to four highly deshielded sp 2 quaternary carbons (δ C 158.4, 170.6, 171.1, 172.1) and one sp 2 quaternary carbon, four sp 2 methines, two sp 3 methines, two sp 3 methylenes including one of oxygenated methylene as well as one methyl carbon.

The planar structure of 5 was determined through a comprehensive analysis of the 1D and 2D NMR spectroscopic data. From the 1 H – 1 H COSY spectrum, three fragments were initially established. The 1,2- disubstituted benzene ring moiety was deduced by its corresponding coupling constants and the COSY correlations of H-3 (δ H 6.97)/H-4 (δ H 7.42)/H-5 (δ H 6.95)/H-6 (δ H 7.43) (Fig. S9 and Fig. 3 ). This moiety was further determined as a 2-substituted phenol substructure according to the HMBC correlations from H-3 and H-5 to C-1 (δ C 115.6) and from H-4 and H-6 to C-2 (δ C 158.4). On observation of the key cross peak of H-6/C- 2 ′ (δ C 172.1) in the HMBC spectrum, 2-substituted phenol fragment was extended to C-2 ′ from C-1. Then, another fragment deduced by COSY correlations of H 2 -5 ′ (δ H 3.56 and 3.26)/H-4 ′ (δ H 4.98)/H 2 -6 ′ (δ H 4.34) coupled with the HMBC correlations from H-4 ′ and H-5 ′ to C-2 ′ revealed connectivity of H-4 ′ and H-5 ′ to C-2 ′ through heteroatoms, which suggested a five-membered heterocyclic ring. In addition, this 2-thiazolylbenzene unit showed high similarity of the carbon chemical shifts with the corresponding moiety of reported watasemycins ( Sasaki et al., 2002), thiazostatins ( Shindo et al., 1989), and pyochelin ( Cox et al., 1981), then these carbons were determined as the thiazoline carbons, therefore the 2-(2-hydroxyphenyl)-2-thiazoline substructure was further determined. The last COSY-defined fragments, H-9 ′ (δ H 3.39)/H 3 -11 ′ (δ H 1.19), together with the following HMBC correlations revealed the connection mode of atoms in the side chain. H-11 ′ and H 2 -6 ′ (δ H 4.34) showed the long-range coupling to C-8 ′ (δ C 170.6), then considering the chemical shifts of C-6 ′ (δ C 64.7) and C-8 ′, it can be conjectured that the connectivity of C-6 ′ and fragment C-9 ′ /C-11 ′ through an ester group. The fragment C-8 ′ /C-9 ′ /C-10 ′ was built by the observation of another cross peak of H 3 -11 ′ in the HMBC spectrum (H 3 -11 ′ /C-10 ′). Finally, on reviewing the molecular formula (C 14 H 16 N 2 O 4 S) given by high-resolution mass spectrometry, the last atom linked to C-10 ′ was allowed to assign as the nitrogen atom, which was further supported by the correlation of NHa (δ H 7.07, s)/C-9 ′ (δ C 45.8). Furthermore, the resonance of another well-resolved amide proton NHb (δ H 7.52, s) was visible ( Table 1). The nonequivalence of these two amide protons revealed hindered rotation of the amine group of 5. The amide group was vicinal with the ester group, a proton acceptor, which provided favorable conditions for forming a six-membered pseudo cycle through an intramolecular H-bonding (Gorobets et al., 2012). Therefore, the full planar structure of 5 was completed as shown in Fig. 2 .

The optical rotation value ([α] 25 D 68.6) revealed that two components of 5 were not enantiomers. Undoubtedly, assigning the absolute configuration of these two diastereomers was hard work as it was a mixture. However, after referring to the report of Tai et al. (2006), a conjecture was proposed in Fig. 4 , the absolute configuration of C-4 ′ should be consistent, and the stereochemistry of the chiral center of C-9 ′ presumably was a reciprocal transformation in equilibrium in two epimers via two processes involving keto-enol tautomerism and Z/E geometry transform. In view of those, there were two cases that can be considered for the epimers, which were [(4 ′ R, 9 ′ S)- 5 +(4 ′ R, 9 ′ R)- 5] and [(4 ′ S, 9 ′ R)- 5 +(4 ′ S, 9 ′ S)- 5]. Thus, we tried to the computed ECD with the methodology of time-dependent density functional theory (TD-DFT) of the two epimers in one case, and mathematical addition them according to the 1:1 ratio given by 1 H NMR. The calculated ECD superposition spectrum for the [(4 ′ R, 9 ′ R)- 5 and (4 ′ R, 9 ′ S)- 5] showed an excellent fit with the experimental data, where the [(4 ′ S, 9 ′ R)- 5 +(4 ′ S, 9 ′ S)- 5] showed positive CEs near 263 and 287 nm. The experimental ECD spectrum showed two positive CEs at 262 and 283 nm. Slight differences between the calculated and experimental data presumably resulted from minor differences between calculated and real conformers. Furthermore, with the comparation of experimental CD values of pulicatin A and B [ Lin et al., 2010], the R -configuration of C-4 ′ was further confirmed. In conclusion, the well-agreement of the TDDFT-calculated ECD curve between and the experimental one ( Fig. 7 ) indicated the two components of 5 to be (4 ′ R, 9 ′ R)- 5 and (4 ′ R, 9 ′ S)- 5.

Thiazostatin C (6) was obtained as a white solid. Its molecular formula was assigned to be C 15 H 19 N 3 O 2 S 2 on the basis of HRESIMS (m/z 338.1004, calcd for [M + H] + 338.0997), requiring eight degrees of unsaturation. Compound 6 showed the characteristic UV absorption of 2-(2-hydroxyphenyl)-2-thiazoline compounds under the process of analysis by HPLC equipped with a DAD detector, which was similar to thiazostatin C. Analysis of 1 H NMR data recorded in DMSO‑ d 6 of 6 ( Table 1) combined with the HSQC spectrum suggested three exchangeable protons at δ H 12.34 (1H, s), 7.27 (1H, s), and 7.15 (1H, s), a typic 1,2-disubstituted benzene ring at δ H 6.96 (1H, d, J =8.1 Hz), 7.43 (1H, t, J =7.7 Hz), 6.95 (1H, t, J =7.7 Hz), 7.41 (1H, d, J =8.0 Hz), two methines [δ H 5.30 (1H, td, J = 9.2, 3.5 Hz); 4.64 (1H, d, J = 3.5 Hz)], two methylenes [δ H 3.62 (1H, t, J =10.4 Hz), 3.50 (1H, t, J =10.4 Hz); 3.04 (1H, d, J = 10.5 Hz), 2.72 (1H, d, J =10.5 Hz)], and two methyls [δ H 2.34 (3H, s), 1.26 (3H, s)]. Observations of the 13 C NMR spectroscopic data ( Table 2) revealed 15 carbon signals attributing to a characteristic 2-substituted phenol moiety [δ C 115.7, 158.3, 116.8, 133.4, 119.3, 130.4], thiazoline ring [δ C 171.6, 78.0, 31.2], thiazolidine ring [δ C 72.8, 73.5, 40.2], one carbonyl carbon [δ C 174.8], and two methyls [δ C 34.9, 13.1]. The aforenoted NMR data showed nearly consistent similarity with those of thiazostatin B ( Shindo et al., 1989), except for the two additional exchangeable protons [δ H 7.27 (1H, s) and 7.15 (1H, s)], then the molecular formulas of thiazostatin B (C 15 H 18 N 2 O 3 S 2) manifested that the carboxyl group in thiazostatin B was replaced by an amide group to form 6. The planar structure was further confirmed by the key COSY and HMBC correlations given in Fig. 3 , especially for the HMBC correlations from NH proton (δ H 7.27, s) to C-4 ′′ (δ C 73.5).

Refer to the report of Rinehart et al. (1995), the anti -relationship between H-4 ′ and H-2 ′′ was assigned by observing to coupling constants and peak splitting of H-4 ′ (1H, td, J =9.2, 3.5 Hz) and H-2 ′′ (1H, d, J = 3.5 Hz). Then, an attempt to obtain nuclear overhauser enhancement to assign relative configuration of C-4 ′′ was unsuccessful, possibly owing to nonrigid conformations of the contiguous rings. For this reason, DFT-GIAO NMR calculations were used to assess the relative configuration of the two isomers [(4 ′ S,2 ′′ S,4 ′′ S)- 6; (4 ′ S,2 ′′ S,4 ′′ R)- 6] respectively, in which isomer 6a was determined as the correct one through classical statistical parameter (R 2, ME, CME, and CMAE) analysis ( Fig. 5 ). To determine its absolute configuration, electronic circular dichroism spectrum of 6 was measured in MeOH, which revealed positive CEs at 231, 262, and 282 nm ( Fig. 7 ). However, the lack of proper reference made assigning its absolute configuration unreliable by directly analyzing the CD curve. Thus, the ECD spectrum of 6 was compared with an ECD simulation of 6 at B3LYP/6-31 + G (d, p) level in MeOH. The match well of experimental ECD spectra of 6 with those of calculated (4 ′ S,2 ′′ S,4 ′′ S)- 6 (Similar positive CEs at 230, 254, and 286 nm) revealed the absolute configuration of 6 was 4 ′ S,2 ′′ S, 4 ′′ S.

Methyl thiazostatin B (7) was isolated as pale-yellow needles, the molecular formula was determined to be C 16 H 20 N 2 O 3 S 2 on the basis of the key HRESIMS ions peaks at m/z 375.0810 [M + Na] + and m/z 353.1011 [M +H] +. Comprehensive analysis of the NMR spectroscopic data (1 H and 13 C NMR data) of 7 suggested that it was also a structural analog of thiazostatin B ( Shindo et al., 1989), except for the presence of an additional highly deshielded methyl signal (δ H 3.75, s; δ C 52.7). In the HMBC spectrum, the correlation from H 3 – OMe to C-6 ′′ revealed the methyl group was linked with the carbonyl carbon C-6 ′′ through an oxygen atom, and formed the ester group ( Fig. 3 ). Consequently, the 2-dimension chemical structure of 7 was unambiguously established. Then configuration of chiral centers in 7 was considered. Same with those of 6, the coupling constants and peak splitting of H-4 ′ (1H, td, J = 9.0, 4.2 Hz) and H-2 ′′ (1H, d, J = 4.2 Hz) prompted us to assign the anti -relationship between H-4 ′ and H-2 ′′. Then, the Nuclear Overhauser Effect between 2 ′′ -H (δ H 4.59) and 8 ′′ - CH 3 (δ H 1.46) was observed in the NOESY spectrum (Fig. S24), suggested that both of them were located on the same side of the thiazolidine ring. Consequently, the relative stereochemistry of 7 was assigned as 4 ′ S *, 2 ′′ S *, 4 ′′ S *. The measured ECD spectrum of 6 revealed positive CEs at 262 and 282 nm, and negative CEs at 243 and 319 nm ( Fig. 7 ). However, the lack of proper reference made assigning its absolute configuration unreliable by directly analyzing the CD curve. Thus, the ECD spectrum of 7 was compared with an ECD simulation of 7 at B3LYP/6-31 + G (d, p) level in MeOH. The excellent agreement of the experimental and calculated (4 ′ S, 2 ′′ S, 4 ′′ S)- 7 ECD curve (similar positive CEs at 259 and 283 nm, negative CEs at 240 and 312 nm) indubitably established its absolute configuration as 4 ′ S, 2 ′′ S, 4 ′′ S.

Compound 8 was isolated as a white powder and assigned the molecular formula C 10 H 9 NO 3 based on HRESIMS analysis (m/z 192.0664 for [M+H] +) and 13 C NMR data ( Table 2). Compound 8 had a lower mass than spoxazomicin C ( Shaaban et al., 2017) by 2 Da, indicating that an additional unsaturated double bond was formed. The 1 H NMR and 13 C NMR spectra of 8 were very similar to those of spoxazomicin C except for having signals corresponding to a vinyl group [δ H 7.65, δ C 134.1 and δ C 140.0], along with for lacking the signals of a methine and a methylene groups. The inference was confirmed by the DFT-GIAO NMR calculation results ( Fig. 6 ) of a high confidence level (with good performance of R 2, MAE, and CMAE). Although its planar structure could be found in the Scifinder database, this was the first time to elucidate its structure and report its NMR data.

Spoxazomicin E (9) was obtained as a brown solid. Its HRESIMS gave to [M+H] + peak at m/z 293.1145 revealed the molecular formula C 14 H 16 N 2 O 5. Compare to 5, the sulfur atom was replaced by an oxygen atom. The 1 H and 13 C NMR spectra of 5 and 9 ( Tables 1 and 2) in DMSO‑ d 6 were similarity especially for the two methyl signals [δ H 1.16, d (7.1); δ H 1.14, d (7.1)] with a ratio of 1:1, revealing that they shared the same side chain. The most obvious differences of 13 C NMR data between compounds 9 and 5 were in the oxazole ring and its adjacent carbon such as [C-5 ′ (δ C 68.85/68.70); C-1 (δ C 109.96)] suggested the thiazole of 5 was replaced by an oxazole ring in 9, which was confirmed by the excellent similarity of 2-oxazolylbenzene unit on the carbon chemical shifts compared to the corresponding moiety of the reported spoxazomicin C. Same as the method for determining the configuration of the two components in 5, by simply adding the calculated ECD curves of the two possible configurations and comparing them with the measured curves ( Fig. 7 ), we assigned the two components of 9 as (4 ′ S, 9 ′ S)- 9 and (4 ′ S, 9 ′ R)- 9.

Streptochelin A (10) was obtained as a white solid. On the basis of the HR-ESI-MS and 13 C NMR spectral data, its molecular formula was assigned as C 13 H 11 N 3 O 3 S, in which the parent ion peak [M+H] + was detected at m/z 290.0603 (calcd for C 13 H 12 N 3 O 3 S), and requiring 10 degrees of unsaturation. The 13 C NMR and HSQC spectra ( Table 2 and Fig. S39) showed 13 carbons classified into one amide carbonyl carbon (δ C 171.7), five sp 2 quaternary carbons (δ C 160.8, δ C 159.6, δ C 156.3, δ C 134.4, δ C 110.0), five sp 2 methine carbons (δ C 140.4, δ C 133.1, δ C 127.5, δ C 119.8, δ C 117.1), one sp 3 methine carbon (δ C 79.2) and one sp 3 methylene carbon (δ C 34.3). The 1 H NMR spectrum revealed the presence of 1,2-disubstituted benzene fragment [δ H 7.08 (1H, d, J =8.4 Hz); 7.45 (1H, t, J = 7.6 Hz); 7.02 (1H, t, J = 7.6 Hz); 7.84 (1H, d, J = 7.8 Hz)]. The spin system from H-3 to H-6 and another spin system of H-4 ′′ / H-5 ′′ were confirmed based on the 1 H – 1 H COSY experiment ( Fig. 3 ). The HMBC correlations from H-4 to C-2 and from H-3 to C-1 along with the chemical shift of C-2 (δ C 156.3) suggested the existence of a 2-hydroxyphenyl moiety.

Comparison of the spectroscopic data of 10 and 8, and of 10 and 7 suggested separately the presence of an oxazole group and a thiazolidine group. The existence of a 4-(2-thiazolinyl)-oxazole moiety was further illustrated by the HMBC correlations from H-4 ′′ and H 2 -5 ′′ to C-2 ′′ and from H-5 ′ to C-2 ′′, C-4 ′ and C-2 ′. Then, it was expanded at C-4 ′′ to one amide through the HMBC correlations of H 2 -5 ′′ (δ H 3.63)/C-6 ′′ (δ C 171.3) and NHa (δ H 7.40)/C-4 ′′ (δ C 76.2). The full planar structure was finally constructed with the key HMBC correlation from H-6 to C-2 ′. Its chiral center C-4 ′′ was assigned the S -configuration due to the computed ECD spectrum (positive CEs at 208, 252, 280 and 288 nm, negative CEs at 233 and 266 nm) of (4 ′′ S)- 10 agreed with the experimental spectrum (positive CEs at 206, 255, and 287 nm, negative CEs at 230 and 260 nm).

By comparing the NMR spectroscopic data and optical rotation values with literature values, other four known compounds were identified as aeruginol (1) ( Yang et al., 1993), (R)-2-(2-Hydroxyphenyl)-4-hydroxymethyl-4,5-dihydrothiazole (2) ( Li et al., 2014), Pulicatin A (3) ( Lin et al., 2010), Pulicatin B (4) ( Lin et al., 2010).

Figures

Fig. 1. Detected chemistries of the EtOAc extract of Streptomyces sp. NBU2966 as generated by LC-MS/MS, which was analyzed using NAP, Dereplicator+, and MolNetEnhancer workflow via the GNPS platform. With this network, wherein nodes represent a precursor ion, and its size is scaled to signal intensity and the thickness of edge between nodes suggests the similarity of fragment pattern. (a) Structural annotation for molecular families, wherein the color of nodes denotes the structural annotation at the superclass level by NAP. (b) Observation of molecular family A allows highlighting dereplicated (R)-2-(2-Hydroxyphenyl)-4-hydroxymethyl-4,5-dihydrothiazole. (c) Observation of molecular family B allowed to highlighting dereplicated pyochelin methyl ester. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 2. Structures of siderophores 1–10. Three types of iron-chelating moieties are marked with blue, red, and purple, respectively. Previously undescribed natural products are highlighted with red subscript numbers. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 3. The key COSY and HMBC correlations of 5, 6, 7, 9, and 10.

Fig. 4. Putative mechanism leading to the formation of epimers in 5 and 9.

Fig. 7. Experimental and calculated ECD spectra of compounds 5, 6, 7, 9, and 10.

Fig. 5. NMR calculation results of two plausible isomers of 6. (a) Linear correlation plots of experimental vs computed 13C NMR chemical shifts. (b) Relative errors between the computed 13C NMR values of two potential structures and experimental 13C NMR chemical shifts. (c) The evaluation of calculation results with statistical parameter ME (Maximum Error), CME (Corrected Maximum Error), and CMAE (Corrected Mean Absolute Error).

Fig. 6. NMR calculation results of 8. (a) Linear correlation plots of computed vs experimental 13C and 1H NMR chemical shifts. (b) Relative errors between the computed NMR values and experimental values. (c) The evaluation of NMR calculation results with statistical parameter MAE and CMAE.