Investigation on Metabolites in Structure and Biosynthesis from the Deep-Sea Sediment-Derived Actinomycete Janibacter sp. SCSIO 52865

For exploring structurally diverse metabolites and uniquely metabolic mechanisms, we systematically investigated the chemical constituents and putative biosynthesis of Janibacter sp. SCSIO 52865 derived from the deep-sea sediment based on the OSMAC strategy, molecular networking tool, in combination with bioinformatic analysis. As a result, one new diketopiperazine (1), along with seven known cyclodipeptides (2–8), trans-cinnamic acid (9), N-phenethylacetamide (10) and five fatty acids (11–15), was isolated from the ethyl acetate extract of SCSIO 52865. Their structures were elucidated by a combination of comprehensive spectroscopic analyses, Marfey’s method and GC-MS analysis. Furthermore, the analysis of molecular networking revealed the presence of cyclodipeptides, and compound 1 was produced only under mBHI fermentation condition. Moreover, bioinformatic analysis suggested that compound 1 was closely related to four genes, namely jatA–D, encoding core non-ribosomal peptide synthetase and acetyltransferase.


Introduction
Deep-sea-derived microorganisms may produce uniquely metabolic mechanisms as a result of surviving in extreme conditions of higher pressure, darkness, lower temperature and lack of oxygen, and it was thought to be as a rich source of structurally diverse and bioactive metabolites for drug discovery [1][2][3]. Some compounds derived from deep-sea bacteria exhibit significant biological properties, such as antimicrobial [4], anti-inflammatory [5], antioxidant [6], antiviral [7], and cytotoxic [8] activities. The genus Janibacter that has been isolated from various environmental sources [9] possesses strong biodegradation abilities for polycyclic aromatic hydrocarbons [10], pentachlorophenol [11], dibenzofuran [12], mono-chlorinated dibenzo-p-dioxins [13] and crude petroleum oil [14], but few metabolites from the genus are reported.
As a continuous effort in finding novel metabolites with structural diversity and biological potentiality from bacteria isolated from the deep-sea sediment, we systematically investigate the chemical constituents and bioinformation of SCSIO 52865 collected from the South China Sea at a depth of 3448 m. The eleven different media were used for cultivating the strain based on the OSMAC (One Strain Many Compounds) strategy [15]. Together with the analysis of molecular networking [16], the mBHI medium was selected as fermentation condition in large scale due to its widespread nodes distributing in molecular networking clusters. The strategy seems to be a simple and efficient for awakening silent biosynthetic gene clusters, and we have reported some new compounds in our previous fermentation condition in large scale due to its widespread nodes distributing in molecular networking clusters. The strategy seems to be a simple and efficient for awakening silent biosynthetic gene clusters, and we have reported some new compounds in our previous work [17,18]. Eventually, the EtOAc extract was obtained from a total of 47 L fermentation broth using Erlenmeyer flasks shaking. One new diketopiperazine, janibatide A (1), along with fourteen known compounds (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15), was further isolated and identified ( Figure 1). The planar structure of 1 was elucidated by detailed 1D/2D NMR spectroscopy and HRESIMS data, and its absolute configuration was further confirmed by Marfey's method.

Analysis of Molecular Networking
To fully uncover the metabolic profiles of SCSIO 52865 isolated from deep-sea sediment, eleven different media (Table S1) were used to explore for metabolites based on OSMAC strategy. In aggregate, except for mM20, other HPLC chromatography ( Figure  S1) of EtOAc extracts exhibited similar HPLC-UV profiles within the scope of 0~12 min. Three EtOAc extracts from mBHI, mA and mISP2 media were selected to further measure their LC-MS/MS spectra under the positive ESI mode and the MS/MS data was used to construct molecular networking (MN), due to slightly different HPLC-UV profiles 12 min later. A visual graph (Figures 2a and S2) was drawn using Cytoscape software [19], in which red, green and pink nodes represented for MS data from EtOAc extracts of mBHI, mA and mISP2 fermentation broths, respectively. The MN contained eighteen network clusters within >two nodes and twenty-three network clusters within two nodes. In addition, we found that the above three EtOAc extracts could produce the same parent mass or structural analogues in MN clusters, and simultaneously they were alone able to generate characteristic parent mass in single nodes. These results were coincident with those of HPLC chromatography ( Figure S1). Additionally, we discovered that the red nodes, corresponding to EtOAc extracts from mBHI fermentation broth, widely distributed in different clusters and possessed some unique parent mass. Consequently, the mBHI medium was used as fermentation condition on a large scale for enough samples. Furthermore, a particular node of parent ion at m/z 303.134 in the MN (Figure 2b) was corresponding to compound 1, a new natural cyclic dipeptide derivative, whose planar structure was recorded on a patent of solid phase apparatus, but without detailed NMR data [20]. Meanwhile, some known cyclodipeptides were further illustrated in the MN ( Figure S2),

Analysis of Molecular Networking
To fully uncover the metabolic profiles of SCSIO 52865 isolated from deep-sea sediment, eleven different media (Table S1) were used to explore for metabolites based on OSMAC strategy. In aggregate, except for mM20, other HPLC chromatography ( Figure S1) of EtOAc extracts exhibited similar HPLC-UV profiles within the scope of 0~12 min. Three EtOAc extracts from mBHI, mA and mISP2 media were selected to further measure their LC-MS/MS spectra under the positive ESI mode and the MS/MS data was used to construct molecular networking (MN), due to slightly different HPLC-UV profiles 12 min later. A visual graph (Figures 2a and S2) was drawn using Cytoscape software [19], in which red, green and pink nodes represented for MS data from EtOAc extracts of mBHI, mA and mISP2 fermentation broths, respectively. The MN contained eighteen network clusters within >two nodes and twenty-three network clusters within two nodes. In addition, we found that the above three EtOAc extracts could produce the same parent mass or structural analogues in MN clusters, and simultaneously they were alone able to generate characteristic parent mass in single nodes. These results were coincident with those of HPLC chromatography ( Figure S1). Additionally, we discovered that the red nodes, corresponding to EtOAc extracts from mBHI fermentation broth, widely distributed in different clusters and possessed some unique parent mass. Consequently, the mBHI medium was used as fermentation condition on a large scale for enough samples. Furthermore, a particular node of parent ion at m/z 303.134 in the MN (Figure 2b) was corresponding to compound 1, a new natural cyclic dipeptide derivative, whose planar structure was recorded on a patent of solid phase apparatus, but without detailed NMR data [20]. Meanwhile, some known cyclodipeptides were further illustrated in the MN ( Figure S2), corresponding to the characteristic signals appeared in HPLC-UV chromatography, which possessed terminal absorption profiles with major polarity. corresponding to the characteristic signals appeared in HPLC-UV chromatography, which possessed terminal absorption profiles with major polarity.

Structural Elucidation of Compounds
Compound 1 was isolated as a white powder. Its molecular formula was assigned as C16H18N2O4 based on the molecular ion peak at m/z 303.1348 [M+H] + (calcd for C16H19N2O4 303.1339), indicating nine degrees of unsaturation. The 1 H NMR spectrum ( Figure S7 , three methylenes (one nitrogenated carbon at δC 52.9; and two carbons at δC 38.2 and δC 36.1), and one methyl (δC 20.9). The abovementioned data suggested the presences of two structural fragments for benzyl and acetyl moieties, which were confirmed by detailed analyses of HMBC and 1 H-1 H COSY correlations (Figure 3), and we further determined the structure of cyclo(Pro-Phe) scaffold based on both the similar NMR data with cyclo(L-trans-Hyp-L-Phe) (compound 2) [21] and 2D NMR spectra ( Figures S9-S12). The acetyl moiety was positioned at C-4 according to HMBC correction from H-4 to C-1′, to establish the planar structure of 1.

Structural Elucidation of Compounds
Compound 1 was isolated as a white powder. Its molecular formula was assigned as C 16 Table 1) showed sixteen resonances for four quaternary carbons (three carbonyl carbons at δ C 172.0, δ C 170.5, and δ C 167.0; one olefinic carbon at δ C 137.0), eight methines (five olefinic carbons at δ C 131.1 × 2, δ C 129.6 × 2, and δ C 128.2; one oxygenated carbon at δ C 71.0; and two nitrogenated carbons at δ C 58.4 and δ C 57.7), three methylenes (one nitrogenated carbon at δ C 52.9; and two carbons at δ C 38.2 and δ C 36.1), and one methyl (δ C 20.9). The above-mentioned data suggested the presences of two structural fragments for benzyl and acetyl moieties, which were confirmed by detailed analyses of HMBC and 1 H-1 H COSY correlations (Figure 3), and we further determined the structure of cyclo(Pro-Phe) scaffold based on both the similar NMR data with cyclo(L-trans-Hyp-L-Phe) (compound 2) [21] and 2D NMR spectra (Figures S9-S12). The acetyl moiety was positioned at C-4 according to HMBC correction from H-4 to C-1 , to establish the planar structure of 1.   , compound 1 was elucidated as cyclo(4R-acetyl-L-Pro-L-Phe), which was strongly supported by the result of Marfey's analysis. The absolute configurations of 4-hydroxy-Pro and Phe moieties were unambiguously determined as 4R-hydroxy-L-Pro and L-Phe by comparison with the HPLC retention times of standard amino acid derivatives (Table 2), in which 4R-hydroxy-L-Pro-L-FDAA derivative was derived from cyclo(L-trans-Hyp-L-Leu) that was determined by single-crystal X-ray diffraction analysis in our previous paper [17]. Thus, the structure of compound 1 was confirmed and named as janibatide A.

Amino Acid
Standard Cyclo(L-trans-Hyp-L-Leu) 1 2 5 6 7  (Table 2), in which 4R-hydroxy-L-Pro-L-FDAA derivative was derived from cyclo(L-trans-Hyp-L-Leu) that was determined by single-crystal X-ray diffraction analysis in our previous paper [17]. Thus, the structure of compound 1 was confirmed and named as janibatide A.  1 The retention times of L-amino acid derivatives. 2 The retention times of D-amino acid derivatives. 3 The retention time of L-FDAA. 4 The retention time of 4(R)-OH-Pro derivative.

Putatively Biosynthetic Pathway of Compound 1
A circular contig of 3,495,359 bp ( Figure S3) with a GC content of 70.95% was produced by genome sequencing. Additionally, four secondary metabolite biosynthetic gene clusters (BGCs) were predicted by antiSMASH bacterial version 6.1.1 with the default settings [31], namely regions 1.1-1.4, corresponding to T3PKS, ectoine, terpene and NRPS-like types (Table S2 and Figure S4). In general, cyclic dipeptides scaffold can be naturally synthesized by either non-ribosomal peptide synthetases (NRPSs) or CDP synthases (CDPSs) [32]. Clearly, the only NRPS-like-type cluster was likely responsible for biosynthesis of cyclic dipeptide derivatives in the strain. The detailed bioinformation analysis revealed that compound 1 was closely related to four core genes, jatA-D ( Figure 4, Scheme 1, and Tables S3 and S4).

Putatively Biosynthetic Pathway of Compound 1
A circular contig of 3,495,359 bp ( Figure S3) with a GC content of 70.95% was produced by genome sequencing. Additionally, four secondary metabolite biosynthetic gene clusters (BGCs) were predicted by antiSMASH bacterial version 6.1.1 with the default settings [31], namely regions 1.1-1.4, corresponding to T3PKS, ectoine, terpene and NRPSlike types (Table S2 and Figure S4). In general, cyclic dipeptides scaffold can be naturally synthesized by either non-ribosomal peptide synthetases (NRPSs) or CDP synthases (CDPSs) [32]. Clearly, the only NRPS-like-type cluster was likely responsible for biosynthesis of cyclic dipeptide derivatives in the strain. The detailed bioinformation analysis revealed that compound 1 was closely related to four core genes, jatA-D ( Figure 4, Scheme 1, and Tables S3 and S4). Firstly, compound 3 was putatively synthesized by a combination of non-ribosomal peptide synthetase (JatA) and 4′-phosphopantetheinyl transferase (JatB). The synergistic Scheme 1. Isolated compounds and proposed biosynthesis pathway for janibatide A (1). Firstly, compound 3 was putatively synthesized by a combination of non-ribosomal peptide synthetase (JatA) and 4 -phosphopantetheinyl transferase (JatB). The synergistic enzymes were likely to possess one specific proline recognition domain to form cyclic dipeptide derivatives containing L-or D-proline moiety according to isolated structures. Subsequently, compound 3 was converted to compound 2 by an oxidation reaction putatively catalyzed by the enzyme JatD, a ferredoxin that can transfer electrons and catalyze the formation of hydroxy in corporation with oxygenase, especially P450 [33,34]. However, we did not seek out any oxygenase in near both upstream and downstream of the ferredoxin, possibly suggesting a particularly biosynthetic mechanism that needs to be studied in future. In addition, we discovered that only cyclo(L-Pro-L-Phe) was hydroxylated from all isolated cyclodipeptides; however, others, including cyclo(D-Pro-L-Phe), were not hydroxylated, which indicated that the enzyme JatD probably possessed substrate specificity. Furthermore, compound 1 was putatively produced by a reaction of compound 2 with acetyl-CoA catalyzed by the enzyme JatC. Meanwhile, two putative prolyl-tRNA synthetase genes (proS) were found by Swissprot annotation (Table S3), in connection with cyclodipeptides containing L-or D-proline moiety. Moreover, analysis of structural characteristics revealed that compounds 1-8 all possessed proline moiety, and compounds 6-8 had the same planar structures with compounds 3-5, respectively; the only difference was the configuration of proline residue. Given the fact that only one NRPS-like gene cluster existed, compounds 4-8 were putatively synthesized by JatA/B. The synergistic enzymes exhibited specificity for Pro, the first amino acid, but have not specificity for another amino acid which is likely as hydrophobic amino acids such as Phe, Leu and Ile.

Biological Activities
All isolated compounds were tested for antibacterial activity against three Grampositive and one Gram-negative bacteria (Bacillus subtilis, Bacillus thuringiensis, Staphylococcus aureus and Escherichia coli), resulting in that none of which exhibited obvious antibacterial activity. Additionally, compound 1 was measured for cytotoxicity against HL-60 human tumor cell line by CCK-8 assay [36], but the result displayed no obvious inhibitory activity at concentration of 100 µM. Moreover, all compounds were evaluated for α-glucosidase inhibitory activity, and all of which did not show inhibitory effects at the concentration of 166.7 µg/mL (Table S13). Instead, the fatty acids, especially compound 12, seemed to significantly increase the conversion of PNPG to PNP catalyzed by α-glucosidase in comparison with the negative control.

Microorganism and Growth Conditions
The strain SCSIO 52865 was isolated and purified from sediment collected from the South China Sea (13 • 08 40 N, 114 • 38 21 E) at a depth of 3448 m. Analysis of the 16S rRNA sequence revealed that the strain was a member of the Janibacter sp. and shared 99.13% identity with Janibacter cremeus HR08-44(T) (GenBank accession no. AB778259). Initially, the strain was cultivated in eleven different liquid media (Table S1). Subsequently, the strain was cultivated in 500 mL Erlenmeyer flasks each containing 200 mL of mBHI culture broth at 28 • C for 7 days with shaking rate at 180 rpm, and a total of 47 L fermentation was obtained.

Whole Genome Sequencing and Bioinformatic Analysis
The genome of SCSIO 52865 was extracted by Oxford Nanopore Technologies (ONT) protocol [37]. In brief, the quality of high molecular weight genomic DNA (gDNA) was controlled by a combination of Nanodrop, Qubit and 0.35% agarose gel electrophoresis, and large fragments were selected by automatic BluePippin system. Subsequently, the library was constructed by the SQK-LSK109 ligation kit (Nanopore, Oxford, UK). The circular contig of 3,495,359 bp with a GC content of 70.95% was assembled by using Canu v1.5 for assembly, Racon v3.4.3 for rectification, Circlator v1.5.5 for cyclization and Pilon v1.22 for correction. The genomic sequence has been deposited in GenBank under accession number CP115184. Genomic sequence annotation was performed by using general database, including KEGG, Pfam, and SwissProt, and the putative secondary metabolite BGCs were predicted by antiSMASH version 6.1.1.

Extraction and Isolation
The culture (47 L) was extracted three times with an equal volume of EtOAc at room temperature. The EtOAc layer was separated from the aqueous phase, and it was evaporated in vacuo to give a dry EtOAc extract (8.  Table 1 and Figures S7-S12).

Molecular Networking
The experimental procedure has been described in our previous paper [17].

GC-MS Analysis
GC-MS analysis was conducted on GCMS-QP2010 Ultra system (Shimadzu). A 1 µL aliquot (2 mg/mL, dissolved in dichloromethane) was injected into analytic system fitted with RXI-5MS (30.0 m × 0.25 mm, 0.25 µm, Shimadzu, Kyoto, Japan) capillary column. Ultra-high-purity helium was used as carrier gas at a constant flow rate of 1.2 mL/min. The injection, transfer line, and ion source temperatures were 250 • C, 280 • C and 220 • C, respectively. The oven temperature was programmed from 50 • C (hold for 3 min) to 320 • C (hold for 5 min) at a rate of 10 • C/min. The first 4 min was the solvent delay, and the mass spectral data were collected from m/z 45-500 for 35 min. These fatty acids were further identified by comparison of their mass spectra with those of reference compounds recorded in the National Institute of Standards and Technology (NIST) mass spectral library.

Biological Assays
Antibacterial evaluation against four indicator bacteria was described in our previous work [17].
Cytotoxicity assay was implemented against the human tumor cell line HL-60 using WST-8 reagent. In brief, when the density of HL-60 cell was near 1 × 10 6 /mL, the HL-60 cell was incubated by adding fresh medium for subculture to maintain 4 × 10 5 /mL density. Then, the compound 1 or Staurosporine as positive control in a gradient descent manner was added in well which has added 5 × 10 5 /mL HL-60 with 50 µL, and the plates were incubated in a 5% CO 2 incubator at 37 • C for 72 h. Subsequently, each cell was treated with 10 µL CCK-8 in a 5% CO 2 incubator at 37 • C for 2 h. Optical density was performed on an EnVision spectrophotometer (PerkinElmer, Waltham, MA, USA) at 450 nm. The inhibition rate was calculated as the equivalent of the following: in which OD S , OD NC and OD BLK were the absorption values of well with the additional test compound, and DMSO as the negative control and blank, respectively. The inhibitory activity of α-glucosidase was tested according to the modified method described in the references [38,39], in which p-nitrophenyl-α-D-glucopyranoside (PNPG) was as a substrate. A total of enzyme solution (20 µL, 0.5 U/mL in 0.2 M PBS, pH 6.8), the tested compound (10 µL, 2 mg/mL in DMSO) and PBS (50 µL) were mixed in a 96-well microplate and preincubated at 37 • C for 15 min. The PNPG solution (20 µL, 5.0 mM in PBS) was added and incubated at 37 • C for 15 min, then Na 2 CO 3 (20 µL, 0.2 M in PBS) was added to stop the reaction. The released PNP was quantified by a microplate reader (Synergy H1, BioTek, Winooski, VT, USA). Acarbose and DMSO were used as the positive control and negative control, respectively.

Conclusions
Initially, we used the OSMAC strategy in combination with the visual MN graph to screen for fermentation condition, and we detected some compounds that were not annotated by GNPS, especially compound 1. Lastly, fifteen compounds were isolated and identified, including one cyclic dipeptide derivative, janibatide A (1), from the EtOAc extract via a combination of extensive spectroscopic analyses, Marfey's method and GC-MS analysis. To our knowledge, this is the first report on the isolation of these metabolites from actinomycete Janibacter sp. The bioactive assays showed that all compounds have not displayed obvious antibacterial and cytotoxic activities as well as inhibitory effect against α-glucosidase. However, cyclo(L-Pro-L-Phe) (3) was reported against Pencillium expansum at 2 µg/mL [23], and unsaturated fatty acids, especially oleic acid (13), presented modulatory effects in a widely physiological functions [40]. Meanwhile, cyclodipeptides were putatively correlated with quorum-sensing [32], suggesting noncompetitive property, but these remain to be investigated in depth. Moreover, the whole-genome sequencing result indicated that SCSIO 52865 possessed 3.49 Mbp sequence, and only four BGCs were predicted by using antiSMASH platform, in which only NRPS-like type BGCs had a lower similarity to known clusters with value of 7%, and others were beyond 66%. Furthermore, the bioinformatic analysis disclosed that cyclodipeptides (1)(2)(3)(4)(5)(6)(7)(8) were bound up with NRPSlike-type BGCs, and four core genes, jatA-D, were putatively responsible for biosynthesis of compound 1.