Streptomyces sp. VN1, a producer of diverse metabolites including non-natural furan-type anticancer compound

Streptomyces sp. VN1 was isolated from the coastal region of Phu Yen Province (central Viet Nam). Morphological, physiological, and whole genome phylogenetic analyses suggested that strain Streptomyces sp. VN1 belonged to genus Streptomyces. Whole genome sequencing analysis showed its genome was 8,341,703 base pairs in length with GC content of 72.5%. Diverse metabolites, including cinnamamide, spirotetronate antibiotic lobophorin A, diketopiperazines cyclo-L-proline-L-tyrosine, and a unique furan-type compound were isolated from Streptomyces sp. VN1. Structures of these compounds were studied by HR-Q-TOF ESI/MS/MS and 2D NMR analyses. Bioassay-guided purification yielded a furan-type compound which exhibited in vitro anticancer activity against AGS, HCT116, A375M, U87MG, and A549 cell lines with IC50 values of 40.5, 123.7, 84.67, 50, and 58.64 µM, respectively. In silico genome analysis of the isolated Streptomyces sp. VN1 contained 34 gene clusters responsible for the biosynthesis of known and/or novel secondary metabolites, including different types of terpene, T1PKS, T2PKS, T3PKS, NRPS, and hybrid PKS-NRPS. Genome mining with HR-Q-TOF ESI/MS/MS analysis of the crude extract confirmed the biosynthesis of lobophorin analogs. This study indicates that Streptomyces sp. VN1 is a promising strain for biosynthesis of novel natural products.

Phylogenetic analysis. Strain Streptomyces sp. VN1 was cultivated in starch-casein broth. Genomic DNA extraction was carried out according to the standard procedure 22 . Cloned 16S rRNA was sequenced using the Sanger sequencing method. A 16S rRNA sequence of 1,467 nucleotides in length was obtained and deposited in GenBank (accession number: KU878019). This sequence was compared with sequences obtained from the online platform EzBioCloud (http://eztaxon-e.ezbiocloud.net/) through BLAST 23 . Phylogenetic analysis of the strain was performed to determine its taxonomic position within the genus of Streptomyces. For the analysis of genomic sequence, automated multi-locus species tree (autoMLST) 24 pipeline was used to generate a phylogenetic based on alignment of default parameters (>100 core genes) present in closely related genomes using default parameters. Selected genomes were subjected to in silico DNA-DNA hybridization (DDH) and values were calculated using Genome-to-Genome Distance Calculator (GGDC) 25   Mass spectrometric analysis of crude extraction sample. Marine broth-malt extract was used as fermentation medium. Briefly, 5 L of fermentation broth was extracted with 10 L of methylene chloride and evaporated under reduced pressure. The concentrate was collected using 50 mL of methanol to give residue A. The supernatant was extracted for the second time with 10 L ethyl acetate and its concentrate was collected using 50 mL MeOH to give residue B. These crude extracts were analyzed by ultra-high-performance liquid chromatography electrospray ionization quadrupole time of flight high-resolution mass spectrometry (UPLC-ESI-Q-TOF-HRMS) analysis using ACQUITY UPLC ® (Waters Corporations, Milford, MA, USA) coupled with SYNAPT G2-S (Waters Corporations). UPLC ® /HRMS analysis was performed on an ACQUITY UPLC ® BEH C18 column (1.7 μm, 2.1 × 100 mm) at a flow rate of 0.3 mL min −1 using a gradient solvent mobile phase A (H 2 O, 0.1% trifluoroacetic acid) and B 0% to 100% ACN (0 to 12 min) at 35 °C. The injection volume was 10 µL. Conditions used for the high-resolution mass spectrometer equipped with an electrospray ionization source were: 3 kV of capillary voltage, 300 °C of desolvation gas temperature, and 600 L/h of desolvation gas flow rate. The mass range was set from m/z 50 to 1,700 in positive mode. MS/MS detection mode was set to a ramp trap collision high energy from 20 V to 40 V.
IR (Infrared Spectroscopy) spectra were obtained using an EQUINOX 55 FT-IR spectrometer (BRUKER Optik GmbH, Wikingerstr. 13 76189 Karlsruhe, Germany). CD (Circular Dichroism) spectra were recorded using a JASCO J-715 spectrophotometer (Ochang, Republic of Korea). The CD spectrum was analyzed with a CDtoolX 31 software. Cytotoxicity assay. Human cancer cell lines AGS, HCT116, A375SM, A549 and U87MG were obtained from the Korean Cell Line Bank (KCLB, Seoul, Korea). Human normal cell lines 267B1 and MRC-5 were purchased from the American Type Culture Collection (ATCC; Manassas, VA, USA). AGS gastric cancer and HCT116 colon cancer cells were maintained in RPMI 1640 medium containing 10% fetal bovine serum (FBS). A375SM melanoma and A549 lung cancer cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% FBS. U87MG glioblastoma cells, normal cell lines of prostate epithelial (267B1) and lung fibroblast (MRC-5) were cultured in Minimum Essential Medium (MEM) containing 10% FBS. All cells were maintained at 37 °C in a humidified 5% CO 2 incubator. For cell-growth assay, various cancer cells were plated into 96-well culture plates at density of 2 × 10 3 cells/well. After compound 4 was added to each well at various concentrations (0, 0.78, 1.56, 3.12, 6.25, 12.5, 25, 50, 100, and 200 µM), cells were incubated at 37 °C for 72 h. Cell growth was measured using a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) colorimetric assay. Briefly, 50 µL of MTT (2 mg/mL stock solution) was added and plates were incubated at 37 °C for an additional 4 h. After removal of medium, 100 µL of dimethyl sulfoxide (DMSO) was added to each well. The absorbance was measured at 540 nm using a microplate spectrophotometer (Thermo Scientific Multiskan ® Spectrum). The cytotoxicity assays were performed on triplets and results are presented as mean values ± standard error (SE).
Wound-healing assay. The migratory potential of AGS gastric cancer cells was analyzed using a wound-healing assay following published methodology 32 . Briefly, a confluent monolayer of AGS cells was scratched using a tip. Each well was then washed with PBS to remove nonadherent cells. Cells were then treated with compound 4 at 25 µM or 50 µM and then incubated at 37 °C for up to 24 h. The perimeter of the area with a central cell-free gap was confirmed under an optical microscope (Olympus). Dotted white line indicates the edge of the gap at 0 h. The wound-healing assays were performed on triplets and results are presented as mean values ± standard error (SE).
Cell invasion assay. The experiment was performed following a previously described methodology 32 . Cell invasion was examined using Transwell chamber inserts with a pore size of 8.0 µm. Lower and upper sides of the polycarbonate filter were coated with gelatin (1 mg/mL) and Matrigel (3 mg/mL), respectively. AGS cells were www.nature.com/scientificreports www.nature.com/scientificreports/ seeded into the upper chamber of the filter while compound 4 at 25 µM or 50 µM was added to the lower chamber filled with medium. The chamber was incubated at 37 °C for 24 h. Cells were then fixed with methanol and stained with hematoxylin/eosin. The total number of cells that invaded the lower chamber of the filter was counted using an optical microscope (Olympus). The cell invasion assays were performed on triplets and results are presented as mean values ± standard error (SE).

Statistical analysis.
Results are presented as mean values ± standard error (SE). Student's t-test was used to determine statistical significance between control and test groups. Statistical significance was considered at p < 0.05.

Results
Morphological and phenotypic characteristics. Scanning electron microscope observations revealed that the substrate mycelium of Streptomyces sp. VN1 was rectiflexible. When the culture of Streptomyces sp. VN1 reached maturity and adequate aerial mycelium was produced, aerial hyphae differentiated into short, straight to flexuous chains ( Fig. 2A) with smooth surfaces. Its spores were found to be regular and smooth (Fig. 2B). Streptomyces sp. VN1 displayed survival on sole carbon sources of L-arabinose, fructose, mannitol, sucrose, xylose, lactose, and starch. Starch and yeast extract were preferred carbon source and nitrogen source, respectively (Table 1). Streptomyces sp. VN1 produced brownish to grayish mycelium with good sporulation on the media (i.e., Marine broth-malt extract medium) used for metabolite isolation. It showed growth at temperatures of 24 °C to 40 °C with pH between 4 and 9. Various salts had different effects on the growth of Streptomyces sp. VN1 (Supplementary Table S1).
Phylogenetic analysis and genome annotation. Whole genome sequencing of Streptomyces sp.
VN1 produced a total of 179,193 sequence reads, yielding a total consensus of 8,341,703 bp with GC content of 72.5% distributed within one main contig. There are 7,176 protein-coding genes, with an average ORF length of 1,018 bp. Coding density was about 87.57%. Within Streptomyces sp. VN1, 86 tRNAs and 18 rRNA operons were predicted (Table 2). Predicted proteins were annotated by blasting the eggNOG database. In eggNOG functional classification, 6,987 (97.36%) of 7,176 proteins were assigned. The following four top categories were classified: transcription, carbohydrate metabolism, amino-acid metabolism, and energy production ( Supplementary  Fig. S1). Genome analysis with autoMLST showed that this new isolate had the highest sequence similarities with Streptomyces sp. FXJ7.023 (GCF_000404005), Streptomyces pactum (GCF_001767375) and Streptomyces olivaceus (GCF_000721235) (Fig. 3 Table S3). These data indicate that Streptomyces sp. VN1 is most closely related to Streptomyces pactum and Streptomyces olivaceus.
Cluster 1 contained a oligosaccharide-T1PKS-T3PKS-NRPS that was highly similar to lobophorin BGCs from Streptomyces sp. FXJ7.023 and Streptomyces sp. SCSIO 01127 (Table 3). A detailed analysis of the lobophorin A BGC of Streptomyces sp. VN1 indicated the involvement of 43 ORFs (open reading frame) and five catalytic domains related to polyketide synthases (Fig. 4). Lobophorin A biosynthesis gene clusters of Streptomyces sp. VN1 is highly similar to previously characterized BGCs which shared around 87-100% identities with corresponding genes in Streptomyces sp. FXJ7.023 and Streptomyces sp. SCSIO 01127 (Supplementary Table S5A). Next to the lobophorin A biosynthetic gene cluster was predicted for the T3PKS-NRPS biosynthetic gene cluster. It contained 48 ORFs, one catalytic domain related to chalcone/stilbene synthase, two catalytic domains related to polyketide synthases and three catalytic domains related to non-ribosomal peptide synthases. Recently, this T3PKS-NRPS biosynthetic gene cluster has been characterized as being responsible for synthesizing totopotensamide 34 . In-depth bioinformatic analysis of the putative biosynthesis gene cluster for totopotensamide in Streptomyces sp. VN1 is highly similar to previously characterized BGC which showed about 96-100% similarity with corresponding genes in Streptomyces sp. SCSIO 02999 (Supplementary Table S5B

Growth on nitrogen sources
Cystine +  www.nature.com/scientificreports www.nature.com/scientificreports/ lobophorin biosynthesis 35,36 . However, lobR1 has been recently annotated as totR5 belonging to the tetR family transcriptional regulator that is responsible for regulating totopotensamides synthesis 37 in Streptomyces sp. SCSIO 02999. Besides, the end of cluster 1 in Streptomyces sp. VN1 contained four ORFs annotated for four regulator genes and a unique ORF (orf98) annotated for N-acetyltransferase (Supplementary Table S5B).
Analysis results suggested that cluster five is highly similar with previously characterized carotenoid biosynthesis gene cluster. It shared 54% similarity with an existing cluster in Streptomyces avermitilis ( Supplementary  Fig. S2). This cluster contained 20 ORFs for biosynthesis of carotenoid in Streptomyces sp. VN1. Core ORFs www.nature.com/scientificreports www.nature.com/scientificreports/ biosynthesis such as phytoene synthase, lycopene cyclase, dehydrogenase, and methyltransferase were annotated. This cluster of Streptomyces sp. VN1 was predicted to have unique ORFs annotated as triacylglycerol lipases and short chain dehydrogenase/reductases (SDR), indicating that Streptomyces sp. VN1 might have the potential to synthesize carotenoid analogs.
Furthermore, two NRPS-hybrid synthases including a T1PKS-NRPS gene cluster showed high similarities to friulimicin and xiamycin biosynthetic gene cluster present in Actinoplanes friuliensis and Streptomyces sp. SCSIO 02999. From the friulimicin biosynthetic gene cluster, 12 core ORFs were annotated. The putative friulimicin biosynthetic gene cluster from Streptomyces sp. VN1 contained an additional condensation domain containing protein, a phosphopantetheine-binding domain containing protein synthetase, and an AMP-dependent protein and ligase ( Supplementary Fig. S3). Therefore, Streptomyces sp. VN1 has high potential to produce friulimicin analogs. The putative xiamycin biosynthesis gene cluster from Streptomyces sp. VN1 was found to contain 33 ORFs (Supplementary Fig. S4). In this gene cluster, core and additional biosynthetic genes were annotated to polyprenyl synthetase, cytochrome P450, acyl-CoA dehydrogenase, and a crotonyl-CoA reductase/alcohol dehydrogenase including additional dehydrogenase. Compared to the most-similar known cluster of Streptomyces sp. SCSIO 02999, this gene cluster of Streptomyces sp. VN1 was found to contain a unique ORF annotated as hybrid polyketide synthetase and non-ribosomal peptide synthase. This indicates that the strain Streptomyces sp. VN1 might have potential to synthesize xiamycin analogs.
In bioinformatics analysis of the Streptomyces sp. VN1 genome, one T2PKS was annotated for an enterocin biosynthesis gene cluster that was 95% similar to the enterocin biosynthesis gene cluster from Streptomyces   www.nature.com/scientificreports www.nature.com/scientificreports/ maritimus ( Supplementary Fig. S5). In this putative enterocin biosynthesis gene cluster, eight core ORFs that contained an additional methyltransferase were annotated. Thus, Streptomyces sp. VN1 might produce enterocin derivatives.
Additionally, Streptomyces sp. VN1 was predicted to harbor T1PKS pathways. The T1PKS gene cluster is related to biosynthesis of divergolide. It showed 100% homology with the existing cluster of Streptomyces sp. HKI0576 ( Supplementary Fig. S6).
Interestingly, biosynthetic gene cluster thirty-two was predicted to be related to bacterocin informatipeptin biosynthesis. The putative informatipeptin biosynthesis cluster of Streptomyces sp. VN1 displayed 42% homology with an existing cluster from Streptomyces viridochromogenes DSM 40736 ( Supplementary Fig. S7). In the genome of Streptomyces sp. VN1, only three ORFs (type A lantipeptide, protease, and PAS/PAC sensor protein) were found to be related to informatipeptin biosynthesis.
Furthermore, the T3PKS gene cluster was predicted to be related to herboxidiene biosynthesis ( Supplementary  Fig. S8). A comparison of the putative herboxidiene biosynthetic gene cluster with Streptomyces chromofuscus A7847 indicated that numerous genes in this cluster could not be annotated. This gene cluster only shared 2% homologies with the herboxidiene biosynthetic gene cluster from Streptomyces chromofuscus A7847.
The putative NRPS-nucleoside biosynthetic gene cluster from Streptomyces sp. VN1 was found to contain 34 ORFs related to nogalamycin biosynthesis ( Supplementary Fig. S9). A comparison of the nogalamycin biosynthetic gene cluster from Streptomyces nogalater indicated that numerous genes in this cluster could not be annotated. This gene cluster only shared 40% homology with the nogalamycin biosynthetic gene cluster from  Fig. S10C). Compounds 1, 2, and 3 were found to be cinnamamide, lobophorin A, and cyclo-L-proline-L-tyrosine, respectively, by comparing 1 H (700 MHz, methanol-d 4 ) and 13 Fig. S11). The IR spectrum of 4 revealed a conjugated carboxylic acid-carbonyl stretching-vibration absorption at 1,664 cm −1 , a CH 3 -asymmetric stretching-vibration absorption at 1,966 cm −1 , a CH 3 -symmetric stretching-vibration absorption at 2,874 cm −1 , a CH 2 -asymmetric stretching-vibration absorption at 2,930 cm −1 , and a carboxylic acid hydrogen bonding broad −OH stretching-vibration absorption at 3,400-2,400 cm −1 (Supplementary Fig. S12). 1 H (700 MHz, methanol-d 4 ) and 13 C (176 MHz, methanol-d 4 ) NMR (Table 4) led to structural conclusions for compound 4. The overall structure of 4 was established mainly based on 1 H-1 H correlation spectroscopy (COSY) and heteronuclear multiple-bond correlation (HMBC) correlations (Fig. 5). In the 1 H-1 H COSY spectrum, cross-peaks between δ H 1.22 ppm and 2.48 ppm, δ H 2.48 ppm and 1.57 ppm, 2.48 ppm and 1.70 ppm, 1.57 ppm and 0.91 ppm, 1.70 ppm and 0.91 ppm indicated three C-C bonds from C-9 to C-12. The correlation peak of between δ H 2.42 and 1.05 indicated C-C bond of C-6 and C-7 ( Supplementary Fig. S13C). In the HMBC spectrum, observed correlated signals between δ H 5.99 (s) and three carbon (δ C 103.67, 166.88 and 167.15) indicated the skeleton of C-2, C-3, C-4, and C5 based on intensities of cross-peaks. In addition, the connection of C-5 and C-10 was confirmed based on the finding of cross-peaks showing that H-9, H-10, H-11′, and H-11″ were correlated to C-5 by C-H long-range coupling. Moreover, the connection of C-2 and C-6 was confirmed from the finding of cross-peaks showing that H-6 was correlated to C-2 by C-H long-range coupling. The cross-peaks of H-6 and δ C 167.48 (C-8) indicated the bonding  www.nature.com/scientificreports www.nature.com/scientificreports/ of C-3 and C-8 as the carbonyl carbon ( Supplementary Fig. S13F). 13 C NMR chemical shift values of C-5 and C-2 were around 167 ppm as the region of the carbonyl carbon of ester, amide, and carboxylic acid group. If two carbons C-5 and C-2 build the ketone form, 13 C chemical shifts should appear at around 200 ppm. However, the 13 C spectrum did not show chemical shifts at around 200 ppm. This observation suggests that C-5 and C-2 are carbons of the double bond with oxygen attached. The number of this oxygen should be one based on HR-MS data. This observation led to the conclusion that it has a shape of five-membered ring formations of furan. Thus, the structure of 4 was completely elucidated to be 5-(sec-butyl)−2-ethylfuran-3-carboxylic acid. Absolute configurations experimental CD spectra of compound 4 showed cotton effect at wavelength of 200, 205, 225, and 290 nm. The positive cotton effect in π−π* transition of -C=C-was strong at 205 nm Δε + 15.12 (mdeg). That of -C=C-C=C-was slightly broad at 290 nm Δε + 1.56 (mdeg). The negative cotton effect negative in the π−π* transition of the -OH was strong at 200 nm Δε −19.4 (mdeg). In the n−π* transition of the -COOH, it was at 225 Δε −6.7 (mdeg) (Supplementary Fig. S14).

Identification of lobophorin A analogs by mass spectrometry.
We further analyzed mass spectra of crude extract to determine whether predicted secondary metabolites could be detected. Mass spectra data were obtained in positive mode. From HR-MS/MS profiles, metabolic substances in crude extract samples were found to be correlated with the mass-to-charge ratio (m/z) of molecular ions. Mass profiles of secondary metabolites existing in isolated strains were compared with genome mining data. Herein, we identified lobophorin A and its analogs in the crude extract through this technique. The proposed fragment of lobophorin A was analyzed based on the fragmentation of kijanimicin 38 (Supplementary Fig. S15A). In the mass profile of lobophorin A presented in Fig. 6A Supplementary Fig. S15B. Compound 2 (lobophorin A, Rt: 6.033 min) and lobophorin A analogs namely compound 5 (demethylated of lobophorin A, retention time, Rt: 5.673 min), compound 6 (dehydroxylated lobophorin A, Rt: 6.341 min), compound 7 (Rt: 6.772 min) and compound 8 (Rt: 7.283 min), were detected (Fig. 6) in fraction A. Possible structures and fragmentation of compounds 5 and 6 were compared with the mass profile of lobophorin A and previously described structures 36,39 . These compounds displayed UV spectrum and fragmentation similar to those of lobophorin A (Supplementary Fig. S15) (Fig. 6B). This analog has been verified and characterized by MS/MS and NMR in a previous study 39 . Peaks generated by MS/MS analysis showed that main fragment ions were at 883.4966, 753.4326, and 517.2950 (Supplementary Fig. S15C) (Fig. 6C). Peaks generated by MS/MS analysis showed that main fragment ions were at 881.4817, 751.4147, 619.3461, and 515.2808 (Supplementary Fig. S15D). Interestingly, the third derivative, compound 7 showed [M + H] + ions at m/z 1,2041.6011 (Fig. 6D). Peaks generated by MS/ MS analysis showed that main fragment ions were at 945.4792, 867.5002, 737.5001, 619.3464, 517.2946, and 499.2837 (Supplementary Fig. S15E). It was noteworthy that the fourth analog, compound 8 showed [M + H] + ions at m/z 1,026.6724 (Fig. 6E). Peaks generated by MS/MS analysis showed that main fragment ions were at 895.5777, 783.4073, 619.3467, 517.2947, and 455.3103 (Supplementary Fig. S15F). To the best of our knowledge, there is no structural reference for compound 7 or 8. They might be new analogs of lobophorin. We also analyzed mass spectra of crude extract residue B. However, we could not correlate the data with the mass of any secondary metabolite (data not shown).

Anticancer activity.
To assess whether compound 4 might have anticancer activity, we evaluated the inhibitory effect of compound 4 on the growth of different tumor cell lines. Compound 4 showed anti-proliferative activities against five types of cancer cell lines (Fig. 7A). This compound exhibited more sensitive growth-inhibitory activities for gastric adenocarcinoma (AGS), glioblastoma (U87MG), and lung cancer (A549) than for melanoma www.nature.com/scientificreports www.nature.com/scientificreports/ (A375SM) and colon cancer cells (HCT116) ( Table 5; Fig. 7A). Compound 4 more effectively inhibited the growth of cancer cell lines than that of normal cell lines (267B1 and MRC-5) at tested doses (Table 5; Fig. 7B). Besides its inhibitory effect against cancer cell growth, the anti-metastatic activity of compound 4 against AGS gastric cancer cell line was also determined through migration and invasion assays. We found that compound 4 effectively suppressed both the migration and invasion of AGS cells after 24 h of treatment at concentrations of 25 and 50 µM    www.nature.com/scientificreports www.nature.com/scientificreports/ ( Figs. 8 and 9). These data suggest that compound 4 possesses an anticancer activity by inhibiting the growth and metastasis abilities of cancer cells.

Discussion
Marine Streptomyces are valuable sources of various secondary metabolites for drug discovery. In this study, we found that Streptomyces sp. VN1 as a new strain produced many active compounds, including cinamamide, lobophorin A, cyclo-L-proline-L-tyrosine, and a furan-type compound based on HR-MS/MS and 2D-NMR studies. Besides these compounds, possible analogs of lobophorin A were found based on HR-MS/MS analysis. We found that fragments of those compounds were similar to a fragment of lobophorin A. However, in mass profiles they appeared in trace amounts. Thus, we were unable to elucidate structures of these compounds.
Cinamamide is the product of phenylalanine ammonia lyase. It has been examined in the biosynthesis of cinnamamide in cultures of Streptomyces verticillatus using L-phenylalanine-carboxyl-[ 14 C] 40 . Cinamamide is a nonlethal chemical repellent 41 . Recently, cinamamide derivatives have been synthesized. They show potential activities in both peripheral and central nervous systems, including antiepileptic, antidepressant, neuroprotective, analgesic, anti-inflammatory, muscle-relaxant and sedative/hypnotic properties 42 . Our results indicate that Streptomyces sp. VN1 is a promising producer of cinamamide compounds for drug discovery.
Lobophorin A and related analogs are members of the spirotetronate family. Their anti-inflammatory and antibacterial properties have been reported 16,43 . Genome sequence analysis results revealed that the putative  www.nature.com/scientificreports www.nature.com/scientificreports/ lobophorin biosynthetic gene cluster in Streptomyces sp. VN1 showed very high similarity with Streptomyces sp. FXJ7.023 and Streptomyces sp. SCSIO 01127 ( Fig. 3; Supplementary Table S5A). Four different analogs of lobophorin (compound 5, 6, 7 and 8) were identified to be produced by Streptomyces sp. VN1 through mass analysis without any genetic modification. The diversity of lobophorin analogs has been previously explored. There has been a report of structural analysis of lobophorin revealing a polycyclic tetronolide core that is further modified by glycosylation and hydroxylation 35 . Through deletion of methyltransferase in Streptomyces sp. SCSIO 01127 the authors obtained two new analogs of lobophorin 39 . In addition, deletion of the glycosylation step in the biosynthesis pathway has achieved novel analogs of lobophorins 36 . The catalytic activity of core enzymes for lobophorin A biosynthesis and post-modification enzymes in Streptomyces sp. VN1 might be different from those in Streptomyces sp. SCSIO 01127. This might lead to different of lobophorin A analogs. In addition, different post-modification steps might convert intermediates to different analogs. Post-modification tailoring of lobophorin A might also follow a different succession leading to accumulation of novel analogs of lobophorin A in Streptomyces sp. VN1. The diversity of lobophorin analogs identified in Streptomyces sp. VN1 might be the result of flexibility of post-modification enzymes such as glycosyltransferase, methyltransferase and hydroxylase which might have different substrate preferences among intermediates of lobophorin. The presence of an N-acetyltransferase gene in cluster 1 of Streptomyces sp. VN1 might have played role in adding to the diversity of lobophorin analogs. The N-acetyltransferase gene might be responsible for transfer of an acetyl group to lobophorin A to form other derivatives such as compound 7 and compound 8 (Supplementary Table S5B). A similar analog of lobophorin has been reported in a previous study where lobophorin G is detected as acetylated lobophorin A 13 . The diversity of lobophorin analogs identified from Streptomyces sp. VN1 highlights the uniqueness of lobophorin biosynthetic pathway compared to other lobophorin producer strains. Thus, Streptomyces sp. VN1 can be developed into a prolific producer of lobophorin and its analogs can be further diversified through genetic engineering. Biosynthetic modifications such as glycosylation 44 , hydroxylation 45 and methylation 46 have been successfully applied for the biosynthesis of non-natural products. These modifications could be applied for Streptomyces sp. VN1 to produce novel lobophorin analogs with potential medical applications.
Cyclo-L-proline-L-tyrosine was first reported to be produced by Streptomyces sp. strain 22-4 17 . It had activities against three economically important plant pathogens, Xanthomonas axonopodis pv. citri, Ralstonia solanacearum and Clavibacter michiganensis.
Although the structure of compound 4 has been reported previously as a synthetic by-product (http://www. ambinter.com/search), unfortunately, no reference NMR data of compound 4 was available. To the best of our knowledge, this is the first report of NMR spectrum and biological activity of furan-type compound 4 produced by a marine Streptomyces. There has been a report of moderate production of a furan-type compound by microorganism and intermediate furan-type compound in Streptomyces that may function as a starting unit for a secondary metabolite, such as A-factor, butyrolactone, and other analogs of methylenomycin furans (MMFs) 47 from Streptomyces celicolor.
In 2008, Jidong et al. reported a furan compound HS071 from Streptomyces sp. HS-HY-071 that showed cytotoxic activity against a cancer cell line 48 . However, there is no report about the biosynthesis pathway of this compound. There are some differences between structures of compound 4 and furan-compound HS071 48 , such as a 4-OH position in the HS071 compound where compound 4 contains a carboxyl group (Supplementary Fig. S16) probably because these two compounds are produced from two different biosynthetic pathways or because compound 4 or compound HS071 is further modified, although they come from the same biosynthesis pathway. The biosynthesis pathway of compound 4 was not explored in this study. Further studies by feeding 13 C in combination with genome sequence data are needed to track down the exact mechanism of compound 4 biosynthesis.
We examined the biological activity of compound 4 against five different types of cancer cell lines. Compound 4 showed growth inhibitory activity most effectively against an AGS cancer cell line with an IC 50 value of 40.5 µM. Although concentration ranges of compound 4 for growth inhibition of cancer cells were much higher than those of known anticancer drugs such as taxol 49 with effective concentrations below 10 µM, compound 4 inhibited the growth of cancer cell lines more effectively as compared to normal cell lines at tested concentrations (Fig. 7). Interestingly, compound 4 also showed anti-migration and anti-invasion activities against AGS cancer cell line after 24 hours of exposure (Figs. 8 and 9). These results confirm compound 4 possesses anticancer activity. However, as shown in cell growth assay the cell viability was high (89%) even after treatment with compound 4 at 25 µM treatment for 72 hours. Therefore, inhibitory effects of compound 4 on the migration and invasion of AGS cancer cells were not merely due to the cytotoxicity of compound 4. Such anti-invasive properties of compound 4 against cancer cell lines showcase the relevance of exploring secondary metabolites biosynthetic pathways for discovering clinically relevant compounds.
Results of in sillico genome analysis showed that Streptomyces sp. VN1 encodes the putative biosynthetic gene clusters of diverse interesting secondary metabolites, such as carotenoid, friulimicin, xiamycin, divergolide, and informatipeptin. However, biosynthetic pathways for these metabolites in Streptomyces sp. VN1 were inactive under normal culture conditions. This study displays the biosynthetic potential of Streptomyces sp. VN1 through isolation and characterization of four compounds. Due to its ability to grow fast and produce various secondary metabolites, it is suitable for further metabolic exploration to produce useful metabolites. We present Streptomyces sp. VN1 as a good candidate for strain optimization to produce therapeutically and industrially relevant compounds. Our results reinforce the need of further exploring marine Streptomycetes as a rich source of novel metabolites relevant for biotechnological applications.