An Unusual Diterpene—Enhygromic Acid and Deoxyenhygrolides from a Marine Myxobacterium, Enhygromyxa sp.

Three new compounds, enhygromic acid (1) and deoxyenhygrolides A (2) and B (3), were isolated from a marine myxobacterium, Enhygromyxa sp. Compound 1 was found to be an acrylic acid derivative with a rare polycyclic carbon skeleton, decahydroacenaphthylene, by spectroscopic analyses. Compounds 2 and 3 were deoxy analogs of the known γ-alkylidenebutenolides, enhygrolides. Compound 1 exhibited cytotoxicity against B16 melanoma cells and anti-bacterial activity against Bacillus subtilis, and enhanced the NGF-induced neurite outgrowth of PC12 cells.


Introduction
Myxobacteria are unique δ-proteobacteria that glide and aggregate to form swarms and multicellular fruiting bodies [1]. Myxobacteria were recently recognized as potential sources of novel secondary metabolites such as polyketides and non-ribosomal peptides as well as their hybrid compounds [2,3]. Marine-derived myxobacteria are especially attractive [4] because their polyketide synthase gene sequences are unique [5], although they are more difficult to cultivate compared to their terrestrial counterparts. In recent years, some bioactive compounds were discovered from marine myxobacteria including haliangicins [6,7] and haliamide [8] from Haliangium ochraceum; miuraenamides produced by Paraliomyxa miuraensis [9,10]; and salimabromides [11], salimyxins [12], and enhygrolides [12] produced by Enhygromyxa salina [13]. Our group has been searching for new myxobacteria species in marine environment for decades [14]. Our research on marine myxobacterial secondary metabolites led to the discovery of three new compounds, enhygromic acid (1) and deoxyenhygrolides A (2) and B (3) (Figure 1), from a new species within the genus Enhygromyxa. Here, we report the isolation, structural elucidation, and biological activities of these compounds, one of which is an acrylic acid derivative with a rare tricyclic carbon skeleton.

Isolation and Structural Elucidation of 1-3
Enhygromyxa sp. SNB-1 was found to be a new species based on the 16S rRNA gene sequence (97% similarity to the closest type strain) and tentatively designated as Enhygromyxa niigataensis ( Figure 2).

Isolation and Structural Elucidation of 1-3
Enhygromyxa sp. SNB-1 was found to be a new species based on the 16S rRNA gene sequence (97% similarity to the closest type strain) and tentatively designated as Enhygromyxa niigataensis ( Figure 2).

Isolation and Structural Elucidation of 1-3
Enhygromyxa sp. SNB-1 was found to be a new species based on the 16S rRNA gene sequence (97% similarity to the closest type strain) and tentatively designated as Enhygromyxa niigataensis ( Figure 2).  The strain was cultured for three weeks in VY/4-SWS (2% NaCl) liquid medium supplemented with 100 mM sodium acetate. Twenty-one liters of cultures containing bacterial cells were extracted with acetone, and aqueous acetone extracts were subsequently extracted with EtOAc. The resulting extract was separated by flash column chromatography on silica gel (hexane-EtOAc), and each fraction was subjected to an MTT assay to determine cytotoxicity against B16 melanoma cells. After a two-step purification of the active fractions by HPLC, Compound 1 (6.9 mg) was obtained as a cytotoxic substance. Meanwhile, Compounds 2 (4.5 mg) and 3 (17.6 mg) were obtained from another fraction.
The molecular formula of enhygromic acid (1) was determined to be C 20 H 30 O 2 by high-resolution electrospray ionization mass spectrometry (ESIMS) analysis using the molecular ion at m/z 303.2299 [M + H] + (calculated for C 20 H 31 O 2 : 303.2319). The infrared (IR) spectrum showed a strong absorption at 1685 cm −1 and a very broad band between 3640 and 2390 cm −1 , and the 13 C nuclear magnetic resonance (NMR) spectrum displayed the carbonyl carbon signal at δ C 170.2, suggesting the presence of a carboxy group. The 1 H and 13 C NMR data are listed in Table 1. The two-dimensional NMR correlations of Compound 1 are summarized in Figure 3. Two partial structures, -CH 2 -CH< and -CH 2 -CH 2 -CH-CH(CH 3 )-CH 2 -CH 2 -CH<, were revealed by DQF-COSY correlations. The connectivity of these substructures and the five quaternary carbons were determined by interpretation of the HMBC spectrum ( Figure 3A). The elucidated planar structure of 1 contained a fused tricyclic skeleton linked with an α-methylacrylic acid chain ( Figure 3). The geometry of the double bond at C-2 was determined to be E on the basis of the NOE correlations of H-20/H-5 and H-20/H-9, and the lack of the NOE of H-3/H-20 ( Figure 3B). The relative configuration of 4S*,5S*,9R*,10R*,13S* was confirmed from the NOE correlations, displayed in Figure 3B.  In order to determine the absolute stereochemistry of 1, conformational analysis was conducted by minimizing the energy of all the conformers generated by changing the torsion angle of the C-3-C-4 bond. Only two energy-minimized conformers, A and B, were obtained ( Figure 4); the conformer A was more stable than B by approximately 14 kcal/mol ( Figure S10). Among them, the more stable conformer A was strongly supported by the NOESY correlations ( Figure 3B). The circular dichroism (CD) spectrum of 1 was subsequently analyzed to determine the spacial relationships between the acrylic acid moiety and the exomethylene group at C-6. On the basis of the splitted negative Cotton effect at 226 nm (Δε − 9.0) and 196 nm (Δε + 8.0) ( Figure 5) and in conjunction with the abovementioned conformational analysis, the absolute configuration of 1 was determined to be 4S,5S,9R,10R,13S. Interestingly, the 13 C NMR signals of C-2 and C-3 indicated weak and broad signals, suggesting that the free rotation of the methylacrylic acid moiety extending from the center (C-4) of the fused tricyclic core is restricted by the steric repulsion between them. These findings corroborate the uniqueness of the structure of 1.  Both Compounds 2 and 3 had the molecular formula C22H22O2 as determined by high-resolution ESIMS. The IR absorption at 1760-1763 cm −1 suggests that they contain a carbonyl group. Their 1 H In order to determine the absolute stereochemistry of 1, conformational analysis was conducted by minimizing the energy of all the conformers generated by changing the torsion angle of the C-3-C-4 bond. Only two energy-minimized conformers, A and B, were obtained ( Figure 4); the conformer A was more stable than B by approximately 14 kcal/mol ( Figure S10). Among them, the more stable conformer A was strongly supported by the NOESY correlations ( Figure 3B). The circular dichroism (CD) spectrum of 1 was subsequently analyzed to determine the spacial relationships between the acrylic acid moiety and the exomethylene group at C-6. On the basis of the splitted negative Cotton effect at 226 nm (∆ε − 9.0) and 196 nm (∆ε + 8.0) ( Figure 5) and in conjunction with the above-mentioned conformational analysis, the absolute configuration of 1 was determined to be 4S,5S,9R,10R,13S. Interestingly, the 13 C NMR signals of C-2 and C-3 indicated weak and broad signals, suggesting that the free rotation of the methylacrylic acid moiety extending from the center (C-4) of the fused tricyclic core is restricted by the steric repulsion between them. These findings corroborate the uniqueness of the structure of 1. In order to determine the absolute stereochemistry of 1, conformational analysis was conducted by minimizing the energy of all the conformers generated by changing the torsion angle of the C-3-C-4 bond. Only two energy-minimized conformers, A and B, were obtained ( Figure 4); the conformer A was more stable than B by approximately 14 kcal/mol ( Figure S10). Among them, the more stable conformer A was strongly supported by the NOESY correlations ( Figure 3B). The circular dichroism (CD) spectrum of 1 was subsequently analyzed to determine the spacial relationships between the acrylic acid moiety and the exomethylene group at C-6. On the basis of the splitted negative Cotton effect at 226 nm (Δε − 9.0) and 196 nm (Δε + 8.0) ( Figure 5) and in conjunction with the abovementioned conformational analysis, the absolute configuration of 1 was determined to be 4S,5S,9R,10R,13S. Interestingly, the 13 C NMR signals of C-2 and C-3 indicated weak and broad signals, suggesting that the free rotation of the methylacrylic acid moiety extending from the center (C-4) of the fused tricyclic core is restricted by the steric repulsion between them. These findings corroborate the uniqueness of the structure of 1.  Both Compounds 2 and 3 had the molecular formula C22H22O2 as determined by high-resolution ESIMS. The IR absorption at 1760-1763 cm −1 suggests that they contain a carbonyl group. Their 1 H In order to determine the absolute stereochemistry of 1, conformational analysis was conducted by minimizing the energy of all the conformers generated by changing the torsion angle of the C-3-C-4 bond. Only two energy-minimized conformers, A and B, were obtained ( Figure 4); the conformer A was more stable than B by approximately 14 kcal/mol ( Figure S10). Among them, the more stable conformer A was strongly supported by the NOESY correlations ( Figure 3B). The circular dichroism (CD) spectrum of 1 was subsequently analyzed to determine the spacial relationships between the acrylic acid moiety and the exomethylene group at C-6. On the basis of the splitted negative Cotton effect at 226 nm (Δε − 9.0) and 196 nm (Δε + 8.0) ( Figure 5) and in conjunction with the abovementioned conformational analysis, the absolute configuration of 1 was determined to be 4S,5S,9R,10R,13S. Interestingly, the 13 C NMR signals of C-2 and C-3 indicated weak and broad signals, suggesting that the free rotation of the methylacrylic acid moiety extending from the center (C-4) of the fused tricyclic core is restricted by the steric repulsion between them. These findings corroborate the uniqueness of the structure of 1.  Both Compounds 2 and 3 had the molecular formula C 22 H 22 O 2 as determined by high-resolution ESIMS. The IR absorption at 1760-1763 cm −1 suggests that they contain a carbonyl group. Their 1 H and 13 C NMR data are summarized in Table 2. Three partial structures, two C 6 H 5 -and a -CH 2 -CH(CH 3 ) 2 , were obtained from DQF-COSY correlations of 2 ( Figure 6). The connectivity of these substructures, five quaternary carbons (δ C 127.5, 133.8, 138.4, 149.2, and 152.0), one carbonyl (δ C 169.8), one methylene (δ C 30.1), and one sp 2 methine (δ C 109.0), was determined on the basis of the HMBC correlations ( Figure 6), furnishing the planar structure of 2. The Z configuration of the C-4-C-5 double bond was determined by the NOESY correlation between H-5 and H-19. The structure of 3 was determined as an isomer of 2 in a similar manner to that for 2 (Table 2 and Figure 6). In contrast to 2, the geometry of the double bond at C-4 was found to be E because of the lack of NOE correlation between H-5 and H-19, concluding that 3 is the E isomer of 2. Deoxyenhygrolides A (2) and B (3) are new enhygrolide analogs, where the phenolic groups in enhygrolides [12] are replaced by hydrogen atoms. Although enhygrolide A (Z isomer) is reported to be more stable than enhygrolide B (E isomer) and that their isomerization is irreversible [12], we did not observe the isomerization from 3 (E) to 2 (Z), and the yield of 3 was approximately four times that of 2. Thus, the phenolic hydroxyl group of enhygrolides may affect the stability of the E/Z isomers for this type of compound. and 13 C NMR data are summarized in Table 2. Three partial structures, two C6H5-and a -CH2-CH(CH3)2, were obtained from DQF-COSY correlations of 2 ( Figure 6). The connectivity of these substructures, five quaternary carbons (δC 127.5, 133.8, 138.4, 149.2, and 152.0), one carbonyl (δC 169.8), one methylene (δC 30.1), and one sp 2 methine (δC 109.0), was determined on the basis of the HMBC correlations (Figure 6), furnishing the planar structure of 2. The Z configuration of the C-4-C-5 double bond was determined by the NOESY correlation between H-5 and H-19. The structure of 3 was determined as an isomer of 2 in a similar manner to that for 2 (Table 2 and Figure 6). In contrast to 2, the geometry of the double bond at C-4 was found to be E because of the lack of NOE correlation between H-5 and H-19, concluding that 3 is the E isomer of 2. Deoxyenhygrolides A (2) and B (3) are new enhygrolide analogs, where the phenolic groups in enhygrolides [12] are replaced by hydrogen atoms. Although enhygrolide A (Z isomer) is reported to be more stable than enhygrolide B (E isomer) and that their isomerization is irreversible [12], we did not observe the isomerization from 3 (E) to 2 (Z), and the yield of 3 was approximately four times that of 2. Thus, the phenolic hydroxyl group of enhygrolides may affect the stability of the E/Z isomers for this type of compound.

Bioactivities of 1-3
Enhygromic acid (1) exhibited cytotoxicity against B16 melanoma cells, and its 50% inhibitory concentration (IC 50 ) of 46 µM is comparable to that of paclitaxel (57 µM) ( Figure 7A). On the other hand, against HeLa-S3 cells, 1 showed no inhibitory activity (IC 50 > 30 µM) although paclitaxel was highly active (IC 50 = 6.0 nM). Furthermore, 1 enhanced the nerve growth factor (NGF)-induced neurite outgrowth of PC12 cells. The neurite outgrowth (22%) induced by a trace amount (1 ng/mL) of NGF was enhanced up to 67% in a dose-dependent manner ( Figure 7B). In anti-microbial tests, 1 inhibited the growth of a Gram-positive bacterium, Bacillus subtilis, at a minimum inhibitory concentration (MIC) of 8 µg/mL, but was inactive against a Gram-negative bacterium (Escherichia coli), fungi (Candida rugosa, Aspergillus niger, Rhizopus oryzae, and Trichophyton mentagrophytes), and an oomycete (Phytophthora capsici). was enhanced up to 67% in a dose-dependent manner ( Figure 7B). In anti-microbial tests, 1 inhibited the growth of a Gram-positive bacterium, Bacillus subtilis, at a minimum inhibitory concentration (MIC) of 8 μg/mL, but was inactive against a Gram-negative bacterium (Escherichia coli), fungi (Candida rugosa, Aspergillus niger, Rhizopus oryzae, and Trichophyton mentagrophytes), and an oomycete (Phytophthora capsici). On the other hand, deoxyenhygrolides A (2) and B (3) did not exhibit any activities in the abovementioned tests. They are classified into γ-alkylidenebutenolides, and some structurally related bioactive metabolites were reported, e.g., enhygrolides [12], nostoclides [15], and cyanobacterin [16]. Enhygrolide A, a phenolic congener of 2, was reported to show an antimicrobial activity (MIC = 4 μg/mL) against a Gram-positive bacterium, Arthrobacter sp. Nostoclides are chlorinated phenols and showed a moderate cytotoxicity (IC50 = 10 μg/mL) against animal cell lines (Neuro-2a and KB cells). Cyanobacterin is a chlorinated phenyl ether and is known to inhibit the growth of cyanobacteria and green algae. The lack of the bioactivities of 2 and 3 suggests that the (halogenated) phenol group(s) may play an important role for the bioactivity of the γ-alkylidenebutenolides.

Putative Biosynthetic Mechanism for 1
Enhygromic acid (1) possesses a unique fused ring structure, decahydroacenaphthylene, which is connected with α-methylacrylic acid. The carbon skeleton of 1 can be divided into four isoprene units [C-1-C-2(C-20)-C-3-C-4, C-18-C-14(C-19)-C-13-C-12, C-11-C-10(C-17)-C-9-C-8, C-7-C-6(C-16)-C-5-C-15], suggesting that this molecule is a diterpene. However, the connectivity of the first unit C-1-C-2(C-20)-C-3-C-4 is inconsistent with the head-to-tail isoprene rule. Therefore, we propose two putative pathways for the biosynthesis of 1: (a) α-humulene formation from farnesyl diphosphate (FPP) followed by a post-prenylation; or (b) cyclization of geranylgeranyl diphosphate (GGPP) followed by the rearrangement of one isoprenyl unit (Figure 8). The resulting intermediate I undergoes repeated 1,3-hydride shift/cyclization and the oxidation of the terminal methyl group. In silico analysis was carried out on the genome sequences of Enhygromyxa salina DSM 15201 (= E. salina SHK-1 T in Figure  2) [4] by antiSMASH (ver. 4.0.0) software (http://antismash.secondarymetabolites.org), revealing the presence of seven terpenoid biosynthetic gene clusters, three of which were found to contain a core gene homologous to a terpene (pentalenene) synthase gene. Since pentalenene is known to be biosynthesized from FPP via αhumulene (Figure 8), one of these genes might be responsible for the biosynthesis of 1, and Route a is more plausible than Route b. However, unfortunately, we were unable to find any related genes such as an FPP (or GGPP) synthase gene, a prenyltransferase gene (Route a), and oxidase gene(s) inside or around these gene clusters to support the biosynthetic route. The biosynthetic mechanism for the construction of this unique carbon skeleton is to be elucidated by cloning the responsible terpene synthase gene in the future. On the other hand, deoxyenhygrolides A (2) and B (3) did not exhibit any activities in the above-mentioned tests. They are classified into γ-alkylidenebutenolides, and some structurally related bioactive metabolites were reported, e.g., enhygrolides [12], nostoclides [15], and cyanobacterin [16]. Enhygrolide A, a phenolic congener of 2, was reported to show an antimicrobial activity (MIC = 4 µg/mL) against a Gram-positive bacterium, Arthrobacter sp. Nostoclides are chlorinated phenols and showed a moderate cytotoxicity (IC 50 = 10 µg/mL) against animal cell lines (Neuro-2a and KB cells). Cyanobacterin is a chlorinated phenyl ether and is known to inhibit the growth of cyanobacteria and green algae. The lack of the bioactivities of 2 and 3 suggests that the (halogenated) phenol group(s) may play an important role for the bioactivity of the γ-alkylidenebutenolides.

Putative Biosynthetic Mechanism for 1
Enhygromic acid (1) possesses a unique fused ring structure, decahydroacenaphthylene, which is connected with α-methylacrylic acid. The carbon skeleton of 1 can be divided into four isoprene units [C-1-C-2(C-20)-C-3-C-4, C-18-C-14(C-19)-C-13-C-12, C-11-C-10(C-17)-C-9-C-8, C-7-C-6(C-16)-C-5-C-15], suggesting that this molecule is a diterpene. However, the connectivity of the first unit C-1-C-2(C-20)-C-3-C-4 is inconsistent with the head-to-tail isoprene rule. Therefore, we propose two putative pathways for the biosynthesis of 1: (a) α-humulene formation from farnesyl diphosphate (FPP) followed by a post-prenylation; or (b) cyclization of geranylgeranyl diphosphate (GGPP) followed by the rearrangement of one isoprenyl unit (Figure 8). The resulting intermediate I undergoes repeated 1,3-hydride shift/cyclization and the oxidation of the terminal methyl group. In silico analysis was carried out on the genome sequences of Enhygromyxa salina DSM 15201 (= E. salina SHK-1 T in Figure 2) [4] by antiSMASH (ver. 4.0.0) software (http://antismash.secondarymetabolites.org), revealing the presence of seven terpenoid biosynthetic gene clusters, three of which were found to contain a core gene homologous to a terpene (pentalenene) synthase gene. Since pentalenene is known to be biosynthesized from FPP via α-humulene (Figure 8), one of these genes might be responsible for the biosynthesis of 1, and Route a is more plausible than Route b. However, unfortunately, we were unable to find any related genes such as an FPP (or GGPP) synthase gene, a prenyltransferase gene (Route a), and oxidase gene(s) inside or around these gene clusters to support the biosynthetic route. The biosynthetic mechanism for the construction of this unique carbon skeleton is to be elucidated by cloning the responsible terpene synthase gene in the future.

Bacterial Strain
The myxobacterial strain SNB-1 was isolated from a sand sample collected in September 1998 at a beach in Kashiwazaki, Niigata, Japan (37°22′ N, 138°33′ E). The isolation procedure was performed according to the method previously described [14]. The strain was cultured and maintained in N1.0-S75-15 agar or VY/2-S75-15 agar [17]. The growth of the strain was observed with 0.0-4.5% (w/v) NaCl (optimum, 1.0-2.0%). The 16S rRNA gene sequence (1505 nt) was deposited under the DDBJ accession number of LC066681. The phylogenetic tree (Figure 2) was constructed with the MEGA6 software by neighbor-joining method (NJ) based on 16S rRNA gene sequences, indicating that the strain SNB-1

Bacterial Strain
The myxobacterial strain SNB-1 was isolated from a sand sample collected in September 1998 at a beach in Kashiwazaki, Niigata, Japan (37 • 22 N, 138 • 33 E). The isolation procedure was performed

Conformational Analysis
The conformational analysis of enhygromic acid (1) was performed by Chem3D Ultra in ChemOffice 15.1 package (PerkinElmer Co., Waltham, MA, USA). The dihedral angle C-2-C-3-C-4-C-5 of 1 was rotated by every 15 • , and MM2 energy minimization was carried out for all angles to obtain only two local minimum conformers. The result was visualized by a graph of the local minimum energy and angle against the input dihedral angle of C-2-C-3-C-4-C-5 ( Figure S10).

NGF-Enhancing Activity for PC12 Cells
Nerve growth factor (NGF)-enhancing activity was evaluated by following the method reported previously [19] with some modifications. Briefly, PC12 cells cultured in DMEM supplemented with 10% fetal bovine serum (Biosera, Kansas City, MO, USA), 5% horse serum (Thermo Fisher Scientific Inc., Waltham, MA, USA), and antibiotics (100 µg/mL streptomycin and 100 units/mL penicillin). A total of 15,000 cultured cells in 1 mL of the same medium were seeded into each well of a 24-well microplate collagen Type I-coated (IWAKI, Tokyo, Japan). After preincubation for 24 h at 37 • C in an atmosphere of 5% CO 2 , the medium was replaced with 1 mL of the serum-free DMEM containing 0.5 ng/mL NGF (PeproTech, Rocky Hill, NJ, USA) and a compound in 5 µL DMSO. The morphological change of the cells was monitored by an inverted microscope FSX100 (OLYMPUS, Tokyo, Japan) after 4 days of incubation. A field that covers approximately 100 cells was randomly chosen and the number of the cells bearing neurite outgrowth longer than the cell diameter was counted. The data were the mean ± SD of triplicate.

Anti-Phytophthora Assay
The growth inhibitory activity against the oomycete Phytophthora capsici was evaluated by following the method reported previously [8].

Conclusions
Three new compounds, enhygromic acid (1) and deoxyenhygrolides (2 and 3), were isolated from a new marine myxobacterial species of the genus Enhygromyxa, and the absolute configuration of 1 was determined by NOESY and CD spectrum. As regards the results of the biological tests, 1 showed the cytotoxicity against B16 melanoma cells, NGF neurite outgrowth promoting activity to PC12 cells, and growth inhibition against B. subtilis. On the other hand, 2 and 3 did not show any biological activities in this study. Compound 1 has a quite unique and rare acrylic acid derivative with a decahydroacenaphthylene skeleton, which had not been reported prior to this study. Therefore, 1 will be noticed in the field of organic synthesis or biosynthesis research in the future.