Introduction

Actinomycetes, especially filamentously growing groups such as Streptomyces and Micromonospora, are excellent producers of various natural products with remarkable bioactivities. However, discovery of new compounds from typical terrestrial actinomycetes is currently getting difficult, while a substantial number of actinomycetal genera are likely unstudied for their secondary metabolites. Rare actinobacteria that are less frequently isolated are now recognized as a promising reservoir of new natural products [1,2,3]. The genus Phytohabitans is a member of the family Micromonosporaceae, first described in 2010 [4] and until recently, habiterpenol was the only one known compound from this genus [5]. Meanwhile, the latest genomic study indicated the presence of type I and III polyketides synthases (PKSs), nonribosomal peptide synthetase (NRPS), and hybrid PKS/NRPS gene clusters in the genome of Phytohabitans strains though the isolation of polyketides or peptidic compounds were not reported [6].

During the course of metabolite profiling in underexplored rare actinomycetes, we recently discovered new δ-lactone-terminated linear polyketides, phytohabitols, from Phytohabitans sp. RD002984 (Fig. 1) [7]. In this study, Phytohabitans sp. RD003013 isolated from a soil collected in Tokyo, Japan, was found to produce a bisprenyl naphthoquinone designated phytohabinone (1) and a calcimycin-class polyketide modified with a dichrolopyrrol and a thiazolecarboxylic acid designated phytohabimicin (2). We herein describe the isolation and structure elucidation of 1 and 2 along with their bioactivities.

Fig. 1
figure 1

Natural products from Phytohabitans

Results and discussion

Phytohabinone (1) was isolated as a yellow amorphous solid (2.1 mg from 1 l of culture). HRESITOFMS gave a deprotonated molecular ion at m/z 339.1600, which defined a molecular formula C21H24O4. The 1H NMR spectrum of 1 revealed the presence of one aromatic singlet at δH 7.05, two olefinic triplets at δH 5.15 and 4.96, two methylene doublets at δH 3.26 and 3.23, five methyl singlets at δH 2.07, 1.72, 1.72, 1.64, and 1.62, one hydrogen-bonded hydroxy proton at δH 12.64, one phenolic hydroxy proton at δH 11.05. The 13C and HSQC spectral data allowed the assignment of 21 carbons to two carbonyls at δC 188.4 and 183.2, two oxygenated sp2 carbons at δC 161.7 and 160.8, seven sp2 carbons at δC 145.3, 142.9, 133.0, 131.5, 130.5, 120.1, and 107.9, three sp2 methines at δC 121.1, 119.4, and 107.2, two sp3 methylenes at δC 25.5 and 21.5, and five methyl carbons at δC 25.4, 25.4, 17.8, 17.7, and 11.8 (Table 1).

Table 1 NMR data for phytohabinone (1) in DMSO-d6

The UV spectrum was found two peaks with the absorption maxima at 270 and 428 nm indicating the presence of a naphthoquinone chromophore [8, 9]. Analysis of the COSY correlations established the connectivities of H2-10/H-11 and H2-15/H-16. HMBC correlations from the aromatic methine H-5 to C-4a, C-6, C-7, C-8a, the phenolic hydroxy proton 6-OH to C-5, C-6, and C-7, hydrogen-bonded proton 8-OH to C-7, C-8 and C-8a established a pentasubstituted benzene substructure bearing two hydroxy groups. Additionally, H3-9 to C-1, C-2, and C-3, H-5 to C-4, 8-OH to C-1 (four bond correlation, Fig. S5) extended the benzene ring to a 1,4-naphthoquinone core (Fig. 1). The remaining carbons were assigned to constitute two prenyl groups connecting at C-3 and C-7. HMBC correlations from H2-10 and H-11 to C-12, C-13, and C-14, H3-13 to C-11, C-12, and C-14, and H3-14 to C-11, C-12, and C-13 established the carbon-carbon connectivity in a prenyl side chain. The prenyl group was then connected at C-3 by HMBC correlations from H2-10 to C-2, C-3, and C-4. Similarly, another prenyl group was determined to be connected at C-7 (Fig. 2). Two methyl groups of the prenyl groups were distinguished by the 13C chemical shift difference. More shielded methyl resonances at δC 17.8 and 17.7 arising from steric compression effect were assigned to C-14 and C-19, respectively.

Fig. 2
figure 2

COSY and key HMBC for 1

Phytohabimicin (2) was isolated as an optically active ([α]22D + 159 (c 0.12, CHCl3)) white powder (2.3 mg from 1 l of culture). Its HRESITOFMS spectrum showed deprotonated molecular ions at m/z 541.1336, 543.1313, and 545.1296 in a peak area ratio of 9:6:1 (Fig. S16), which inferred the presence of two chlorine atoms in this molecule, thereby establishing the molecular formula C25H32Cl2N2O5S. The 1H NMR spectrum exhibited two sp2 methines (δH 8.27, 6.81), seven sp3 methines (δH 3.56, 3.13, 3.01, 3.00, 1.66, 1.63, 1.38), six methylenes (δH 1.78, 1.65, 1.35, 1.25, 1.22, 0.73), five methyl doublets (δH 1.32, 0.95, 0.91, 0.77, 0.75) (Table 2). The 13C and HSQC spectral data confirmed the presence of 25 carbons assignable to one ketone (δC 194.2), two carbony-like deshielded carbons (δC 177.7, 164.5), four sp2 carbons (δC 147.2, 132.3, 120.9, 111.4), two sp2 methines (δC 128.3, 116.9), one acetal/hemiacetal carbon (δC 99.4), two oxygenated sp3 methines (δC 77.3, 75.3), five sp3 methines (δC 43.0, 40.9, 33.9, 32.3, 29.7), three sp3 methylenes (δC 36.7, 33.2, 28.1), and five methyl carbons (δC 17.5, 16.9, 13.5, 12.5, 11.5) (Table 2).

Table 2 NMR data for phytohabimicin (2) in CD3OD

COSY analysis clarified three spin systems: a seven-carbon fragment from H3-21 to H2-9 with a methyl substitution at C-7, a six-carbon fragment from H3-23 to H-14 with a methyl group at C-13, and a two-carbon fragment H-15/H3-25 (Fig. 3). These three fragments were joined into one carbon chain from C-5 to C-15 by HMBC correlations from H-6, H2-9, H2-12, and H3-23 to C-10 and H-15 and H3-25 to C-14. Though HMBC correlations that connect C-10 and C-14 were not observed, formation of a 6,6-spiroacetal ring system by the carbons from C-6 to C-14 was inferred by a NOESY correlation between H-6 and H-14 and a deshielded resonance of C-10 (δC 99.4), completing an aliphatic part of 2.

Fig. 3
figure 3

COSY and key HMBC and NOESY correlations for 2

HMBC correlations from the deshielded singlet proton H-3 to C-1, C-2, and C-4 and their 13C chemical shifts suggested a 2-substituted thiazole-carboxylic acid moiety. The position of the carboxyl group was unable to be assigned only by NMR analysis because thiazole-4-carboxylic acid and thiazole-5-carboxylic acid show almost the same chemical shifts for thiazole carbons and protons (Fig. S15). Bacterial thiazole carboxylic acids are generally synthesized from cysteine by NRPS. The genomic report, BLAST search, and antiSMASH [10] analysis of the closest strain P. suffuscus NBRC 105367 T detected a thiazole generating NRPS in a calcimycin-class type I PKS gene cluster. Additionally, thiazole-5-carboxylic acid is not known in natural products (https://dnp.chemnetbase.com). Based on these considerations, the carboxyl group was proposed to be connected at C-2.

The remaining four carbons could be assigned to constitute a pyrrole ring which is commonly present in the structures of calcimycins as a PKS starter unit [11]. Chemical shifts of the carbons from C-17 to C-20 showed close similarity to those for the pyrrole carbons in calcimycins and HMBC correlations shown by an sp2 methine H-18 to C-16, C-17, and C-20 established the connectivity of this ring system to C-16. Though no HMBC correlation was observed to C-19, the position of this carbon in a pyrrole ring was determined by comparing the reported NMR data for the related compounds (Fig. 4). The remaining two chlorine atoms were connected to C-19 and C-20 to satisfy the molecular formula. The position of chlorine atoms was further verified by 13C chemical shift comparison with known dichloropyrroles. 2,3 and 3,5-dichloropyrroles show a clear contrast in the 13C chemical shifts of a methine and its neighboring chlorinated carbon: the methine carbons of 5-acetyl-2,3-dichloropyrrole and 2-acetyl-3,5-dichloropyrrole resonates at δC 115.6 and 110.4 while the chlorinated C-3 carbons at δC 110.5 and 119.7, respectively (Fig. 4) [12]. Several natural products bearing a 2,3-dichloropyrrole show similar chemical shift pattern to 2 [13, 14]. The 3,5-dichloropyrrole moiety in nai414-A shows similar chemical shifts to 2-acetyl-3,5-dichloropyrrole [15]. Finally, these substructures were connected to spiroacetal core structure by HMBC correlations from H3-21 to C-4, COSY fragment H-15/H3-25 to C-14 and C-16 (Fig. 3).

Fig. 4
figure 4

13C chemical shift comparison of acyldichloropyrroles

The relative configuration of 2 was determined by analyzing 3JHH coupling constants and NOESY correlation data (Fig. 5). A large vicinal coupling constant between H-6 and H-7 (10.4 Hz) indicated the diaxial orientation of these protons and thus an equatorial orientation of the C-5 substituent. A small coupling constant for H-5 and H-6 (1.8 Hz) together with NOE correlations H-7/H3-21 and H-5/H3-22 allowed the placement of H3-21 methyl group anti to H-6 and H-5 methine directing to the same side as H-7. The anti relationship for H-14/H-15 and H-14/H3-24 were evidenced by a large coupling constant 3JH14,H15 10.2 Hz which was obtained by 1D selective homonuclear decoupling experiment and an NOE correlation H-15/H3-24. Additional NOE correlations H-11/H3-24 and H-13/H3-25 suggested the equatorial orientation of H3-23 and H-13 and the orientation of H3-25 directing opposite to the C-14 ether oxygen. An NOE between H-6 and H-14 assured the relative configuration of the spiroacetal ring system and these data established the chair conformation of tetrahydropyran rings, thereby establishing the overall relative stereochemistry of 2 as shown in Fig. 5.

Fig. 5
figure 5

Relative stereochemical analysis for 2. Thz thiaxolecarboxilic acid, Dcp dichloropyrrole

The absolute configuration of 2 was analyzed by comparing the experimental and calculated ECD simulation using time-dependent density functional theory (TDDFT). Conformational search using molecular mechanics and subsequent DFT optimization afforded four stable conformers within an energy threshold of 3.0 kcal mol−1 (Fig. S13). The calculated spectrum based on those geometries agreed with the experiment ECD of 2 (Fig. 6). The absolute configuration of 2 was thus determined as 5 R,6 S,7 S,10 R,11 R,13 R,14 S and 15 S, which were identical with that for calcimycin-class metabolites except for the C-7 methyl group. Large chemical shift differences for the carbons C-6, C-7, C-22, and C-8, neighboring C-7, support the inversed configuration of C-7 (Fig. S14).

Fig. 6
figure 6

Comparison of the experimental ECD spectrum of 2 in CHCl3 (black line), the calculated ECD spectra of (5 R,6 S,7 S,10 R,11 R,13 R,14 S,15 S)-2 (dotted line), and its enantiomer (5 S,6 R,7 R,10 S,11 S,13 S,14 R,15 R) (gray line)

Biological activities of 1 and 2 were evaluated in antimicrobial, cytotoxicity, and cancer cell migration assays. Antimicrobial activity was tested against Gram-positive and -negative bacteria and yeasts in comparison with calcimycin (Table 3). Compound 1 was weakly active against Gram-positive bacteria while inactive against Gram-negative bacteria and yeasts. Compound 2 showed moderate activity against Gram-positive bacteria, R. solanacearum, and yeast while no activity against E. coli and R. radiobacter (Table 3). Compounds 1, 2, and calcimycin exhibited moderate to potent cytotoxic activity against P388 murine leukemia cells with IC50 values of 1.7, 4.4, and 0.19 μΜ, respectively. In addition, 2 significantly inhibited the migration of EC17 cancer cells at 10 μΜ, which is clearly more effective than the 30 μΜ positive control, LY294002 (Fig. 7).

Table 3 Antimicrobial activity of compound 1, 2, and calcimycin
Fig. 7
figure 7

Inhibitory effect of 2 on EC17 cells migration

Two classes of new polyketides, phytohabinone (1) and phytohabimicin (2), were discovered from a rare actinomycete Phytohabitans sp. RD003013. Compound 1 is a meroterpenoid, mixed with polyketide and terpenoid, which is widely distributed in eukaryotes. Several types of naphthoquinone-based meroterpenoids are known from Streptomyces [16] but they were rarely from non-Streptomyces. Structurally similar compounds of 1, phosphatoquinone B [8], fumaquinone [9] (Fig. 7), and other prenylated naphthoquinones were reported from Streptomyces; however, prenylation position is different with 1. The naphthoquinones are prenylated on a quinone ring at C-2 and/or C-3; therefore, one of the prenyltransferases of strain RD003013 seems to be a different lineage. Fumaquinone is the only reported metabolite in the same prenylated position (Fig. 8).

Fig. 8
figure 8

Related natural products of 1 and 2 isolated from actinomycetes

Compound 2 is a new member of calcimycin-class polyketides. The calcimycin family is characterized by a spiroacetal core structure modified with a benzoxazole and pyrrole moieties. Known calcimycin congeners were isolated from Streptomyces [17,18,19,20], Dactylosporangium [21], and Frankia [22] and structural variation is seen in the methyl substitution in the polyketide chain and hydroxylation of the benzoxazole (Fig. 8). By contrast, 2 possesses a chlorinated pyrrole and a thiazolecarboxylic acid at both ends of the polyketide backbone and the stereochemically inversed methyl group at C-7 (Fig. 9), which makes this compound distinctive from the previously known calcimycins. These structural differences likely affect the biological activity: calcimycin is inactive against Gram-negative bacterium, R. sclanacearum, whereas 2 exhibits moderate activity.

Fig. 9
figure 9

Putative biosynthetic pathway for phytohabimicin (2)

The family Micromonosporaceae is one of the most prominent families of actinomycetes. They are recognized as a rich source of various natural products; however, most genera remain unexplored. According to the BLAST search, both P. suffuscus and P. houttuyneae have an almost complete set of compound 2 biosynthetic genes except for the genes for benzoxazole biosynthesis (Table 4 and S1). In contrast, the calcimycin-producing strain, Streptomyces chzrteusis NRRL 3882 lacks halogenase and thiazole biosynthetic genes in its calcimycin cluster. The result suggests that even though core biosynthetic genes are the same, modification genes differ according to the genera. Besides, strain RD003013 shares 100% of 16 S rRNA gene sequences with the strain of P. suffuscus K07-0523T and phytohabitols producing strain RD002984 [7]. Although strain RD003013 produced phytohabitols, strain RD002984 did not produce compounds 1 and 2; therefore, the ability to produce secondary metabolites is difficult to determine by 16 S rRNA gene similarity. Thus, unstudied rare actinomycetes for natural products are worth detailed screening even though they are the same species in the 16 S rRNA gene. Our recent analysis indicated the high potential of the genus Phytohabitans for secondary metabolites. We will report the detail of the metabolite analysis of Phytohabitans in the next paper.

Table 4 Blast search of compound 2 biosynthetic gene clusters in P. suffuscus

Materials and methods

General experimental procedures

Optical rotations were measured using a DIP-3000 polarimeter (JASCO, Tokyo, Japan). UV spectra were recorded on a UV-1900 spectrophotometer (Shimazu, Kyoto, Japan). ECD spectra were recorded on a J-720W spectropolarimeter (JASCO). IR spectra were measured by a spectrum 100 spectrometer (PerkinElmer, MA, USA). NMR spectra were obtained on an AVANCE NEO 500 spectrometer (Bruker, MA, USA) in DMSO-d6 (δH 2.50, δC 39.5) and CD3OD (δH 3.31, δC 49.0). HRESITOFMS spectra were recorded on a compact QTOF mass spectrometer (Bruker) Cosmosil 75C18-PREP (Nacalai Tesque, Inc. Kyoto, Japan) for flash ODS column chromatography. HPLC separations were performed using a COSMOSIL 5C18-PAQ Packed Column (10 × 250 mm, Nacalai Tesque, Inc.). The computational study was performed using MacroModel implemented in the Maestro 12.8 software package [23] and the Gaussian16 Rev C.01 program [24]. A part of these computations was conducted using the SuperComputer System, Institute for Chemical Research, Kyoto University. Molecular structures were visualized using the Maestro 12.8 software package. ECD spectra were visualized using GaussView 6.0.16 and Microsoft Excel.

Microorganism

Strain RD003013 was obtained from Biological Resource Center, National Institute of Technology and Evaluation (NBRC), Chiba, Japan. The results of phylogenetic analysis based on 16 S rRNA gene sequences indicated that strain RD003013 belonged to the genus Phytohabitans (Fig. S17). The highest similarity value was observed with Phytohabitans suffuscus K07-0523T (AB490769, 100.0%). The DDBJ accession number for the 16 S rRNA gene sequence of strain RD003013 is LC688267 (1425 nucleotides).

Fermentation and isolation

Strain RD003013 growing on an ISP 2 agar medium consisting of 0.4% yeast extract (Kyokuto Pharmaceutical Industrial, Tokyo, Japan), 1.0% malt extract (Becton Dickinson, NJ, USA), 0.4% glucose (pH 7.2) was inoculated into 500-ml K-1 flasks (custom-ordered cylindrical flask) each containing 100 ml of the V-22 seed medium consisting of 1.0% soluble starch, 0.5% glucose, 0.3% N-Z-case (Wako Pure Chemical Industries, Tokyo, Japan), 0.2% yeast extract, 0.5% Tryptone (Becton Dickinson), 0.1% K2HPO4, 0.05% MgSO4·7H2O, and 0.3% CaCO3 in distilled water (pH 7.0). The flasks were placed on a rotary shaker (200 rpm) for 7 days at 30 °C. Then, the seed cultures (3 ml) were transferred to 500-ml K-1 flasks, each containing 100 ml of the A-11M production medium consisting of 2.5% soluble starch, 0.2% glucose, 0.5% N-Z-Amine A (Wako Pure Chemical Industries, Tokyo, Japan), 0.5% yeast extract, 0.3% CaCO3, and 1.0% Diaion HP-20 resin (Mitsubishi Chemical, Tokyo, Japan) in distilled water. The pH of the medium was adjusted to 7.0 before sterilization. All the media were sterilized by autoclaving at 121 °C for 20 min. The inoculated 20 flasks were placed on a rotary shaker (200 rpm) at 30 °C for 14 days. After incubation, 100 ml of 1-BuOH was added to each flask, and the flasks were allowed to shake for 1 h. The mixture was centrifuged at 6000 rpm for 10 min, and the organic layer was separated from the aqueous layer containing the mycelium. The BuOH layer was evaporated to give 11.1 g of extract from 6.3 l of culture. The extract was fractionated using silica gel column chromatography with a step gradient of CHCl3 and MeOH (1:0, 20:1, 10:1, 4:1, 2:1, 1:1, and 0:1 v/v). The fraction 4:1 containing 1 and 2 was concentrated to give 3.0 g, which was subjected to ODS flash column chromatography with a gradient of MeCN and 0.1% HCO2H solution (2:8, 3:7, 4:6, 5:5, 6:4, 7:3, 8:2, and MeOH v/v), and then washed with MeOH. The washing fraction was evaporated, and the remaining aqueous layer was extracted three times with EtOAc and concentrated to give a red solid (974.6 mg). 314 mg of the fraction was subjected to ODS flash column chromatography with a gradient of MeCN/0.1% HCO2H solution (5:5, 6:4, 7:3, 8:2, 8:2, 8:2, 8:2, 8:2, 8:2, 9:1, 9:1, 9:1, 9:1, 9:1, and 9:1 v/v), and then washed with MeOH. The third fraction of 9:1 was evaporated, and the remaining aqueous layer was extracted three times with EtOAc and concentrated to give a red solid (17.9 mg). The final purification was achieved by preparative HPLC using an isocratic condition of 66% MeCN/0.1% HCO2H solution at 4 ml min−1, yielding 1 (5.5 mg) and 2 (6.0 mg) with a retention time of 21.0 min and 17.9 min, respectively.

Phytohabinone (1): yellow amorphous solid; UV (MeOH) λmax (log ε) 270 (4.23), 428 (3.62); IR (ATR) νmax 3392, 2912, 1624, 1588, 1442, 1329, 1226, 1055, 795 cm−1; 1H and 13C NMR data, Table 1 and Supporting Information; HRESITOFMS m/z 339.1600 [M − H] (calcd for C21H23O4, 339.1602).

Phytohabimicin (2): white powder; [α]22D + 159 (c 0.12, CHCl3); UV (MeOH) λmax (log ε) 238 (3.40), 296 (3.55); ECD (9.2 × 10–5 M, CHCl3) λmaxε) 305.6 (+13.0), 242.8 (−5.6) nm; IR (ATR) νmax 2932, 1650, 1401, 1083, 987 cm−1; 1H and 13C NMR data, Table 1 and Supporting Information; HRESITOFMS m/z 541.1336 [M − H] (calcd for C25H3135Cl2N2O5S, 541.1336).

Computational analysis

The conformational sampling of structure 2 was performed by applying 100,000 steps of the Monte Carlo Multiple Minimum (MCMM) method with PRCG energy minimization by the OPLS4 force field to obtain 124 conformational isomers within 10.0 kcal mol−1 from the minimum energy conformer. Geometries of the conformers were then optimized at the M06-2X/6-31 G(d) level of theory with the SMD solvation model (CHCl3). Frequency calculations were carried out at the same level of theory to confirm the absence of imaginary frequencies and obtain thermal corrections for the Gibbs free energy. After eliminating duplicated structures with the threshold of 0.01 Å RMSD, the single-point energy was calculated at the M06-2X/def2-TZVP-SMD(CHCl3) level of theory, affording four conformers within 3.0 kcal mol−1 from the minimum Gibbs free energy. The ECD spectrum of each conformer was simulated by the TDDFT calculation of 25 excited states at the ωB97X-D/def2-TZVP-IEFPCM(CHCl3) level of theory. The spectrum of 2 was created by the weighted average of the above-obtained spectra (half-width: 0.24 eV) according to the Boltzmann distribution, applied the UV correction, and scaled the vertical axis.

Biological assays

Antimicrobial, cytotoxicity, and cell migration assays were carried out according to the method described previously [25, 26].