Introduction

Terpenoid metabolites are an important class of natural products that act as antibiotics, hormones, flavor and odor constituents, and pigments. Some of these metabolites also possess other physiologically or commercially important properties as vitamins and antimalarial and antitumor agents. Most of these metabolites have been isolated from terrestrial and marine higher plants, liverworts and fungi. Actinomycetales microorganisms are prolific producers of a large number of natural products; however, the discovery of terpenoid metabolites from Actinomycetales microorganisms is rare. Previous studies reported the production of the volatile monoterpene and degraded sesquiterpene alcohols, 2-methylisoborneol and geosmin, respectively, from soil bacteria, actinomycetes,1, 2 cyanobacteria3, 4 and myxobacteria.5, 6 Although a few terpenoid metabolites have been isolated from Actinomycetales microorganisms, many genes encoding presumptive terpene synthases, which were in a cryptic state in the microorganisms, have recently been identified in public bacterial genome databases.7 Furthermore, many novel terpenes have been isolated by the expression of these silent genes in the heterologously engineered Streptomyces host, Streptomyces avermitilis SUKA22.8

Several cyclic terpene hydrocarbons and alcohols are formed by variations in the common terpene cyclization mechanism that is initiated by enzyme-catalyzed ionization to form allylic cations from the universal acyclic precursors geranyl diphosphate (for monoterpenes), farnesyl diphosphate (for sesquiterpenes), and geranylgeranyl diphosphate (GGPP; for diterpenes). The intramolecular electrophilic attack of the intermediate allylic cations on the central or distal double bonds of the substrate, followed by well-precedented cationic transformations, including hydride shifts and carbon–carbon backbone rearrangements, and the ultimate quenching of the positive charge either by deprotonation or by the capture of a nucleophilic water molecule, may account for the formation of several hundred cyclic terpene compounds. In the formation of mono- and sesquiterpenes, these cyclization reactions are accomplished by a single polypeptide synthase, whereas the formation of diterpenes is achieved by the above single polypeptide synthase (type-A or type-B terpene synthase) or a combination of two fundamental types of terpene synthases (type-A, type-B, type-A–type-B, and type-B–type-A).9 The type-A diterpene synthase reaction is initiated by ionization of the diphosphate of GGPP to give an allylic cation, followed by deprotonation to generate an olefinic diterpene hydrocarbon or the capture of a water molecule to produce a diterpene alcohol. The cyclization reaction catalyzed by type-B diterpene synthases is initiated by protonation at the C-14,15-double bond of GGPP distal to the allylic diphosphate.

In bacterial terpene synthases, more than 260 presumptive synthases have been classified into type-A synthase; however, the combination of type-A and type-B diterpene synthases is extremely rare. Only three pairs of type-A and type-B diterpene synthases, which are involved in the formation of the diterpenoid metabolites viguiepinol,10 oxaloterpin C11 and terpentecin,12 had been characterized from a bacterial origin. We recently characterized two pairs of type-A and type-B diterpene synthases from S. cyslabdanicus K04-0144 and S. anulatus GM95, respectively, and they catalyzed the formation of the labdane-type bicyclic diterpene hydrocarbon, labda-8(17),12(E),14-triene.13 We herein determined whether other Actinomycetales microorganisms harbor the labdane-type diterpene synthases from public databases and characterized them.

Materials and methods

Preparation of recombinant type-A and type-B diterpene synthases in Escherichia coli

Four genes, cldB, cldD, sclav_p0491 and sclav_p0490, were amplified by PCR with cosmid clones containing the cld cluster and sclav_p0491/sclav_p0490 as template DNA, respectively, using the following primer pairs: forward for cldB: 5′-GCTGGCACAT ATGAGCTCCATATCCCACCAGGCGGC-3′ (underlined and bold characters indicate the NdeI site and start codon of cldB, respectively) and reverse for cldB: 5′-CTGCAGAAGCTT TCAGGGCCGACGGGCCTGCAGCA-3′ (underlined and bold characters indicate HindIII and the stop codon of cldB, respectively), forward for cldD: 5′-GGGAGAGCAT ATGACCCGCACCGGCGATGCGGTGAC-3′ (underlined and bold characters indicate the NdeI site and start codon of cldD, respectively) and reverse for cldD: 5′-GTCTCCAAGCTTGGTTCAACGACTGGTTGGGGCCGCG-3′ (underlined and bold characters indicate HindIII and the stop codon of cldD, respectively), forward for sclav_p0490: 5′-CCTGTCCCAT ATGCCCGTCGACGTCGGCACCCTTC-3′ (underlined and bold characters indicate the NdeI site and start codon of sclav_p0490, respectively) and reverse for sclav_p0490: 5′-GACCACAAGCTT TCAGTCGGCGGGCCCGGTCCGAG-3′ (underlined and bold characters indicate HindIII and the stop codon of sclav_p0490, respectively), and forward for sclav_p0491: 5′-GGGAGTGCAT ATGAGGTCTGGGGACCGTGGCGAG-3′ (underlined and bold characters indicate the NdeI site and start codon of sclav_p0491, respectively) and reverse for sclav_p0491: 5′-GAGCAGAAGCTT CTACTCCTCCCGCCAGCGGTCGC-3′ (underlined and bold characters indicate HindIII and the stop codon of sclav_p0491, respectively). Initial denaturation at 96 °C for 180 s was followed by 25 cycles of amplification (95 °C for 30 s, 60 °C for 30 s and 72 °C for 60 s (cldD and sclav_p0491) or 100 s (cldB and sclav_p0490)) and then a final incubation at 72 °C for 5 min. After amplification, the reaction mixture was treated with DpnI to remove template DNA. Amplified DNA was digested with NdeI and HindIII, and the NdeI/HindIII amplicon was then ligated with the large NdeI/HindIII-pET28a(+). After confirming the sequence of each amplicon, each recombinant plasmid was expressed as the N-terminal His6-tagged fusion protein in E. coli JM109 (DE3). Transformants were grown at 37 °C in Luria broth with 50 μg of kanamycin per ml. Expression of the recombinant protein was induced by adding 0.1 mM isopropyl-β-d-thiogalactopyranoside when optical density at 600 nm reached approximately 0.8. Cultivation was continued for an additional 12 h at 18 °C. After the induced cells were harvested by centrifugation at 3000 r.p.m. at 4 °C for 10 min, the cells were disrupted by sonication 10 times for 15 s (20-s cooling intervals). Since transformants carrying pET28a(+)::cldB and pET28a(+)::sclav_p0490 produced insoluble proteins, these proteins were sedimented by centrifugation at 5000 r.p.m. at 4 °C for 10 min. Insoluble proteins were dissolved in buffer containing 50 mm Tris-HCl (pH 7.5), 1 mm ETDA, 5 mm dithiothreitol (DTT) and 8 m urea at 37 °C for 20 h. After insoluble materials had been removed by centrifugation at 10 000 r.p.m. at 4 °C for 10 min, the supernatant was dialyzed against buffer containing 50mm Tris-HCl (pH 7.5), 1 mm EDTA, 1 mm DTT and 20 v/v% glycerol. Dialyzed solution was centrifuged at 10 000 r.p.m. at 4 °C for 20 min in order to remove insoluble materials. The purification of His6-tagged recombinant proteins was performed in accordance with the manufacturer’s (Qiagen K. K., Tokyo, Japan) protocol using Ni-NTA agarose. Purified proteins were analyzed by sodium dodecyl sulfate 12.5 w/v% polyacrylamide gel electrophoresis.

Enzymatic synthesis of diterpenes by type-A and type-B diterpene synthases

The reaction mixture (0.5 ml) contained 50 mm Tris-HCl (pH 7.5), 10 mm MgCl2, 1 mm DTT, 0.1 w/v% Tween 80, 20 w/v% glycerol, 30 μm IPP, 10 μm DMAPP, 100 μg of purified recombinant GGPP synthase (N-terminus His6-tagged CldA prepared from E. coli JM109 (DE3) harboring pET28a(+)::cldA) and 150 μg each of the purified type-A and type-B diterpene synthases per ml. The reaction was carried out at 30 °C for 16 h, was stopped by the addition of 50 μl of 0.5 m EDTA (pH 7.5), and the product was extracted two times with 0.5 ml of n-pentane. The upper organic layers were combined and concentrated under reduced pressure to remove n-pentane. The residue was dissolved in 100 μl of methanol and a 2-μl portion of the extract was subjected to GC-MS.

Expression of genes encoding type-A and type-B diterpene synthases in the engineered Streptomyces host

Each gene encoding cldB, cldD, sclav_p0490 or sclav_p0491 was amplified by PCR with the cosmid clone containing the cld cluster or sclav_p0490/sclav_p0491 using the following primer pairs: forward for cldB: 5′-CGCTCGGGTCTAGACGGAAGGCAGTCATGAGCTCCATATCCC-3′ (underlined and bold characters indicate the XbaI site and start codon of cldB, respectively) and reverse for cldB: 5′-TTAATTAAGCAACCTTTGC TCAGGGCCGACGGGCCTGCAGCAG-3′ (underlined and bold characters indicate the BstAPI site and stop codon of cldB, respectively), forward for cldD: 5′-TGCCTGCCGCAAAGGTTGCAGGCCGCATGACCCGCACC-3′ (underlined and bold characters indicate the BstAPI site and start codon of cldD, respectively) and reverse for cldD: 5′-CCCGAGAAGCTT TCAACGACTGGTTGGGGCCGCGCCGCG-3′ (underlined and bold characters indicate the HindIII site and stop codon of cldD, respectively), forward for sclav_p0490: 5′-CCTCTAGAGAGAAAGGCAGTCATGCCCGTCGACGTCGGCAC-3′ (underlined and bold characters indicate the XbaI site and start codon of sclav_p0490, respectively) and reverse for sclav_p0490: 5′-TTAATTAAGCAACCTTTGC TCAGTCGGCGGGCCCGGTCCGAGC-3′ (underlined and bold characters indicate the BstAPI site and stop codon of sclav_p0490, respectively), or forward for sclav_p0491: 5′-TGCCTGCCGCAAAGGTTGCGAGGAGTGTCCGTGAGGTCTGGGG-3′ (underlined and bold characters indicate the BstAPI site and start codon of sclav_p0491, respectively) and reverse for sclav_p0491: 5′-CCCTCGAGAAGCTTGCTACTCCTCCCGCCAGCGGTCGC-3′ (underlined and bold characters indicate the HindIII site and stop codon of sclav_p0491, respectively). Initial denaturation at 96 °C for 180 s was followed by 25 cycles of amplification (96 °C for 30 s, 60 °C for 30 s and 72 °C for 60 s (cldD and sclav_p0491) or 100 s (cldB and sclav_p0490)) and then a final incubation at 72 °C for 5 min. After amplification, the reaction mixture was treated with DpnI to remove template DNA. Each amplified DNA was digested with XbaI and BstAPI (cldB and sclav_p0490) or BstAPI and HindIII (cldD and sclav_p0491). The large XbaI/HindIII-cut pKU1021ggs7 was ligated with the XbaI/BstAPI cldB amplicon and BstAPI/HindIII cldD amplicon, XbaI/BstAPI cldB amplicon and BstAPI/HindIII sclav_p0491 amplicon, XbaI/BstAPI sclav_p0490 amplicon and BstAPI/HindIII cldD amplicon, or XbaI/BstAPI sclav_p0490 and BstAPI/HindIII sclav_p0491amplicon to generate pLDT101 (pKU1021ggs::cldB::cldD), pLDT102 (pKU1021ggs::cldB::sclav_p0491), pLDT103 (pKU1021ggs::sclav_p0490::cldD), or pLDT104 (pKU1021ggs::sclav_p0490::sclav_p0491), respectively. Each amplified region was confirmed by sequencing. After these integrating plasmids were transformed into E. coli GM2929 hsdS::Tn10 to prepare unmethylated DNA preparations, the DNAs prepared were introduced into an engineered S. avermitilis SUKA22 by PEG-assisted protoplast transformation.14 Transformants were obtained by the selection of resistance to neomycin and spores of the transformants were grown on YMS15 agar medium containing 1 μg of neomycin per ml to prepare spores. The spores of S. avermitilis SUKA22 transformants carrying pLDT101, pLDT102, pLDT103 or pLDT104 were grown in vegetative medium16 at 30 °C for 2 days on a reciprocal shaker (200 strokes per min). A 0.15-ml portion of the vegetative culture was inoculated into 15 ml of the production medium16 in a 125-ml flask. The whole culture was centrifuged at 3000 r.p.m. for 10 min. Sedimented mycelia were extracted with 5 ml of methanol and removed by centrifugation. A 2-μl portion of the methanol extract was directly subjected to GC-MS.

Isolation of diterpenes by heterologous expression in the engineered Streptomyces host

Two liters of the culture (20 × 100 ml medium in a 500-ml Erlenmeyer flask) growing S. avermitilis SUKA22 carrying pLDT101, pLDT102, pLDT103 or pLDT104 in production medium16 at 28 °C for 5 days was centrifuged at 6000 r.p.m. for 20 min to separate mycelia and the supernatant. Sedimented mycelia were suspended in 500 ml of water and the suspension was re-centrifuged at 6000 r.p.m. for 20 min. Washed mycelia were extracted twice with 500 ml of methanol and mycelia were then removed by filtration. The methanol extract was concentrated under reduced pressure to make approximately 100 ml. The concentrate was extracted two times with 50 ml of n-hexane. The upper organic phases were combined and extracted with 100 ml of 0.1 n NaOH in order to remove free fatty acids. The organic phase was separated, dried over anhydrous Na2SO4 and concentrated to yield a brownish oily material (0.5–1.2 g). The oily material was dissolved in 20 ml of n-pentane and insoluble materials were removed by filtration. The filtrate was evaporated to reduce the volume and the concentrate obtained was subjected to silica gel column chromatography (20 × 150 mm). The column was developed with n-pentane and the eluates were fractionated. The metabolites eluted were monitored by silica gel TLC developed with n-hexane. Diterpene compounds on silica gel TLC were detected by heating at 110 °C after spraying 5 w/v% phosphomolybdic acid in ethanol solution. The fractions containing the objective metabolites were combined and concentrated by evaporation. Diterpene hydrocarbon 1 (20.2 mg, λmaxMeOH nm (logɛ) 232 (4.32), [α]25D +29.5 (c 0.5, CHCl3)) was obtained as transparent oil from S. avermitilis SUKA22 carrying pLDT101. The HR-MS (ESI mode) of 1 showed a molecular ion peak at m/z 272.2506 [M]+, corresponding to C20H32 (calcd. for 272.2504). Diterpene hydrocarbons 2 (13.4 mg, λmaxMeOH nm (logɛ) 232 (4.36), [α]25D +28.0 (c 0.5 CHCl3)) and 3 (7.2 mg, λmaxMeOH nm (logɛ) 232 (4.35), [α]25D; +3.6 (c 0.5 CHCl3)) were obtained as transparent oil from S. avermitilis SUKA22 carrying pLDT102 and pLDT103, respectively. The HR-MS (ESI) of 2 and 3 showed molecular ion peaks at m/z 272.2507 [M]+ and 272.2506 [M]+, respectively, corresponding to C20H32 (calcd. for 272.2504). Diterpene hydrocarbon 4 (8.1 mg, λmaxMeOH nm (logɛ) 232 (4.32), [α]25D; +3.5 (c 0.5 CHCl3)) was obtained as transparent oil from S. avermitilis SUKA22 carrying pLDT104. The HR-MS (ESI) of 4 showed molecular ion peaks at m/z 272.2505 [M]+, corresponding to C20H32 (calcd. for 272.2504).

Physicochemical analysis

The crude and purified preparations of metabolites were analyzed by GC-MS (Shimadzu GC-17 A; 70 eV, electron ionization, positive ion mode; 30 m × 0.25 mm InterCap 5 capillary column (diphenyl (5%)-dimethyl (95%) polysiloxane; GL Sciences, Tokyo, Japan) using a temperature program of ~50–280 °C and temperature gradient of 20 °C/min). The compounds produced were evaluated by comparisons with the mass spectra of the corresponding reference compounds in the GC-MS database, the NIST/EPA/NIH MS Library (2014 version). 1D and 2D NMR (1H: 500 MHz; 13C: 125 MHz, in CDCl3) spectra were obtained on a JEOL JNM-ECP 500 FT NMR system. Abbreviations in 1H NMR signals were as follows: singlet (s), multiplet (m), doublet (d), triplet (t), double doublet (dd), double triplet (dt) and broad singlet (brs) signals. HR-MS by the ESI mode were obtained on a JEOL JMS-700 Mstation. Optical rotations were recorded on a Horiba SEPA-300 polarimeter.

Results

Searching pairs of type-A and type-B diterpene synthases frompublic databases

In our previous studies, more than 260 presumptive terpene synthases were discovered in public databases and in-house databases using a search method based on the use of hidden Markov models and the protein families database (Pfam) search.7 Of these, the type-A diterpene synthases with E-value ranges from 6 × 10−9 to 2.9 × 10−50, terpenete-3,13,15-triene synthase (311-aa; BAB39207) and pimara-9(11),15-diene synthase (295-aa; BAD86798), located downstream of the type-B diterpene synthases, terpentedienyl diphosphate synthase (499-aa; BAB39206) of terpentecin-producing Kitasatospora griseola MF730-N6 and ent-copalyl diphosphate synthase (511-aa; BAD86797) of viguiepinol-producing Streptomyces sp. KO-3988, respectively, had already been characterized. We recently characterized new genes encoding the labda-8(17),12(E),14-triene synthases, CldD (329-aa; BAR97453) of S. cyslabdanicus K04-0144 and RmnD (322-aa; BAR97463) of S. anulatus GM95, which were also located downstream of genes encoding the (+)-copalyl diphosphate synthases, CldB (532-aa; BAR97451) and RmnB (549-aa; BAR97461), respectively. Since the genes encoding each pair of type-A and type-B diterpene synthases formed an operon, a new pair of type-A and type-B diterpene synthases were searched for in the public databases using a combination of the recent bacterial type-A terpene synthase hidden Markov models and bacterial type-B diterpene synthase hidden Markov models, as reported previously.17 The pairs of diterpene synthases selected by their genes were located in the same operon or gene cluster. In consideration of the pair of type-A and type-B diterpene synthases in the diterpene synthase family from bacterial proteins in the public databases, Sare_1288 (531-aa)/Sare_1287 (295-aa) of Salinispora arenicola CNS-205, SCLAV_5672 (529-aa)/SCLAV_5671 (311-aa) and SCLAV_p0490 (545-aa)/SCLAV_p0491 (343-aa) of Streptomyces clavuligerus ATCC 27064, AHK61133 (311-aa)/AHK61132 (510-aa) of S. griseus CB00830 (viguiepinol, oxaloterpins C and E producer), and Sros_3192 (560-aa)/Sros_3191 (295-aa) of Streptosporangium roseum DSM 43021 were selected as the combinations of pairs of type-A and type-B diterpene synthases.

In order to evaluate these diterpene synthase pairs, the deduced amino acid sequences of type-A and type-B diterpene synthases were aligned and subjected to a phylogenetic analysis. Type-A diterpene synthases were classified into at least two groups (group 1: Sros_3191, Sare_1287, SCLAV_5671, AHK61133 and BAD86798; group 2: SCLAV_p0491, PmnD, CldD and BAB39207) in Figure 1a. Since two pimara-9(11),15-diene synthases, AHK61133 and BAD86798, belong to group 1, the other synthases, Sros_3191, Sare_1287 and SCLAV_5671, may exhibit similar catalytic activities. SCLAV_5671 was located in the pimara-9(11),15-diene synthase clade. The presumptive type-A diterpene synthase, SCLAV_p0491, was very similar to the labda-8(17),12(E),14-triene synthases, RmnD and CldD. All predicted type-A terpene synthases possessed a pair of characteristically conserved metal-binding domains consisting of the acidic amino acid-rich motif DW[LQ]VD (Sros_3191 and Sare_1287), DD[HRY][TVW]E (SCLAV_5671, AHK61132, BAD86798 and BAB39207) or DD[AMV]H[GV]E (SCLAV_p0491, CldD and RmnD) located within 65–114 amino acids of the N-terminus, while the downstream triad Nxxx[ST]xxx[DE] found other 140 amino acids downstream, closer to the C-terminus (Figure 2a). The phylogenetic analysis of type-B synthases shown in Figure 1b suggested that the synthases were classified into three groups (group 3: Sros_3192, Sare_1288, SCLAV_5672, AHK61132 and BAD86797; group 4: SCLAV_p0490, RmnB and CldB; group 5: BAB39206). Five synthases, including the ent-copalyl diphosphate synthases, AHK61132 and BAD86797, belonged to the group 3 clade. Since SCLAV_5672 was similar to AHK61132 and BAD86797, the presumptive synthase may exhibit ent-copalyl diphosphate synthase activity. On the other hand, the presumptive synthase, SCLAV_p0490, was classified into two (+)-copalyl diphosphate synthases, RmnB and CldB, suggesting that SCLAV_p0490 exhibited similar catalytic activity to these synthases. Sequence alignment with the predicted type-B diterpene synthases revealed that the motifs DxD[DT]T[AS] and Qxx[DNS]G[SG]W, which were proposed to stabilize an intermediate cation during the cyclization process and mediate substrate binding by the chelation of divalent metal ions,18, 19 respectively, were highly conserved in all predicted type-B diterpene synthases, except for SCLAV_p0490, which possessed two motifs 315DADDTA and 444RRTDGSW (Figure 2b).

Figure 1
figure 1

Phylogenetic analysis of (a) type-A diterpene synthases and (b) type-B diterpene synthases from Actinomycetales microorganisms. A phylogenetic analysis of aligned sequences was conducted using the bootstrap method (bootstrap number: 1000; seed number: 111) of CLUSTALW version 2.10 (ftp://ftp.ebi.ac.uk/pub/software/clustalw2/2.1/). A bootstrap tree was drawn by njplot (http://doua.prabi.fr/software/njplot). Abbreviations: AHK61132 (510-aa, accession # AHK61132), AHK61133 (311-aa, accession # AHK61132), BAB39206 (499-aa, accession # BAB39206), BAB39207 (311-aa, accession # BAB39207), BAD86797 (511-aa, accession # BAD86797), BAD86798 (295-aa, accession # BAD86798), CldB (532-aa, accession # BAR97451), CldD (329-aa, accession # BAR97453), RmnB (549-aa, accession # BAR97461), RmnD (322-aa, accession # BAR97463), Sare_1287 (295-aa, accession # ABV97190), Sare_1288 (531-aa, accession # ABV97191), SCLAV_5671 (311-aa, accession # EFG10738), SCLAV_5672 (529-aa, accession # EFG10739), SCLAV_p0490 (545-aa, accession # EFG03980), SCLAV_p0491 (343-aa, accession # EFG03981), Sros_3191 (295-aa, accession # ACZ86136) and Sros_2192 (560-aa, accession # ACZ86137).

Full size image
Figure 2
figure 2

Alignment of amino acid sequences of (a) type-A diterpene synthases and (b) type-B diterpene synthases. Shadow boxes in the upper alignment indicate metal-binding domains consisting of an acidic amino acid-rich motif (left box) and downstream triad (right box). Shadow boxes in the lower alignment indicate an acidic amino acid-rich motif (left box) and downstream highly conserved motif (right box).

Full size image

A phylogenetic analysis of these aligned sequences of type-A and type-B diterpene synthases revealed that the pairs of Scros_3191/Sros_3192, Sare_1287/Sare_1288, SCLAV_5671/SCLAV_5672, AHK61133/AHK61132 and BAD86798/BAD86797 possessed the catalytic function for the formation of the tricyclic diterpene skeleton including pimara-9(11),15-diene, whereas the pairs of SCLAV_p0491/SCLAV_p0490, RmnD/RmnA, CldD/CldB and BAB39207/BAB39206 are involved in the formation of the bicyclic diterpene skeleton including labda-8(17),12(E),14-triene and terpente-3,13,15-triene. The presumptive diterpene synthases, SCLAV_p0491 and SCLAV_p0490, were classified into the clade of labda-8(17),12(E),14-triene synthases, RmnD and CldD, and the clade of (+)-copalyl diphosphate synthases, RmnB and CldB, respectively. Therefore, the pair of SCLAV_p0491 and SCLAV_p0490 may be involved in the formation of labdane-type diterpenes.

Detection of diterpenes by enzymatic synthesis

At least 17 genes encode type-A mono-, sesqui- and diterpene synthases, including sclav_p0491, in S. clavuligerus ATCC 27064; however, when grown under various culture conditions, the microorganism did not produce any detectable terpenes.7 Therefore, these two genes were also defined as a cryptic state in this microorganism and the function of either gene product, SCLAV_p0490 or SCLAV_p0491, was not expected from cultural metabolites. In order to verify the biochemical functions of SCLAV_p0490 and SCLAV_p0491 and similar diterpene synthases, CldB and CldD, their recombinant enzymes were prepared in E. coli. Two recombinant type-A synthases, CldD and SCLAV_p0491, were obtained as soluble forms, whereas recombinant CldB and SCLAV_p0490 both formed insoluble proteins under the various induction conditions (concentration of isopropyl-β-d-thiogalactopyranoside, growth phase for induction and the incubation temperature after induction). Furthermore, although a recombinant C-terminal His6-tagged protein was prepared, the resultant proteins were still insoluble. These insoluble proteins were dissolved in 8 m urea containing 5 mm DTT at 37 °C for 20 h, and a very small amount of soluble proteins was recovered by dialysis against buffer containing 1 mm DTT followed by centrifugation to remove insoluble proteins. The reaction products of four combinations of pairs of type-A and type-B diterpene synthases (CldD/CldB, CldD/SCLAV_p0490, SCLAV_p0491/CldB and SCLAV_p0491/SCLAV_p0490, respectively) were examined. As shown in Figure 3, the elution time (10.74 min) and MS fragmentation pattern of the enzymatic product (1) generated by the catalytic reaction of CldD and CldB in the presence of magnesium ions, IPP, DMAPP and GGPP synthase (CldA) were identical to those of the authentic sample of labda-8(17),12(E),14-triene (biformene), which was prepared from the mycelial products of S. avermitilis SUKA22 carrying pCLD10413 (cldA-cldB-cldD). The enzymatic reaction product by SCLAV_p0491/SCLAV_p0490 was eluted at 10.42 min, with a molecular ion peak at m/z 272 [M+], which was presumed to be diterpene hydrocarbon, and MS fragmentation patterns different from those of 1. Other two-enzymatic reaction products were eluted at 10.12 min (2) from CldD/SCLAV_p0490 and at 10.91 min (3) from SCLAV_p0491/CldB, and the MS fragmentation patterns of these two products with a molecular ion peak at m/z 272 [M]+ were different from those of 1 and 4. These results indicated that all enzymatic products were diterpene hydrocarbons and these three diterpene hydrocarbons (2, 3 and 4) were not identical to 1. Since the soluble forms of the type-B diterpene synthases, CldB and SCLAV_p0490, were inefficiently obtained from the T7-based E. coli expression system, it was difficult to prepare a large amount of recombinant enzyme for structural analysis of these diterpene hydrocarbons (2, 3 and 4). The preparation of each product was accomplished by the heterologous expression of two genes encoding type-A and type-B diterpene synthases using the Streptomyces expression system.

Figure 3
figure 3

GC-MS analysis of n-pentane extracts of recombinant enzyme reactions. After the enzyme reactions containing type-B and type-A diterpene synthases (A: CldB and CldD; B: CldB and SCLAV_p0491; C: SCLAV_p0490 and CldD; and D: SCLAV_p0490 and SCLAV_p0491) were performed at 30 °C for 20 h, they were stopped by the addition of EDTA, and the products were extracted with n-pentane. A 5-μl portion of the n-pentane extract was subjected to GC-MS. The reaction products of CldB and CldD were identical to labda-8(17),12(E),14-triene.

Full size image

Expression of the pair of type-A and type-B diterpene synthases in the engineered Streptomyces host

We previously developed an efficient expression system for the gene encoding terpene synthase in an engineered Streptomyces host,7, 14, 17, 20 and discovered many novel terpene compounds from bacterial origins.8 Furthermore, we examined the overexpression of the cldD and cldB pair by co-expressing with a gene encoding GGPP synthase, which was controlled by the constitutively expressed promoter in S. avermitilis SUKA22, and a large amount of 1 was obtained from the transformants.13 Each gene with its ribosome-binding sequence was prepared by PCR amplification. All gene pairs, cldD/cldB, cldD/sclav_p0490, sclav_p0491/cldB and sclav_p0491/sclav_p0490, were introduced downstream of the cldA-encoding GGPP synthase in pKU1012,21 the operon of which was controlled by the constitutively expressed promoter of ribosomal protein S10 (PrpsJ),21 in order to obtain pLDT101, pLDT102, pLDT103 and pLD104, respectively. As shown in Figure 4, all transformants produced a single component of diterpene hydrocarbon in the mycelium and each peak was identical to that of the enzymatic reaction described in Figure 3. The productivity of 1 by S. avermitilis SUKA22 transformants carrying pLDT101 was similar to that of transformants carrying pCLD104, as previously reported.13 The S. avermitilis SUKA22 transformant carrying pLDT101, pLDT102, pLDT103 or pLDT104 was grown in a 2-l culture and the structure of each diterpene hydrocarbon was determined by detailed 1H and 13C NMR analyses including DEPT, 1H-1H COSY, 1H-1H NOESY, 1H-13C HMDC and 1H-13C HMBC, as measured in CDCl3.

Figure 4
figure 4

GC-MS analysis of n-hexane extracts of S. avermitilis SUKA22 transformants carrying (A) pLDT101 (cldA::cldB::cldD), (B) pLDT102 (cldA::cldB::sclav_p0491), (C) pLDT103 (cldA::sclav_p0490::cldD) and (D) pLDT104 (cldA::sclav_p490::sclav_p0491). Peaks eluted from 9.0 to 9.7 min were identical to middle chain fatty acids.

Full size image

The diterpene hydrocarbon 1 was obtained from S. avermitilis SUKA22 carrying pLDT101 and the structure of 1 was identical to labda-8(17),12(E),14-triene (biformene) based on comparisons with the 1H and 13C NMR spectra of previous studies13 and published data.22 Two diterpene hydrocarbons generated by the expression of the heterogeneous combination of genes encoding type-A and type-B diterpene synthases, 2 and 3, were obtained from S. avermitilis SUKA22 carrying pLDT102 and pLDT103, respectively. In consideration of NMR spectral data (Supplementary Figures S1–S5), the structure of 2 was identical to the known diterpene labda-8(17),13(16),14-triene (sclarene; from the foliage of Podocarpus cunninghamii).23 Although the molecular formula of 3 was identical to those of 1 and 2, the MS fragmentation pattern of 3 was different (Figure 3). The structure of 3 was identical to the known diterpene labda-7,12(E),14-triene by comparisons with published data on plant metabolites (from the stem bark of Croton oblongiflolius)24 and spectra of 3 (Supplementary Figures S6–S10).

A diterpene hydrocarbon that possesses the characteristics of 4 has not yet been described in the literature, indicating that 4 is a novel diterpene. The MS fragmentation pattern of 4 was similar to that of 3 (Figure 3) and the molecular formula was estimated to be C20H32 for a diterpene hydrocarbon with five degrees of unsaturation. The 1H NMR spectrum of 4 showed two pairs of two olefinic protons, H-15 (δH 5.23, d, J=10.6 Hz, 1H and 5.05, d, J=7.4 Hz, 1H) and H-16 (δH 4.99, d, J=10.6 Hz, 2H), assignable to exomethylenes. An additional olefinic proton (δH 5.36, brs, 1H) was observed. An analysis of the 13C NMR and DEPT spectra of 4 showed 20 resolved signals and confirmed the presence of four methyls, six sp3 methylenes, two sp2 methylenes, two sp3 methines, two sp2 methines, and two sp2 and two sp3 quaternary carbons (Table 1). A comparison between the 1 and 4 data of 1H and 13C NMR spectra (Supplementary Figure S11) indicated that the following four different points were observed: the first was that the C-16 of 4 was assigned as an exomethylene (sp2 methylene), whereas that of 1 was a methyl carbon; the second was that the C-12 of 4 was assigned as a sp3 methylene carbon, whereas that of 1 was an sp2 methine carbon; the third was that the C-17 of 4 was assigned as a methyl carbon, whereas that of 1 was an exomethylene carbon; and the fourth was that C-7 was assigned as an sp2 methine, whereas that of 1 was an sp2 methylene carbon. Three double bonds accounted for the three degrees of unsaturation, indicating that 4 possessed a bicyclic skeleton. The connectivity of protons and carbons was assigned by an analysis of the 1H-13C HMQC spectrum (Supplementary Figure S13), as shown in Figure 5. An analysis of 1H-1H COSY (Supplementary Figure S12) revealed four partial structures: C-1–C-3, C-5–C-7, C-9–C-11 and C-14–C-15 (Figure 5), and the 1H-13C long-range coupling of 2 J and 3 J observed in the 1H-13C HMBC experiments (Supplementary Figure S14) provided the following information: the cross peaks from H-14 (δH 6.36) to C-15 (δC 113.3) from H-15 (δH 5.23) to C-13 (δC 147.1) and C-14 (δC 139.0), from H-16 (δH 4.99) to C-12 (δC 54.9), C-13 and C-14, from H-17 (δH 1.72) to C-7 (δC 122.4), C-8 (δC 135.4) and C-9 (δC 54.9), H-18 (δH 0.85) to C-3 (δC 42.4), C-4 (δC 33.3), C-5 (δC 50.3) and C-19 (δC 21.9), from H-19 (δH 0.86) to C-3, C-4, C-5 and C-18 (δC 33.1), and from H-20 (δH 0.73) to C-1 (δC 39.2), C-5, C-9 and C-10 (δC 36.9) supported the structure shown in Figure 5. The stereochemical assignment of two methyl protons (H3-19 and H3-20) was deduced from 1H-1H NOESY data (Supplementary Figure S15), whereas the decaline skeleton from C-1–C-10 was assigned a chair–chair configuration based on the observed correlation between H-5 and H-1, H-3 and H-9. Thus, the structure of the novel diterpene hydrocarbon (4) was elucidated as labda-7,13(16),14-triene.

Table 1 1H and 13C NMR data of 4
Full size table
Figure 5
figure 5

Structure of labda-7,13(16),14-triene (4). The left panel shows 1H-1H NOESY (dashed arrows) data for each compound. The middle panel shows 1H-1H COSY (bold lines) and 1H-13C HMBC (arrows) and double arrows indicate a cross peak in 1H-13C HMBC.

Full size image

Discussion

Terpenoid metabolites are a relatively minor fraction in bacterial natural products. Since diterpenoid metabolites are extremely rare in bacterial metabolites, studies on the isolation of diterpenoid metabolites, diterpene synthases and their genes are limited. On the other hand, diterpenoid metabolites and diterpene synthases, which have been cloned predominantly from higher plants and fungi, have been extensively examined.9 Many bi- and tricyclic diterpenes of plants have been generated by the cyclization of GGPP under type-B and type-A diterpene synthases. In most of the type-A and type-B diterpene synthases of plant and fungal origins, metal-binding domains consisting of the acidic amino acid-rich motif sequences, -DDxxD- and -DxDD-, have been observed in type-A and type-B diterpene synthases, respectively. Furthermore, the highly conserved sequence, -SAYDTAW- (the function of which remains unknown) located within 80–150 aa of the N-terminus, was found in all diterpene synthases of plants, whereas the corresponding motif was not observed in any bacterial diterpene synthases. The type-A terpene synthases of bacterial, fungal and plant origins possess a pair of characteristically conserved metal-binding domains consisting of the acidic amino acid-rich motif -DDxxx[DE]- or -DDxxxE-, located within 60–120 aa (bacterial and fungal terpene synthases) or 230–270 aa (plant terpene synthases) of the N-terminus, and a three-amino acid triad -Nxxx[ST]xxxE-, found 140 aa downstream, closer to the C-terminus. Since the two synthases of Sros_3191 and Sare_1287 displayed the incomplete motif, -DxxxD-, these proteins may not catalyze ionization by the removal of a diphosphate residue. Two type-A diterpene synthases, AHK61133 and BAD86798, which were classified into group 1 in the phylogenetic analysis in Figure 2, have been characterized as pimara-9(11),15-diene synthases and the acidic amino acid-rich motif -DDxxE- was present in both synthases. The same motif sequence was also observed in SCLAV_5671. On the other hand, the labda-8(17),12(E),14-triene synthases, CldD and RmnD, possessed the slightly longer motif, -DDxxxE-, which was also observed in SCLAV_p0491. Furthermore, another characteristically conserved metal-binding domain, the triad, was present in all bacterial type-A diterpene synthases, and was classified into two groups. The downstream triad -NxxxSxxxE- in the C-terminus region was present in four synthases: the labda-8(17),12(E),14-triene synthases, CldD and RmnD, and terpenete-3,13,15-triene synthases, BAB39207 and SCLAV_p0491. Another downstream triad motif -NxxxTxxxE- was in the pimara-9(11),15-diene synthases, AHK61133 and BAD86798, and SCLAV_5671. Two conserved motifs of type-B synthases, an acidic amino acid-rich motif -DxDDT[AS]- and downstream motif -QxxDGSW-, were present and conserved in all type-B synthases. Pimara-9(11),15-diene synthases displayed the acidic amino acid-rich motif -DxDDTS-, whereas the -DxDDTA- motif was observed in the two labda-8(17),12(E),14-triene synthases, CldB and RmnB, and SCLAV_p0490. Another downstream motif, -QxxDGSW-, was present and conserved in the pimara-9(11),15-diene synthases and labda-8(17),12(E),14-triene synthases. Based on these alignment and phylogenetic analyses of type-A and type-B diterpene synthases, a pair of presumptive diterpene synthases SCLAV_p0491 and SCLAV_p0490 may be involved in the generation of labdane-type bicyclic diterpenes.

The structural elucidation of diterpene products generated by the heterologous expression of genes encoding presumptive labdane-type diterpene synthases provided the biochemical functions of gene products. Since SCLAV_p0490/SCLAV_p0491 appeared to be involved in the formation of labdane-type diterpenes by a phylogenetic analysis of aligned sequences, the function of each gene product was examined. Protonation of the C-14,15-double bond of GGPP by type-B diterpene synthases, followed by the attack of C-10 on C-15, then C-7 on C-11 (Figure 6), gave four possible products ((+)-copalyl, ent-copalyl, syn-copalyl and syn-ent-copalyl diphosphates were generated from the chair–chair—'normal', chair–chair—'antipodal', chair–boat—'normal' and chair–boat—'antipodal' conformation of GGPPs, respectively), depending on the conformation of the prochiral substrate. The normal chair–chair conformation of GGPP gave the C8-carbonium ion of (+)-copalyl diphosphate with the normal anti,anti absolute stereochemistry in the generation of bicyclic labdane-type diterpenes. However, the antipodal chair–chair conformation of GGPP gave the C8-carbonium ion of ent-copalyl diphosphate with the enantiomeric anti,anti absolute stereochemistry in the generation of the tricyclic diterpenes, pimaradiene and ent-kaurene. Thus, CldB and RmnB, which are involved in the formation of the bicyclic labdane-type diterpene, labda-8(17),12(E),14-triene, have been defined as (+)-copalyl diphosphate synthases, while AHK61132 and BAD86797, which are involved in the formation of the tricyclic diterpene, pimara-9(11),15-diene, are ent-copalyl diphosphate synthases. Four different bicyclic labdane-type diterpene hydrocarbons were produced by the heterologous expression of four pairs of genes encoding type-A and type-B diterpene synthases: cldB/cldD, cldB/sclav_p0491, sclav_p0490/cldD and sclav_p0490/sclav_p0491, respectively. Transformants carrying the pair of genes cldB/cldD or cldB/sclav_p0491 produced bicyclic labdane-type diterpene hydrocarbons consisting of the 8(17)-methylenedecaline skeleton, while bicyclic diterpene hydrocarbons with the 7,8-dehydrodecaline skeleton were produced by transformants carrying sclav_p0490/cldD or sclav_p0490/sclav_p0491 (Figure 5). These results indicated that CldB catalyzed the generation of (+)-copalyl diphosphate from GGPP through protonation at the C-14,15-double bond of GGPP, in which the C8-carbonium ion of (+)-copalyl diphosphate was generated, and (+)-copalyl diphosphate was then formed by the release of a proton at H-17 in the C8-carbonium ion (Figure 6). On the other hand, SCLAV_p0490 catalyzed the generation of labda-8(17),13-dienyl diphosphate by a similar reaction, in which the diterpene diphosphate, labda-7,13-dienyl diphosphate, was generated by the release of a proton at H-7 in the C8-carbonium ion of (+)-copalyl diphosphate (Figure 6). The assignment of the actual biochemical functions of these type-B diterpene synthases, CldB and SCLAV_p0490, requires direct experimental verification.

Figure 6
figure 6

Cyclization mechanisms underlying the generation of labdane-type bicyclic diterpene hydrocarbons 1, 2, 3 and 4 by combinations of type-B and type-A diterpene synthases (CldB/CldD, CldB/SCLAV_p0490, SCLAV_p0490/CldD and SCLAV_p0490/SCLAV_p0491).

Full size image

Type-A diterpene synthases catalyze the generation of diterpene hydrocarbons or alcohols from copalyl diphosphate. This reaction is initiated by the enzyme-catalyzed ionization of copalyl diphosphate to form an allylic cation and the resultant cation is converted to a diterpene hydrocarbon by deprotonation. Transformants carrying the pair of genes cldB/cldD or sclav_p0490/cldD produced bicyclic diterpene hydrocarbons possessing two olefins at C-12 and C-14 in the side chain of the labdane skeleton, while bicyclic diterpene hydrocarbons possessing two olefins at C-13(16) and C-14 in the side chain of the labdane skeleton were produced by transformants carrying cldB/sclav_p0491 or sclav_p0490/sclav_p0491 (Figure 5). The ionization of (+)-copalyl diphosphate (generated by CldB) or labda-7,13-dienyl diphosphate (generated by SCLAV_p0490) by CldD was initiated by the removal of diphosphate, and the resultant cation was then quenched by the deprotonation at H-12, whereas the cation was quenched by deprotonation at H-16 in the reaction of SCLAV_p0491 (Figure 6).