Diterpene synthases from Leonurus japonicus elucidate epoxy-bridge formation of spiro-labdane diterpenoids

Abstract

Spiro-9,13-epoxy-labdane diterpenoids are commonly found in Leonurus species, particularly in Leonurus japonicus Houtt., which is a medicinal herb of long-standing use in Asia and in which such spiro-heterocycles are present in at least 38 diterpenoids. Here, through generation of a transcriptome and functional characterization of six diterpene synthases (diTPSs) from L. japonicus, including three class II diTPSs (LjTPS1, LjTPS3, and LjTPS4) and three class I diTPSs (LjTPS5, LjTPS6, and LjTPS7), formation of the spiro-9,13-epoxy-labdane backbone was elucidated, along with identification of the relevant diTPSs for production of other labdane-related diterpenes. Similar to what has been found with diTPSs from other plant species, while LjTPS3 specifically produces the carbon-9 (C9) hydroxylated bicycle peregrinol diphosphate (PPP), the subsequently acting LjTPS6 yields a mixture of four products, largely labda-13(16),14-dien-9-ol, but with substantial amounts of viteagnusin D and the C13-S/R epimers of 9,13-epoxy-labda-14-ene. Notably, structure–function analysis identified a critical residue in LjTPS6 (I420) in which single site mutations enable specific production of the 13S epimer. Indeed, extensive mutagenesis demonstrated that LjTPS6:I420G reacts with PPP to both specifically and efficiently produce 9,13S-epoxy-labda-14-ene, providing a specialized synthase for further investigation of derived diterpenoid biosynthesis. The results reported here provide a strong foundation for future studies of the intriguing spiro-9,13-epoxy-labdane diterpenoid metabolism found in L. japonicus.

Elucidated epoxy-bridge formation of spiro-9,13-epoxy-labdane diterpenoids through functional characterization of diterpene synthases from L. japonicas.

Introduction

Over 18,000 diterpenoids compounds have been identified (Hu et al., 2021), many with important biological activities, and these have emerged as innovative lead compounds for development of modern drugs. An intriguing group of diterpenoids are labdanes containing an oxy group at C9, with those containing a spiro-9,13-epoxy ring providing a particularly distinct structural feature (Zhang et al., 2018; Miao et al., 2019). A considerable number of these diterpenoids have strong biological activity, such as the anticancer activity of LS-1/2 (Satoh et al., 2003), analgesic and anti-diabetic activity of marrubiin (Popoola et al., 2013), and anti-platelet aggregation and coagulant activity exhibited by prehispanolone and related spiro-epoxy-labdane diterpenoids (Lee et al., 1991; Moon, 2010; Miao et al., 2019). This type of diterpenoid is prevalent in the genera Leonorus, Otostegia, Ballota, and Marrubium from the Lamiaceae plant family (Meyre-Silva and Cechinel-Filho, 2010; Zhang et al., 2018; RosselLi et al., 2019). Spiro-epoxy-labdane diterpenoids are particularly common in the genus Leonorus, with about 70 spiro-epoxy-labdane diterpenoids having been identified from these plant species (Zhang et al., 2018).

The C9-oxy group of these labdane diterpenoids seems to generally stem from addition of water during initial (bi)cyclization of the general diterpenoids precursor (E,E,E)-geranylgeranyl diphosphate (GGPP) catalyzed by a class II diterpene cyclase. These enzymes utilize a conserved DxDD motif to catalyze a protonation-initiated carbocation cascade reaction with GGPP that forms a decalin bicycle, initially labda-13-en-8-yl⁺ diphosphate (Peters, 2010). Deprotonation of this carbocationic intermediate at the neighboring methyl generates labda-8(17),13-dien-15-yl diphosphate, which has been termed copalyl diphosphate (CPP), with enzymes producing this termed CPP synthases (CPSs). CPP can be produced in various stereoisomeric forms, with that defined as the enantiomeric form, ent-CPP (1), produced by all vascular plants as a necessary precursor for gibberellin (GA) phytohormone biosynthesis (Zi et al., 2014). However, particularly relevant here, this carbocationic intermediate also can undergo a 1,2-hydride shift from C9 to C8, with addition of water to the ensuing carbocation at C9 prior to concluding deprotonation, to yield 9α-hydroxy-labda-13-en-15-yl diphosphate, which has been termed peregrinol diphosphate (PPP, 3) (Zerbe et al., 2014; Heskes et al., 2018; Johnson et al., 2019; Karunanithi and Zerbe, 2019).

Class II diterpene cyclase products such as ent-CPP and PPP are generally further reacted upon by more prototypical class I terpene synthases (TPSs) (Peters, 2010). Class I TPSs are characterized by DDxxD and DTE/NSE motifs that chelate trio of divalent metal ion co-factors that assist lysis of the allylic diphosphate ester to initiate another carbocationic cascade leading to cyclization and/or rearrangement, and that also can include the addition of water prior to concluding deprotonation (Christianson, 2017). Those that act on class II diterpene cyclase products almost invariably arise from the TPS-e subfamily, which is anchored by the requisite member for production of ent-kaurene (2) from ent-CPP for GA biosynthesis, then termed kaurene synthases (KSs) (Zi et al., 2014). Again of relevance here, it is notable that a number of class I diterpene synthases (diTPSs) that catalyze further cyclization of CPP, involving addition of the 8(17)-ene to the initially formed 13-yl tertiary allylic carbocation (such as KSs), have been shown to efficiently (hetero)cyclize the stereochemically relevant C8 hydroxylated variant (i.e. 8-hydroxy-labda-13-ene-15-yl diphosphate), forming the 8,13-epoxy derivative manoyl oxide with high stereoselectivity (Mafu et al., 2015). With regards to nomenclature, due to an ancestral fusion event, class II diterpene cyclases are considered members of the TPS family, typically falling within the TPC-c subfamily in angiosperms (Chen et al., 2011). Thus, these and the subsequently acting class I TPSs are both termed diTPSs.

Arguably somewhat surprisingly, while previous studies found that PPP is produced quite specifically by the relevant class II diTPSs, subsequent production of 9,13-epoxy-labda-14-ene (4) is catalyzed much less selectively by the relevant class I diTPSs. In particular, all those identified to date produce not only a mixture of the two C13 epimers of this spiro-epoxy, but generally also the exo-ene and/or tertiary alcohol derivatives stemming from deprotonation, either immediately of the neighboring methyl or following the addition of water to the initially generated 13-yl tertiary allylic carbocation, generating labda-13(16),14-dien-9α-ol (5) or viteagnusin D (6), respectively (Figure 1A; Zerbe et al., 2014; Heskes et al., 2018; Johnson et al., 2019). Examples of such enzymes are found throughout the Lamiaceae plant family, as originally reported from Marrubium vulgare, with MvCPS1 producing PPP (3), which 9,13-epoxy-labda-14-ene synthase (MvELS) further converts to a mixture, largely 9,13S-epoxy-labda-14-ene (4b), but with significant amounts of the 13R epimer (4a) as well as 5 (Zerbe et al., 2014). Similar catalytic promiscuity in product outcome is observed with class I diTPSs from Ajuga reptans (ArTPS3), Leonotis leonurus (LlTPS4), Perovskia atriplicifolia (PaTPS3), Salvia officinalis (SoTPS1), and Origanum majorana (OmTPS5), while another from this species (OmTPS4) exhibits additional promiscuity and yields all four products, and yet another (OmTPS3) produces largely 4a along with significant amounts of 4b, with one from Prunella vulgaris (PvTPS1) producing an analogous epimeric mix along with substantial amounts of 5 (Johnson et al., 2019). Class I TPSs from Vitex agnus-castus also react with PPP, with one (VacTPS2) predominantly producing 4a along with significant amounts of 6, while another (VacTPS6) produces largely 5 along with an equimolar mixture of 4a and 4b (Heskes et al., 2018). There are class I TPSs that react with 3 much more selectively to produce either 5, such as one from Mentha spicata (MsTPS1) and another from the bacterium Kitasatospora griseola (KgTS), or 6, such as the sclareol synthase from Salvia sclarea (SsSS) (Johnson et al., 2019; Jia et al., 2016). L. japonicus is a perennial medicinal shrub of the mint family (Lamiaceae) that has been utilized in Asia for over 1,800 years and has a wide range of pharmacological activities (Miao et al., 2019; Zhang et al., 2018; Shang et al., 2014). More than 60 C9-oxy containing labdane diterpenoids have been isolated from this plant (Miao et al., 2019; Li et al., 2020), and 38 exhibit a spiro-9,13-epoxy ring, including examples derived from both 13R- and 13S-epimers (Miao et al., 2019; Shang et al., 2014; Li et al., 2019). Given the prevalent nature of these diterpenoids in L. japonicus, it was expected that investigation of this species would help further elucidate such spiro-9,13-epoxy-labdane diterpenoid biosynthesis. In this study, a transcriptome was generated for L. japonicus, from which seven diTPSs, four class II and three class I diTPSs, were identified. These LjTPSs were biochemically characterized, and structure–function investigation carried out with the 9,13-epoxy-labda-14-ene producing LjTPS6, leading to identification of a critical residue. Notably, this enabled identification of LjTPS6:I420G as providing catalytic efficient and selective production of 9,13S-epoxy-labda-14-ene (4b). Altogether this report provides a strong foundation for future investigation of the biosynthesis of derived diterpenoids in L. japonicus.

Figure 1.

The diTPSs in the diterpenoid biosynthetic pathway, along with their possible downstream natural products in L. japonicus. A and B, Catalytic function of L. japonicus diTPSs. diTPSs in boxes were characterized in this study. A, Catalytic function of TPS-c and TPS-e subfamily LjTPSs. B, Catalytic function of TPS-f subfamily LjTPSs. C, Diterpene scaffolds found in L. japonicus. Geranylgeranyl diphosphate (GGPP), farnesyl diphosphate (FPP), geranyl diphosphate (GPP), copalyl diphosphate (CPP) and peregrinol diphosphate (PPP).

Results

Accumulation of diterpenoids in L. japonicus

More than 60 C9-oxy containing labdane diterpenoids (e.g. 9,13-epoxy-labdane diterpenoids) have been reported in L. japonicus (Miao et al., 2019; Shang et al., 2014; Li et al., 2019). Given that only 11 authentic standards were available, in order to analyze tissue-specific accumulation of diterpenoids in L. japonicus a m/z mass fragmentation library (DLJ library) was constructed from the results of previous studies (Miao et al., 2019; Shang et al., 2014; Li et al., 2019; Yang et al., 2020; Peng et al., 2017). The diterpenoid profile of the root, stem and leaf of L. japonicus were analyzed using ultra performance liquid chromatography coupled with quadrupole time of flight mass spectrometry (UPLC-Q/TOF-MS), with comparison to the available authentic standards and searches of the DLJ library assisted by the UNIFI information system. In particular, the [M + H]⁺, [M + Na]⁺, and [M + K]⁺ adducts were observed and analyzed (Supplemental Table S1). While only galeopsin could be identified by comparison to an authentic standard, 10 other diterpenoids were identified by UNIFI software matching to the DLJ library. All of these identified diterpenoids were mainly accumulated in leaves, and five (including galeopsin) were C9-oxy containing labdane diterpenoids. This relatively high accumulation of C9-oxy diterpenoids in leaves was consistent with the leaf-specific expression pattern of the relevant class II diterpene synthase LjTPS3 (see below).

Identification of multiple diTPSs in an L. japonicus transcriptome

To enable investigation of diterpenoid biosynthesis in L. japonicus, RNA from root, leaf, and stem, from three independent individuals, was extracted and utilized to generate a transcriptome for this medicinal herb. Briefly, a nonnormalized complementary DNA (cDNA) library was prepared from separated and mixed RNA from different tissues of L. japonicus, and sequenced using the Illumina novaseq6000 platforms. A total of 279,710,836 reads were obtained. The Trinity de novo assembler found 133,725 contigs with an average length of 1,119 bp, and lengths varying from 201 to 15,425 bp (Accession number: RPJCA, CRA004698).

Homology searches and annotation information of this L. japonicus transcriptome revealed seven diTPSs (designated LjTPS1 to LjTPS7) (Supplemental Table S2), of which six (excluding LjTPS2) had full-length Open Reading Frames (ORFs). To understand the physiological roles of the candidate diTPS in L. japonicus, the transcript levels in various organs, including root, stem, and leaf were show by the Reads Per Kilobase Million (FPKM) value (Figure 2A). This reveals that all candidates are expressed in all tissues, with the exception of LjTPS3. Three LjTPSs (e.g. LjTPS4, LjTPS5, and LjTPS6) had higher transcript levels in root, and two (e.g. LjTPS1 and LjTPS7) in stem, with only LjTPS3 specifically expressed in leaves. To verify these expression profiles, LjTPS transcripts in leaf, stem, and root were quantified using reverse transcription quantitative PCR (RT-qPCR). The expression profiles of LjTPSs by RT-qPCR are similar to that from the transcriptome, with the exception of LjTPS4, which was found to be more highly expressed in leaf (Figure 2A).

Figure 2.

Phylogenetic analysis of diTPSs from L. japonicus and relative expression of diTPSs in different tissues. A, Relative expression level of L. japonicus diTPSs. Relative expression level of L. japonicus diTPSs in leaf, stem, and root by RT-qPCR and the reads per kilobase million value from the L. japonicus transcriptome. The expression level was normalized to that of actin, and the error bars show the sds from mean value (n = 3). Statistically significant differences between the transcript levels of leaf and stem or root based on a Student’s t-test (P < 0.05). B, Phylogeny of L. japonicus diTPSs. Constructed using the NJ algorithm with a representative set of TPSs (Supplemental Table S3). The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree (Nei and Kumar, 2000). Bootstrap values (1,000 replicates, shown if >70%) are given. PpCPS/KS was used as outgroup. Red-marked enzymes show diTPSs from L. japonicus in this study.

Six LjTPSs were used for phylogenetic analysis with a representative set of TPSs. Three of these putative diTPSs contained the DDxD motif (LjTPS1, LjTPS3, and LjTPS4) and were clustered in the TPS-c subfamily with the class II diTPS of other plants (Figure 2B). The other three putative diTPSs (LjTPS5, LjTPS6, and LjTPS7) were tentatively identified as Class I diTPSs based on the conserved DDxxD and NSE/DTE motifs, and fell within the TPS-e/f subfamily (Figure 2B;  Supplemental Figure S1). LjTPS1 and LjTPS5 clustered with the known ent-CPSs (e.g. VacTPS5 and IrCPS5) and ent-EKSs (e.g. MvEKS and VacTPS4), suggesting roles in GA biosynthesis (Zerbe et al., 2014; Jin et al., 2017; Heskes et al., 2018). LjTPS3 was grouped with the three known PPP synthases (e.g. LlTPS1, MvCPS1, and VacTPS1) (Meyre-Silva and Cechinel-Filho, 2010; Zerbe et al., 2014; Heskes et al., 2018; Johnson et al., 2019; ). LjTPS6 was clustered with multiple diTPS that can react with PPP (Johnson et al., 2019). This suggested that LjTPS3 and LjTPS6 are involved in the synthesis of C9 oxidized diterpenoids in L. japonicus. LjTPS4 was most closely related to CPSs that produce CPP of normal stereochemistry (e.g. MvCPS3, IrCPS1, and CfTPS1), which also is widespread in the Lamiaceae plant family (Meyre-Silva and Cechinel-Filho, 2010; Pateraki et al., 2014; Jin et al., 2017). Finally, LjTPS7 falls within the TPS-f subfamily, members of which often play a role in homoterpene biosynthesis (Herde et al., 2008; Figure 2B).

Functional characterization of putative class II LjTPSs

For functional characterization of the three class II diTPSs (LjTPS1, LjTPS3, and LjTPS4), these were heterologously expressed in Escherichia coli with a C-terminal 3His epitope tag. Crude recombinant proteins were assayed in vitro with GGPP, and subsequent dephosphorylation carried out to enable extraction and analysis of the reaction products by GC–MS. When compared to the products of known CPSs (Fleet et al., 2003; Gao et al., 2009), LjTPS1 and LjTPS4 were found to produce CPP (Figure 3). To determine the absolute stereochemistry of their products LjTPS1 and LjTPS4 were incorporated into coupled assays. In particular, with class I diTPSs specific for either ent- or normal CPP, either the KS from Arabidopsis (Arabidopsis thaliana) (AtKS) (Fleet et al., 2003) or miltiradiene synthase from Salvia miltiorrhiza (Gao et al., 2009), respectively. These assays were compared with those utilizing CPSs of known stereochemistry. In this manner, LjTPS1 was shown to produce ent-CPP (1) (Figure 3a), while LjTPS4 produced normal CPP (7) (Figure 3C). LjTPS3 was shown to produce PPP (3), as evidenced by the analogous mixture of products as observed with MvCPS1—that is, upon dephosphorylation to peregrinol (3′), there is spontaneous formation of a small amount of a mixture of the C13 epimers of 9,13-epoxy-labda-14-ene (4a and 4b). The latter two products were identified by comparison to coupled assays with a class I diTPS that produces 5 (KgTS) and 6 (SsSS), respectively (Jia et al., 2016), with equivalent results from assays with either MvCPS1 or LjTPS3 (Figure 1A). Thus, LjTPS3 is a PPP synthase.

Figure 3.

GC–MS analysis of L. japonicus Class II diterpene cyclases with comparison to known diTPSs from in vitro assays. A–C, Extracted ion chromatograms (EICs) of the dephosphorylated reaction products obtained by Class II LjTPSs and known cyclases, or the products obtained by coupling with class I diTPSs with known substrate stereospecificity. A, Dephosphorylated products from LjTPS1, A. thaliana (At) copalyl diphosphate synthase (AtCPS), or either coupled with At KS (AtKS). B, Dephosphorylated reaction products obtained by LjTPS3, Marrubium vulgare (Mv) copalyl diphosphate synthase (MvCPS1), or either coupled with Kitasatospora griseola (Kg) terpene synthase (KgTS) or Salvia sclarea (Ss) sclareol synthase (SsSS). C, Dephosphorylated reaction products obtained by LjTPS4, Salvia miltiorrhiza (Sm) copalyl diphosphate synthase (SmCPS), or either coupled with Sm miltiradiene synthase.

Functional characterization of putative class I diTPSs

For the functional characterization of the three class I diTPSs (LjTPS5, LjTPS6, and LjTPS7), these also were heterologously expressed in E. coli with a C-terminal 3His epitope tag, and the crude recombinant proteins used for in vitro assays, in the case of LjTPS5 and LjTPS6 these were coupled with various class II diTPSs. By comparison to the known activity of the ent-CPP producing AtCPS and AtKS, LjTPS5 was shown to produce ent-kaurene (2) when coupled with either AtCPS or LjTPS1 (Figure 4A). Thus, LjTPS5 was identified as a KS, presumably involved in GA biosynthesis in L. japonicus. On the other hand, LjTPS6 was found to react with PPP (3), as produced by either MvCPS or LjTPS3, and produce largely labda-13(16),14-dien-9α-ol (5), along with smaller amounts of viteagnusin D (6) and a mixture of the C13 epimers of 9,13-epoxy-labda-14-ene (4), with somewhat more of 13S (4b) than 13R (4a) (Figures 1, A and 4, B). Accordingly, LjTPS6 appears to be a catalytically promiscuous labda-13(16),14-diene-9-ol synthase that may enable production of C9-oxy containing labdanes in L. japonicus more generally. In contrast, LjTPS7 was found to react directly with GGPP, producing (E,E)-geranyllinalool (Figure 4C).

Figure 4.

GC–MS analysis of L. japonicus class I terpene synthase products from (coupled) in vitro assays. A–C, EICs of the products obtained by class I LjTPSs and identified synthases coupled with class II TPSs in vitro. A, LjTPS5 or AtKS coupled with LjTPS1 (producing ent-CPP). B, LjTPS6, KgTS, SsSS, and MvELS coupled with LjTPS3 (producing PPP). C, LjTPS7 with GGPP as substrate.

Substrate specificity of class I LjTPSs

Many diTPSs exhibit substrate promiscuity, particularly class I diTPSs that react with a variety of class II diTPS products to yield distinct diterpenoid skeletons as described by previous studies (Pateraki et al., 2014; Mafu et al., 2015; Jia et al., 2016). To investigate the substrate promiscuity of class I LjTPSs in L. japonicus, these were each co-expressed with the previously described set of 12 catalytically distinct class II diTPSs in a modular metabolic engineering system (Jia et al., 2016). Both LjTPS5 and LjTPS6 displayed catalytic activity when combined with two other class II diTPSs. In particular, LjTPS5 could functionally couple with MvCPS1 (producing PPP) and AtCPS:H263A (producing 8β-hydroxy-ent-CPP) to produce viteagnusin D (6) and 13R-ent-manoyl oxide (10) (respectively), as identified by comparison to previously identified diTPS products (Figures 1, A, 5, A, and B). LjTPS6 was active in combination with AgAS:D621A (producing normal CPP) and NgCLS (producing 8α-hydroxy-CPP). LjTPS6 coverts normal CPP largely to sandaracopimaradiene (11), and two by-products, the oxygenated diterpene nezukol (12) (Pelot et al., 2017) and an unidentified product (13), again with product identification by comparison to known diTPS products (Figure 5C). Similarly, LjTPS6 was found to convert 8α-hydroxy-CPP to 13R-manoyl oxide (14) (Figures 1, A and 5, D).

Figure 5.

Diterpenes produced by class I LjTPSs with alternative substrates in vivo or in vitro. A–D, The EIC of the product(s) obtained with LjTPS5 or LjTPS6 coupled with other class II diTPS in a previously described metabolic engineering system (Jia et al., 2016), termed here in vivo. A, LjTPS5 or SsSS coupled with MvCPS1 (producing PPP) in vivo.B, LjTPS5 or TwTPS16 coupled with AtCPS:H263A (producing 8β-hydroxy-ent-CPP) in vivo. C, LjTPS6 or CfTPS3 coupled with NgCLS (producing 8α-hydroxy-CPP) in vivo. D, LjTPS6 or Isodon rubescens TPS2 (IrTPS2), Mentha spicata TPS1 (MsTPS1) coupled with AgAS:D621 (producing (+)-CPP) in vivo. E and F, The EICs of the product obtained by LjTPS7 with FPP or GPP as substrate in vitro. E, LjTPS7 with FPP as substrate in vitro. F, LjTPS7 with GPP as substrate in vitro.

No product was found upon co-expression of LjTPS7 with this set of class II diTPSs. However, previous studies reported that most geranyllinalool synthases can react with isoprenyl diphosphate precursors of other chain lengths (e.g. (E,E)-farnesyl diphosphate [FPP, C15] and (E)-geranyl diphosphate [GPP, C10]) (Falara et al., 2014; Su et al., 2017). Thus, LjTPS7 assayed in vitro with either FPP or GPP as substrates. The results revealed that LjTPS7 could also react with FPP to produce nerolidol (15), and GPP to produce linalool (16) (Figures 1, B, 5, E and F). Thus, LjTPS7 is capable of producing tertiary alcohols of varied isoprenoid chain length in L. japonicus.

Single residue mutation switches the catalytic specificity of LjTPS6

LjTPS6 produces a mixture of products from PPP (3), much as reported for related class I diTPSs—for example, LlTPS4, OmTPS3, OmTPS5, and MvELS (Johnson et al., 2019; Zerbe et al., 2014). Notably, it has been previously shown that single residue changes can dramatically alter product outcome in diTPSs (Wilderman and Peters, 2007; Xu et al., 2007; Morrone et al., 2008; Mann et al., 2010; Jia et al., 2019). Given the interest in spiro-9,13-epoxy-labdane diterpenoids, it was hypothesized that single residue mutations in LjTPS6 may impart more specificity for such heterocyclization. Among such diTPSs OmTPS3 was the most specific, albeit still producing both C13 epimers (Johnson et al., 2019). Thus, LjTPS6 was aligned with these related diTPSs, and its protein structure modeled, with the hallmark DDxxD and DTE/NSE motifs used to define the active site cavity into which PPP was docked. This led to identification of two candidate sites for mutagenesis in LjTPS6, T315, and I420, for which the residues found in the related diTPSs were substituted (Figure 6, a and b). These mutants were characterizing by co-expression with MvCPS1 in the modular metabolic engineering system. Notably, mutation of I420 but not T315 can substantially increase the specificity of LjTPS6 product outcome with PPP as substrate. In particular, substitution of either Asn or Val for I420 led to production of only the 13S epimer (4b) of 9,13-epoxy-labda-14-ene, with the 4a, 5, or 6 also produced by the wild-type (WT) enzyme no longer observed (Figure 6c).

Figure 6.

Single residue mutations affect the products of LjTPS6 with PPP. A, Homology modeling and docking with PPP as ligand. B, Multiple sequence alignments of LjTPS6 with related diTPSs. C, EIC of the products obtained by LjTPS6:I420V/N coupled with MvCPS1 (producing PPP) in vivo. D, Relative peak area of diterpenes produced by LjTPS6 mutants with PPP as the substrates, and the error bars show the sds from mean value (n = 3). Statistically significant differences between the product abundance of LjTPS6 mutants and LjTPS6 WT based on a Student's t-test (P < 0.05).

It also has been reported that replacement of the key residue by different amino acid can have distinct effects on diTPSs activity (Jia et al., 2017; Potter et al., 2016; Jia and Peters, 2016). Thus, saturation mutagenesis was carried out at residue 420, revealing that substitution with seven other amino acids (A/C/D/G/P/S/T) can increase LjTPS6 catalytic specificity, similarly imparting selective production of 4b, albeit with varied catalytic efficiency. The activity of LjTPS6 mutants with nonpolar amino acid substitution (i.e. Ala, Gly, Val, and Pro) was higher than those with polar amino acids (i.e. Cys, Asp, Asn, Ser, and Thr) (Figure 6d). Strikingly, LjTPS6:I420G was found to not only selectively produce 4b but also exhibit the highest catalytic efficiency.

To determine if changes at I420 altered reactivity with alternative substrates these LjTPS6 mutants also were tested with normal CPP and 8α-hydroxy-CPP using the modular metabolic engineering system (i.e. via co-expression with AgAS:D21A and NgCLS). The results indicated that these changes also affect reactivity with these substrates (Figure 7). Eight of these LjTPS6:I420 mutants (A/C/G/N/P/S/T/V) reacted with normal CPP to produce the endo olefin derivative trans-biformene (17) (Figure 7, a and b), and with 8α-hydroxy-CPP to produce 13S-manoyl oxide (13b) as well as the same 13R epimer selectively produced by WT (Figure 7, c and d).

Figure 7.

GC–MS analysis of extracts from in vivo assay of LjTPS6:I420 mutants coupled with AgAS:D621 or NgCLS. A and B, EICs of the products obtained by LjTPS6:I420V coupled with (A) AgAS:D621A (producing (+)-CPP) or (B) NgCLS (producing 8α-hydroxy-CPP) in vivo. C and D, Relative area of product peaks obtained from LjTPS6:I420 with (C) AgAS:D621A or (D) NgCLS A in vivo, and the error bars show the sds from mean value (n = 3). Statistically significant differences between the product abundance of LjTPS6 mutants and LjTPS6 WT based on a Student's t-test (P < 0.05).

Discussion

Here, an L. japonicus transcriptome was generated, enabling identification of six diTPSs, which were then functionally characterized, elucidating the relevant enzymes for diterpenoid metabolism in this medicinal plant. In addition to the ent-CPP synthase (LjTPS1) and subsequently acting ent-KS (LjTPS5) required for GA phytohormone biosynthesis, diTPSs that act in more specialized diterpenoid biosynthesis also were identified. One of these serves as a geranyllinalool synthase (LjTPS7), suggesting the possibility that L. japonicus produces homoterpenes, which are often involved in tri-trophic plant-(insect herbivore)-predator/parasitoid interactions (Herde et al., 2008; Brillada et al., 2013; Falara et al., 2014). In addition, these include another class II diTPS that produces normal CPP (LjTPS4), as well as one that produces a C9 hydroxylated derivative, PPP (LjTPS3). The final class I diTPS (LjTPS6) reacts with both normal CPP and PPP, producing sandaracopimaradiene and a mixture of C9-oxy derivatives, respectively (Figure 1, A and B). While L. japonicus is not known to produce any sandaracopimaradiene-derived diterpenoids, these results encourage further examination for such natural products. The gene expression and metabolites analysis showed that only LjTPS3 has a close relationship with C9-oxy diterpenoids distribution, which was mainly accumulated in aerial parts (Figure 2A), consistent with a role for LjTPS3 in C9-oxy diterpenoids biosynthesis in L. japonicas.

Despite the preponderance of derived diterpenoids in this plant, LjTPS6 does not primarily produce 9,13-epoxy-labda-14-ene. While similar observations have been made for other class I diTPSs from related plant species that also produce such diterpenoids, this catalytic promiscuity complicates both further biosynthetic investigations, as well as use of these enzymes for metabolic engineering directed at specific production of derived diterpenoids. Further elucidation of biosynthesis is additionally complicated by the often-observed promiscuity of the subsequently acting cytochrome P450 monooxygenases (Banerjee and Hamberger, 2017). For example, currently only MvCYP71AU87 has been identified as acting on 9,13-epoxy-labda-14-ene, but catalyzes hydroxylation of either of the C4 geminal methyl substituents to yield C18 or C19 hydroxylated products (Karunanithi et al., 2019). Notably, while a diTPS has been found to specifically produce 9,13-epoxy-labda-14-ene (4a and 4b; OmTPS3), this yields significant amounts of both C13 epimers (Johnson et al., 2019). Thus, the enzymatic engineering undertaken here that resolves this key issue, stereospecifically producing the 13S epimer (4b), with LjTPS6:I420G further exhibiting strong catalytic efficiency as well, provides a key reagent for further studies of the biosynthesis of derived bioactive diterpenoids. Together with the L. japonicus transcriptome generated here, there is now a strong foundation for such biosynthetic investigations, which are expected to enable access to important diterpenoids such as prehispanolone and similar spiro-9,13S-epoxy ring containing diterpenoids that exhibit anti-platelet and anti-coagulant properties (Lee et al., 1991; Moon, 2010; Miao et al., 2019).

Materials and methods

Plant materials, RNA isolation, and cDNA synthesis

The plants of L. japonicus were received from Nanjing, Jiangsu, China (store at -80°C). Three plants of similar conditions were selected and divided into three parts (root, stem, and leaf) and stored at -80°C. Total RNA was extracted from these organs using a Quick RNA Isolation Kit (HuaYueYang Biotechnology, Beijing, China) following the manufacturer’s instructions. The integrity and concentration of total RNA were detected by 1.0% (w/v) agarose gel electrophoresis and NanoDrop ND-3000 (Thermo NanoDrop, Wisconsin, USA). A PrimeScript RT Reagent Kit with gDNA Eraser (Takara Bio., Dalian, USA) was used to reverse transcribed into cDNA with 1 µg of total RNA from three tissues.

Transcriptome sequencing, de novo assembly, and annotation

Total RNA from L. japonicus tissues (e.g. leaf, root, and stem) was extracted using a Quick RNA Isolation Kit (HuaYueYang biotechnology, Beijing, China) and the integrity checked using a 2,100 RNA Nano 6,000 Assay Kit (Agilent Technologies, Palo Alto, CA, USA) following the manufacturer’s instructions. RNA was prepared for sequencing using Illumina TruSeq sample preparation kit version 2 (Illumina, San Diego, CA, USA). The fragments were sequenced with paired ends (2 × 150 bp) on a Illumina NovaSeq6000 (Illumina San Diego, CA, USA) by the Annoroad Gene Tech.Co., Ltd, Beijing, China. A total 279 million raw reads were generated. Adapter sequences were removed from raw reads and reads were trimmed at the ends to Q30, using the fastq-mcf tool from ea-utils (https://github.com/Expression Analysis/ea-utils). Processed reads were de novo assembled and estimated using the Trinity (Trinity Release version 2.4.0) pipeline according to guidelines in the published protocols (Grabherr et al., 2011), resulting in a total of 133,725 assembled putative transcripts. ORFs were predicted by TransDecoder. Trinotate was used for the functional annotation of unigenes and ORFs to known sequence data (BLAST+/SwissProt), protein domain identification (HMMER/PFAM), protein signal peptide and transmembrane domain prediction (singalP/tmHMM), and comparison to currently current annotation databases (EMBL Uniprot eggNOG/GO Pathways databases).

LjTPSs screening in L. japonicus and bioinformatic analysis

The genes were filtered according to their transcriptome information to comprehensively identify TPS genes in L. japonicus. The nucleotide and deduced amino acid sequences were analyzed, and the BLASTP tool on the NCBI online database (http://www.ncbi.nlm.nih.gov/) was used to compare the sequences. Multiple sequence alignment was implemented by ClustalW software, and MEGA version 7.0 software was used to construct a phylogenetic tree using the neighbor-joining (NJ) method (bootstrap = 1,000; Kumar et al., 2016).

Cloning of LjTPS coding sequences

According to the transcriptome sequence information, specific primers were developed to amplify the LjTPS ORFs (the primers are listed in Supplemental Table S4). A GeneJET Gel Extraction Kit (Thermo Scientific, San Jose, CA, USA) was used to purify the PCR products, after which the products were cloned into a T-Vector (pEASY-Blunt Zero Simple Vector) (TransGen, Beijing, China), which was subsequently transformed into E. coli DH5α (Covin Biosciences, Beijing, China) competent cells. Sangon Biotech (Shanghai) Co., Ltd. isolated the rebuilt plasmids and sequenced their nucleotides.

Functional characterization of LjTPSs

The ORF regions of the LjTPSs were subcloned into a pET-32a (+) (EMD Biosciences, Novagen) expression vector via PCR amplification with a pEASY-Uni Seamless Cloning and Assembly Kit (TransGen, Beijing, China), and the amplicons were digested with BamH I restriction enzymes (New England Biolab, Ipswich, MA, USA) (the primers are listed in Supplemental Table S4). The verified vectors were subsequently transformed into Rosetta (DE3) in vitro. The biochemical functional characterization of LjTPS was performed as the follows: the 500-μL reaction volume contained 200-μL solution of the crude enzyme, 50-mM HEPES buffer (pH 7.5), 100 mM KCl, 7.5-mM MgCl₂, 1-mM DTT, 5% glycine, 40 μM GPP, FPP, GGPP, or couple with class I diTPS. Class II LjTPSs were estimated with 40 μM GGPP as a substrate for 8 h at 30°C, class I LjTPSs were estimated with GPP, FPP, GGPP, or couple with class I diTPS. The in vitro products were hydrolyzed for 10 h at 30°C using two units of calf intestinal alkaline phosphatase and two units of potato apyrase in hydrolysis buffer (100 mM Tris–HCl, pH 9.5), and extracted thrice using an equal volume of n-hexane. The extracts were dried with N2 and dissolved in 100-μL n-hexane for GC–MS analysis (Su et al., 2018). The substrate specificity of class I LjTPSs were coupled with the reported 12 class II diTPSs in the E. coli OverExpress C41 strain (Lucigen, Middleton, WI, USA), using the previously developed modular metabolic engineering system, which can produce 12 different substrates of class I diTPSs (Morrone et al., 2010; Anthony Cyr et al., 2007).

The products of diTPSs analysis by GC–MS

The GC–MS analysis was performed by a Trace 1310 instrument equipped with a TSQ 8000 mass detector and then separated by a TG-5 MS column (30 m × 0.25 mm I.D., DF = 0.25 μm; Thermo Scientific, USA). Helium was used as a carrier gas at a flow rate of 1.0 mL⁻¹·min. The oven program was as follows: 50°C for 2 min, an increase of 20°C min⁻¹ to 300°C, and then holding for 10 min (for GGPP as substrate) (Hansen et al., 2017), or 50°C for 2 min, an increase of 5°C min⁻¹ to 150°C holding for 5 min, an increase of 10°C min⁻¹ to 220°C and holding for 2 min, an increase of 30°C min⁻¹ to 300°C and holding for 5 min (for GPP as substrate), or 50°C for 2 min, an increase of 8°C min⁻¹ to 250°C, an increase of 10°C min⁻¹ to 320°C and then holding for 10 min (for FPP as substrate). The ion trap temperature was 250°C, and the scan range was 40–450 m/z (Hansen et al., 2017).

Metabolite analysis of L. japonicus by UPLC-Q/TOF-MS

For metabolite analyses, 100 mg finely powdered tissues of L. japonicus were ultrasound extracted with 0.5-mL methanol (contain 50-μg·mL⁻¹ Andrographolide as internal standard) at 25°C for 1 h, and then centrifuged at 12,000g for 15 min at room temperature, and three independent experiments of the tissues were performed for each analysis. The supernatant was directly analyzed by UPLC-Q/TOF-MS. UPLC-Q/TOF-MS was carried out using an Acquity UPLC system (Waters Corp., Milford, MA, USA) with an Acquity UPLC BEH C18 column (50 × 2.1 mm, 1.7 μm). The column temperature was set at 40°C. The flow rate was kept at 400 μL·min⁻¹. Mobile phases were water (with 0.1% Formic acid)(A) and acetonitrile (B). The gradient was as follows: 0–8 min, 5%–20% B; 8–16 min, 20%–100% B; 16–21 min, 100% B; 21–21.5 min, 100%–5% B; 21.5–24.5  min, 5% B. Time-of-flight MS detection was performed with a Xevo G2-S MS system (Waters Corp., Manchester, UK). The data acquisition range was from 50 to 1,000 Da. The source temperature was set at 100°C, and the desolvation temperature was set at 450°C, with desolvation gas flow set at 900 L·h⁻¹. The lock mass compound used was leucine enkephalin at a concentration 200 pg·μL⁻¹. The capillary voltage was set at 2.0 kV. The cone voltage was set at 500 V. The collision energy was set as 6 eV for a low-energy scan, with a 35–60 eV ramp for a high-energy scan. The instrument was controlled by MassLynx version 4.1 software (Waters Corp., Milford, MA, USA).

RT-qPCR

Total RNA was extracted from L. japonicus tissues using a Quick RNA Isolation Kit (HuaYueYang Biotechnology, Beijing, China) following the manufacturer’s instructions. First-strand cDNA was synthesized using the PrimeScript RT Reagent Kit with gDNA Eraser (Takara Bio., Dalian, USA). Relative transcript abundance was determined by RT-qPCR using the TransStart Top Green qPCR SuperMix (TransGen Biotech, Beijing, China) on a Roche LightCycler480 (Roche, Basel, Switzerland). The primers used for RT-qPCR analysis are listed in Supplemental Table S4. The gene for actin was used as the endogenous control. At least three independent experiments were performed for each analysis.

Molecular docking and site-directed mutagenesis

Within SWISS-MODEL Workspace, homology modeling was utilized to create the three-dimensional structure of LjTPS6. After searching the PDB (https://www.rcsb.org/) for templates based on alternate sequence alignments, templates were chosen based on their GMQW values, and multiple models were generated using the chosen templates. To relate sequence position, a multiple sequence alignment was created using the sequences. The PHENIX suite’s elBow algorithm was used to refine the substrate. Molecular docking was used to model LjTPS6 and its substrates (CPP, PPP, and 8-hydroxy-CPP). AutoDock4 was used to perform docking calculations, as previously stated (Morris et al., 2009). The results of this docking were shown by PyMOL.

The mutations of the LjTPS6 were cloned into a pET-32a (+) expression vector via PCR amplification with a pEASY-Uni Seamless Cloning and Assembly Kit (TransGen, Beijing, China), the primers were digested with BamH I restriction enzymes (New England Biolab, USA) and shown in Supplemental Table S4. Using the previously described modular metabolic engineering system (Morrone et al., 2010; Anthony Cyr et al., 2007; Jia et al., 2016), functional assessment of LjTPS6 mutations was coupled with MvCPS1, AgAS:D621A, and NgCLS in the E. coli Over Express C41 strain (Lucigen).

Accession numbers

The sequence information has been deposited at GenBank: LjTPS1 (MZ557361), LjTPS3 (MZ557362), LjTPS4 (MZ557363), LjTPS5 (MZ557364), LjTPS6 (MZ557365), and LjTPS7 (MZ557366), and listed in Supplemental Table S2. On reasonable request, others can obtain information on the LjTPSs from the appropriate author. The transcriptome reported in this paper has been deposited in China National Center for Bioinformation under accession number PRJCA CRA004698, which is publicly accessible for all researchers at http://bigd.big.ac.cn/gsa.

Supplemental data

The following materials are available in the online version of this article.

Supplemental Figure S1. Alignment of diTPSs from L. japonicus.

Supplemental Table S1. Ions detected by UPLC-Q/TOF-MS in the tissues of L. japonicus.

Supplemental Table S2. RNA-seq information of diTPSs from L. japonicus.

Supplemental Table S3. The information of TPSs used in this study.

Supplemental Table S4. Primers used in this study.

Funding

This work was supported by the National Key R&D Program of China (2020YFA0908000, 2018YFA0900600), the National Natural Science Foundation of China (81822046, 82011530137, 81891010, 82003904), Key project at central government level: the ability to establish sustainable use of valuable Chinese medicine resources (2060302, China), and a grant from the NIH to R.J.P. (GM131885), Scientific and technological innovation project of China Academy of Chinese Medical Sciences (CI2021A04110 and ZZ13-YQ-083), Innovation Team and Talents Cultivation Program of National Administration of Traditional Chinese Medicine (ZYYCXTD-D-202005).

Conflict of interest statement. None declared.

 

J.W., J.G., G.C., and L.H. designed the study. J.W. performed most of the experiment and analyzed the data. Y.M., Y.M., and B.J. conducted the molecular docking. J.Y. and H.L. in participated in analyzing the data. R.P., J.T., W.Z., W.G., and Y.Z. participated in manuscript preparation. J.G., G.C., and L.H. were responsible for overall inspection and supervision of the work. All authors revised and approved the final manuscript.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plphys/pages/general-instructions) is: Juan Guo (guojuanzy@163.com).